Biomaterials 24 (2003) 1759–1770

Engineering cell adhesive surfaces that direct integrin a5b1 bindingusing a recombinant fragment of fibronectin

Sarah M. Cutlera, Andr!es J. Garc!ıab,*aWallace H. Coulter School of Biomedical Engineering, Petit Institute for Bioengineering and Bioscience, Georgia Institute of Technology,

Atlanta, GA, 30332-0363, USAbWoodruff School of Mechanical Engineering, Petit Institute for Bioengineering and Bioscience, 315 Ferst Drive, IBB Room 2314,

Atlanta, GA, 30332-0363, USA

Received 7 August 2002; accepted 11 November 2002

Abstract

Integrin receptors mediate cell adhesion to extracellular matrices and trigger signals that direct cell function. While many integrins

bind to the arginine–glycine–aspartic acid (RGD) motif present in numerous extracellular proteins, integrin a5b1 requires both the

PHSRN synergy site in the 9th and the RGD site in the 10th type III repeat of fibronectin (FN). Binding of a5b1 to FN is critical to

many cellular processes, including osteoblast and myoblast differentiation. This work focused on engineering integrin-specific

bioadhesive surfaces by immobilizing a recombinant FN fragment (FNIII7–10) encompassing the a5b1 binding domains of FN.

Model hybrid surfaces were engineered by immobilizing FNIII7–10 onto passively adsorbed, non-adhesive albumin. Homo- and

hetero-bifunctional crosslinkers of varying spacer-arm length targeting either the cysteine or lysine groups on FNIII7–10 were

investigated in ELISA and cell adhesion assays to optimize immobilization densities and activity. FN-mimetic surfaces presenting

controlled densities of FNIII7–10 were generated by varying the concentration of FNIII7–10 in the coupling solution at a constant

crosslinker concentration. Cells adhered to these functionalized surfaces via integrin a5b1 and blocking with integrin-specific

antibodies completely eliminated adhesion. In addition, adherent cells spread and assembled focal adhesions containing a5b1;vinculin, and talin. This biomolecular engineering strategy represents a robust approach to increase biofunctional activity and

integrin specificity of biomimetic materials.

r 2003 Elsevier Science Ltd. All rights reserved.

Keywords: Fibronectin; Integrins; Cell adhesion; RGD; Biomimetic

1. Introduction

Cell adhesion to extracellular matrix (ECM) compo-nents is necessary for the development, organization,and maintenance of tissues [1–3]. The critical impor-tance of ECM ligands, adhesion receptors, and signalinginteractions is underscored by the absolute lethality atearly embryonic development in models with targetedgene deletions [4–7]. Moreover, cell adhesion to ECMligands is crucial to numerous biotechnological andbiomedical applications, including biomaterials, artifi-cial organs, tissue engineering, and synthetic supportsfor in vitro cell culture [8–10]. Adhesive cells utilizeextracellular matrix proteins to attach to and migrate on

substrates, exchange signals that prevent apoptosis andmaintain cell cycle progression, and determine tissue-specific phenotypes [11–15].Cell adhesion to ECM proteins is primarily mediated

by integrins, transmembrane receptors composed of aand b subunits [1]. Upon ligand binding, integrinscluster together and organize into focal adhesioncomplexes that contain structural and signaling proteins[16,17]. These complexes act in mechanical and chemicalcapacities to anchor the cell and transmit signalsregulating migration, survival, proliferation, and differ-entiation [15,18–21]. Integrins bind to specific aminoacid sequences on ECM proteins, such as the arginine–glycine–aspartic acid (RGD) motif present in manyECM ligands including fibronectin (FN), vitronectin,and thrombospondin [22,23]. While RGD supportsbinding of many integrins, several integrins, includinga5b1 [24] and a11b b3 [25], require specific domains in

*Corresponding author. Tel.: +1-404-894-9384; fax: +1-404-385-

1397.

E-mail address: andres.garcia@me.gatech.edu (A.J. Garc!ıa).

0142-9612/03/$ - see front matter r 2003 Elsevier Science Ltd. All rights reserved.

doi:10.1016/S0142-9612(02)00570-7

addition to the RGD motif for effective receptor-ligandinteractions. In particular, a5b1 integrin, which controlsproliferation and differentiation of osteoblasts andmyoblasts [15,26], cell cycle progression [27], and FNmatrix assembly [28], will only bind to the RGD motifon the 10th type III repeat of FN in the presence of theproline-histidine-serine-arginine-asparagine (PHSRN)‘‘synergy’’ domain on the 9th type III repeat [29]. Eachdomain independently contributes little to binding, butin combination, they synergistically bind to the integrinto produce significant increases in adhesion strength[30,31]. Furthermore, the structural orientation of thesebinding domains is crucial to the synergistic effects asincreases in the relative distance between the PHSRNand RGD sites completely abrogate a5b1 binding, cellspreading, and integrin-mediated signaling [32].The identification of recognition sequences, such as

RGD for FN, that mediate integrin-mediated adhesionhas stimulated the development of bio-inspired adhesivesurfaces. This biomolecular strategy involves incorpora-tion of short adhesive peptides onto appropriatesynthetic or natural materials, often non-fouling andnon-adhesive supports that reduce background effectsarising from non-specific protein adsorption, in order toproduce biofunctional surfaces [33,34]. Several groupshave demonstrated that incorporation of RGD peptidesinto synthetic polymers, alginate hydrogels, silk films,and silicon-based substrates promotes integrin-mediatedcell adhesion and migration [35–47]. Although thisapproach remains one of the most promising strategiesfor the engineering of bioactive materials, the biologicalactivity of these short peptides is significantly lower thanthat of the complete protein [48,49]. The loss in activityprimarily results from conformation-dependent effectsand the absence of crucial modulatory domains, such asthe PHSRN site.The objective of this work was to engineer surfaces

that support a5b1-mediated adhesion using recombinantpeptide technology. A model non-adhesive support,bovine serum albumin (BSA), was functionalized with aFN fragment encompassing the RGD and PHSRNdomains. Using this approach, we created FN-mimeticsurfaces presenting controlled ligand densities thatdirect integrin a5b1-mediated adhesive interactions.

2. Materials and methods

2.1. Materials

LB agar, LB broth, ampicillin, and IPTG used forrecombinant protein production were obtained fromGibco BRL (Grand Island, NY). Chemical reagents,CellLytic B-Clear II and DNAase I used for FNIII7–10purification were obtained from Sigma Chemical (St.Louis, MO). HiTrap Q chromatography columns wereobtained from Amersham Pharamcia (Piscataway, NJ),and centrifugal concentration devices were purchasedfrom Gelman Laboratory (Ann Arbor, MI). Cross-linking agents, 3,30-dithiobis[sulfo-succinimidyl propio-nate] (DTSSP), dimethyl 3,30-dithiobispropionimidate(DTBP), N-[e-maleimidocaproyloxy]sulfosuccinimideester (sulfo-EMCS), sulfosuccinimidyl-4-(P-maleimido-phenyl) butyrate (sulfo-SMPB), N-[k-maleimidoundeca-noyloxy]sulfosuccinimide ester (sulfo-KMUS), andsulfosuccinimidyl 6-[30(2-pyridyldithio)-propionamido]hexanoate (sulfo-LC-SPDP) (Table 1) were purchasedfrom Pierce Chemical (Rockford, IL).MC3T3-E1 murine immature osteoblast-like cells and

NIH3T3 fibroblasts were obtained from Riken CellBank (Tokyo, Japan) and ATCC (Manassas, VA),respectively. Cell culture reagents, Dulbecco’s PBS(DPBS), PBS without Ca+2 or Mg+2, and humanplasma fibronectin (pFN) were purchased from GibcoBRL. Fetal bovine and newborn calf sera were obtainedfrom Hyclone (Logan, UT). Corning tissue culture-treated 100mm dishes and 96 well plates and Dynexblack 96 U-shape well plates were used.

2.2. Antibodies

HFN7.1 (Developmental Hybridoma Studies, IowaCity, IA), directed against the central cell bindingdomain of FN, was used for ELISA and cell adhesionassays. Hamster anti-rat b1 chain, hamster anti-mouseb3 chain, and rat anti-mouse a5 chain antibodies wereobtained from PharMingen International (San Diego,CA) and used to block integrin binding. Rabbit anti-mouse integrin a5 subunit (AB1921, Chemicon), mouseanti-vinculin (V284, Upstate Biotechnology, Lake

Table 1

Characteristics of homo- and hetero-bifunctional crosslinkers used to immobilize FNIII7–10 onto BSA supports. Coupling index represents the ratio

of immobilized FNIII7–10 for the highest coating concentration (20mg/ml) to background levels (0mg/ml)

Cross-linker Reactive groups Reactive towards Spacer arm ( (A) pH/concentration Coupling index

DTBP Imidoester NH2/NH2 11.9 8.5/20mm (PBS) 83

DTSSP NHS ester NH2/NH2 12.0 7.4/1mm (PBS) 44

Sulfo-KMUS NHS ester/maleimide NH2/SH 19.5 7.4/1mm (PBS, no Ca+2/Mg+2) 8.2

Sulfo-LC-SPDP NHS ester/pyridyldithio NH2/SH 15.6 7.4/1mm (PBS, no Ca+2/Mg+2) 3.3

Sulfo-SMPB NHS ester/maleimide NH2/SH 14.5 7.4/1mm (PBS, no Ca+2/Mg+2) 2.2

Sulfo-EMCS NHS ester/maleimide NH2/SH 9.6 7.4/1mm (PBS, no Ca+2/Mg+2) 1.1

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–17701760

Placid, NY), and mouse anti-talin (8D4, Sigma) wereused as primary antibodies for immunofluorescencestaining. Alkaline phosphatase-conjugated donkey anti-mouse IgG (Jackson Immunoresearch, West Grove, PA)was used as a secondary antibody for ELISA. Goat anti-mouse and anti-rabbit Alexa Fluor 488-conjugated IgG(Molecular Probes, Eugene, OG), rhodamine red-con-jugated donkey anti-rabbit IgG (Jackson Immunore-search), Hoescht 33258 (Molecular Probes), andrhodamine-conjugated phalloidin (Molecular Probes)were used as secondary antibodies and stains forimmunofluorescence imaging.

2.3. Recombinant FNIII7–10 production and purification

FNIII7–10, a recombinant FN fragment spanning the7–10th type III repeats of FN (Fig. 1), was producedand purified using standard techniques. E. coli stocktransformed with cDNA coding for human FNIII7–10and ampicillin resistance [50] was streaked onto an LBagar plate containing 100 mg/ml ampicillin and incu-bated overnight at 371C. One colony was isolated anddynamically cultured in 10ml of LB broth with 50 mg/mlampicillin for 3–5 h at 371C. 5ml of concentratedbacterial culture was added to 500ml LB+ampicil-lin+0.4mm IPTG and cultured overnight at 281C toinduce FNIII7–10 production. The culture was spundown at 10,000g for 10min and the pellet was frozen at�801C. Upon thawing, CellLytic B-Clear II was addedto the pellet at 5ml/g with 5 mg/ml DNAase I andagitated for 30min. The extract was centrifuged at25,000g for 15min. SDS-PAGE gel electrophoresis wasused to verify that FNIII7–10 was located in thesupernatant. Proteins were then precipitated throughcentrifugation of 5ml of supernatant in 20ml ofsaturated 40% ammonium sulfate for 1 h at 25,000g.The pellet was resuspended in 10ml of start buffer (0.2mTris, pH 8.0) and purified by anion exchange chromato-graphy using a BioRad Econo Gradient Pump, UVmonitor, and fraction collector (Hercules, CA) with 5mlHiTrap Q columns. After binding and washing with fivecolumn volumes of start buffer, elution was done via saltgradient from 0.2 to 0.4m NaCl elution buffer (0.5mTris, pH 7.7). The resulting product was verified to be>95% pure FNIII7–10 by SDS-PAGE and Westernblotting. Relevant fractions were concentrated usingMicrosep 10K Omega centrifugal devices. ConcentratedFNIII7–10 was dialyzed overnight in CAPS buffer(10mm CAPS, 150mm NaCl, pH 11.0) and flash frozenfor storage at �801C.

2.4. Cell model

Murine MC3T3-E1 immature osteoblast-like cellswere chosen as the cell model because of their expressionof multiple integrins, including a5b1; their preference for

adhesion to FN, and the necessity of adhesion to FN fordifferentiation [51]. Cells were cultured in a-MEMsupplemented with 10% fetal bovine serum and 1%penicillin/streptomycin and used prior to passage 20.Cultures were maintained at 70–80% confluence bysubculturing every 2–3 days using standard techniques.Murine NIH3T3 fibroblasts were also used as this well-defined cell line adheres and spreads on FN usingintegrin a5b1 [21]. Cells were maintained in DMEMsupplemented with 10% newborn calf serum and 1%penicillin/streptomycin.

2.5. Immobilization of proteins onto model non-adhesive

supports

Homo-bifunctional and hetero-bifunctional crosslin-kers, targeting free amines (NH2/NH2) or free aminesand sulfhydryls (NH2/SH), respectively, were usedfollowing the manufacturer’s instructions. Bovine serumalbumin (BSA) was passively adsorbed onto syntheticsubstrates at saturating levels to provide a modelsupport with non-fouling/non-adhesive properties andavailable free NH2 moieties. Polystyrene wells or cleanglass coverslips were incubated with 200 ml of 1% BSAfor 1 h, rinsed twice with DI water, and then incubated100 ml of crosslinker solution at the indicated concentra-tion (Table 1) for 1 h. Serial dilutions (1:2) of protein,starting at a concentration of 40 mg/ml, were thenincubated on the activated supports for 1 h. Unreactedcrosslinker was quenched for 30min with 1mm glycinein DPBS. Human plasma FN (pFN, dimer 440–480 kDa) and passively adsorbed proteins were used aspositive controls whereas proteins incubated on BSA-coated substrates without crosslinking agents were usedas negative controls.

2.6. Enzyme linked immunosorbent assay (ELISA)

Surfaces were prepared in 96-well black U-well Dynexplates as described above. Following surface prepara-tion, blocking buffer (0.25% BSA, 0.00125% NaN3,0.1m EDTA, 2.5% Tween-20) was applied for 1 h,followed by a 1 h incubation in HFN7.1 antibody(0.6 mg/ml). After washing in blocking buffer for10min, substrates were incubated in alkaline phospha-tase-conjugated donkey anti-mouse IgG (0.6 mg/ml) for

Fig. 1. Model of FNIII7–10 crystal structure backbone (PBD 1FNF)

showing PHSRN and RGD binding domains. Also shown are cysteine

(cys) and lysine (lys) residues targeted for immobilization.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–1770 1761

1 h, washed in blocking buffer, and incubated in4-methylumbelliferyl phosphate (25mg/ml in 10mm

diethanolamine, pH 9.5) for 30min at 371C. Reactionproducts were quantified using an HTS 7000 Plusfluorescence plate reader (Perkin Elmer, Foster City,CA) at 360 nm excitation/465 nm emission.

2.7. Centrifugation cell adhesion assay

Surfaces were prepared as above in tissue culture-grade polystyrene 96-well plates and blocked with 1%BSA for 1 h. Cells were labeled with calcein-AM, a cellmembrane permeable fluorescent marker (485 nm/535 nm), at 4mm in DPBS+2mm dextrose for 45min,detached from the culture support using 0.05% trypsin,and centrifuged for 5min at 1200g in a-MEM with 10%fetal bovine serum. Cells were washed and resuspendedin DPBS/dextrose and seeded onto substrates at 12,000cells/well for 1 h at 371C. After initial fluorescencereadings were taken, the plates were sealed, inverted,and centrifuged at 20 g on a Beckman Allegra 6centrifuge (GH 3.8 rotor) for 5min. Post-spin fluores-cence readings were used to calculate the density ofadherent cells. In addition, cells were photographed witha Nikon TE-300 fluorescence microscope at 4.5�magnification. For adhesion blocking studies, cells wereincubated for 15min in 0.02mg/ml (0.1mg/ml for anti-a5) integrin-specific antibodies under gentle agitationprior to seeding on surfaces. For ligand blocking studies,surfaces were pre-incubated for 15min with HFN7.1antibody (1.2 mg/ml). After antibody pre-incubation,cells were seeded for 1 h and adhesion was evaluatedas described above.

2.8. Immunofluorescence staining for focal adhesions

Surfaces were prepared on glass coverslides asdescribed above. Cells were plated serum-free for 2 hat 371C. For integrin staining, cells were incubated in1mm sulfo-BSOCOES (Pierce) at 41C for 15min tocrosslink bound integrins to the underlying extracellularmatrix [15]. Cells were then extracted in 0.1% SDSsupplemented with protease inhibitors (350 mg/mlPMSF) to remove uncrosslinked cellular components.Samples were blocked in 10% fetal bovine serum for 1 h,and incubated with primary antibodies against integrinsubunits followed by an 1 h incubation in fluorochrome-labeled secondary antibodies. For visualization of focaladhesion components, cells were extracted in 0.5%Triton X-100 in ice-cold cytoskeleton buffer (50mm

NaCl, 150mm sucrose, 3mm MgCl2, 20 mg/ml aprotinin,1 mg/ml leupeptin, 1mm PMSF, 50mm Tris, pH 6.8) for10min and fixed in cold formaldehyde (3.7% in PBS) for5min [21]. After blocking in 10% fetal bovine serum inDPBS, substrates were incubated in primary antibodies(1:250) in DPBS with 10% serum+0.2% saponin for

1 h. Following rinsing in DPBS+serum, samples wereincubated in fluorochrome-labeled secondary antibodies(1:200). After rinsing, samples were mounted onmicroslides with Gel/Mount mounting media (Biomeda,Foster City, CA).

2.9. Statistics

Graphs represent characteristic results from a parti-cular run (at least three independent runs wereconducted) and error bars reflect standard errorcalculated from one standard deviation from the meanof three replicates. Differences in cell adhesion for cellcentrifugation and blocking assays were analyzed viaANOVA and Tukey’s test for pairwise comparisonsusing SYSTAT (SPSS Science, Chicago, IL).

3. Results

3.1. FNIII7–10 activity compared to pFN

FNIII7–10 (Fig. 1), which spans the 7–10th type IIIrepeat of FN and contains the PHSRN and RGDdomains, was used engineer FN-mimetic surfaces. Thefragment was produced using standard recombinantprotein technology and recovered in greater than 95%purity. Fig. 2 shows characteristic results of salt gradientchromatography purification, where FNIII7–10 elutedbetween 0.27 and 0.29m NaCl, as indicated by the39 kDa band. Additional confirmation of the fragmentwas established by Western blotting (data not shown).Fragment activity was evaluated upon passive ad-

sorption to polystyrene via antibody binding and celladhesion. The FN-specific HFN7.1 monoclonal anti-body was used as a probe of active protein. HFN7.1reacts with the cell-binding domain in the 9–10th typeIII repeats of FN and blocks cell adhesion to FN [52].

39 kD

lysa

te

MW

mar

ker

was

h 1

was

h 2

NaCl elution gradient0.20 – 0.29 M

Fig. 2. FNIII7–10 purification from bacterial lysate. Coomassie blue-

stained SDS-PAGE gel showing lysate, washes, and fractions collected

from salt gradient chromatography.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–17701762

We have recently demonstrated that the bindingefficiency of this antibody for FN correlates with thebinding affinity of integrin a5b1 for FN and thisantibody can therefore be used as a probe of FNactivity [53]. For both FNIII7–10 and pFN, HFN7.1binding increased with protein coating concentrationfollowing the expected hyperbolic trend and there wereno significant differences in adsorbed density betweenFNIII7–10 and pFN (Fig. 3a). Moreover, FNIII7–10activity was demonstrated by equivalent levels ofMC3T3-E1 cell adhesion to FNIII7–10 and pFN(Fig. 3b). Cell adhesion increased with FNIII7–10 and

pFN coating concentration (ANOVA, po0:005) andblocking antibodies directed against the cell bindingdomain of FN or the b1 integrin subunit reducedadhesion to background levels (BSA-coated surfaces).Furthermore, FNIII7–10 exhibited equivalent levels ofcell spreading compared to pFN (Fig. 3c).

3.2. Immobilization of FNIII7–10 to BSA: crosslinker

efficiency

BSA was chosen as a model support for immobilizingFNIII7–10 because it conveniently adsorbs from solution

coating concentration (ug/ml)BSA 0.7 ug/ml 2.5 ug/ml 10 ug/ml anti-FN anti-b1

0

500

1000

1500

2000

2500

pFNFNIII7-10

adhe

rent

cel

ls (

rel.u

nits

)

0 5 10 15 20

boun

d an

tibo

dy (

rel.u

nits

)

0

5000

10000

15000

20000

25000

pFNFNIII7-10

5 ug

/ml

0 ug

/ml

pFN FNIII7-10

(a) (b)

(c)

Fig. 3. FNIII7–10 exhibits equivalent activity to pFN: (a) HFN7.1 antibody binding for passively adsorbed proteins as a function of coating

concentration showing hyperbolic increases in antibody binding; (b) MC3T3-E1 cell adhesion to FNIII7–10 and pFN showing comparable dose-

dependent levels. Adhesion was reduced to background levels (BSA) by blocking antibodies against FN (anti-FN) and b1 integrin subunit (anti-b1);(c) Micrographs of fluorescence-labeled cells showing robust adhesion and spreading to FNIII7–10 and pFN but not to BSA controls.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–1770 1763

onto synthetic surfaces to render them non-fouling andnon-adhesive for short term studies. In addition, BSAhas multiple lysine residues available for anchoringpeptides/proteins using conventional peptide chemistry.We evaluated a series of water-soluble homo- andhetero-bifunctional crosslinkers (Table 1) with onereactive group targeting lysine residues in BSA and theother functionality targeting either the single cysteine orseven lysine residues in FNIII7–10 (Fig. 1). Followingpassive adsorption, BSA was activated by incubating incrosslinker prior to adding the target protein. Forhetero-bifunctional reagents, which could react witheither lysine or cysteine residues on the support, controlexperiments demonstrated that the crosslinker reactedprimarily with lysine groups in BSA. This result isconsistent with the observation that most cysteineresidues in BSA are present as disulfide bridges and

are not available for further coupling reactions [54].After activating the BSA, solutions of varying FNIII7–10concentration were added for 1 h and the density ofimmobilized FNIII7–10 was determined via ELISA usingHFN7.1 antibody.For most crosslinkers, immobilized FNIII7–10 in-

creased with FNIII7–10 solution concentration, exhibit-ing linear increases at low concentrations and reachingsaturation levels at higher (B5 mg/ml) concentrations(Fig. 4). However, density of immobilized FNIII7–10varied tremendously among crosslinking agents. Inorder to compare crosslinker activity, we defined thecoupling index as the ratio of immobilized FNIII7–10 forthe maximum FNIII7–10 concentration (20 mg/ml) tobackground levels (0 mg/ml FNIII7–10) (Table 1). Thehomo-bifunctional agents DTSSP and DTBP exhibited5–10 times higher immobilized densities than any of the

FNIII7-10 concentration (ug/ml)

FNIII7-10 concentration (ug/ml)

0 5 10 15 20

boun

d an

tibo

dy (

rel.

unit

s)bo

und

anti

body

(re

l. un

its)

0

5000

10000

15000

20000

25000 DTSSPDTBPsulfo-KMUSsulfo-LC-SPDPsulfo-SMPBsulfo-EMCS

0 5 10 15 200

500

1000

1500

2000

2500

3000

3500 sulfo-KMUSsulfo-LC-SPDPsulfo-SMPBsulfo-EMCS

(a)

(b)

Fig. 4. Immobilized FNIII7–10 density as determined by HFN7.1 binding for different crosslinking agents: (a) Immobilized FNIII7–10 density

exhibited hyperbolic increases with FNIII7–10 solution concentration and was strongly dependent on the crosslinker used; (b) Immobilization density

for hetero-bifunctional crosslinkers.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–17701764

hetero-bifunctional reagents (Fig. 4a). We attribute thehigher immobilization levels for the homo-bifunctionallinkers to the larger number of target lysine moieties inFNIII7–10 compared to the single cysteine residue, whichmay also be partially buried, targeted by the hetero-bifunctional crosslinkers. The higher coupling index andlower variability of DTBP compared to DTSSP isprobably due to the higher stability of the imidoesterreaction over the NHS-ester acylation reaction [55]. Thelength of the crosslinker spacer arm also influencedcoupling efficiency as illustrated by the increases inimmobilized FNIII7–10 density with spacer arm lengthfor the series of homologous crosslinkers with identicalreactive groups (sulfo-EMCS, sulfo-SMPB, and sulfo-KMUS) (Fig. 4b and Table 1).

3.3. FN-mimetic surfaces supporting a5b1-mediated

adhesion

Based on the crosslinker efficiency results, we focusedon DTBP as the crosslinking agent to engineer non-adhesive surfaces presenting controlled densities ofFNIII7–10. FNIII7–10 and pFN (positive control) wereimmobilized onto pre-adsorbed BSA and analyzed viaELISA with HFN7.1 as described above. Fig. 5 showshyperbolic increases in immobilized FNIII7–10 and pFNdensity with protein coating concentration. Controlexperiments in which the crosslinker was omitteddisplayed background levels of either immobilizedFNIII7–10 or pFN, demonstrating that the pre-adsorbedBSA provided a non-fouling surface. For equivalentFNIII7–10 coating concentrations, the density of im-mobilized FNIII7–10 was approximately 2–3 times lowerthan the density of FNIII7–10 passively adsorbed onto

the plastic substrate (ANOVA, p o 0.01). This result isnot surprising because passively adsorbed proteins areexpected to occupy random sites within the entiresubstrate area whereas surface coverage for the im-mobilized protein is limited by availability of anchoringgroups on the support and is in agreement withpublished work [56]. Although exhibiting equivalentsurface densities when passively adsorbed, FNIII7–10displayed significantly lower immobilized levels com-pared to pFN (po0:01), especially at low coatingconcentrations. We attribute these differences to thesignificantly higher number of lysine residues on thecomplete molecule (158 lysine residues, GENBANKP02751).The HFN7.1 antibody binding measurements re-

vealed that surfaces presenting well-controlled densitiesof FNIII7–10 could be generated by varying the FNIII7–10 solution concentration. In order to determine whetherthe FN fragment was immobilized in a conformationand density that mediate specific cellular responses, wequantified cell adhesion at 1 h to these engineeredsurfaces using a centrifugation assay. We previouslydemonstrated that this assay applies controlled andreproducible detachment forces to adherent cells [53,57].Consistent with antibody binding measurements,MC3T3-E1 cell adhesion to immobilized FNIII7–10increased in a dose-dependent manner, indicating thatthe immobilized fragment supported adhesion over arange of ligand densities (Fig. 6) (ANOVA, po0:002).Control surfaces (no DTBP) displayed backgroundvalues of adhesion, demonstrating that the BSA supportprovided a non-adhesive substrate. Cell adhesion topassively adsorbed FNIII7–10 (positive control) alsoexhibited dose-dependent increases with FNIII7–10

protein concentration (ug/ml)0 10 20 30 40

boun

d an

tibo

dy (

rel.

unit

s)

0

5000

10000

15000

20000

25000

FNIII7-10 + DTBP

pFN + DTBPadsorbed FNIII7-10

adsorbed pFNFNIII7-10 no DTBP

pFN no DTBP

Fig. 5. Immobilization density for FNIII7–10 and pFN (positive control) as determined by HFN7.1 binding as a function of solution concentration.

Immobilized densities exhibited linear increases at low solution concentrations and reached saturation values at high concentrations. Omission of

DTBP crosslinker resulted in background levels of ligand, demonstrating the non-fouling nature of the support. Profiles for passive adsorption onto

synthetic support are shown for reference.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–1770 1765

coating concentration, but at levels higher than thosefor the immobilized fragment (po0:01). These higherlevels of cell adhesion probably result from higherdensities of adsorbed FNIII7–10 compared to immobi-lized FNIII7–10 (Fig. 5). Finally, equivalent results wereobtained with the NIH3T3 fibroblastic cell line (data notshown).The cell adhesion experiments demonstrated that

model surfaces presenting FNIII7–10 supported dose-dependent levels of cell adhesion. We next examined themechanism of adhesion to these FN-mimetic surfacesusing function-perturbing monoclonal antibodies.Blocking antibodies directed against the a5 or b1 integrinsubunit eliminated adhesion to background levels (BSA-coated substrates) (ANOVA, po0:00004), whereasantibodies against the b3 integrin subunit reducedadhesion by B30% but to levels higher than back-ground (po0:002) (Fig. 7). These results demonstratethat the engineered surfaces primarily support integrina5b1-mediated adhesion. Furthermore, cell adhesion tothese surfaces was completely abrogated by HFN7.1antibody (Fig. 7). Notably, HFN7.1 is specific forhuman FN and does not cross-react with mouse FN,indicating that the observed adhesion is mediated by theimmobilized FN and not matrix secreted by the cells.Similar results were obtained for NIH3T3 fibroblasts(data not shown).Finally, we examined assembly of focal adhesions for

MC3T3-E1 cells at 2 h. Cells adhering to immobilizedFNIII7–10 spread and clustered a5b1 integrin intocomplexes similar to those observed in cells adheringto passively adsorbed FNIII7–10 or pFN (Fig. 8a). Inaddition, cells attaching to FNIII7–10-fuctionalizedsurfaces displayed actin cytoskeleton reorganizationand assembled robust focal adhesions containing a5b1integrin, vinculin, and talin (Figs. 8a and 8b).

4. Discussion

We engineered model non-adhesive surfaces present-ing controlled densities of recombinant FNIII7–10, a39 kDa fragment of human fibronectin encompassingthe PHSRN synergy site and the RGD motif, usingconventional peptide chemistry techniques. Functiona-lized surfaces directed integrin a5b1-mediated cell adhe-sion, spreading, and assembly of focal adhesionscontaining a5b1; vinculin, and talin. An importantconsideration in this study is the use of BSA as asupport for protein immobilization. As demonstrated inthe present work, BSA provides a convenient non-fouling/non-adhesive support presenting high density oflysine residues available for immobilization. However,

control anti-FN anti-a5 anti-b1 anti-b3 BSA

adhe

rent

cel

ls (

rel.

unit

s)

0

500

1000

1500

2000

* * *

*

†, §

Fig. 7. Integrin- and FN-specific antibodies block cell adhesion to

FNIII7�10-functionalized surfaces (ANOVA po0:00004; * different

from control, po0:0009; w different from control, po0:006; y differentfrom BSA, po0:002).

protein concentration (ug/ml)0 10 20 30 40

adhe

rent

cel

ls (

rel.

unit

s)

0

500

1000

1500

2000

2500 FNIII7-10 + DTBP

adsorbed FNIII7-10

FNIII7-10 no DTBP

Fig. 6. MC3T3-E1 cell adhesion to immobilized FNIII7–10 showing dose-dependent adhesion levels over a range of ligand densities. Control

experiments (no DTBP crosslinker) exhibited background levels of adhesion. Adhesion to passively adsorbed FNIII7–10 was higher than adhesion to

immobilized FNIII7–10 due to differences in ligand density.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–17701766

this substrate is only effective for short term studies(o4 h under serum-free conditions for the cell lines usedin this study) because BSA can eventually be reorga-nized by cells and displaced from the substrate by serumproteins. Synthetic non-adhesive substrates, such asethylene glycol-functionalized materials [37,43,58,59],would provide more robust supports for long-termstudies. Nevertheless, BSA provided an appropriatemodel support to analyze the effects of FNIII7–10immobilization and crosslinker characteristics on theengineering of bioadhesive FN-mimetic substrates.Our biomolecular engineering strategy of using

recombinant protein fragments represents a versatileapproach to increase the biofunctional activity and,more importantly, integrin specificity of biomimeticmaterials. Numerous groups have functionalized sur-faces with short biomimetic peptides (e.g. RGD) thatsupport integrin-mediated adhesion. However, thebiological activity of these short peptides is significantlylower than that of the complete protein due to theabsence of complementary domains [48,49]. This loss inactivity for short peptides compared to the nativeextracellular matrix protein is illustrated by a recentstudy by Oreffo and colleagues examining the differ-entiation capacity of osteoprogenitor cells cultured onunmodified, RGD-functionalized, and FN-coated poly-meric substrates [47]. While several osteoblastic-specificmarkers and matrix mineralization were upregulated incells adhering to RGD-functionalized supports com-pared to unmodified substrates, osteoblasts activities

were significantly enhanced for cells cultured on FN-coated polymers compared to RGD-decorated surfaces.Similarly, other groups have reported improved cellularactivities on RGD-functionalized substrates that alsoincorporate proteoglycan-binding sequences [42,60].A major limitation of using short RGD peptides is the

lack of integrin specificity. While the RGD motif bindsto multiple integrins [23], several integrins, notably a5b1;require additional domains for binding [25,29]. Integrinspecificity is critical to directing higher order cellfunctions, such as proliferation and differentiation,since different integrins trigger diverse signaling path-ways [61]. For instance, we have demonstrated thatmodulating the binding of a5b1 vs. avb3; both of whichcompete for the RGD site in FN, regulates switchingbetween myogenic proliferation and differentiation [15].Similarly, we have observed upregulated expression ofthe osteoblastic phenotype and matrix mineralization onsubstrates that enhance a5b1 binding [51].Given the importance of domains outside the RGD

recognition sequence in specific integrin binding andsignaling, biomolecular approaches to develop bioadhe-sive surfaces should not be limited to the RGD site, butextend to include additional domains required for fullactivity, such as the PHSRN synergy site in FN. Kaoand collaborators have engineered oligopeptides con-taining both PHSRN and RGD binding motifs anddemonstrated peptide-dependent differences in foreignbody giant cell fusion but not macrophage adhesion[59]. These investigators, however, did not compare the

FNIII7-10 + DTBP adsorbed FNIII7-10 adsorbed pFN(a) integrin αα5

talin vinculin F-actin

(b) focal adhesions on FNIII7-10 + DTBP

Fig. 8. Focal adhesion assembly on FNIII7–10-engineered surfaces: (a) Cells adhering to FNIII7–10 clustered integrin a5b1 into complexes equivalent

to those assembled on passively adsorbed FNIII7–10 and pFN; (b) Focal adhesions for cells adhering to FNIII7–10 showing recruitment of talin and

vinculin and reorganization of the actin cytoskeleton.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–1770 1767

activity of this engineered peptide to complete FN or theability of the engineered peptide to support a5b1 integrinbinding. Due to the exquisite sensitivity of a5b1 bindingto small perturbations in the orientation of thesedomains [15,31,32], reconstitution of the proper struc-tural orientation using synthetic peptides remains achallenging task. The present work provides an alter-native approach to engineer FN-mimetic surfaces usinga recombinant fragment of FN to present integrinbinding domains in the appropriate structural orienta-tion. The use of recombinant FN fragments offersseveral advantages over the entire molecule, includingreduced antigenicity, elimination of domains that mayelicit adverse reactions, such as complement-, fibrino-gen-, collagen-, and heparin-binding domains [62–65],and enhanced cost efficiency. Finally, recombinant frag-ments provide flexibility in engineering of specific char-acteristics on the fragment via site-directed mutagenesis inorder to enhance protein immobilization and activity.

5. Conclusion

Model FN-mimetic surfaces were engineered byimmobilizing FNIII7–10, a recombinant fragment ofFN encompassing the PHSRN and RGD bindingdomains, onto BSA supports. Cells adhered to thesefunctionalized surfaces over a range of ligand densitiesvia integrin a5b1: Furthermore, adherent cells spreadand assembled focal adhesions containing a5b1; vinculin,and talin. This biomolecular engineering strategyrepresents a robust approach to increase biofunctionalactivity and integrin specificity of biomimetic materials.

Acknowledgements

The authors gratefully acknowledge Harold P.Erickson (Duke University) for providing E.coli trans-formed with cDNA for FNIII7–10. This research wassupported by the NSF-sponsored Georgia Tech/EmoryEngineering Research Center on the Engineering ofLiving Tissues (EEC-9731643), NSF CAREER (BES-0093226), and the Arthritis Foundation. SMC receivedadditional support from the Medtronic Foundation.HFN7.1 monoclonal antibody was obtained from theDevelopmental Studies Hybridoma Bank developedunder the auspices of the NICHD and maintained bythe University of Iowa, Department of BiologicalSciences, Iowa City, IA 52242.

References

[1] Hynes RO. Integrins: versatility, modulation, and signaling in cell

adhesion. Cell 1992;69:11–25.

[2] Damsky CH. Extracellular matrix-integrin interactions in osteo-

blast function and tissue remodeling. Bone 1999;25:95–6.

[3] De Arcangelis A, Georges-Labouesse E. Integrin and ECM

functions: roles in vertebrate development. Trends Genet 2000;

16:389–95.

[4] George EL, Georges-Labouesse EN, Patel-King RS, Rayburn H,

Hynes RO. Defects in mesoderm, neural tube and vascular

development in mouse embryos lacking fibronectin. Development

1993;119:1079–91.

[5] Stephens LE, Sutherland AE, Klimanskaya IV, Andrieux A,

Meneses J, Pedersen RA, Damsky CH. Deletion of beta 1

integrins in mice results in inner cell mass failure and peri-

implantation lethality. Genes Dev 1995;9:1883–95.

[6] Furuta Y, Ilic D, Kanazawa S, Takeda N, Yamamoto T, Aizawa

S. Mesodermal defect in late phase of gastrulation by a targeted

mutation of focal adhesion kinase, FAK. Oncogene 1995;11:

1989–95.

[7] Xu W, Baribault H, Adamson ED. Vinculin knockout results in

heart and brain defects during embryonic development. Devel-

opment 1998;125:327–37.

[8] Ziats NP, Miller KM, Anderson JM. In vitro and in vivo

interactions of cells with biomaterials. Biomaterials 1988;9:

5–13.

[9] Koller MR, Papoutsakis ET. Cell adhesion in animal cell culture:

physiological and fluid-mechanical implications. Bioprocess

Technol 1995;20:61–110.

[10] Langer R, Vacanti JP. Tissue engineering. Science 1993;260:

920–6.

[11] Menko AS, Boettiger D. Occupation of the extracellular matrix

receptor integrin is a control point for myogenic differentiation.

Cell 1987;51:51–7.

[12] Werb Z, Tremble PM, Behrendtsen O, Crowley E, Damsky CH.

Signal transduction through the fibronectin receptor induces

collagenase and stromelysin gene expression. J Cell Biol 1989;109:

877–89.

[13] Streuli CH, Schmidhauser C, Bailey N, Yurchenco P, Skubitz AP,

Roskelley C, Bissell MJ. Laminin mediates tissue-specific gene

expression in mammary epithelia. J Cell Biol 1995;129:591–603.

[14] Zhu X, Ohtsubo M, Bohmer RM, Roberts JM, Assoian RK.

Adhesion-dependent cell cycle progression linked to the expres-

sion of cyclin D1, activation of cyclin E-cdk2, and phosphoryla-

tion of the retinoblastoma protein. J Cell Biol 1996;133:391–403.

[15] Garc!ıa AJ, Vega MD, Boettiger D. Modulation of cell prolifera-

tion and differentiation through substrate-dependent changes in

fibronectin conformation. Mol Biol Cell 1999;10:785–98.

[16] Jockusch BM, Bubeck P, Giehl K, Kroemker M, Moschner J,

Rothkegel M, Rudiger M, Schluter K, Stanke G, Winkler J. The

molecular architecture of focal adhesions. Annu Rev Cell Dev

Biol 1995;11:379–416.

[17] Burridge K, Chrzanowska-Wodnicka M. Focal adhesions, con-

tractility, and signaling. Annu Rev Cell Dev Biol 1996;12:

463–518.

[18] Almeida EA, Ilic D, Han Q, Hauck CR, Jin F, Kawakatsu H,

Schlaepfer DD, Damsky CH. Matrix survival signaling: from

fibronectin via focal adhesion kinase to c-Jun NH(2)-terminal

kinase. J Cell Biol 2000;149:741–54.

[19] Sieg DJ, Hauck CR, Ilic D, Klingbeil CK, Schaefer E, Damsky

CH, Schlaepfer DD. FAK integrates growth-factor and integrin

signals to promote cell migration. Nat Cell Biol 2000;2:249–56.

[20] Sastry SK, Lakonishok M, Wu S, Truong TQ, Huttenlocher A,

Turner CE, Horwitz AF. Quantitative changes in integrin and

focal adhesion signaling regulate myoblast cell cycle withdrawal. J

Cell Biol 1999;144:1295–309.

[21] Gallant ND, Capadona JR, Frazier AB, Collard DM, Garc!ıa AJ.

Micropatterned surfaces for analyzing cell adhesion strengthen-

ing. Langmuir 2002;18:5579–84.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–17701768

[22] Ruoslahti E, Pierschbacher MD. Arg-Gly-Asp: a versatile cell

recognition signal. Cell 1986;44:517.

[23] van der Flier A, Sonnenberg A. Function and interactions of

integrins. Cell Tissue Res 2001;305:285–98.

[24] Obara M, Kang MS, Yamada KM. Site-directed mutagenesis of

the cell-binding domain of human fibronectin: separable, syner-

gistic sites mediate adhesive function. Cell 1988;53(4):649–57.

[25] Bowditch RD, Hariharan M, Tominna EF, Smith JW, Yamada

KM, Getzoff ED, Ginsberg MH. Identification of a novel integrin

binding site in fibronectin. Differential utilization by beta 3

integrins. J Biol Chem 1994;269:10856–63.

[26] Moursi AM, Globus RK, Damsky CH. Interactions between

integrin receptors and fibronectin are required for calvarial

osteoblast differentiation in vitro. J Cell Sci 1997;110:2187–96.

[27] Roovers K, Davey G, Zhu X, Bottazzi ME, Assoian RK.

Alpha5beta1 integrin controls cyclin D1 expression by sustaining

mitogen-activated protein kinase activity in growth factor-treated

cells. Mol Biol Cell 1999;10:3197–204.

[28] Wu C, Bauer JS, Juliano RL, McDonald JA. The alpha 5 beta 1

integrin fibronectin receptor, but not the alpha 5 cytoplasmic

domain, functions in an early and essential step in fibronectin

matrix assembly. J Biol Chem 1993;268:21883–8.

[29] Aota S, Nomizu M, Yamada KM. The short amino acid sequence

Pro-His-Ser-Arg-Asn in human fibronectin enhances cell-adhesive

function. J Biol Chem 1994;269:24756–61.

[30] Redick SD, Settles DL, Briscoe G, Erickson HP. Defining

fibronectin’s cell adhesion synergy site by site-directed mutagen-

esis. J Cell Biol 2000;149:521–7.

[31] Garc!ıa AJ, Schwarzbauer JE, Boettiger D. Distinct activation

states of alpha5beta1 integrin show differential binding to RGD

and synergy domains of fibronectin. Biochemistry 2002;41:

9063–9.

[32] Grant RP, Spitzfaden C, Altroff H, Campbell ID, Mardon HJ.

Structural requirements for biological activity of the ninth and

tenth FIII domains of human fibronectin. J Biol Chem 1997;

272:6159–66.

[33] Hubbell JA. Bioactive biomaterials. Curr Opin Biotechnol 1999;

10:123–9.

[34] Shakesheff K, Cannizzaro S, Langer R. Creating biomimetic

micro-environments with synthetic polymer-peptide hybrid mole-

cules. J Biomater Sci Polym Ed 1998;9:507–18.

[35] Matsuda T, Kondo A, Makino K, Akutsu T. Development of a

novel artificial matrix with cell adhesion peptides for cell

culture and artificial and hybrid organs. ASAIO Trans 1989;

35:677–9.

[36] Massia SP, Hubbell JA. Covalent surface immobilization of Arg-

Gly-Asp- and Tyr-Ile-Gly-Ser-Arg- containing peptides to obtain

well-defined cell-adhesive substrates. Anal Biochem 1990;187:

292–301.

[37] Drumheller PD, Hubbell JA. Polymer networks with grafted cell

adhesion peptides for highly biospecific cell adhesive substrates.

Anal Biochem 1994;222:380–8.

[38] Dee KC, Rueger DC, Andersen TT, Bizios R. Conditions which

promote mineralization at the bone-implant interface: a model in

vitro study. Biomaterials 1996;17:209–15.

[39] Bearinger JP, Castner DG, Healy KE. Biomolecular modification

of p(AAm-co-EG/AA) IPNs supports osteoblast adhesion and

phenotypic expression. J Biomater Sci Polym Ed 1998;9:629–52.

[40] Hern DL, Hubbell JA. Incorporation of adhesion peptides into

nonadhesive hydrogels useful for tissue resurfacing. J Biomed

Mater Res 1998;39:266–76.

[41] Neff JA, Caldwell KD, Tresco PA. A novel method for surface

modification to promote cell attachment to hydrophobic sub-

strates. J Biomed Mater Res 1998;40:511–9.

[42] Rezania A, Healy KE. Biomimetic peptide surfaces that regulate

adhesion, spreading, cytoskeletal organization, and mineraliza-

tion of the matrix deposited by osteoblast-like cells. Biotechnol

Prog 1999;15:19–32.

[43] Mann BK, Gobin AS, Tsai AT, Schmedlen RH, West JL. Smooth

muscle cell growth in photopolymerized hydrogels with

cell adhesive and proteolytically degradable domains: synthetic

ECM analogs for tissue engineering. Biomaterials 2001;22:

3045–51.

[44] Alsberg E, Anderson KW, Albeiruti A, Franceschi RT, Mooney

DJ. Cell-interactive alginate hydrogels for bone tissue engineer-

ing. J Dent Res 2001;80:2025–9.

[45] Sofia S, McCarthy MB, Gronowicz G, Kaplan DL. Functiona-

lized silk-based biomaterials for bone formation. J Biomed Mater

Res 2001;54:139–48.

[46] Massia SP, Stark J. Immobilized RGD peptides on surface-

grafted dextran promote biospecific cell attachment. J Biomed

Mater Res 2001;56:390–9.

[47] Yang XB, Roach HI, Clarke NM, Howdle SM, Quirk R,

Shakesheff KM, Oreffo RO. Human osteoprogenitor growth

and differentiation on synthetic biodegradable structures after

surface modification. Bone 2001;29:523–31.

[48] Pierschbacher MD, Hayman EG, Ruoslahti E. Synthetic peptide

with cell attachment activity of fibronectin. Proc Natl Acad Sci

USA 1983;80:1224–7.

[49] Akiyama SK, Aota S, Yamada KM. Function and receptor

specificity of a minimal 20 kilodalton cell adhesive fragment of

fibronectin. Cell Adhes Commun 1995;3:13–25.

[50] Aukhil I, Joshi P, Yan Y, Erickson HP. Cell- and heparin-binding

domains of the hexabrachion arm identified by tenascin expres-

sion proteins. J Biol Chem 1993;268:2542–53.

[51] Stephansson SN, Byers BA, Garc!ıa AJ. Enhanced expression of

the osteoblastic phenotype on substrates that modulate fibronec-

tin conformation and integrin receptor binding. Biomaterials

2002;23:2527–34.

[52] Schoen RC, Bentley KL, Klebe RJ. Monoclonal antibody against

human fibronectin inhibits cell attachment. Hybridoma 1982;

1:99–108.

[53] Keselowsky BG, Collard DM, Garc!ıa AJ. Surface chemistry

modulates fibronectin conformation and directs integrin binding

and specificity to control cell adhesion. J Biomed Mater Res, in

press.

[54] Minchiotti L, Galliano M, Kragh-Hansen U, Watkins S, Madison

J, Putnam FW. A genetic variant of albumin (albumin Asola;

Tyr140-Cys) with no free -SH group but with an additional

disulfide bridge. Eur J Biochem 1995;228:155–9.

[55] Hermanson GT. Bioconjugate techniques. San Diego, CA:

Academic Press, 1996.

[56] Biran R, Webb K, Noble MD, Tresco PA. Surfactant-immobi-

lized fibronectin enhances bioactivity and regulates sensory

neurite outgrowth. J Biomed Mater Res 2001;55:1–12.

[57] Reyes CD, Garc!ıa AJ. Engineering integrin-specific surfaces with

a triple-helical collagen-mimetic peptide. J Biomed Mater Res, in

press.

[58] Roberts C, Chen CS, Mrksich M, Martichonok V, Ingber DE,

Whitesides GM. Using mixed self-assembled monolayers present-

ing RGD and (EG)3OH groups to characterize long-term

attachment of bovine capillary endothelial cells to surfaces.

J Am Chem Soc 1998;120:6548–55.

[59] Kao WJ, Lee D, Schense JC, Hubbell JA. Fibronectin modulates

macrophage adhesion and FBGC formation: the role of RGD,

PHSRN, and PRRARV domains. J Biomed Mater Res 2001;

55:79–88.

[60] Dee KC, Andersen TT, Bizios R. Design and function of novel

osteoblast-adhesive peptides for chemical modification of bioma-

terials. J Biomed Mater Res 1998;40:371–7.

[61] Giancotti FG, Ruoslahti E. Integrin signaling. Science 1999;285:

1028–32.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–1770 1769

[62] Grinnell F, Feld M, Minter D. Fibroblast adhesion to fibrinogen

and fibrin substrata: requirement for cold-insoluble globulin

(plasma fibronectin). Cell 1980;19:517–25.

[63] Lahav J, Hynes RO. Involvement of fibronectin, Von

Willebrand factor, and fibrinogen in platelet interaction with

solid substrata. J Supramol Struct Cell Biochem 1981;17:

299–311.

[64] Stanton H, Ung L, Fosang AJ. The 45 kDa collagen-binding

fragment of fibronectin induces matrix metalloproteinase-13

synthesis by chondrocytes and aggrecan degradation by aggreca-

nases. Biochem J 2002;364:181–90.

[65] Yasuda T, Poole AR. A fibronectin fragment induces type II

collagen degradation by collagenase through an interleukin-1-

mediated pathway. Arthritis Rheum 2002;46:138–48.

S.M. Cutler, A.J. Garc!ıa / Biomaterials 24 (2003) 1759–17701770