Proc. Natl. Acad. Sci. USAVol. 90, pp. 3280-3283, April 1993Cell Biology

Pattern formation and handedness in the cytoskeleton ofhuman platelets

(actin/vinculin/talin/gpIlb-IIIa)

J6RG HAGMANNFriedrich Miescher-Institut, P.O. Box 2543, 4002 Basel, Switzerland

Communicated by Sheldon Penman, January 11, 1993

ABSTRACT The cytoskeletal patterns of human plateletsspread on a glass surface are analyzed. F-actin is arranged inpatterns of parallel microframents, microfilaments formingtriangles, or microrflaments radiating tangentially from acentral ellipse or circle. Vinculin, a cytoskeletal protein, islocated at both ends of the filaments. In platelets with tangen-tiadly radiating microframents, vinculin patches are aligned onthe branches of a two-armed spiral. The spirals are alwaysleft-handed. Talin and two integrins (gplIb-IIIa, vitronectinreceptor), proteins usually associated with focal contacts intissue culture cells, are not concentrated at the ends of micro-filaments in human platelets. It is suggested that the distribu-tion of vinculin is due to competitive aggregation of vinculinclose to the inner leaflet of the ventral plasma membrane andthat sites of cytoskeleton-membrane linkage are important forgenerating supramolecular asymmetries of biological systems.

Comparing biological patterns to purely chemical systemscan lead to a better understanding of the physicochemicalmechanisms at the basis of self-organizing living systems.Cytoskeletal proteins form functionally important patternsthat determine cell shape, but there is very little informationabout how these patterns are created at the physicochemicallevel. A striking example of cytoskeletal pattern formation isprovided by the actin skeleton oftissue culture cells. Bundledmicrofilaments span the cells and are anchored in the ventralplasma membrane in focal contacts. Focal contacts areenriched in a number of proteins, some of which are believedto play a role in linking the actin skeleton to the plasmamembrane and to the extracellular matrix (e.g., vinculin,talin, and integrins; for a review see ref. 1).Human platelets provide a system with many advantages

for analyzing the mechanisms that organize cytoskeletalelements into global structures. When attaching and spread-ing on glass surfaces, they generate essentially two-dimensional compartments in which actin filaments and othercytoskeletal proteins are arranged in a small number ofcharacteristic patterns. Here, patterns of actin, vinculin,talin, and integrin distribution in platelets are examined. Twoaspects of the regularities observed are discussed. First,based on the similarities between the distribution of vinculinand the patterns generated by periodic precipitations inchemical systems, such as those described by Liesegang (2),it is suggested that vinculin self-organizes at the plasmamembrane by competitive aggregation. Second, the surpris-ing discovery that some of the actin and vinculin patterns, butnot the corresponding patterns ofproteins known to be linkedto the extracellular matrix, exhibit handedness implies thatasymmetry in biological systems is generated at the interfacebetween the plasma membrane and the cytosol.

MATERIALS AND METHODSPlatelets. Fresh blood containing 10 mM citrate was ob-

tained from the blood bank. Platelets were isolated bydifferential centrifugation and suspended in Hepes-bufferedTyrode's solution (137 mM NaCl/2.6 mM MgCl2/5.5 mMglucose/15 mM Hepes, pH 7.4) at a concentration of 5 x 107per ml. They were maintained at 37°C and used within 2 hr.

Immunofluorescence. Samples (7 ,ul) of the platelet suspen-sion were placed above glass coverslips in Hepes-bufferedTyrode's solution containing 2 mM CaCl2 at 37°C. At thetimes indicated, platelets that had spread on the coverslipswere fixed in 3.7% formaldehyde and permeabilized with0.1% Triton X-100 for 5 min. They were then double-stainedwith rhodamine-phalloidin (Molecular Probes) and one of thefollowing antibodies: (i) a monoclonal antibody againstchicken gizzard vinculin, (ii) a monoclonal antibody againstplatelet glycoprotein complex gpIlb-IIIa (complex-specific),(iii) a monoclonal antibody against the integrin chain 83, and(iv) a monoclonal antibody against chicken gizzard talin. Theantibody against vinculin was purchased from ICN Immu-nologicals. The antibodies against integrin were a gift from B.Steiner (Hoffmann-LaRoche, Basel). The antibody againsttalin was raised by us in mice and specifically recognizesavian and mammalian talin. The secondary antibody was afluorescein-labeled affinity-purified goat anti-mouse IgMfrom Jackson ImmunoResearch.The coverslips were examined and photographed through

an upright microscope (Zeiss Axiophot) equipped with ax 100 Plan-Neofluar objective.Other Materials. Type I collagen from rat tail was from

Serva; fibronectin, a gift from R. Chiquet (Friedrich MiescherInstitute, Basel), had been isolated from newborn calf serum;fibrinogen and fibrinogen-related peptide (Gly-Gln-Gln-His-His-Leu-Gly-Gly-Ala-Ly s-Gln-Ala-Gly-Asp-Val,GQQHHLGGAKQAGDV) were from Sigma; and Gly-Arg-Gly-Asp-Ser-Pro (GRGDSP) was from Novabiochem(Laufelfingen, Switzerland).

RESULTS AND DISCUSSIONMicrofilament bundles in the cytoskeletons of fully spreadplatelets on glass surfaces resemble stress fibers of culturedcells. The process of spreading preceding this stage has beenwell documented and occurs by radial growth offilopodia andthe extension of veils between them (3-5). Corresponding tothe radial outgrowth of filopodia, the actin skeletons ofplatelets fixed during the spreading phase are arranged radi-ally (Fig. 1A). After completion of the spreading phase, alimited number of actin patterns are observed: parallel bun-dles of microfilaments traversing the length of a platelet,bundles arranged in triangles, and bundles radiating tangen-tially from a center which is most often ellipsoid but can bea circle (Figs. 1 and 2). In addition, mixed forms are found.These patterns are not related to the actin patterns of thespreading phase, nor do they seem to undergo any further

3280

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

4, 2

021

Proc. Natl. Acad. Sci. USA 90 (1993) 3281

FIG. 1. Actin patterns of platelets 2 min (A) and 30 min (B) after settling on glass coverslips. (Bar = 10 ,um.)

changes (data not shown). Previous work has shown that intissue culture cells, stress fibers are anchored in the plasmamembrane through linking proteins assembled in focal con-tacts (6). In spread platelets the focal contact protein vinculinis localized, as expected, at both ends of individual filament

bundles (ref. 7 and Fig. 2). I also observed that both vinculinand actin bundles are situated close to the ventral plasmamembrane. Fig. 2 A and a show an example of parallelfilaments, and Fig. 2B and b a triangle ofmicrofilaments withvinculin accumulated in the corners. The most striking pat-

FIG. 2. Actin (A-F) and vinculin (a-f) patterns of individual platelets fixed 30 min after settling on a glass coverslip. For details see text.(Bar = 5 Am.)

Cell Biology: Hagmann

'.:.. ..

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

4, 2

021

Proc. Natl. Acad. Sci. USA 90 (1993)

terns are found in platelets with actin filaments radiating froma central ellipse or circle: in this case vinculin accumulates inconcentric circles (Fig. 2 C and c) or, more often, along lineswhich form a two-armed spiral (Fig. 2 D-F and d-f). Inter-estingly, all the spirals are left-handed when the objective ofthe microscope faces the ventral plasma membrane. At thelevel of the actin skeleton, this is reflected in the rightwardslant of microfilaments radiating from the central area (e.g.,see Fig. 4A). Vinculin is not distributed homogeneously alongthe arms ofthe spiral, but in discrete patches with their longeraxis corresponding to the direction and location of individualmicrofilament bundles. Mapping of actin filaments onto cor-responding vinculin patches shows that sections ofthe spiral,where distal ends of microfilaments are inserted, are contig-uous with sections of proximal insertion points (Fig. 3C,arrow).The organization of platelet cytoskeletons provides a

unique opportunity to examine the physicochemical princi-ples underlying the assembly of cytoskeletal structures. In18% Liesegang (2) described concentric rings ofprecipitationobtained by letting an electrolyte diffuse from a central wellinto a gel containing a second electrolyte with which it formeda poorly soluble salt. Muller et al. (8) described details ofsuchprecipitations that resemble the spiral depositions of vinculinshown here. Similarities include not only the formation ofspirals, which had already been described by Lieseganghimself (9), but also the substructure ofthe precipitation rings(8). Like vinculin patches, rings of Liesegang precipitatesconsist of small parallel stripes aligned radially on neighbor-ing rings. Moreover, rings and spiral arms are subdivided intosegments (e.g., Fig. 2c). Patterned precipitation has also beenobserved in the absence of a macroscopic gradient (10, 11).In this case, a particularly dominant aggregate ("greedygiant") might organize its surroundings into concentric ringsor spirals (11). Such precipitation patterns belong to the classof dissipative structures described by Nicolis and Prigogine(12), and can be explained by competition between growingaggregates. As the radius of an aggregate increases, theequilibrium concentration of electrolyte decreases locally.The resulting depletion of electrolyte sets up a microscopicconcentration gradient, preventing the formation of newaggregates in the immediate vicinity.

While mere similarity is no proof that two phenomena aregenerated by the same mechanism, its observation maynonetheless lead to a better understanding of biological

_w _~~~~~~~~~~

structures. The similarities reported suggest the possibilitythat in spread platelets vinculin is organized as a result ofpatterned precipitation. Most likely, binding of vinculin toother vinculin molecules (13-15) plays a major role byproviding the self-enhancing interaction. In those caseswhere spirals or concentric rings are observed, cytoskeletalorganization might be dominated by a large early centralaggregate (a physiological "greedy giant"). The other pat-terns are then generated by one or several large peripheralaggregates. A mechanism of patterned precipitation wouldexplain the observations that (i) in rat myotubes, acetylcho-line receptors and vinculin assemble into alternating stripesof aggregates (16) and (ii) binding ofvinculin to focal contactsin vitro is not saturable (17).What other molecules contribute to the symmetry-

breaking phenomena described? Focal contact sites of tissueculture cells are characterized by an accumulation of, besidesvinculin, integrin-type receptors, talin, and various otherintracellular proteins (1). This led to models offocal contactsthat have in common a chain of molecules linking theextracellular matrix via integrins, talin, and vinculin to theends of actin filaments (e.g., see ref. 18). However, extra-cellular matrix and integrins are probably not involved in theperiodic cytoskeletal structures of platelets. All the patternsobserved on uncoated glass coverslips are also found oncoverslips coated with collagen, fibronectin, or fibrinogen(data not shown). Peptides interfering with the binding ofextracellular matrix proteins to integrins (GRGDSP andGQQHHLGGAKQAGDV, the carboxyl-terminal peptide offibrinogen) have no effect on spreading and pattern types, nordoes an antibody blocking the activity of gplIb-IIIa, themajor platelet integrin (data not shown). Immunostainingplatelets spread on glass with antibodies against gplIb-IIIa oragainst 83 (a chain common to gpIIb-IIIa and the vitronectinreceptor) revealed that these integrins do not accumulate atthe ends of microfilaments but are homogeneously distrib-uted (Fig. 4 a and b). Finally, talin, which binds to integrins(19) and to vinculin (20) in vitro, shows the same homoge-neous distribution as integrins (Fig. 4c; see also ref. 7). Takentogether, these results suggest that symmetry breaking oc-curs independently of extracellular components at the inter-face between the ventral plasma membrane and the cytosol.A close interaction between vinculin and anionic phospho-lipids has been demonstrated in vitro and in vivo (21-23).Possibly, interaction with phospholipids enhances self-

C

FIG. 3. Mapping of actin filaments (A and black lines in C) onto vinculin patches (B and red areas in C). For explanation see text.

3282 Cell Biology: Hagmann

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

4, 2

021

Proc. Natl. Acad. Sci. USA 90 (1993) 3283

FIG. 4. Actin (A-C), gpllb-IIIa (a), (33 integrin (b), and talin (c) localization in platelets 30 min after settling on coverslips. (Bar = 10 ,um.)

aggregation of vinculin by increasing its concentration closeto the lipid bilayer. A similar mechanism has been proposedas an explanation for the self-organization of laminin onsulfatide-rich planar lipid bilayers (24).To what extent direct or indirect interactions between actin

and vinculin are required cannot be decided at present. Actinhas been shown to self-organize at lipid interfaces (25), but itmight not be required for periodic precipitations of vinculinas such. However, since actin polymers are asymmetrical,actin could be involved in selecting the left-handed spiral.Handedness is well documented at the molecular level and atthe level ofwhole organs and organisms. But the gap betweenthe two levels is wide, and how (or whether) they are relatedis an open question (26, 27). The observation that handedcytoskeletal structures are generated at the interface betweenthe attached plasma membrane and the cytosol of plateletscould be an important step in filling that gap. In analogy to theresults obtained with platelets, spreading ofa cell on a surfaceof extracellular matrix (or on another cell) might be sufficientto determine asymmetry. In that case, the bias does not haveto be provided by an extracellular signal. Interestingly,disturbing the interaction between cells and extracellularmatrix during blastula or early gastrula stages in Xenopuslaevis results in random lateralization ofthe viscera (i.e., 50oof the embryos exhibit situs inversus) (28). Although it is tooearly to draw conclusions, it is tempting to speculate thatsevering the first cells of a nascent organ from their naturalsurface leaves the choice of handedness to chance.

I thank Prof. M. M. Burger for his support and discussion of theproject, M. Grob for her excellent technical assistance, and I.Obergfoll for her photographic work. I am also indebted to Drs. D.Dagan, A. Matus, and T. Schafer for their critical reading of themanuscript and to Dr. S. C. Muller for discussing properties ofdissipative structures.

1. Burridge, K., Fath, K., Kelly, T., Nuckolls, G. & Turner, C.(1988) Annu. Rev. Cell Biol. 4, 487-525.

2. Liesegang, R. E. (1896) Naturwiss. Wschr. 11, 353.

3. Allen, R. D., Zacharski, L. R., Widirstky, S. T., Rosenstein,R., Zaitlin, L. M. & Burgess, D. R. (1979) J. Cell Biol. 83,126-142.

4. Karlsson, R., Lassing, I., Hoglund, A. & Lindberg, U. (1984)J. Cell. Physiol. 121, 96-113.

5. White, J. G. (1987) Ann. N. Y. Acad. Sci. 509, 156-176.6. Geiger, B. (1979) Cell 18, 193-205.7. Nachmias, V. T. & Golla, R. (1991) Cell Motil. Cytoskeleton

20, 190-202.8. Muller, S. C., Kai, S. & Ross, J. (1982) Science 216, 635-637.9. Liesegang, R. E. (1939) Kolloid Z. 87, 57-58.

10. Muller, S. C. & Hess, B. (1989) in Cooperative Dynamics inComplex Physical Systems, ed. Takayama, H. (Springer, Ber-lin), pp. 307-317.

11. Feinn, D., Ortoleva, P., Scalf, W., Schmidt, S. & Wolff, M.(1978) J. Chem. Phys. 69, 27-39.

12. Nicolis, G. & Prigogine, I. (1977) Self-organization in Non-equilibrium Systems (Wiley Interscience, New York).

13. Otto, J. J. (1983) J. Cell Biol. 97, 1283-1287.14. Molony, L. & Burridge, K. (1985) J. Cell. Biochem. 29, 31-36.15. Fringeli, U. P., Leutert, P., Thurnhofer, H., Fringeli, M. &

Burger, M. M. (1986) Proc. Natl. Acad. Sci. USA 83, 1315-1319.

16. Bloch, R. J. & Geiger, B. (1980) Cell 21, 25-35.17. Zafrira, A., Small, J. V. & Geiger, B. (1983) J. Cell Biol. 96,

1622-1630.18. Geiger, B. (1989) Curr. Opin. Cell Biol. 1, 103-109.19. Horwitz, A., Duggan, K., Buck, C., Beckerle, M. C. & Bur-

ridge, K. (1986) Nature (London) 320, 531-533.20. Burridge, K. & Mangeat, P. (1984) Nature (London) 308,

744-746.21. Niggli, V., Dimitrov, D. P., Brunner, J. & Burger, M. M. (1986)

J. Biol. Chem. 261, 6912-6918.22. Niggli, V., Sommer, L., Brunner, J. & Burger, M. M. (1990)

Eur. J. Biochem. 187, 111-117.23. Meyer, R. K. (1989) Eur. J. Cell Biol. 50, 491-499.24. Kalb, E. & Engel, J. (1991) J. Biol. Chem. 266, 19047-19052.25. Weisenhorn, A. L., Drake, B., Prater, C. B., Gould, S. A. C.,

Hansma, P. K., Ohnesorge, F., Egger, M., Heyn, S. P. &Gaub, H. E. (1990) Biophys. J. 58, 1251-1258.

26. Galloway, J. W. (1989) Experientia 45, 859-872.27. Brown, N. A. & Wolpert, L. (1990) Development 109, 1-9.28. Yost, H. J. (1992) Nature (London) 357, 158-161.

Cell Biology: Hagmann

I

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

4, 2

021