Morphologie (2014) 98, 53—64

Disponible en ligne sur

ScienceDirectwww.sciencedirect.com

ORIGINAL ARTICLE

Diversity of bone matrix adhesion proteinsmodulates osteoblast attachment andorganization of actin cytoskeletonLes différents types de molécules d’adhésion modulentl’ancrage des ostéoblastes et l’organisation des fibresd’actine du cytosquelette

V. Demaisa, C. Audraina,b, G. Mabilleaua,b, D. Chapparda,∗,b,M.F. Basléa,b

a Groupe études remodelage osseux et biomatériaux (GEROM), LHEA, IRIS-IBS (institut de biologie ensanté), LUNAM université, CHU d’Angers, 49933 Angers cedex, Franceb Service commun d’imageries et d’analyses microscopiques (SCIAM), IRIS-IBS (institut de biologie ensanté), LUNAM université, CHU d’Angers, 49933 Angers cedex, France

Available online 13 April 2014

KEYWORDSMatrix protein;Osteoblast;Cytoskeleton;Scanning electronmicroscopy;Immunogold labeling;Fractal analysis;Vitronectin;Integrin;Fibronectin

Summary Interaction of cells with extracellular matrix is an essential event for differentia-tion, proliferation and activity of osteoblasts. In bone, binding of osteoblasts to bone matrixis required to determine specific activities of the cells and to synthesize matrix bone proteins.Integrins are the major cell receptors involved in the cell linkage to matrix proteins such asfibronectin, type I collagen and vitronectin, via the RGD-sequences. In this study, cultures ofosteoblast-like cells (Saos-2) were done on coated glass coverslips in various culture conditions:DMEM alone or DMEM supplemented with poly-L-lysine (PL), fetal calf serum (FCS), fibronectin(FN), vitronectin (VN) and type I collagen (Col-I). The aim of the study was to determine thespecific effect of these bone matrix proteins on cell adherence and morphology and on thecytoskeleton status. Morphological characteristics of cultured cells were studied using scanningelectron microscopy and image analysis. The heterogeneity of cytoskeleton was studied usingfractal analysis (skyscrapers and blanket algorithms) after specific preparation of cells to exposethe cytoskeleton. FAK and MAPK signaling pathways were studied by western blotting in these

various culture conditions. Results demonstrated that cell adhesion was reduced with PL andVN after 240 min. After 60 min of adhesion, cytoskeleton organization was enhanced with FN,VN and Col-I. No difference in FAK phosphorylation was observed but MAPK phosphorylationwas modulated by specific adhesion on extracellular proteins. These results indicate that

∗ Corresponding author.E-mail address: daniel.chappard@univ-angers.fr (D. Chappard).

1286-0115/$ – see front matter © 2014 Elsevier Masson SAS. All rights reserved.http://dx.doi.org/10.1016/j.morpho.2014.02.003

54 V. Demais et al.

culture conditions modulate cell adhesion, cytoskeleton organization and intracellular proteinpathways according to extracellular proteins present for adhesion.© 2014 Elsevier Masson SAS. All rights reserved.

MOTS CLÉSProtéinesmatricielles ;Ostéoblaste ;Cytosquelette ;Microscopieélectronique àbalayage ;Immunomarquage àl’or ;Analyse fractale ;Vitronectine ;Intégrine ;Fibronectine

Résumé L’interaction des cellules avec la matrice extracellulaire est un événement essentielpour la différenciation, la prolifération et l’activité des ostéoblastes. Dans l’os, l’adhérencedes ostéoblastes à la matrice osseuse est nécessaire pour déterminer les activités spécifiques deces cellules et pour qu’elles synthétisent les protéines matricielles. Les intégrines sont les prin-cipaux récepteurs cellulaires impliqués dans la liaison des cellules aux protéines matriciellestelles que la fibronectine, le collagène de type I et la vitronectine (via la séquence RGD).Dans cette étude, des cultures de cellules de type ostéoblastique (Saos-2) ont été réaliséessur des lamelles de verre recouvertes de différents supports de culture : DMEM seul ou DMEMsupplémenté avec de la poly-L-lysine (PL), sérum de veau fœtal (FCS), fibronectine (FN), vit-ronectine (VN) et collagène de type I (Col-I). Le but de l’étude était de déterminer l’effetspécifique de ces protéines matricielles sur l’adhérence des cellules, leur morphologie et surleur cytosquelette. Les caractéristiques morphologiques des cellules en culture ont été étudiéesen microscopie électronique à balayage et analyse d’images. L’hétérogénéité du cytosquelettea été étudiée par analyse fractale (algorithmes des gratte-ciel et des couvertures) après unepréparation spécifique des cellules pour exposer le cytosquelette. Les voies de signalisationFAK et MAPK ont été étudiées par western blot dans ces différentes conditions de culture. Lesrésultats ont montré que l’adhésion cellulaire a été réduite avec PL et VN après 240 minutes.Après 60 minutes, l’organisation du cytosquelette a été optimisée sur les lamelles recouvertesde FN, VN et Col-I. Aucune différence dans la phosphorylation de FAK n’a été observée maisla phosphorylation MAPK était modulée par l’adhérence spécifique aux protéines extracellu-laires. Ces résultats indiquent que les conditions de culture modulent l’adhésion cellulaire,l’organisation du cytosquelette et les voies de signalisation intracellulaire en fonction desprotéines extracellulaires présentes qui permettent l’adhérence des cellules ostéoblastiques.

. Tou

I

OTmOitcdOraottlessutaiv(gaahe

tI

bkvcfps∝fipAadifia

bmfctcfis

© 2014 Elsevier Masson SAS

ntroduction

steoblasts are the cells responsible for bone formation.hey synthesize numerous proteins of the extracellular boneatrix and are involved in the process of mineralization.steoblasts are generated from bone marrow osteoprogen-

tors and several hormones and growth factors contributeo their differentiation and activity. However, interaction ofells with the extracellular matrix is an essential event forifferentiation, proliferation and activity of osteoblasts [1].steoblasts are linked to the bone matrix via cell surface

eceptors (mainly integrin) that bind the specific arg-gly-aspmino-acid sequence (RGD) shared by some glycoproteinsf the matrix (thrombospondin, fibronectin, vitronectin,ype I collagen, fibrilline and osteopontin, bone sialopro-ein [belonging to the SIBLING family: small integrin-bindingigand, N-linked glycoproteins]) [2]. Integrins are ∝� het-rodimers with an extracellular domain that binds the RGDequence, a transmembranous part and a short intracellularequence linked to cytoskeleton proteins. At least 16 ∝ sub-nits and 8 � subunits have been identified leading to morehan 20 different types of integrins. Each type of integrin isble to bind a specific RGD-containing protein but severalntegrins may also bind the same protein [3]. The major cellitronectin receptor is ∝V�3 that can also bind fibronectinFN); conversely, vitronectin (VN) may also link other inte-rins. Numerous integrins mediate cell attachment to FN,mong them ∝3�1, ∝4�1 and ∝V�3 [4]. All these integrins

re expressed at the osteoblast surface [5]. However, in alluman osteosarcoma cells some of them may be absent. Forxample, Saos-2 cells express ∝1�1 but not ∝2�1 [6] contrary

cve

s droits réservés.

o MG-63 where only ∝2�1 (and not ∝1�1) is present as type collagen receptor [7].

When bound to their ligands, integrins transmit signalsy recruiting cytoskeleton and signaling proteins to sitesnown as focal adhesions. Formation of focal adhesion pro-ides the physical contact sites of attachment betweenells and extracellular matrix. Clustering of integrins inocal adhesion site induces the recruitment and the phos-horylation of tensin and focal adhesion kinase (FAK) thatubsequently induces the recruitment of talin, vinculin and-actinin [8,9]. Talin, vinculin and ∝-actinin link the F-actinbers to the plasma membrane and interact with severalroteins such as vaso-dilatator phosphoprotein (VASP) andrp2/3 proteins which are implicated in the regulation ofctin polymerization [10,11]. Rearrangement of F-actin bun-les induced by their liaison to the integrin � subunit cannduce changes in the cell shape [12,13]. This can producenger-like or sheet-like protrusions (respectively filopodiand lamellipodia) [14].

Synthesis and distribution of cytoskeleton proteins haveeen involved as a cause or a consequence, in osteoblastorphology, growth, migration, attachment, signaling and

unctions [15]. The growth of bone cells is associated withhanges in organization and synthesis of cytoskeleton pro-eins and migration implicates dynamic reorganization of theytoskeleton. Rearrangement and polymerization of actinlaments are a consequence of the linkage between cellurface receptor and bone matrix adhesion protein. As a

onsequence of the involvement of osteoblast attachmentia the integrin-cytoskeleton system, signal transductionvents are activated allowing control of gene expression

C

SD55citef

••••••

g2jactd

K

FvcictMa

i+hfica0taa�

A

Ccwi

Effects of matrix proteins on osteoblasts

in osteoblasts [15]. Orientation of matrix collagen fibersdeposited by osteoblasts is controlled by cell orienta-tion, which depends of the cytoskeleton organization. Inosteoblasts, actin filaments are abundant and mainly dis-tributed at the periphery of the cytoplasm and in the core ofthe cell processes [15]. Two other cytoskeleton elements arealso present in osteoblasts, microtubules of tubulin (whichcross the cytoplasm as continuous structure for long dis-tance between perinuclear space and plasma membrane)and intermediate filaments involved in the nuclear anchor-age [16]. Actin is a major cytoskeleton protein that playsa central role in cell shape determination. Changes in thepattern of the actin fibers distribution induce cell morpho-logical changes, particularly in term of cell spreading. It mayalso determine or reflect functional modifications. A grow-ing consensus of experimental data now suggests that thechanges within cytoarchitecture of the cell and activationof specific signal transduction pathways are linked [17].

In this study, osteoblast-like cells from human osteosar-coma (Saos-2) were cultured on FN, VN and collagen type I(Col-I) coated coverslips. Kinetic of cell attachment, mor-phology and proliferation were studied and the organizationof the actin cytoskeleton was determined using light andelectron microscopy.

Materials and methods

Materials and reagents

Human osteoblast-like cells, Saos-2, derived fromosteogenic osteosarcoma (HTB 85) were purchasedfrom ATCC (Manassas, USA). Dulbecco’s Modified Eagle’sMedium (DMEM), L-glutamine, penicillin/streptomycin andtrypsin/EDTA were all from Biomedia (Boussens, France) andfetal calf serum (FCS) from Polylabo (Strasbourg, France).Poly-L-lysine (PL), human fibronectin (F 0895), calf collagentype I (C9791), hexamethyldisilazane (HMDS), 3,3′ dithio bis(propionic acid N. hydroxysuccinimide ester) (DTSP), TritonX-100, EGTA, Piperazine-N, N′-bis 2-ethanesulfonic acid(Pipes), Dnase I (D 7291), Rnase A (R 6513), Phalloidin-TRITC(P1951), paraformaldehyde, sodium borohydride, rabbitpolyclonal antibody against-actin (A 2066), rabbit polyclonalantibody against — focal adhesion kinase (F-2918) were pur-chased from Aldrich-Sigma (Saint Quentin Fallavier, France).The mouse monoclonal antibody against phospho-p44/42MAPK was from Ozyme (Saint-Quentin-en-Yvelines, France).Chloroform was from Aldrich. Human VN (12165064) and6-well plates came from Life Technology (Eragny, France).Bovine serum albumin (BSA) was from Euromedex (Souf-felweyershein, France), PEG 6000 and glutaraldehydefrom Merck (Fontenay-sous-bois, France). Glass coverslipswere from CML (Nemours, France), and 100 and 250 mmplastic dishes from Fisher Scientific (Elancourt, France).DNA quantification was done on a Shimatzu spectrometer(Kyoto, Japan). Carbon-sputtering was done with a MED 020,

Bal-Tec (Chatillon sur Cher, France). The scanning electronmicroscopy (SEM) was done using a JEOL JSM-6301 F fromJEOL (Paris, France) and the fluorescent microscopy withan Olympus Fluoview confocal microscope (Paris, France).

s1aw

55

ell culture

aos-2 were routinely grown on 250 mm plastic dishes inMEM supplemented with 10% FCS, 4 mm L-glutamine and0 U/ml penicillin/streptomycin, at 37 ◦C in humidified and% CO2 atmosphere. At confluence and before experiments,ells were FCS-deprived overnight at 37 ◦C with 5% CO2

n a humidified atmosphere and then passaged with 0.1%rypsin/EDTA. For experiments, the supports of culture (cov-rslips or plastic dishes) were coated according to theollowed conditions:

DMEM medium culture alone; DMEM with 0.001% PL as unspecific substrate; DMEM supplemented with 10% FCS; DMEM supplemented with 2 �g/ml FN; DMEM supplemented with 1 �g/ml VN; DMEM supplemented with 10 �g/ml type I collagen.

For each condition, 100 mm plastic dishes and 25 mmlass coverslips were sterilized by ultraviolet irradiation for

h and then coated by pre-incubation at 4 ◦C, overnight,ust before experiment. The coating medium was removednd cells were plated on coated glass coverslips at 2.5 × 105

ells per coverslip and at 2 × 105 per cm2 in 100 mm plas-ic dishes. For all the following experiments, cultures wereone in DMEM alone without any supplementation.

inetic of cell adherence

or each cultured condition, cells were seeded and culti-ated in triplicate onto coated glass coverslips (2.5 × 105

ells/coverslip) for 5, 30, 60, 120 and 240 min at 37 ◦Cn humidified and 5% CO2 atmosphere. Before harvesting,ultures were rinsed in PBS and then treated with 0.1%rypsin/EDTA. The number of cells was counted using aalassez cell and results were expressed as percentage ofdherent cells vs. seeded cells

For DNA content measurements, cell cultures were rinsedn PBS and lysed in potassium hydroxide 0.6 N by 3 cycles37 ◦C/−196 ◦C. Suspensions were incubated 1 h at 37 ◦C toydrolyze RNA. Hydrogen perchloride was then added at thenal concentration of 0.6 N and incubated 1 h at 37 ◦C to pre-ipitate total DNA and proteins. After centrifugation 10 mint 30,000 g, pellets were solubilized in hydrogen perchloride.6 N and incubated 10 min at 80 ◦C for DNA hydrolyze andhen 1 h at 4 ◦C. After centrifugation, proteins are pelletednd the DNA in the supernatants was quantified by dosaget 260 nm using a spectrometer. Results were expressed ing of DNA per coverslip.

ctin visualization with TRITC-labelled phalloidin

ells were cultured onto coated coverslips (2.5 × 105

ells/coverslip) for 5, 30, 60, 120, 240 min and 24 h. Afterashing in PBS, cells were fixed with 4% paraformaldehyde

n PBS for 20 min, at room temperature. Following exten-

ive rinsing with PBS, cells were treated in the dark with2.7 �M TRITC-labeled phalloidin in PBS, for 30 min. Thectin cytoskeleton was identified under confocal microscopeith a highpass BA610IF filter.

5

S

Fcc2dtpoepS

cfipcnraSat5iawPtol

wbPwtigcwt

I

T(c

TWepaonobTt

Dab

TPsTfosaiebrtd

Immunoprecipitation and western blot analysis

Cells were plated onto 100-mm coated plastic Petri dishes(2 × 105 per cm2) for 5, 30, 60 and 120 min at 37 ◦C in

Figure 1 Effects of matrix proteins on DNA contents. Cellswere grown during 5, 30, 60, 120 and 240 min in DMEM alone(DMEM) and in presence of poly-L-lysine (PL), fetal calf serum(FCS), fibronectin (FN), vitronectin (VN) and type I collagen(Col-I). *P < 0.05 vs. 5 min, § vs. 30 min, © vs. 60 min, vs.120 min.Effet des protéines matricielles sur le contenu en ADN. Les cel-lules ont été cultivées pendant 5, 30, 60, 120 et 240 minutesdans du milieu DMEM seul (DMEM) et en présence de poly-

6

canning electron microscopy (SEM)

or morphological study and quantitative measurements,ells were plated onto coated glass coverslips (2.5 × 105

ells/coverslip) and cultured for 5, 30, 60, 120, 240 min and4 h. Coverslips were fixed in 2.5% glutaraldehyde in PBS,ehydrated in graded ethanol series, desiccated in hexame-hyldisilazane and air dried. Samples were carbon-coatedrior SEM examination at 5 kV. Analysis and measurementsf cell surface area were performed on at least 10 cells, fromach culture condition and time, by image analysis using SEMhotographs at the original magnification of ×1500 and NIHcion image for Windows.

For cytoskeleton examination, cells were plated ontooated glass coverslips (2.5 × 105 cells/coverslip), culturedor 5, 30, 60, 120, 240 min and 24 h and prepared accord-ng to the method of Bell et al. [18—20]. Briefly, cells werere-fixed during 5 min at 37 ◦C in 0.02% DTSP in PBS. Theytoskeleton was then prepared by extracting cells with aon-ionic detergent, which solubilized the membrane andevealed the cytoskeleton. Coverslips were treated for 5 mint 37 ◦C, with 1% Triton X-100 and 0.02% DTSP in Micro Tubuletabilizing Buffer, MTSB (MTSB: EGTA 1 mM, PEG 6000 4%nd Pipes 100 mM, at pH 6.9 in deionised water), and thenreated twice with 1% Triton X-100 in MTSB, without DTSP, for

min each, at room temperature. Cells were then incubatedn 100U/ml Dnase I and 100 �g/ml Rnase A in MTSB, 30 mint 37 ◦C. Coverslips were post-fixed in 2% paraformaldehydeith 0.1% glutaraldehyde in PBS for 5 min, rinsed 3 times inBS and treated by sodium borohydride for 5 min at roomemperature to block aldehyde sites. For SEM examinationf the cytoskeleton, cells were then treated as for morpho-ogical study.

Proteins of the cytoskeleton (actin, tubulin and vimentin)ere identified using immunogold labeling. After sodiumorohydride incubation, cells were treated with 5% BSA inBS to block non-specific antigenic sites and rinsed in PBSith 0.1% BSA. Cells were incubated for 2 h at 37 ◦C with

he following first antibodies diluted at 1/100 in 0.1% BSAn PBS: goat anti human actin, rabbit anti human tubulin oroat anti human vimentin. Protein A conjugated with 20 nmolloidal gold beads, diluted at 1/100 in PBS with 0.1% BSA,as applied overnight at 4 ◦C. The cultured coverslips were

hen treated as for SEM examination.

mage analysis

wo-texture analysis were developed in VisualBasicMicrosoft) and used sequentially on the SEM images of theytoskeleton.

he ‘‘blanket’’ fractal analysis [21]e calculated the fractal dimension by using dilatation androsion of an image, the mathematical details have beenresented elsewhere [22]. The structural element used was

cross rod-shaped element. Given this structuring elementf size �, a dilation and an erosion of the image provided twoew covering images: respectively, the upper and the lower

nes. The volume of the blanket, i.e., the volume enclosedetween the dilatation and erosion images, was measured.he number of dilations and erosions ranged from 1 to 10 andhe volume as measured each time. The fractal dimension

Lvnm

V. Demais et al.

blank was computed by plotting the logarithm of the volumegainst the logarithm of � and searching the slope coefficienty the least-squares method.

he ‘‘skyscrapers’’ fractal analysis [23]ixels that constituted an image were considered askyscrapers whose height was represented by the grey level.he roof of the skyscrapers was a square of side �. The sur-ace area of the image was obtained by measuring the sumf top surfaces and sum of the exposed lateral sides of thekyscrapers. Gray levels of adjacent pixels were then aver-ged in squares of �: 2, 4, 8, 16, and 32 pixels to produce newmages; the surface area of each image was calculated forach �. The fractal dimension of the surface was determinedy plotting the logarithm of the surface against the loga-ithm log �. A linear regression line was computed only onhe aligned points by the least-squares method. The fractalimension was obtained as Dsky = 2 + slope [22].

-lysine (PL), sérum de veau fœtal (FCS), fibronectine (FN),itronectine (VN) et collagène de type I (Col-I). Différences sig-ificatives à *p < 0,05 vs à 5 minutes, § vs 30 minutes, © vs 60inutes, vs 120 minutes.

Effects of matrix proteins on osteoblasts 57

Figure 2 Scanning electron microscopy of the effects of matrix proteins on the morphological aspect of cells cultured during 5,30, 60 min and 24 h. Coverslips coated with poly-L-lysine (PL), DMEM alone, fetal calf serum (FCS), fibronectin (FN), vitronectin (VN)and type I collagen (Col-I). Cell spreading increased with time, excepted on PL.Microscopie électronique à balayage des effets des protéines matricielles sur l’aspect morphologique des cellules cultivées pendant5, 30, 60 minutes et 24 heures. Lamelles couvre-objet revêtues de poly-L-lysine (PL), DMEM seul, sérum de veau fœtal (FCS),

-I). L’

pw5(g1

taapCao

fibronectine (FN), vitronectine (VN) et collagène de type I (Colde la PL.

serum-free medium. Cells were then solubilized in 1 mlof lysis buffer (50 mM Tris pH 7, 100 mM NaCl, 50 mMNaF, 3 mM Na3PO4, 2 �g/ml leupeptin, 5 �g/ml aprotinin,1 �g/�l pepstatin, 1 mM PMSF, 1% NP40) for 1 h at 4 ◦C. Thelysate was centrifuged for 30 seconds at 4 ◦C at 12,000 rpmand the lysate protein concentrations were determinedby a colorimetric assay (Bio-Rad Laboratories, Hercules,California) and standardized prior to further analysis.

For MAPK analysis, 15 �l of sample buffer (1 M Tris,0.025% bromophenol blue, 10% SDS, 10% (v/v) glycerol, 5%mercaptopropanediol) was added to 25 �l of lysate. Afterboiling 10 min at 100 ◦C, cell lysates were ready to westernblot analysis. For FAK analysis, cell lysates must previouslyundergo immunoprecipitation. For immunoprecipitation,

cell lysates were incubated with primary antibody againstfocal adhesion kinase overnight at 4 ◦C with rocking,followed by incubation with protein A-sepharose for anadditional 2 h at 4 ◦C. Protein A-sepharose complexes were

a7ba

étalement cellulaire a augmenté avec le temps, à l’exception

elleted and washed 2 times with lysis buffer, one timeith lysis buffer without detergent and one time with Tris0 mM. Bound proteins were eluted in 40 �l sample buffer1 M Tris, 0.025% bromophenol blue 10% SDS, 10% (v/v)lycerol, 5% mercaptopropanediol) by boiling 10 min at00 ◦C.

Proteins in the sample buffer were subjected to elec-rophoresis on a 7.5% SDS-polyacrylamide gels for FAKnalysis and on a 10% SDS-polyacrylamide gels for MAPKnalysis. Proteins were electrophoretically transferred ontoolyvinylidene fluoride membranes (Immobilon P, Milliporeorp., Bedford, MA). Blots were blocked with 6% BSA for FAKnd 5% milk for MAPK in 200 mM NaCl and 50 mM Tris pH 7.6vernight at 4 ◦C, and subsequently incubated with primary

ntibody (0.5 �g/ml) in 1% BSA/200 mM NaCl/50 mM Tris pH.6/0.05% Tween 20, for 5 h at room temperature. Mem-ranes were then washed briefly and incubated with goatnti-mouse IgG or goat anti-rabbit IgG peroxidase conjugates

58 V. Demais et al.

Figure 3 Effects of matrix proteins on cell surface area determined on cells cultured during 5, 30, 60 min and 24 h. DMEM alone(DMEM), and in presence of poly-L-lysine (PL), fetal calf serum (FCS), fibronectin (FN), vitronectin (VN) and type I collagen (Col-I).Cell surface significantly increased with time excepted on PL (*P < 0.05 vs. all time).Effets des protéines matricielles sur la surface des cellules en culture pendant 5, 30, 60 minutes et 24 heures. DMEM seul (DMEM),et en présence de poly-L-lysine (PL), sérum de veau fœtal (FCS), fibronectine (FN), vitronectine (VN) et collagène de type I (Col-I).La surface cellulaire a considérablement augmenté avec le temps sauf avec la PL (différences significatives à *p < 0,05 vs tous lest

fd(fi

S

S6vims

R

C

Te(RoTdsccdd

M

NfcCFwscsgw

Ansoifs

w(cwcoe

emps).

or 1 h. Following extensive washing, immunoreactivity wasetected on film using chemiluminescent immunodetectionECL kit, Amersham Corp. Arlington Heights, IL). Imagesrom ECL autoradiograms were captured using ImageJ (NIHmage software).

tatistical analysis

tatistical analysis was done using Systat software, release.0.1 (SPSS, Chicago, Ltd) with a nonparametric analysis ofariance (ANOVA) with Bonferroni’s probability least signif-cant difference post-hoc test. All data were expressed asean and standard deviation. Differences were considered

ignificant when P < 0.05.

esults

ell attachment assays

he number of adherent cells onto the different coated cov-rslips was evaluated by counting with the Malassez cellresult no shown) and the dosage of DNA content (Fig. 1).esults were similar with the two methods. For each typef coated coverslips, cell adherence was time dependent.he comparison at each time did not show any importantifferences until 60 min. After 60 min, cell adherence wasignificantly decreased on VN and PL as compared with other

onditions. At 240 min, adherence on VN was decreasedompared to FCS, FN and Col-I and adherence on PL wasiminished versus on FN or Col-I. There were no significantifferences for other coatings.

cofia

orphological aspects

umerous adherent cells were observed in SEM as soon asrom 5 min onto coated supports. However on FCS-coatedoverslips scarcely distributed cells were observed (Fig. 2).ells were small and globulous until 30 min on DMEM, FCS,N, VN and Col-I but larger on PL. After 1 h, very few cellsere observed on FCS coatings. Cells appeared rounded but

ome flat cells were presents on FN, VN and Col-I. On PL,ells were numerous, large and very flat. At 4 h (results nothown) and 24 h of adhesion, cells cultured on PL showed alobular appearance and on the other coatings, large cellsere in majority (Fig. 2).

Cell surfaces were measured by image analysis (Fig. 3).fter 30 min, cells cultivated on the PL coating were sig-ificantly larger than on all other supports. However, theirurface significantly decreased with time while the surfacef cells cultivated on all others coated supports significantlyncreased with time. After 24 h of culture, the cell sur-ace area was significantly increased on FN vs. all otherupports.

After treatment with TRITC-phalloidin, the actin net-ork was identified in confocal microscopy as red filaments

Fig. 4). After 5 min of adhesion, cells were round in allulture conditions but no organized cytoskeleton networkas observed. Actin filaments were observed first in cellsultivated on VN, FN and Col-I from 1 h of culture. Cellsn VN showed actin filaments mainly distributed in periph-ry while the actin filaments were observed throughout the

ytoplasm on FN and Col-I. At 24 h, flat cells were presentn FCS, FN, VN and Col-I with an important network of actinbers. On PL and DMEM, cells are smaller, rounder and noctin fibers were observed (Fig. 4).

Effects of matrix proteins on osteoblasts

Figure 4 Effects of matrix proteins on actin cytoskeletonanalyzed by confocal microscopy. Cells were grown during 5,60 min and 24 h. DMEM alone, and onto coverslips coated withpoly-L-lysine (PL), fetal calf serum (FCS), fibronectin (FN), vit-ronectin (VN) and type I collagen (Col-I). Clearly organized actincytoskeleton was detected with FCS at 24 h, FN, VN and Col-Ifrom 60 min.Effets des protéines matricielles sur les fibres d’actine ducytosquelette ; analyse par microscopie confocale. Les cel-lules ont été cultivées pendant 5, 60 minutes et 24 heures.DMEM seul, et sur des lamelles recouvertes de poly-L-lysine(PL), sérum de veau fœtal (FCS), fibronectine (FN), vitronec-tine (VN) et collagène de type I (Col-I). Les fibres d’actinedu cytosquelette sont clairement détectées avec le FCS à24 heures, FN, VN et Col-I à partir de 60 minutes.

gavatcassfitzi(tttaodTOatve

tcwacash(aaa

W

Fbeeg

cFdt

D

Iosca

59

After extraction of the plasma membrane and immuno-old labeling, elements of the cytoskeleton were observed:ctin microfilaments, microtubules of �- and �-tubulin andimentin intermediate filaments (Fig. 5). On FCS, cellsppeared very small and round until 240 min so no cytoskele-on could be analyzed on such samples. Two regions in theell cytoskeleton were observed: one perinuclear and onet the cell periphery. The perinuclear cytoskeleton did nothow any differences whatever time of culture and coatedupport and was mainly represented by vimentin as identi-ed by immunogold (Fig. 5a). After 1 h, the cytoskeleton athe periphery of the cell showed different aspects of organi-ation according to the coating used. Immunogold labelingdentified actin (Fig. 5b), vimentin (Fig. 5c), and tubulinFig. 5d) but it was not possible to characterize each struc-ure of the cytoskeleton so it was not possible to determinehe degree of organization of each element of the cytoskele-on (Figs. 5 and 6). On PL and FCS, cytoskeleton appeared as

compact mesh with no individualized filament, no filipodiar lamellipodia. On DMEM, the cytoskeleton was not reallyeveloped, but cells send long filipodia towards other cells.hese filipodia contained a single and very long filament.n VN, the filaments seemed to be much more polymerizednd oriented parallel to the cell profile. On FN and Col-I,he cytoskeleton seemed to be much more oriented; indi-iduals filaments were packaged together in large bundlesxtending throughout the cell (Figs. 5 and 6).

Texture analysis provided different results according tohe algorithm used. Dblank showed few differences betweenoatings and times. The heterogeneity of the cytoskeletonas significantly different between 5 min and 24 h on PLnd Col-I (P < 0.05). At 24 h VN heterogeneity was signifi-antly lower than on cells grown on PL or FCS (P < 0.05). Dsky

nalysis revealed much more differences. The Dsky dimen-ion, reflecting cytoskeleton heterogeneity was significantlyigher at 24 h than at 5 min for PL, DMEM, FN and Col-IP < 0.05). No differences were observed with FCS and VNt any time. Dsky measured on VN was lower than on DMEMnd FCS at 240 min and 24 h and also lower than on PL, FNnd Col-I at 24 h (P < 0.05).

estern blot analysis

AK phosphorylation (Fig. 7) showed no significant differenceetween culture conditions at all times. However, a weakerxpression was observed on PL and DMEM until 60 min. Forach cultured condition, FAK phosphorylation seemed torow up with time.

MAPK phosphorylation (Fig. 8) was lower at each times inells plated on PL until 60 min. In cells cultured with DMEM,CS and Col-I, MAPK phosphorylation showed no significantifference with time. MAPK phosphorylation decreased withime in cells plated on FN and VN.

iscussion

n this study, the influence of extracellular proteins on

steoblastic adhesion and shape was evaluated. Osteoblastsynthesize numerous extracellular proteins including type Iollagen, the principal organic component of bone matrixnd a wide variety of non-collagen proteins such as

60 V. Demais et al.

Figure 5 Immunogold labeling of cytoskeleton proteins for (A) perinuclear vimentin, (B) actin, (C) vimentin intermediate fila-ments, (D) tubulin microtubules, (E) negative control for vimentin and (F): negative control for actin. White arrows indicate someof 20 nm gold nanoparticles and arrowhead points out to the nucleus.Immunomarquage des protéines du cytosquelette (A) vimentine périnucléaire, (B) l’actine, (C) filaments intermédiaires vimentine,( entii he po

fiibaccc

oeTosa[aoin

Fuoatie

Rocrom

D) microtubules de tubuline, (E) contrôle négatif pour la vimndiquent des nanoparticules d’or de 20 nm et les têtes de flèc

bronectin, osteopontin, and bone sialoprotein. Adhesiventeractions between cells and their substrata, mediatedy cell surface integrin and extracellular matrix proteins,ppear to result in massive rearrangement of the cellytoskeleton. These early events promote several structuralhanges such as cell spreading and initiate signaling cas-ades, which modulate genes expression [24,25].

This study strongly supported the hypothesis that therganization of the cytoskeleton was mainly related to thextracellular matrix protein on which osteoblast adhered.he kinetics of osteoblast-cell adherence and the aspectsf the cell shape were dependent of the nature of the sub-trate without any influence of cell density which was similarfter long time culture whatever matrix protein used was26]. This is in agreement with reports showing different

dhesive and spreading kinetics of cell cultured on FN, Col-Ir serum-free medium [27—29]. Quantification of the SEMmages of the cytoskeleton was only possible by using tech-iques developed to evaluate the complexity of an image.

toao

ne et (F) contrôle négatif pour l’actine. Les flèches blanchesintent vers le noyau.

ractal methods for measuring the texture of an image areseful and can provide more information than those basedn Euclidean geometry. It is also well known that the fractallgorithms should be adapted to the texture to measure. Inhe present study, the skyscraper algorithm provided morenformation than the blanket because there was no prefer-ntial distribution of the cytoskeleton fibers [30,31].

Cells cultured on purified adhesion proteins that containGD-sequences had the highest surface area and only cellsn PL showed a decrease with time in cell surface area. Theell rounding on PL might be related to disassociation of celleceptor-substratum complexes, which causes a dissipationf isotonic tension at cell surface and retraction of corticalicrofilaments. This might lead to a growth arrest.The cellular attachment in serum-free medium indicates

hat osteoblast binding can occur in the absence of serumr exogenously added cell adhesion proteins. When cellsdhere to glass, they use presumably other mechanismsr non-integrin receptors to mediate cell attachment. Our

Effects of matrix proteins on osteoblasts 61

Figure 6 Effects of matrix proteins on ultrastructural cytoskeleton aspect. Cells were cultivated in DMEM alone for 5, 60 minand 24 h onto coverslips coated with poly-L-lysine (PL), DMEM, fetal calf serum (FCS), fibronectin (FN), vitronectin (VN) and type Icollagen (Col-I).Effets des protéines matricielles sur l’aspect ultrastructural du cytosquelette. Les cellules ont été cultivées dans du milieu DMEMseul pendant 5, 60 minutes et 24 heures sur des lamelles recouvertes de poly-L-lysine (PL), DMEM, sérum de veau fœtal (FCS),

-I).

cs

fibronectine (FN), vitronectine (VN) et collagène de type I (Col

results confirm previous observations, which did not reportmarked changes in attachment in presence or absence ofa specific or preferred ligand such as fibronectin [32]. For

other matrix proteins (FN, VN or Col-I), the osteoblast-cell had a relatively flat shape. These findings demonstratethe specific role of extra cellular matrix molecules tothe cell binding, the actin polymerization triggering, the

bpaF

ontact adhesion formation and the stabilization of cellhape [16,33].

Previous authors suggested that fibronectin promoted

oth osteoblast attachment and spreading [34] and thatlating on matrices such as type I collagen and fibronectinccelerated osteoblastic cell maturation [35—37]. Moreover,N functions as a survival factor for mature osteoblasts [38].

62 V. Demais et al.

Figure 7 Effects of matrix proteins on FAK phosphorylation.Cells were grown in DMEM alone during 5, 30, 60 and 120 minonto coverslips coated with poly-L-lysine (PL), DMEM, fetal calfserum (FCS), fibronectin (FN), vitronectin (VN) and type I col-lagen (Col-I) and FAK phosphorylation study was performed bywestern blot.Effets des protéines matricielles sur la phosphorylation de FAK.Les cellules ont été cultivées dans du DMEM seul pendant 5,30, 60 et 120 minutes sur des lamelles recouvertes de poly-L-lysine (PL), DMEM, sérum de veau fœtal (FCS), fibronectine(d

Oaftmeoofitts

iracigcsgfetb

Figure 8 Effects of matrix proteins on MAPK phosphorylation.Cells were grown in DMEM alone during 5, 30, 60 and 120 minonto coverslips coated with poly-L-lysine (PL), DMEM, fetal calfserum (FCS), fibronectin (FN), vitronectin (VN) and type I colla-gen (Col-I) and MAPK phosphorylation study was performed bywestern blot.Effets des protéines matricielles sur phosphorylation de MAPK.Les cellules ont été cultivées dans du DMEM seul pendant 5,30, 60 et 120 minutes sur des lamelles recouvertes de poly-L-lysine (PL), DMEM, sérum de veau fœtal (FCS), fibronectine(d

Mscadvbwdda

eaFiMolodtcmif

FN), vitronectine (VN) et collagène de type I (Col-I) et l’étudee la phosphorylation de FAK a été réalisée par western blot.

steoblastic cells cultured on a collagen substratum adherednd spread over it. Col-I, like FN anchorage is importantor osteogenesis but may function at different stages ofhe osteoblast differentiation program [39]. The attach-ent of osteoblasts to the collagenous matrix had important

ffects on the expression of the osteoblast phenotype andsteoblast marker genes such as alkaline phosphatase andsteocalcin [40]. In cell cultivated on VN, polymerized actinbers are observed but they are mainly at the periphery ofhe cell. Some authors supposed that, in epithelial cells,his circumferential band of actin filaments provided thetructural support for cell-cell junctions [41].

Results reported in this study seem to confirm thatntegrins convey information from the extracellular envi-onment by linkage between the extracellular matrix andctin cytoskeleton. They transmit signals by organizing theytoskeleton and regulating cell shape [42,43]. Such changesn cell shape and cytoskeleton organization might regulateene expression and cell biosynthesis [44,45] and therebyontrol the differentiation and growth of cells [46]. Currentpeculation postulates that the formation of large aggre-ates of signal transduction molecules on a cytoskeleton

ramework provides high local concentrations of receptors,nzymes, substrates, and structural molecules. This pro-ein clustering facilitates activation of signaling pathwaysy stimulating tyrosine phosphorylation cascades linked to

m

cg

FN), vitronectine (VN) et collagène de type I (Col-I) et l’étudee la phosphorylation de MAPK a été réalisée par western blot.

APK and other pathways [47] and organization of adhesiveites. Adhesion on various matrix proteins induces structuralhanges in cells, and may alter the conformation and inter-ctions of integrin receptor thereby leading to activation ofifferent signal transduction pathways, than would be acti-ated by cell adhesion or receptor occupancy. Finally it maye necessary that the integrin receptors could be engagedith an extracellular ligand in order to mediate signal trans-uction. Such a result is clearly consistent with the uniqueependency of mechanotransduction on the maintenance of

intact cytoskeleton [32].Binding of integrin and the subsequent cell adher-

nce stimulate the formation of membrane-associated focaldhesion complexes [32]. Some authors have reported thatAK might play a role in the regulation of MAPK in induc-ng tyrosine phosphorylation of p44MAPK in NIH 3T3. So,APK activity may be related to cell shape and cytoskeletonrganization [48]. In this study FAK and MAPK are phosphory-ated after cell adhesion. If no significant differences werebserved for FAK phosphorylation, the MAPK phosphorylationiffers with time and type of matrices. These data suggestedhat FAK and MAPK phosphorylations are part of regulation inytoskeleton organization but other cell signaling pathwaysight be involved. That is in agreement with findings show-

ng that Rho pathway was implicated in integrin-mediatedocal adhesion formation and may also enhance actin poly-

erization [49].In this study the influence of matrix proteins on

ell attachment and cytoskeleton organization was investi-ated. These results show that binding to different adhesion

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

Effects of matrix proteins on osteoblasts

proteins induces changes in cytoskeleton organization. Thissuggests that cells are able to distinguish specific integrin-ligand binding, to induce different intracellular pathwaysin response and in consequence to modulate cytoskeletonorganization.

Disclosure of interest

The authors declare that they have no conflicts of interestconcerning this article.

Acknowledgments

This work was made possible by grants from Contrat RegionPays de la Loire: Bioregos2 program.

References

[1] Ruoslahti E. Integrins. J Clin Invest 1991;87:1—5.[2] Staines KA, MacRae VE, Farquharson C. The importance of the

Sibling family of proteins on skeletal mineralisation and boneremodelling. J Endocrinol 2012;214:241—55.

[3] Kirchhofer D, Languino LR, Ruoslahti E, Pierschbacher MD. a2b1integrins from different cell types show different binding speci-ficities. J Biol Chem 1990;265:615—8.

[4] Nissinen L, Pirilä L, Heino J. Bone morphogenetic protein-2 isa regulator of cell adhesion. Exp Cell Res 1997;230:377—85.

[5] Omar O, Lenneras M, Svensson S, Suska F, Emanuelsson L, HallJ, et al. Integrin and chemokine receptor gene expression inimplant-adherent cells during early osseointegration. J MaterSci Mater Med 2010;21:969—80.

[6] Vihinen P, Riikonen T, Laine A, Heino J. Integrin alpha 2 beta1 in tumorigenic human osteosarcoma cell lines regulates celladhesion, migration, and invasion by interaction with type Icollagen. Cell Growth Differ 1996;7:439—47.

[7] Riikonen T, Westermarck J, Koivisto L, Broberg A, Kahari VM,Heino J. Integrin alpha 2 beta 1 is a positive regulator of colla-genase (MMP-1) and collagen alpha 1(I) gene expression. J BiolChem 1995;270:13548—52.

[8] Craig SW, Johnson RP. Assembly of focal adhesions: progress,paradigms, and portents. Curr Opin Cell Biol 1996;8:74—85.

[9] Kuo JC. Mechanotransduction at focal adhesions: integrat-ing cytoskeletal mechanics in migrating cells. J Cell Mol Med2013;17:704—12.

[10] Calderwood DA, Shattil SJ, Ginsberg MH. Integrins and actinfilaments: reciprocal regulation of cell adhesion and signaling.J Biol Chem 2000;275:22607—10.

[11] Reinhard M, Jarchau T, Walter U. Actin-based motility: stopand go with Ena/VASP proteins. Trends Biochem Sci 2001;26:243—9.

[12] David FS, Zage PE, Marcantonio EE. Integrins interact with focaladhesions through multiple distinct pathways. J Cell Physiol1999;181:74—82.

[13] Schlie-Wolter S, Ngezahayo A, Chichkov BN. The selective roleof ECM components on cell adhesion, morphology, proliferationand communication in vitro. Exp Cell Res 2013;319:1553—61.

[14] Zigmond SH. Signal transduction and actin filament organiza-tion. Curr Opin Cell Biol 1996;8:66—73.

[15] Lomri A, Marie PJ. The cytoskeleton in the biology of bonecells. Cytoskeleton 1996;3:229—63.

[16] Yu H, Tay CY, Leong WS, Tan SC, Liao K, Tan LP. Mechanicalbehavior of human mesenchymal stem cells during adipogenicand osteogenic differentiation. Biochem Biophys Res Commun2010;393:150—5.

[

63

17] Leisner TM, Wencel-Drake JD, Wang W, Lam SC. Bidirectionaltransmembrane modulation of integrin �IIb�3 conformations.J Biol Chem 1999;274:12945—9.

18] Bell PB, Lindroth M, Fredriksson BA. Preparation of cytoskele-tons of cells in culture for high-resolution scanning andscanning transmission electron microscopy. Scanning Microsc1988;2:1647—61.

19] Bell PB, Lindroth M, Fredriksson BA. Use of sputter coating toprepare whole mounts of cytoskeletons for transmission andhigh-resolution scanning and scanning transmission electronmicroscopy. J Electron Microsc Tech 1987;7:149—59.

20] Bell PB, Lindroth M, Fredriksson BA, Liu XD. Problems associ-ated with the preparation of whole mounts of cytoskeletons forhigh-resolution electron microscopy. Scanning Microsc Suppl1989;3:117—35.

21] Peleg S, Naor J, Hartley R, Avnir D. Multiple resolution textureanalysis and classification. IEEE Trans Pattern Anal Mach Intel1984;6:518—23.

22] Chappard D, Degasne I, Huré G, Legrand E, Audran M, Baslé MF.Image analysis measurements of roughness by texture and frac-tal analysis correlate with contact profilometry. Biomaterials2003;24:1399—407.

23] Caldwell CB, Stapleton SJ, Holdsworth DW, Jong RA, WeiserWJ, Cooke G, et al. Characterization of mammographicparenchymal pattern by fractal dimension. Phys Med Biol1990;35:235—47.

24] Farrell E, O’Brien FJ, Doyle P, Fischer J, Yannas I, Harley BA,et al. A collagen-glycosaminoglycan scaffold supports adult ratmesenchymal stem cell differentiation along osteogenic andchondrogenic routes. Tissue Eng 2006;12:459—68.

25] Shi S, Kirk M, Kahn AJ. The role of type I collagen in theregulation of the osteoblast phenotype. J Bone Miner Res1996;11:1139—45.

26] Lomri A, Marie PJ, Escurat M, Portier MM. Cytoskeletal proteinsynthesis and organization in cultured mouse osteoblastic cells.FEBS Lett 1987;222:311—6.

27] Kornu R, Maloney WJ, Kelly MA, Smith RL. Osteoblast adhe-sion to orthopaedic implant alloys: effects of cell adhesionmolecules and diamond-like carbon coating. J Orthop Res1996;14:871—7.

28] Sinha RK, Morris F, Shah SA, Tuan RS. Surface compositionof orthopaedic implant metals regulates cell attachment,spreading, and cytoskeletal organization of primary humanosteoblasts in vitro. Clin Orthop Relat Res 1994;305:258—72.

29] Sinha RK, Tuan RS. Regulation of human osteoblast integ-rin expression by orthopedic implant materials. Bone1996;18:451—7.

30] Guggenbuhl P, Bodic F, Hamel L, Baslé MF, Chappard D. Textureanalysis of X-ray radiographs of iliac bone is correlated withbone micro-CT. Osteoporos Int 2006;17:447—54.

31] Mallard F, Bouvard B, Mercier P, Bizot P, Cronier P, ChappardD. Trabecular microarchitecture in established osteoporosis:relationship between vertebrae, distal radius and calcaneusby X-ray imaging texture analysis. Orthop Traumatol Surg Res2013;99:52—9.

32] Carvalho RS, Schaffer JL, Gerstenfeld LC. Osteoblasts induceosteopontin expression in response to attachment on fibronect:demonstration of a common role for integrin receptors in thesignal transduction processes of cell attachment and mechan-ical stimulation. J Cell Biochem 1998;70:376—90.

33] Rychly J, Pommerenke H, Dürr F, Schreiber E, Nebe B. Analysisof spatial distributions of cellular molecules during mechanicalstressing of cell surface receptors using confocal microscopy.Cell Biol Int 1998;22:7—12.

34] Gehron Robey PG, Boskey AL. The biochemistry of bone. In:Marcus R, Feldman D, Bilizekian JP, Kelsey J, editors. Osteo-porosis. New York: Academic Press; 1996. p. 95—183.

6

[

[

[

[

[

[

[

[

[

[

[

[

[

[

4

35] Hristova K, Pecheva E, Pramatarova L, Altankov G. Improvedinteraction of osteoblast-like cells with apatite-nanodiamondcoatings depends on fibronectin. J Mater Sci Mater Med2011;22:1891—900.

36] Lynch MP, Stein JL, Stein GS, Lian JB. The influence of type I col-lagen on the development and maintenance of the osteoblastphenotype in primary and passaged rat calvaria osteoblasts:modification of expression of genes supporting cell growth,adhesion, and extracellular matrix mineralisation. Exp Cell Res1995;216:35—45.

37] Shah AK, Sinha RK, Hickok NJ, Tuan RS. High-resolution mor-phometric analysis of human osteoblastic cell adhesion onclinically relevant orthopedic alloys. Bone 1999;24:499—506.

38] Globus RK, Doty SB, Lull JC, Holmuhamedov E, Humphries MJ,Damsky CH. Fibronectin is a survival factor for differentiatedosteoblasts. J Cell Sci 1998;111:1385—93.

39] Jikko A, Harris SE, Chen D, Mendrick DL, Damsky CH. Colla-gen integrin receptors regulate early osteoblast differentiationinduced by BMP-2. J Bone Miner Res 1999;14:1075—83.

40] Andrianarivo AC, Robinson JA, Mann KG, Tracy RP. Growth ontype I collagen promotes expression of the osteoblastic phe-

notype in human osteosarcoma MG-63 cells. J Cell Physiol1992;153:256—65.

41] Small JV, Rottner K, Hahne P, Anderson KI. Visualising the actincytoskeleton. Microsc Res Tech 1999;47:3—17.

[

V. Demais et al.

42] Ingber D. Integrins as mechanochemical transducers. Curr OpinCell Biol 1991;1:841—8.

43] Ruwhof C, Van Der Laarse A. Mechanical stress-induced cardiachypertrophy: mechanisms and signal transduction pathways.Cardiovasc Res 2000;47:23—37.

44] Maniotis AJ, Chen CS, Ingber DE. Demonstration of mechani-cal connections between integrins, cytoskeletal filaments, andnucleoplasm that stabilize nuclear structure. Proc Natl AcadSci U S A 1997;94:849—54.

45] Pavalko FM, Chen NX, Turner CH, Burr DB, Atkinson S, Hsieh YF,et al. Osteoblast requires cytoskeleton-integrin interactions.Am J Physiol 1998;275:C1591—601.

46] Kuppuswamy D, Kerr C, Narishige T, Kasi VS, Menick DR,Cooper G. Association of tyrosine-phosphorylated c-Src withthe cytoskeleton of hypertrophying myocardium. J Biol Chem1997;272:4500—8.

47] Xiao G, Wang D, Benson MD, Karsenty G, Franceschi RT.Role of the a2-integrin in osteoblast-specific gene expressionand activation of the Osf2 transcription factor. J Biol Chem1998;273:32988—94.

48] Chiquet M, Gelman L, Lutz R, Maier S. From mechanotrans-

duction to extracellular matrix gene expression in fibroblasts.Biochim Biophys Acta 2009;1793:911—20.

49] Longhurst CM, Jennings LK. Integrin-mediated signal transduc-tion. Cell Mol Life Sci 1998;54:514—26.