tumor cell locomotion and metastatic spread

Tumor Cell Locomotion and Metastatic SpreadGUNTER SIEGEL,1* MARTIN MALMSTEN,2 AND DIETRICH KLUßENDORF1

1Institute of Physiology, Biophysical Research Group, The Free University of Berlin, Berlin, Germany2Institute of Surface Chemistry, Stockholm, Sweden

KEY WORDS tumor cell migration and motility; metastatic spread; cytoskeletal filamentproteins; a-actinin; filamin; desmin; cross-linked filamin-desmin; actin cytoskel-eton; syndecan; perlecan; ellipsometry; molecular length

ABSTRACT The cytoskeletal filament proteins a-actinin, filamin, desmin, and filamin–desminaggregates were adsorbed to a hydrophobic silica surface. The adsorbed amount as measured byellipsometric methods after rinsing and equilibration was 2.7 mg/m2 for a-actinin and 0.4 mg/m2 forfilamin plus desmin, respectively. Adsorbed layer thicknesses in physiological salt solution wereabout 107 nm, 89 nm, 108 nm and 93 nm for a-actinin, filamin, desmin, and cross-linkedfilamin-desmin, respectively. Ca21 ions in a concentration of 1024, 1023, and 2.52 mmol/l had noeffect on the adsorbed amount, refractive index, and adsorbed layer thickness of the individualintermediate filament proteins. Cross-linked filamin–desmin, however, reacted markedly upon theaddition of these Ca21 concentrations with a change in refractive index and adsorbed layerthickness. The layer formed by the filamin–desmin complex contracted by 2–3, 6–7, and 6–7 nm,respectively. The maximum shortening occurred at 1 µmol/l Ca21. The Ca21-dependent adsorbedlayer changes of cross-linked filamin–desmin supports the contractile mechanisms in musculartissues and forms the basis for migration and motility in nonmuscular cells. These motional eventsare crucially involved in peripheral organ perfusion, inflammation, and tumor invasion andmetastasis. Microsc. Res. Tech. 43:276–282, 1998. r 1998 Wiley-Liss, Inc.

INTRODUCTIONThe flexible, thin mats of basement membranes are

the strategic sites at which cells traverse this surface,subbase, and interfacial structure in inflammation andtumor invasion. In inflammatory processes they do sounder complete control, and in tumor invasion andmetastasis they do so by avoiding and excluding thecontrol systems (Siegel, 1996a).

The three-step development of the invasive pheno-type of cancer cells comprises cell attachment, localproteolysis, and cell migration (Siegel and Malmsten,1997). Adhesion is mediated by a contribution patternof several classes of cell adhesion molecules, such asvitronectin and laminin receptors amplified and dis-persed over the entire cell surface. In addition, cytokine-inducible members of the immunoglobulin superfamilyand the cell surface hyaluronan receptor facilitate themetastatic process in serving as tumor cell attachmentsites to the vascular or lymphatic endothelium. Mol-ecules of the cadherin family, however, suppress tumorinvasion. Overexpression of E-cadherin in highly inva-sive clones resulted in a loss of invasive potency. Matrixmetalloproteinases, constitutively overexpressed or in-duced by cytokines in tumor cell invasion, promote theextracellular matrix proteolysis. Growth factors suchas EGF, TGFa, and PDGF upregulate the transcriptionof interstitial collagenase and stromelysin (Matrisian,1992). Moreover, the balance of active enzymes andtheir inhibitors (TIMPs) has shifted in favor of proteolysis.

Tumor cell motility and metastatic spread are domi-nated by chemokinesis, chemotaxis, and haptotaxis(Aznavoorian et al., 1990; Stetler-Stevenson et al.,

1993). Matrix proteolysis and directional migration areprerequisites for intravasation, extravasation, and prop-agation into the target tissue. While the initial stages ofmetastasis encompass haptotactic migration over in-soluble matrix proteins, chemotactic responses to par-tially degraded matrix components characterize themigratory phenotype later (Stetler-Stevenson et al.,1993). Cell locomotion in eukaryotic cells resides in thecortical actin cytoskeleton, which is interwoven withthe cytoplasmic syndecan (hybrid heparan sulfate/chondroitin sulfate proteoglycan) as well as the cell-and matrix-binding perlecan (heparan sulfate proteogly-can) domains (Siegel and Malmsten, 1997). Therefore,the actin-based cytoskeletal proteins a-actinin, filamin,and desmin (Siegel, 1996b) were adsorbed as monomo-lecular layers to a hydrophobic silica surface in order toinvestigate the effect of Ca21 ions by ellipsometry. TheCa21-dependent contraction of cross-linked filamin–desmin as highlighted in this study supports the contrac-tile mechanisms in muscular tissues, and forms thebasis for migration and motility in nonmuscular cells.These motional events are crucially involved in periph-eral organ perfusion, inflammation, and tumor inva-sion and metastasis.

Abbreviations: bFGF, basic fibroblast growth factor; cAMP, cyclic adenosine38,58-monophosphate; EGF, epidermal growth factor; IGF-1, insulin-like growthfactor-1; PDGF, platelet-derived growth factor; PKC, protein kinase C; PTX,pertussis toxin; TGFa, transforming growth factor a; TGFb, transforming growthfactor b; TIMP, tissue inhibitor of metalloproteinases

*Correspondence to: Prof. Dr. med. Gunter Siegel, Institute of Physiology,Biophysical Research Group, The Free University of Berlin, Arnimallee 22,D-14195 Berlin, Germany.

Received 26 May 1998; accepted in revised form; 8 June 1998

MICROSCOPY RESEARCH AND TECHNIQUE 43:276–282 (1998)

r 1998 WILEY-LISS, INC.

MATERIALS AND METHODSPreparations and Solutions

Water was purified by a Milli-RO 10PLUS unit(Millipore Corp., Bedford, MA, USA), including depthfiltration, carbon adsorption, and decalcination preced-ing reverse osmosis. Subsequently, it was passedthrough a Milli-Q PLUS185 unit (UV light, 185 nm and254 nm) and a Q-PAK unit consisting of an activecarbon unit, a mixed bed ion exchanger, an Organexcartridge, and a final 0.22 µm Millipak 40 filter.

The protein a-actinin (Sigma Chemical Co., St. Louis,MO, USA) was dissolved in a Ca21-free K1 Krebssolution to a final concentration of 0.1 mg/ml (,1.015µmol/l). The pH was kept constant at 7.31 6 0.01 bycontinuously aerating the cuvette solution with a 95%O2/5% CO2 gas mixture (carbogen). The individualproteins desmin (Fluka Chemie, Buchs, Switzerland)and filamin (Sigma) were also used at a 0.1 mg/mlconcentration (desmin ,1.887 µmol/l, pH 7.30 6 0.003;filamin ,0.357 µmol/l, pH 7.29 6 0.02) in Ca21-free K1

Krebs solution. Cross-linked filamin–desmin Ca21-freeKrebs solution (pH 7.27 6 0.01) contained filamin anddesmin in a concentration of 0.05 mg/ml each. Thenormal K1 Krebs solution simulated the intracellularionic microenvironment of the intermediate filamentsand was of the following composition: K1 145.85, Na1

10.0, Ca21 0.0001, Mg21 1.1, Cl2 140.36, HCO32 16.31,

H2PO42 1.38 mmol/l (21°C, pH 7.29). After protein

adsorption, the cuvette solution was diluted by a factorof ten (D in the figures) with respect to protein to reducenonadsorbed excess protein. Thereafter, a 30-minuteequilibration time (E in the figures) was allowed. Silicasurfaces were prepared from polished silicon slides(p-type, boron-doped, resistivity 7–13 Vcm; Okmetic,Finland). In short, these were oxidized thermally inpure and saturated oxygen, followed by annealing andcooling in argon flow to generate an oxide layer thick-ness of about 40 nm. The slides were then cleaned asdescribed previously (Malmsten, 1994) and rinsed twicewith water and ethanol. This procedure rendered thesurfaces hydrophilic, with a water–air contact angle ofless than 10°. Methylated silica surfaces were preparedfrom the silica surfaces by double rinsing with water,ethanol, and trichloroethylene (pro Analysi, Merck,Darmstadt, Germany), followed by treatment with 0.1wt% solution of Cl2(CH3)2Si (Merck) in trichloroethyl-ene for 90 minutes (Jonsson et al., 1982). Finally, theywere rinsed four times in trichloroethylene and etha-nol. This procedure rendered the slides hydrophobic,with an advancing and receding contact angle of 95°and 88°, respectively. They were then stored in ethanoluntil use.

Ellipsometry MeasurementsEllipsometry measurements were performed in situ

by means of null ellipsometry (Azzam and Bashara,1989), using an automated Rudolph thin-film ellipsom-eter, type 436, controlled by a personal computer. Axenon lamp filtered to 401.5 nm was used as the lightsource. A thorough description of the experimentalsetup has been given previously (Landgren and Jons-son, 1993). Prior to adsorption, the ellipsometry mea-surements require a determination of the complexrefractive index of the substrate (Malmsten, 1994). In

the case of a layered substrate such as oxidized silicon,a correct determination of the adsorbed layer thicknessand mean refractive index requires a determination ofthe silicon bulk complex refractive index (N2 5 n2 2 ik2)as well as the thickness (d1) and the refractive index(n1) of the oxide layer. This is done by measuring theellipsometric parameters C and D in two differentmedia, e.g., air and buffer. From the two sets of C and D,n2, k2, d1, and n1 can be determined separately (Malm-sten, 1994; Tiberg and Landgren, 1993). The methyllayer is neglected for the methylated silica surfaces(Azzam and Bashara, 1989). Electrophoretic studiessuggest such silanization only adds 0.5 nm or less to thedouble layer surface of shear (Burns et al., 1995; vanAlstine et al., 1993). Similar results were obtained withellipsometry. All measurements were performed byfour-zone null ellipsometry in order to reduce the effectsof optical component imperfections (Azzam and Bashara,1989). After optical analysis of the bare substratesurface, the protein solution was added to the cuvetteand the values of C and D recorded. The adsorption wasmonitored in one zone, since the four-zone procedure istime-consuming and since corrections for componentimperfections had already been performed. The maxi-mal time-resolution between two measurements is 3–4seconds. Stirring was performed by a magnetic stirrerat about 300 rpm.

From C and D, the mean refractive index (nf) andaverage thickness (del) of the adsorbed layer werecalculated numerically according to an optical four-layer model (Malmsten, 1994). It was previously shownthat both adsorbed amounts and adsorbed layer thick-nesses obtained with ellipsometry agree well with thoseobtained with other methods for proteins, polymers,and surfactants at model surfaces (Malmsten, 1994,1995; Malmsten et al., 1992; Tiberg and Landgren,1993; Tiberg et al., 1994). The adsorbed amount (G) wasobtained using values of the molar refractivity and thespecific volume of 4.1 and 0.75, respectively. Bulkconcentrations used in the measurements were 0.05–0.1 mg/ml. Throughout, the pH was kept around 7.29 60.01 by the bicarbonate/phosphate buffer, and by acontinuous aeration of the cuvette solution with a 95%O2–5% CO2 gas mixture (Aga, Stockholm, Sweden). Thelatter is necessary to keep the pH well-regulatedthroughout the experiments in order to avoid bothprotein degradation and calcium phosphate precipita-tion (cf. Siegel et al., 1996a).

RESULTSIn Figure 1 the adsorbed amount of a-actinin from a

Ca21-free Krebs solution to a hydrophobic surface wasmeasured over time with ellipsometry. The initial courseof the adsorption is similar to that in proteoheparansulfate, even if the adsorbed amount is about twice ashigh (Malmsten and Siegel, 1995; Siegel et al., 1996b).Dilution of the concentration by a factor of ten and thefollowing equilibration insignificantly affected the ad-sorption. In contrast to heparan sulfate proteoglycan,the addition of Ca21 ions had no effect on adsorption inthe physiologic intracellular (1027 to 1026 mol/l) nor inthe physiologic extracellular concentration range (2.52mmol/l). Also, the adsorbed layer thickness of 107 nmwas not changed through dilution, equilibration, or

277TUMOR CELL LOCOMOTION AND METASTATIC SPREAD

Ca21 ions (Fig. 2). The adsorbed amount and themolecular dimensioning of filamin (Fig. 3) and desmin(Fig. 4) behaved qualitatively identically with the sameCa21 additions as in the a-actinin solution.

A completely different picture resulted when a mixedsolution of filamin plus desmin was adsorbed to methyl-ated silica (Fig. 5). The adsorbed amount was initiallyonly 0.4 mg/m2, doubled with dilution and decreasedagain with equilibration. The origin of this ratherunusual behavior might be that the initial adsorptionwas performed in 2.5% glycerol, which was removed onrinsing with Krebs solution. Therefore, this maximumrepresents at least partially a refractive index-basedartifact. However, since we were interested in the Ca21

effects, this is of no significance. Unlike heparan sulfateproteoglycan, with which Ca21 ions initiate a promotionin adsorption (Siegel et al., 1996a), these ions here

caused a reduction of the primary adsorption amountby about 25% in a concentration of 1024 mmol/l Ca21;however, by 85% with 1023 and 2.52 mmol/l. The effectsof Ca21 ions on the adsorbed layer thickness andrefractive index of the filamin–desmin aggregates weresubstantial (Fig. 6). Whereas Ca21 had no effect on theadsorption of the individual compounds to a hydropho-bic silica surface nor on their conformation and molecu-lar length, 1024, 1023, and 2.52 mmol/l Ca21 caused achange of refractive index as well as a reduction of thefilamin–desmin adsorbed layer thickness of 93 nm by2–3, 6–7, and 6–7 nm, respectively (<22.6%, 29.9%,29.7%). Strangely enough, the Ca21-induced effect waslargely transient, indicating several competing pro-cesses. Nevertheless, the shortening triggered immedi-ately by the Ca21 addition lasted for 12 minutes.

Fig. 1. Amount of a-actinin (0.1 mg/ml) adsorbed at hydrophobicsilica from a Ca21-free K1-Krebs solution as a function of time. ArrowD indicates dilution of a-actinin by a factor of ten, arrow E equilibra-tion. Ca21 additions were 1024, 1023, and 2.52 mmol/l.

Fig. 2. Adsorbed layer thickness versus time for a-actinin (0.1mg/ml) at methylated silica from Ca21-free Krebs solution (pH 7.31).D, E and Ca21 additions as described in Figure 1.

Fig. 3. Adsorbed layer thickness versus time for filamin (0.1mg/ml) at methylated silica from Ca21-free Krebs solution (pH 7.29).D, E and Ca21 additions as described in Figure 1.

Fig. 4. Adsorbed layer thickness versus time for desmin (0.1mg/ml) at methylated silica from Ca21-free Krebs solution (pH 7.30).D, E and Ca21 additions as described in Figure 1.

278 G. SIEGEL ET AL.

DISCUSSIONWith respect to the last step within the trias cell

attachment, local proteolysis, and cell migration, it hadto be accepted that active cell motility is a necessaryfeature of carcinoma cells in invasion and metastasis(Siegel, 1996a). Matrix proteolysis and directional mi-gration are required for penetration of extracellularmatrices, intravasation, and extravasation. A variety ofagents, such as host-derived scatter factors (Weidner etal., 1990), growth factors (Aznavoorian et al., 1990),extracellular matrix components (Lester et al., 1991),hyaluronan (Turley, 1992), and tumor-secreted cyto-kines (Atnip et al., 1987; Liotta et al., 1986), promotetumor cell motility. Furthermore, tumor cells are in-volved in several forms of motility: 1) chemokinesis(random in nature), 2) chemotaxis (cell migration di-rected by concentration gradients of soluble attractants),

and 3) haptotaxis (cell migration directed by insolubleextracellular matrix proteins) (Aznavoorian et al., 1990;Stetler-Stevenson et al., 1993). Autocrine motility fac-tors are tumor cell-derived cytokines that are secretedby the primary tumor to stimulate both chemokineticand chemotactic motility through PTX-sensitive recep-tors on the responding cells (Liotta et al., 1986; Strackeet al., 1989). Extracellular matrix proteins such as typeIV collagen, laminin, fibronectin, and thrombospondinexert chemotactic and haptotactic influences(Aznavoorian et al., 1990; Taraboletti et al., 1987).Chemotaxis and haptotaxis are mediated by distinctcell surface receptors that recognize different domainsof the same glycoprotein. Haptotactic migration overinsoluble matrix proteins appears to be the significantresponse during the initial stages of metastasis (Stetler-Stevenson et al., 1993). In contrast, chemotactic re-sponses dominate the migratory phenotype later, whenproteolytic processing of the connective tissue hasborne partially degraded matrix components.

The molecular machinery for cell locomotion in eu-karyotic cells resides in the cortical actin cytoskeleton.During amoeboid movement, the actin filament net-work of tumor cells must be reversibly disassembledand then reassembled to allow both directional pseudo-pod protrusion and subsequent stabilization of theresulting extension (Condeelis et al., 1992). Within thiscontext, we designed experiments with ellipsometrictechniques, where we could study the behavior of thecytoskeletal filament proteins a-actinin, filamin, anddesmin after adsorption to a hydrophobic silica surface(Malmsten, 1994; Malmsten and Siegel, 1995). As men-tioned elsewhere (Siegel, 1996a), many cell surfacereceptors and adhesion molecules, but also cell surfaceand subbase proteoglycans, are connected with thecytoskeleton. The elucidation of the functional bindingto and involvement in the actin cytoskeleton of both thecytoplasmic syndecan and the cell and matrix proteinbinding perlecan domains discloses completely newaspects of the mechanotransduction process and of theendothelial contribution to tumor cell motility duringmetastatic spread.

In order to elucidate the linkage between the cytoplas-mic and cell/matrix-binding domains of the proteohepa-ran sulfates and the cytoskeleton, ellipsometric investi-gations on the influence of Ca21 ions on a-actinin,filamin, and desmin were conducted. Actin-based cyto-skeletal and contractile proteins appear to be distrib-uted in two domains (Siegel, 1996b): 1) longitudinalintermediate filaments free of myosin, but containingdesmin, vimentin, filamin, actin, and a-actinin, and 2)contractile filaments containing actin, myosin, tropo-myosin, and caldesmon. The tension developed by thecontractile filaments of a myocyte is proportionatelytransmitted to the cell interior and exterior by theproteins of intermediate filaments. These latter appearnot only in muscle cells but also in normal and cancer-ously transformed endo- and epithelial cells, neurons,and fibroblasts. The intermediate filaments build anintracellular network which is anchored in the densebands and dense bodies (Fig. 7), corresponding function-ally to the Z disks in obliquely striated musculature(Stephens, 1977). They might be their phylogeneticprecursors, since both structures contain a-actinin. Incontrast to the dense bodies, dense bands contain the

Fig. 5. Amount of cross-linked filamin-desmin (0.05 mg/ml forfilamin and desmin, respectively) adsorbed at hydrophobic silica froma Ca21-free K1-Krebs solution as a function of time. D, E and Ca21

additions as described in Figure 1.

Fig. 6. Adsorbed layer thickness (del) and mean adsorbed layerrefractive index (nf) versus time for cross-linked filamin-desmin (0.05mg/ml for filamin and desmin, respectively) at methylated silica fromCa21-free Krebs solution (pH 7.27). Ca21 additions as listed inFigure 1.


protein vinculin. As the intermediate filaments linkdense bands and dense bodies throughout the cell, theyform an actin-based cytoskeleton. Via vinculin andtalin, but also via fibulin and a-actinin, this cytoskel-eton is connected with transmembrane integrins. Theirextracellular domains interact with proteoglycans (per-lecan, decorin) and glycoproteins (laminin, fibronectin)of the basement membrane and extracellular matrices,

which in turn communicate with extracellular collagenand elastin fibers (inside-out signaling) (Hynes, 1992).

On the other hand, the cell-surface proteoglycans ofthe novel gene family of syndecans (Greek syndein, tobind together) are polymorphic, membrane-interca-lated, proteoheparan/chondroitin sulfates abundant onepi/endothelial and induced mesenchymal cells duringembryogenesis and in adult tissues (Mali et al., 1990;Saunders et al., 1989). Their highly conserved cytoplas-mic protein sequence contains three tyrosine residuesthat may provide phosphorylation sites for proteinkinases, which are key enzymes in many signal trans-duction pathways, or they may form the sites forcross-linkage with cytoskeletal elements (Hardinghamand Fosang, 1992). This entodomain has protein bind-ing activity because the intact proteoglycan can binddirectly or indirectly to F-actin (Rapraeger and Bern-field, 1982) and associates with the actin-containingcytoskeleton (Rapraeger et al., 1986) when the ectodo-main undergoes a conformational change, whether byshear stress or by cross-linking at the cell surface(outside-in signaling) (Hynes, 1992). Since a cell typecan contain both syndecans and integrins as structur-ally and functionally distinct receptors, syndecan hasbeen called a coreceptor collaborating with integrins insignal transduction (Hardingham and Fosang, 1992).Thus, this manifoldly interwoven structure of copoly-mers builds up a functional cytoskeletal organizationand mechanical syncytium (Izzard et al., 1986; Siegel,1996b). This aspect appears to be even more significant,as the dynamic regulation of the assembly and turnoverof intermediate filaments takes place by means of thephosphorylation of desmin by cAMP-dependent proteinkinase (cA-PK) or protein kinase C (PKC) (Kamm andStull, 1989).

The striking observation in our investigations wasthat Ca21 ions affected only adsorbed layers formed byfilamin–desmin aggregates, whereas they had no effecton the single elements, i.e., filamin and desmin alone.More precisely, on addition of even extremely low Ca21

concentrations, there were changes in both the ad-sorbed amount and the adsorbed layer thickness. Anattractive interpretation of the latter findings would bea Ca21-dependent contraction of the molecules. How-ever, since these measurements give information on theadsorbed layer thickness rather than on moleculardimensions, and since the former generally decreaseswith a decreasing interfacial crowding (decreasing ad-sorbed amount), this interpretation is not unambigu-ous. Despite this, however, it is clear that Ca21 didindeed have pronounced effects on the stratified layer.Apparently, a cross-linkage of both molecules is theprerequisite to this effect. Filamin has been describedas a cross-linker and bundling protein (Gorlin et al.,1990; Hartwig and Kwiatkowski, 1991). The free Ca21

concentration in a tumor cell is about 1027 mol/l. Eventhis low concentration had a slightly contracting effect.The physiological elevation of [Ca21]i by a factor of tenwas followed by an adsorbed layer thickness reductionof approximately 10%. Comparably, a vascular smoothmuscle cell contracts maximally with a rise in [Ca21]iby a factor of 2.5–9 (Siegel, 1996b). The unphysiologicelevation of the Ca21 concentration to its extracellular

Fig. 7. Vascular smooth muscle cells. 1, Cell nucleus; 2, sarcoplas-mic reticulum; 3, caveola; 4, nexus (gap junction); 5, myosin filaments;6, actin filaments; 7, intermediate filaments; 8, dense bands; 9, densebodies; 10, basement membrane; 11, terminal axon; 12, axonal varicosi-ties containing dense-cored vesicles with neurotransmitters; 13, collag-enous fibers; 14, elastic fibers.


value (2.52 mmol/l; factor 2,520) revealed no furthereffect. Thus, the ‘‘contraction curve’’ exhibits saturationbehavior dependent on the Ca21 concentration. Themaximum contraction is attained at 1 µmol/l Ca21 orlower values. In contrast, the measurement of theadsorption G and thickness del of a-actinin, filamin, anddesmin turned out to be Ca21-independent. This is notsurprising, since a-actinin is a constituent of the densebands and dense bodies, strategic anchoring sites ofintermediate filament proteins. It may be mentionedthat we determined with ellipsometric techniques theadsorbed layer thickness of a-actinin in physiologicalsalt solution to be 107 nm, while electron microscopicimages have revealed these macromolecules to be 40nm long (Blanchard et al., 1989). The origin of thisdifference is unknown at present, but may indicate theoccurrence of multilayer adsorption or the formation ofstaggered multimers in the ellipsometry experiments.

Kjellen and Lindahl (1991) reported on the functionalsignificance of the intracellular domain of syndecanthat it interacts with the actin cytoskeleton, resultingin the formation or elimination of focal cell adhesions.Therefore, syndecan is involved in the cellular regula-tion of shape and tissue organization (Bernfield andSanderson, 1990; Kreis and Vale, 1993). Structuralassociation of the entodomain with actin-based interme-diate filaments provides a network through whichcytoskeletal contractile elements can apply tractileforces for subsequent cell movement or deformation(Watson, 1991). Furthermore, the syndecans may par-ticipate in a variety of cellular phenomena such asproliferation, differentiation, transformation, migra-tion, and motility (Mali et al., 1990). This multifunction-ality might serve to integrate cellular behavior inresponse to physical stimuli (shear stress, electrostaticattractions), growth factors (bFGF, TGFb, IGF-1) andextracellular matrices (laminin, fibronectin, thrombo-spondin) (Saunders et al., 1989; Siegel et al., 1996b).

The findings of Ca21-dependent effects on stratifiedcross-linked intermediate filament proteins may indi-cate that filamin–desmin stands for assistance of thecontraction mechanisms in muscular cells, although,due to the changes in the adsorbed amount, this is notentirely clear at present. But nonmuscular cells, liketumor cells, can contract in this way as well. A Ca21

increase could result in a retraction of the tumor cell viaits effect on the cytoskeleton. Finally, the filamin–desmin contraction could be a trigger of nonmuscularcell locomotion, for example, of the amoeboid migration,or the extra- and intravasation in inflammation ortumor invasion and metastasis.

ACKNOWLEDGMENTThe authors are grateful to Mrs. E. Hofmann for the

translation, editorial elaboration, and writing of themanuscript.

REFERENCESAtnip, K.D., Carter, L.M., Nicolson, G.L., and Dabbous, M.K. (1987)

Chemotactic response of rat mammary adenocarcinoma cell clonesto tumor-derived cytokines. Biochem. Biophys. Res. Commun.,146:996–1002.

Aznavoorian, S., Stracke, M.L., Krutzsch, H., Schiffmann, E., andLiotta, L.A. (1990) Signal transduction for chemotaxis and hapto-taxis by matrix molecules in tumor cells. J. Cell Biol., 110:1427–1438.

Azzam, R.M.A., and Bashara, N.M. (1989) Ellipsometry and PolarizedLight. North-Holland, Amsterdam.

Bernfield, M., and Sanderson, R.D. (1990) Syndecan, a developmen-tally regulated cell surface proteoglycan that binds extracellularmatrix and growth factors. Philos. Trans. R. Soc. Lond. (Biol.),327:171–186.

Blanchard, A., Ohanian, V., and Critchley, D. (1989) The structure andfunction of a-actinin. J. Muscle Res. Cell Motil., 10:280–289.

Burns, N.L., van Alstine, J.M., and Harris, J.M. (1995) Poly(ethyleneglycol) grafted to quartz: Analysis in terms of a site-dissociationmodel of electroosmotic fluid flow. Langmuir, 11:2768–2776.

Condeelis, J., Jones, J., and Segall, J.E. (1992) Chemotaxis of meta-static tumor cells: Clues to mechanisms from the Dictyosteliumparadigm. Cancer Metastasis Rev., 11:55–68.

Gorlin, J.B., Yamin, R., Egan, S., Stewart, M., Stossel, T.P., Kwiat-kowski, D.J., and Hartwig, J.H. (1990) Human endothelial actin-binding protein (ABP-280, nonmuscle filamin): A molecular leafspring. J. Cell Biol., 111:1089–1105.

Hardingham, T.E., and Fosang, A.J. (1992) Proteoglycans: Many formsand many functions. FASEB J., 6:861–870.

Hartwig, J.H., and Kwiatkowski, D.J. (1991) Actin-binding proteins.Curr. Opin. Cell Biol., 3:87–97.

Hynes, R.O. (1992) Integrins: Versatility, modulation, and signaling incell adhesion. Cell, 69:11–25.

Izzard, C.S., Radinsky, R., and Culp, L.A. (1986) Substratum contactsand cytoskeletal reorganization of BALB/c 3T3 cells on a cell-binding fragment and heparin-binding fragments of plasma fibronec-tin. Exp. Cell Res., 165:320–336.

Jonsson, U., Ivarsson, B., Lundstrom, I., and Berghem, L. (1982)Adsorption behavior of fibronectin on well-characterized silica sur-faces. J. Colloid Interface Sci., 90:148–163.

Kamm, K.E., and Stull, J.T. (1989) Regulation of smooth musclecontractile elements by second messengers. Annu. Rev. Physiol.,51:299–313.

Kjellen, L., and Lindahl, U. (1991) Proteoglycans: Structures andinteractions. Annu. Rev. Biochem., 60:443–475.

Kreis, T., and Vale, R. (1993) Guidebook to the Extracellular Matrixand Adhesion Proteins. Oxford University Press, Oxford, UK.

Landgren, M., and Jonsson, B. (1993) Determination of the opticalproperties of Si/SiO2 surfaces by means of ellipsometry, usingdifferent ambient media. J. Phys. Chem., 97:1656–1660.

Lester, B.R., Weinstein, L.S., McCarthy, J.B., Sun, Z., Smith, R.S., andFurcht, L.T. (1991) The role of G-protein in matrix-mediated motil-ity of highly and poorly invasive melanoma cells. Int. J. Cancer,48:113–120.

Liotta, L.A., Mandler, R., Murano, G., Katz, D.A., Gordon, R.K.,Chiang, P.K., and Schiffmann, E. (1986) Tumor cell autocrinemotility factor. Proc. Natl. Acad. Sci. USA, 83:3302–3306.

Mali, M., Jaakkola, P., Arvilommi, A.-M., and Jalkanen, M. (1990)Sequence of human syndecan indicates a novel gene family ofintegral membrane proteoglycans. J. Biol. Chem., 265:6884–6889.

Malmsten, M. (1994) Ellipsometry studies of protein layers adsorbedat hydrophobic surfaces. J. Colloid Interface Sci., 166:333–342.

Malmsten, M. (1995) Ellipsometry studies of the effects of surfacehydrophobicity on protein adsorption. Colloids Surf. B: Biointer-faces, 3:297–308.

Malmsten, M., and Siegel, G. (1995) Electrostatic and ion-bindingeffects on the adsorption of proteoglycans. J. Colloid Interface Sci.,170:120–127.

Malmsten, M., Linse, P., and Cosgrove, T. (1992) Adsorption ofPEO-PPO-PEO block copolymers at silica. Macromolecules, 25:2474–2481.

Matrisian, L.M. (1992) The matrix-degrading metalloproteinases.BioEssays, 14:455–463.

Rapraeger, A.C., and Bernfield, M. (1982) An integral membraneproteoglycan is capable of binding components of the cytoskeletonand the extracellular matrix. In: Extracellular Matrix. S. Hawkes,and J.L. Wang, eds. Academic Press, New York, pp. 265–269.

Rapraeger, A., Jalkanen, M., and Bernfield, M. (1986) Cell surfaceproteoglycan associates with the cytoskeleton at the basolateral cellsurface of mouse mammary epithelial cells. J. Cell Biol., 103:2683–2696.

Saunders, S., Jalkanen, M., O’Farrell, S., and Bernfield, M. (1989)Molecular cloning of syndecan, an integral membrane proteoglycan.J. Cell Biol., 108:1547–1556.

Siegel, G. (1996a) Connective tissue: More than just a matrix for cells.In: Comprehensive Human Physiology. R. Greger, and U. Wind-horst, eds. Springer-Verlag, Berlin, Vol. 1, pp. 173–224.

Siegel, G. (1996b) Vascular smooth muscle. In: Comprehensive HumanPhysiology. R. Greger, and U. Windhorst, eds. Springer-Verlag,Berlin, Vol. 2, pp. 1941–1964.


Siegel, G., and Malmsten, M. (1997) The role of the endothelium ininflammation and tumor metastasis. Int. J. Microcirc. Clin. Exp.,17:257–272.

Siegel, G., Malmsten, M., Klußendorf, D., Walter, A., Schnalke, F., andKauschmann, A. (1996a) Blood-flow sensing by anionic biopolymers.J. Auton. Nerv. Syst., 57:207–213.

Siegel, G., Walter, A., Kauschmann, A., Malmsten, M., and Buddecke,E. (1996b) Anionic biopolymers as blood flow sensors. Biosens.Bioelectron., 11:281–294.

Stephens, N.L. (1977) The Biochemistry of Smooth Muscle. UniversityPark Press, Baltimore, MD.

Stetler-Stevenson, W.G., Aznavoorian, S., and Liotta, L.A. (1993)Tumor cell interactions with the extracellular matrix during inva-sion and metastasis. Annu. Rev. Cell Biol., 9:541–573.

Stracke, M.L., Engel, J.D., Wilson, L.W., Rechler, M.M., Liotta, L.A.,and Schiffmann, E. (1989) The type I insulin-like growth factorreceptor is a motility receptor in human melanoma cells. J. Biol.Chem., 264:21544–21549.

Taraboletti, G., Roberts, D.D., and Liotta, L.A. (1987) Thrombospondin-induced tumor cell migration: Haptotaxis and chemotaxis are

mediated by different molecular domains. J. Cell Biol., 105:2409–2415.

Tiberg, F., and Landgren, M. (1993) Characterization of thin nonionicsurfactant films at the silica/water interface by means of ellipsom-etry. Langmuir, 9:927–932.

Tiberg, F., Jonsson, B., Tang, J., and Lindman, B. (1994) Ellipsometrystudies of the self-assembly of nonionic surfactants at the silica—water interface: Equilibrium aspects. Langmuir, 10:2294–2300.

Turley, E.A. (1992) Hyaluronan and cell locomotion. Cancer Metasta-sis Rev., 11:21–30.

van Alstine, J.M., Burns, N.L., Riggs, J.A., Holmberg, K., and Harris,J.M. (1993) Electrokinetic characterization of hydrophilic polymercoatings of biotechnical significance. Colloids Surf. A: Physicochem.Eng. Aspects, 77:149–158.

Watson, P.A. (1991) Function follows form: Generation of intracellularsignals by cell deformation. FASEB J., 5:2013–2019.

Weidner, K.M., Behrens, J., Vandekerckhove, J., and Birchmeier, W.(1990) Scatter factor: Molecular characteristics and effect on theinvasiveness of epithelial cells. J. Cell Biol., 111:2097–2108.


tumor cell locomotion and metastatic spread

Documents