energy-filtered electron microscopy reveals that talin is a highly flexible protein composed of a...

Eur. .I. Biochem. 243, 430-436 (1997) 0 FEBS 1997

Energy-filtered electron microscopy reveals that talin is a highly flexible protein composed of a series of globular domains Jorg WINKLER I, Heinrich LUNSDORF' and Brigitte M. JOCKUSCH I

I Cell Biology, Zoological Institute, Technical University of Rraunschweig, Germany Department of Microbiology, GBF-National Research Center for Biotechnology, Braunschweig, Germany

(Received 9 Septembedl9 November 1996) - EJB 96 13460

Talin is a multidoinain cytoskeletal protein containing discrete binding sites for acidic phospholipids, [I-integrin, actin and vinculin. Hence, it is thought to link microfilaments to the cytoplasmic membrane in cell-matrix adhesion sites, and this should critically depend on talin structure. To obtain more information on the latter, we used energy-filtered transmission electron microscopy of negatively stained talin purified from chicken smooth muscle. We show that in buffers of physiological ionic strength. talin adopts an elongated shape (56 +- 7 nm in length), consisting of a series of globular masses. While these compact elements, arranged like beads on a string, were of rather uniform dimensions (3.8 nm in diameter), their center-to-center spacings varied, indicating the flexibility of the connecting strands. The ends of the elongated molecules frequently formed loops. The images obtained are consistent with the assumption that, under the conditions used, the majority of the talin molecules are monomeric. A minor fraction appeared as dimers, composed of two chains only partially intertwined, thus giving rise to Y-shaped particles. Electron micrographs revealed that the biochemically defined 50-kDa N-terminal talin head domain is composed of two globular subunits, while chemical cross-linking provided evidence that the C- terminal 220-kDa fragment is solely responsible for dimerization. These results imply that in the dimeric molecules, the polypeptide chains are arranged in parallel, in contrast to what has been described for human-platelet talin. In buffers of low ionic strength (0.02 M instead of 0.15 M KCI), the molecules collapsed into a compact shape. By showing the high flexibility and versatility of its morphology, our data favour the concept of talin as an important resilient link in microfilament-plasma-membrane attachment.

Keywcri-ds: cytoskeleton; talin; structure ; electron microscopy; electron-spectroscopic imaging.

Talin is a major component of cell-matrix junctions, concen- trated at the cytoplasmic face of focal contacts, dense plaques of smooth muscle and myotendinous junctions (Burridge and Connell. 1983a.b; Tidball et al., 1986). The amino acid sequence of mouse talin consists of 2541 amino acids, from which the molecular mass of the protein was determined as 269 854 Da (Rees et al., 1990). The molecule can be cleaved by thrombin between residues 433 and 434, yielding a 47-kDa N-terminal and a 200-kDa C-terminal fragment (Fox et al., 1985 ; Beckerle et al., 1987). The N-terminal fragment, which has sequence sim- ilarities to the membrane-binding domains of protein band 4.1 and ezrin (Rees et al.. 1990). can interact with liposomes containing acidic phospholipids (Niggli et al., 1994; Tempe1 et al., 1995), while the C-terminal portion binds to the cytosolic domains of integrin (Horwitz et al., 1986). vinculin (Burridge and Mangeat, 1984; Gilmore et al.. 1993) and actin (Muguruma et al., 1990; Kaufmann et al., 1991: Goldmann and Isenberg, 1991). These multiple binding sites for integral and peripheral components of the plasma membrane, together with the findings that talin concentrates very early in nascent focal contacts ( h a r d , 1988), have prompted the hypothesis that talin is an

Corr~.sporidrrrc.r r o J. Winkler, Cell Biology, Zoological Institute, Technical Univer5ity of Braunxhweig. Spielrnannstrasse 7, D-38092 Braunschweig, Germany

F a r : +49 531 391 8203. Ahhr-rviarioris. ESI, electron spectroscopic imaging; TEM, transmis-

sion electron microxopy.

essential element in the establishment of discrete cell-matrix junctions (Jockusch et al., 1995). Hence, structural information on talin should contribute to our understanding of the assembly of these elements.

Transmission electron microscopy (TEM) of rotary-shadowed preparations of chicken gizzard talin revealed highly asymmetric particles approximately 60 nm in length, which were rod shaped in buffers of physiological ionic strength, but appeared globular in low-ionic-strength buffer (Molony et al., 1987). In contrast. rotary-shadowed human platelet talin, in buffers of low ionic strength, showed a mixture of horseshoe-like, elongated particles and dumb-bell - shaped structures which were interpreted as dimers (Goldmann et al., 1994). Fourier- transformed computer-assisted analysis of a C-terminal portion of talin, comprising residues 481 -2541, revealed that this region consists of 50-60 copies of an irregular structural motif approximately 34 amino acid residues long (McLachlan et al., 1994). This work suggested a pattern of repetitive units in the talin rod domain.

In the present study, we used conventional negative-staining techniques combined with energy-filtered TEM to determine further structural details of isolated gizzard talin. This combina- tion of techniques has already proven useful in unraveling the structure of vinculin, another cytoskeletal component concen- trated in cellular-adhesion sites (Winkler et al., 1996). We show that talin is composed of a regular pattern of discrete globular subunits arranged as beads on a string. Electron micrographs and data obtained by chemical cross-linking provide evidence

Winkler et al. (Eur: J. Biochenz. 243) 431

a b c 205

45 29

Fig. 1. SDSlPAGE analysis of purified 215-kDa talin (lane a), and its 190-kDa (lane b) and 47-kDa fragments (lane c) generated by thrombin cleavage. Coomassie-blue-stained protein profiles on gradient (from 3 % to 18 %) gels are shown. The numbers on the left indicate molecular masses of standard proteins in kDa.

that these apparently monomeric particles can form dimers that are tail-to-tail associated, yielding Y-shaped molecules. These data enlarge and support the hypothesis of talin being an important element in microfilament- plasma-membrane linkages by demonstrating the high degree of flexibilty and versatility of this multidomain protein.

MATERIALS AND METHODS

Protein purification and thrombin digestion. Talin was isolated from chicken gizzard as described by O'Halloran et al.

(1986). The protein was purified to homogeneity as judged by SDS/PAGE (Fig. I), and the protein concentration was determined by the method of Bradford (1976). Proteolytic cleavage of talin and isolation of fragments (Fig. 1) was achieved accord- ing to published procedures (Niggli et al., 1994).

Chemical cross-linking. Protein samples were cross-linked in SO mM sodium phosphate, pH 6.8 with 10 mM l-ethyl-3-(3- dimethylaminopropyl) carbodiimide/HCl and 30 mM N-hy- droxysulfosuccinimide. Samples were incubated for 90 min at 30"C, and the reaction was stopped with sample buffer. For molecular-mass analysis, aliquots were applied to gradient (from 3% to 18%) SDS/PAGE gels (Laemmli, 1970).

Electron microscopy. Purified talin was dialyzed in 0.02 M Tris/HCI. 0.02 M KCI, 0.1 mM EGTA, 0.1 mM phenylmethyl- sulfonyl fluoride, 0.1 mM dithioerythritol, pH 7.6. For high- ionic-strength conditions, the KCl concentration was increased to 0.15 M. The protein (50 pg/ml) was adsorbed to an ultrathin carbon film and negatively stained with 4% (masdvol.) uranyl acetate as described previously (Valentine et al., 1968). Carbon foils were mounted on perforated carbon-collodion films (Liinsdorf and Spiess, 1986). Specimens were examined in a Zeiss EM 902 transmission electron microscope in the bright field mode and in the electron-spectroscopic imaging (ESI) mode at an energy loss of 115 eV (Bauer, 1988). Exposures were carried out at X50000 magnification. Length measurements of elongated talin molecules were carried out by means of Sigma Scan Image software provided by Jandel Scientific.

Fig.2. Field views of purified talin from chicken gizzard smooth muscle. The protein samples in 0.15 KCI (high-ionic-strength buffer) were negatively stained with 4% aqueous uranyl acetate. (a, c) Bright-field views at AE = 0 eV. (b, d) Dark-field images at A E = 115 eV. Identical molecules or protein subunits detectable with both methods are encircled or inarked with arrowheads, respectively. Bar = 30 nm.

432 Winkler et al. (Eur J. Biochern. 243)

b Fig. 3. Gallery of negatively stained talin molecules in high-ionic-strength buffer. The molecules reveal an elongated, flexible structure with a predominantly U-shaped overall morphology. They are composed of a series of globular masses, organized like beads on a string. Each molecule is shown in a set of two images, taken in the ESI mode (a) or under bright-field conditions (b). Bar = 20 nm.

a

b

Fig. 6. High magnifications of talin molecules in high-ionic-strength buffer imaged the ESI mode. The elongated structure of the t& molecules and appropriate stain conditions allowed for a regular projection of single globular masses (arrowheads) revealing different center- to-center spacings, even within the same molecule. Both molecules con- sist of at least ten globular subunits. Bar = 10 nm.

Fig. 4. Gallery of negatively stained talin molecules in bigh-ionic- strength buffer and imaged in the ESI mode. (a) The terlninal region of the molecules appears as rings or loops. with a protein-deficient central part (anowheads), (b) The of single globular masses (arrows). Bar = 20 nm.

structures are

RESULTS

Molecular structure of talin. Survey electron micrographs of talin in buffers of physiological ionic strength (0.25 M KCl) yielded inhomogeneous images of' small granules (Fig. 2). As SDSlPAGE revealed homogeneity of the protein preparations without indication of fragmentation (Fig. I), we concluded that the majority of the molecules contained in these preparations must be full-length talin, which might be composed of smaller particles or granules. On closer inspection, rows of granules were discernible, which were frequently bent into arcs or U- shaped structures (Fig. 2).

A gallery of elongated talin molecules that were discernibly projected from the background were chosen for the gallery shown in Fig. 3. Each molecule is presented in a set of two images taken either in the ESl mode (Fig. 3a) or under elastic bright field conditions (Fig. 3 b). Selective filtering of the inelas- tic-scattered, element-specific signals derived from the uranium salt enhanced significantly the contrast between the amorphous

7.9 6.5 6.0 5.8

layer of stain and the protein particles. The talin molecules iden- tified this way appeared predominantly U shaped and were composed of globular subdomains of similar size. They measured 56 -C 7 nm in length (n = 20). This value corresponds well with the figures previously obtained by Molony et al. (1987) for rotary-shadowed talin (60 -C 8 nm). Many molecules displayed terminal loops (Fig. 4 a), and appropriate staining conditions revealed that these loops were also composed of discrete globular masses (Fig. 4 b).

The dimensions and arrangement of the globular subdomains were further analysed in selected examples. Fig. 5 shows a gallery of molecules displaying 6-9 of these subdomains. Their diameter was determined to 3.8 i 0.2 nm (n = 62). While they appeared fairly uniform in size, their center-to-center spacing varied markedly between individual molecules (Figs 5 and 7) and within a given molecule. The latter became evident at higher magnification (Fig. 6). The average center-to-center distance was determined as 4.8? 1.2 nm (n = 270). From these data it can be concluded that the globular subunits of talin molecules

5.7 5.0 4.6 4.4 3.5 Fig. 5. Gallery of negatively stained talin in high-ionic-strength buffer, imaged in the ESI mode, revealing their composition of single globular subunits. The numbers undernedth the individual images refer to the average center-to-center apacings of the subdomains Bar = 30 iim

Winkler et al. (Eur: J. Biochem. 243) 433

2 3 4 I I 5 (I 6 L

7 8 9 1 0

Center-to-center distance (nm)

Fig. 7. Relative frequencies of center-to-center distances of globular subunits in talin. 270 values were obtained from extended molecules in high-salt buffer. About 13% of the measured distances were above 6 nm while about 21 % were below 4 nm.

Fig. 8. Negatively stained sample of purified talin head fragments. (a) Dark-field survey view at an energy loss of A E = 115 eV. (b) Gallery of individual bead particles. Each particle is composed of two globular masses. Bar = 20 nm.

are bound to each other by flexible connective polypeptide strands not visible by electron microscopy.

Total number of repetitive centers of mass in talin. Reliable data on the total number of discrete subdomains are difficult to obtain by visual inspection of such highly flexible structures. Two requirements should be fulfilled for negatively stained preparations. The staining conditions and the orientation of the adsorbed molecules must allow for a projection of complete molecules, and only extended molecules can be considered for such measurements. The talin molecules depicted in Fig. 6 largely comply to these conditions. Ten discrete globular subunits are discernible in each of the two molecules shown, but since the terminal masses seem somewhat larger and oval- shaped, they may be composed of two adjacent subunits. Thus, we concluded that talin consists of at least 10, possibly 12, subdomains. The validity of this assumption was tested by an inde- pendent method.

The molecular mass of a single subunit was calculated on the assumption that a globular particle with an average diameter

a

C

Fig. 9. Molecular shape of talin molecules obtained under low-salt conditions. Protein samples were negatively stained and imaged in the ESI mode. The molecules seemed more compact than under physiological conditions, and appeared either rolled-up (a), rod-like (b), or roughly ring-shaped (c). Bar = 20 nm.

Fig. 10. Gallery of presumptive talin dimers in low-salt buffer imaged in the ESI mode. Y-shaped, rather-compact structures can be seen. Isolated globular masses are not well resolved. Bar = 20 nm.

of 3.8 nm, as determined here, corresponds to a volume of 29 nm' (Wrighley, 1988). Based on a partial specific volume of 0.73 g/ml (Molony et al., 1987), this value yields an average molecular mass of approximately 24 kDalsubunit. Considering 270 kDa for the complete talin polypeptide (Rees et al., 1990), one arrives at 11 -12 subdomains for each talin molecule.

We estimated the number of subdomains by relating their dimensions to the maximal and minimal values measured for elongated molecules and the average center-to-center spacing. By this method, one arrives at a number of discrete protein masses between 10-13. Taken all evidence together, we propose that the talin molecule consists of 10- 1 2 discrete globular subdomains.

The ultrastructure of head and tail fragments. The N-terminal 50-kDa and C-terminal 220-kDa fragments, obtained by thrombin cleavage and purified by gel chromatography, were negatively stained and analysed by ESI. Images of the N-terminal fragment are shown in Fig. 8. The head particles are composed of two subdomains, with a center-to-center distance of 5.1 2 0 . 4 nm (n = 40). Depending on the staining conditions, these masses appeared either as two separate subunits, indicating immersion in deep stain, or as a dumb-bell-shaped particle, as a result of shallow staining. As expected, the large C-terminal fragments consisted of a series of subdomains. However, the images appeared less sharp and more blurred compared with that of intact talin molecules, which may be the result of proteolysis at room temperature and handling during purification (data not shown). Hence, the determination of the exact number of subunits' in this fragment was not possible.

Ultrastructure of talin in low-salt buffer. When isolated gizzard talin was examined in a buffer of low ionic strength

434 Winkler et al. (ELM J . Biochem. 243)

Fig. 11. Gallery of negatively stained presumptive dimeric talin molecules in high-ionic-strength buffer imaged in the ESI mode. The overall structure of the complexes is Y-shaped, but less compact than in Fig. 10. The individual chains associate with each other at one end only (arrowheads), suggesting tail-tail interactions of chains arranged in parallel. Bar = 20 nm.

a b c d e

116 914 66 45 29

Fig. 12. Cross-linking analysis of purified talin and its fragments as seen on a Coomassie-stained gradient (from 3 % to 18%) SDSl polyacrylamide gel. Non-cross-linked (lane a) and l-ethyl-3-(3-dimeth- ylarninopropy1)-carbodiimide-cross-linked talin (lane b), cross-linked 190-kDa fragment (lane c), cross-linked sample of a mixture of the 190- kDa and the 47-kDa fragments (lane d). cros+linked 47-kDa fragment (lane e). Monomeric talin displays an apparent molecular mass of about 215 kDa (lane a), whereas cross-linked talin bands (arrowheads) show mobilities equivalent to about 350 kDa and 500 kDa (lane b). The cross- linked 190-kDa fragment yielded an additional band with an apparent molecular mass of about 360 kDa (lanes c, d, arrowhead). The 47-kDa fragment neither cross-linked with itself (lane e) nor with the 190-kDa fragment (lane d). The numbers on the left indicate molecular masses of standard proteins in kDa.

(0.02 M KCI), the pictures obtained differed significantly from the images described above. In general, the molecules appeared much shorter (Fig. 9). This was due to several effects. The center-to-center spacings were considerably reduced, so that the individual masses seemed to merge (Fig. 9a-c), and the molecules appeared curled up (Fig. 9a) or were totally collapsed into twisted, compact structures (Fig. 9 b,c). Thus, by changing the buffer conditions, the overall shape of talin could be drastically altered, demonstrating the high flexibility of this protein.

Dimeric talin molecules. The relation of the number of subdomains to the overall molecular inass of talin, as discussed above, argues for a model in which the beads-on-a-string configuration is composed of a monomeric talin polypeptide. This was observed for most of the individual molecules seen in our preparations. However, in both types of buffers used, we found a small number of molecules whose conformation suggested the formation of dimers (Figs 10, l l ). While the compactness induced by low-salt buffers made an unequivocal identification of such Y-shaped particles as being composed of two chains difficult (Fig. lo), this interpretation was more plausible under physiological conditions. Extended chains were found intertwined to various degrees, yielding Y-shaped structures with different angles (Fig. 11).

These images suggested that, in the presumptive dimers, the chains formed close contacts in only a defined region, and that they were arranged in a parallel orientation. This hypothesis was

tested by chemical cross-linking. Using a zero-length crosslinker [l-ethyl-3-(3-dimethylaminopropyl) carbodiimide], we con- firmed previous results that talin can form dimers (Goldmann et al., 1994; Zhang et al., 1996). In addition, by performing cross- linking experiments with isolated talin fragments, we found that only the large C-terminal portion can be cross-linked, while the N-terminal heads are unaffected (Fig. 12). Similar results were reported by Muguruma et al. (1995). These data support the interpretation of the Y-shaped particles as being composed of two talin polypeptides arranged in parallel.

DISCUSSION Although the negative-staining technique offers the advan-

tage of high resolution, the contrast provided by simple negative staining alone is inadequate in many cases, particularly for elongated, slender proteins (Glenney, 1987). Thus, metal shadowing became the method of choice to investigate proteins such as spectrin, fodrin, filamin (Tyler et al., 1980; Glenney et al., 1982) and talin (Molony et al., 1987; Goldmann et al., 1994). The application of energy-filtered TEM in conjunction with negative staining resulted in a significant enhancement of contrast at the O,,,-edge of uranium, while reducing the background noise. Both effects allow the projection of delicate protein structures, such as the elongated talin molecule. Substructures can thus be revealed, as seen here for talin, and as demonstrated before for vinculin (Winkler et al., 1996).

Survey images obtained from purified talin samples revealed considerable heterogeneity, as is typical for highly elongated, flexible proteins. The talin samples displayed granules and groups of particles differing in size and shape (Fig. 2). As the biochemical analysis proved the protein purification to be ho- mogenous, these images might either result from submersion of parts of the molecules into the deeper layers of stain, or, less likely, from fragmentation during processing for electron microscopy. Since the survey images were dominated by this background, we had to select for structures that might represent full-length talin molecules. This selection was based on two criteria: rows of particles which clearly projected from the background and were closely spaced, and thus were seemingly aligned; and a total length of such rows compatible with published data for rotary-shadowed talin, i.e. approximately 60 nm (Molony et al., 1987). Only structures fulfilling both criteria were classified as full-length molecules (Figs 3 and 6).

In this study, we have shown that chicken smooth muscle talin is composed of a series of repeating centers of mass. Elon- gated multidomain proteins are quite frequent in the microfilament system, as for example spectrin, a-actinin and dystrophin (Gilmore et al., 1994) and many of these proteins have similar repetetive-sequence motifs, but correlations between such sequence repeats to structural and functional domains are mostly unknown. For the talin rod domain, we face an analogous diffi- culty: 50-60 copies of a repetive motif were recognized by fourier sequence analysis (McLachlan et al., 1994), but their re-

Winkler et al. (Eur: J. Biochem. 243) 435

lation to the 8- 10 subdomains contained in the tail, as described here, remains unclear.

As documented here, the entire talin particle is arranged as globular subdomains, interconnected by fine strands. In contrast to vinculin (Jockusch and Rudiger, 1996), monomeric talin apparently is not restricted to a few discrete conformations. The high degree of flexibility, giving rise to a multitude of molecular shapes, might be useful for absorbing physical stress at the microfilament/membrane interface, such as shearing or compres- sion of cells, which might endanger the maintenance of cell- matrix contacts, and for the selection of specific ligands at the nascent contact site. For example, talin-actin interaction was found to be sensitive to ionic conditions (Zhang et al., 1996). It is conceivable that the accessibility of ligand-binding sites such as that for actin corresponds to the overall talin conformation. In the cell, the changes seen here with different buffer conditions might be induced by more physiological parameters.

In addition, we found evidence for dimer formation. Talin dimers have been biochemically described by Molony et al. (1987), and these authors suggested that dimer formation re- quires relatively high protein concentrations ( 3 0.72 mg/ml). Evidence for a dimeric structure of rotary-shadowed talin molecules, even under low protein concentrations (0.1 -0.4 mg/ml), was given by Goldmann et al. (1994). These authors described human platelct talin as being predominantly a homodimer with an antiparallel dumb-bell-shaped structure and a length of 51 nm t 2.4 nm. In addition to straight molecules, they observed a whole range of bent configurations, including horseshoe-like molecules, which they also interpreted as dimers. However, these correspond quite well to particles seen in our preparations (Figs 3, 5 and 6), which we consider as monomers. A balance shift between two discrete populations of particles, each being composed of a defined number of polypeptide chains, has been described before for the dimerketramer equilibrium of spectrin. This equilibrium, and the overall dimensions of the spectrin particles, is strongly affected by ionic strength (Ungewickell and Gratzer, 1978; LaBrake et al., 1993). It is conceivable that human-platelet and chicken-gizzard talin differ with respect to dimerization-favoring conditions. In both cases, however, the pre- ferred state in the living cell is unknown.

Another discrepancy remains to be explained. While Gold- mann et al. ( 3 994) imply that the particles seen i n their electron micrographs are composed of two polypeptides arranged in an antiparallel orientation, the Y-shaped particles shown in Figs 10 and 11 suggest a parallel orientation of the chains, and this interpretation is supported by the cross-linking data. Such a Y-shaped configuration with variable angles and flexible terminal portions might allow for very effective ligand binding, as has, for example, been shown for filamin (Hartwig and Stossel, 1981 ; Lebart et al., 1994). Further comparative studies involving electron microscopy and biochemistry will be needed to understand the causal relationship between flexibility, dimerization and function of talin.

The expert technical assistance of Eva Saxinger is gratefully ac- knowledged. This work was financially supported by the Deirtsche For- scliungsRemeinschuft.

REFERENCES Bauer, J. (1 988) Electron spectroscopic imaging: an advanced technique

for imaging and analysis in transmission electron microscopy, Mrth- ods Microbid. 20, 11 3 - 146.

Beckerle, M. C., Burridge, K., DeMartino, G. N. & Croall, D. (1987) Colocalization of calcium-dependent protease I1 and one of its sub- strates at sites of cell adhesion, Cell 51, 569-577.

Bradford, M. M. (1976) A rapid and sensitive method for the quantifica- tion of microgram quantities of protein utilizing the principle of protein-dye binding, Anal. Biochem. 72, 248-254.

Burridge, K. & Connell, L. (1983a) A new protein of adhesion plaques and membrane ruffles, J. Cell Biol. 97, 359-367.

Burridge, K. & Connell, L (1983b) Talin: a cytoskeletal component con- centrated in adhesion plaques and other sites of actin-membrane interaction, Cell Motil. 3 , 405-417.

Burridge, K. & Mangeat, P. (1984) An interaction between vinculin and talin, Nature 308, 744-746.

Fox, J. E. B., Goll, D. E., Reynolds, C. C. & Philips, D. R. (1985) Identification of two proteins (actin-binding protein and P235) that are hydrolyzed by endogenous Ca2+-dependent protease during platelet aggregation, J. B id . Chem. 260, 1060- 1066.

Gilmore, A. P., Wood, C., Ohanian, V., Jackson, P., Patel, B., Rees, D. J . G., Hynes, R. 0. & Critchley, D. R. (1993) The cytoskeletal protein talin contains at least two distinct vinculin binding domains, J.

Gilmore, A. P., Parr, T., Patel, B., Gratzer, W. B. & Critchley, D. R. (1994) Analysis of the phasing of four spectrin-like repeats in (1-

actinin, ELM J. Biochem. 225, 235-242. Glenney, J. R., Glenney, J. R. P., Osbom, M. & Weber, K. (1982) An

F-actin- and calmodulin-binding protein from isolated intestinal brush borders has a morphology related to spectrin, Cell 28, 843- 854.

Glenney, J. R. (1987) Rotary shadowing for visualizing rod-shaped proteins, in Electron microscopy in molecular biology: a practical ap- proach (Sommerville, J. & Scheer, U., eds) pp. 167-178, IRL Press Limited, Oxford.

Goldmann, W. H. & Isenberg, G. (1991) Kinetic determination of talin- actin binding, Biochem. Biophys. Rex Commun. 178, 71 8-723.

Goldmann, W. H., Bremer, A,, Haner, M., Aebi, U. & Isenberg, G. (1994) Native talin is a dumbbell-shaped hoinodimer when it in- teracts with actin, J . struct. Biol. 112, 3-10.

Hartwig, J. H. & Stossel, T. P. (1981) Structure of macrophage actin- binding protein molecules in solution and interacting with actin filaments, J. Mol. B i d . 145, 563-581.

Horwitz, A., Duggan, K., Buck, C., Beckerle, M. C. & Burridge, K. ( 1 986) Interaction of plasma membrane fibronectin receptor with talin - a transmembrane linkage, Nature 320, 531 -533.

Izzard. C. S. (1988) A precursor of the focal contact in cultured fibro- blasts, Cell Molil. Cytoskeleton 10, 137-142.

Jockusch, B. M., Bubeck, P., Giehl, K., Kroemker, M., Moschner, J., Rothkegel, M., Rudiger, M., Schluter, K., Stanke, G. & Winkler, J. ( 1 995) The molecular architecture of focal adhesions, Anncc. Rev. Cell Dev. Biol. 11, 379-416.

Jockusch, B. M. & Rudiger, M. (1996) Crosstalk between cell adhesion molecules, Trends Cell B i d . 6 , 31 1-315.

Kaufmann, S., Piekenbrock, T., Goldmann, W. H., Barmann, M. & Isenberg, G. (1991) Talin binds to actin and promotes filament nucle- ation, FEBS Lett. 284, 187 - 191.

LaBrake, C. C., Wang, L., Keiderling, T. A. & Fung, L. W.-M. (1993) Fourier transformed infrared spectroscopic studies of the secondary structure of spectrin under different ionic strength, Biochemistry 32,

Laemnili, U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophages T4, Nature 227, 680-685.

Lebart, M. C., Mejcan, C., Casanova, D., Audemard, E., Dcrancourt, J., Roustan, C. & Benyamin, Y. (1994) Characterization of the actin binding site on smooth muscle filamin, J. Bid. Chem. 269, 4279- 4284.

Lunsdorf, H. & Spiess, E. (1986) A rapid method for preparing perforated supporting foils for the thin carbon films used in high resolution transmission elcctron microscopy, J. Microsc. (Oxf) 144, 21 1-213.

McLachlan, A. D., Stewart, M., Hynes, R. 0. & Rees, J. D. G. (1994) Analysis of repeated motifs in the talin rod, J . Mol. Bid. 235, 1278- 1290.

Molony, L., McCaslin, D., Abernethy, J., Paschal, P. & Burridge, K. (1987) Properties of talk from chicken gizzard smooth muscle, J. Bid. Cheni. 262, 7790-7795.

Muguruina, M., Matsumura, S. & Fukazawa, T. (1990) Direct interation between talin and actin, Biochem. Biophys. Res. Commun. 171,

Cell Biol. 122, 337-347.

10296- 10 302.

1217- 1223.

436 Winkler et al. ( E m J. Biochem. 243)

Muguruma, M., Nishimuta, S. , Tomisaka, Y., Ito, T. & Matsumura, S . (1995) Organization of the functional domains in membrane cytoskeletal protein talin, J. Biochem. 1/7, 1036-1042.

Niggli, V., Kaufmann, S., Goldmann, W. H., Weber, T. & Isenberg, G. (1 994) Identification of functional domains in the cytoskeletal protein talin, EUK J. Bioch~nz. 227, 951-957.

O’Halloran, T., Molony, L. & Burridge, K. (1986) Purification and assay of vinculin, metavinculin, and talin, Methods Enzymol. 134, 69 -77.

Rees, D. J. G., Ades, S. E., Singer, S. J. & Hynes, R. 0. (1990) Sequence and domain structure of talin, Nuture 347. 685-689.

Tempel, M., Goldmann, W. H., Isenberg. H. & Sackmann, E. (1995) Interaction of the 47 kDa talin fragment and the 32 kDa vinculin fragment with acidic phospholipids : a computer analysis, Biophys.

Tidball, J. G., O’Halloran, T. & Burridge, K. (1986) Talin at myotendi- J. 69, 228-241.

nous junctions, J . Cell B i d . /03. 1465 - 1472.

Tyler, J., Anderson, J. M. & Branton, D. (1980) Structural comparison of several actin-binding macromolecules, J. Cell Biol. 85, 489 -495.

Ungewickell, E. & Gratzer, W. (1978) Self-association of human spectrin, EUI: J . Biochem. 88, 379-385.

Valentine, R. C., Shapiro, B. M. & Stadtman, E. R. (1968) Regulation of glutamine synthetase. XII. Electron microscopy of the enzyme from Lkherichiu coli, Biochemisrrji 7. 2143-2152.

Winkler, J., Lunsdorf, H. & Jockusch, B. M. (1996) The ultrastructure of chicken gizzard vinculin as visualized by high-resolution electron microscopy, J. Struct. Biol. 116, 270-277.

Wrighley, N. G. (1988) Electron micrographs of protein molecules: in- terconversion of size, shape and molecular weight made easy, Proc.

Zhang, J., Robson, R. M., Schmidt, J. M. & Stromer, M. H. (1996) Talin can crosslink actin filaments into both networks and bundles, Biochem. Biophys. Res. Commun. 218. 530-537.

R. Microsc. SOC. 23, 299-302.

energy-filtered electron microscopy reveals that talin is a highly flexible protein composed of a...

Documents