RGD-containing peptide GCRGYGRGDSPG reducesenhancement of osteoblast differentiation by poly(L-lysine)-graft-poly(ethylene glycol)-coated titanium surfaces
S. Tosatti,1 Z. Schwartz,2,3,4 C. Campbell,3 D. L. Cochran,3 S. VandeVondele,5 J. A. Hubbell,5 A. Denzer,6
J. Simpson,6 M. Wieland,6 C. H. Lohmann,7 M. Textor,1 B. D. Boyan2,4
1BioInterfaceGroup, Laboratory for Surface Science and Technology, Department of Materials, ETH, CH-8900 Zurich,Switzerland2Wallace H. Coulter Department of Biomedical Engineering at Georgia Tech and Emory University, Atlanta, Georgia 303323Department of Periodontics, University of Texas Health Science Center at San Antonio, San Antonio, Texas 782844Department of Periodontics, Hebrew University Hadassah, IL-91010 Jerusalem, Israel5Institute for Biomedical Engineering, Department of Materials, ETH , CH-8900 Zurich, Switzerland6Institut Straumann, CH-4473 Waldenburg, Switzerland7Department of Orthopaedics, University of Hamburg-Eppendorf, D-20255 Hamburg, Germany
Received 24 April 2003; revised 27 August 2003; accepted 3 September 2003
Abstract: Osteoblasts exhibit a more differentiated morphol-ogy on surfaces with rough microtopographies. Surface effectsare often mediated through integrins that bind the RGD motifin cell attachment proteins. Here, we tested the hypothesis thatmodulating access to RGD binding sites can modify the re-sponse of osteoblasts to surface microtopography. MG63 im-mature osteoblast-like cells were cultured on smooth (Ti sput-ter-coated Si wafers) and rough (grit blasted/acid etched) Tisurfaces that were modified with adsorbed monomolecularlayers of a comb-like graft copolymer, poly-(l-lysine)-g-poly-(ethylene glycol) (PLL-g-PEG), to limit nonspecific protein ad-sorption. PLL-g-PEG coatings were functionalized with vary-ing amounts of an integrin-receptor-binding RGD peptideGCRGYGRGDSPG (PLL-g-PEG/PEG-RGD) or a nonbindingRDG control sequence GCRGYGRDGSPG (PLL-g-PEG/PEG-RDG). Response to PLL-g-PEG alone was compared with re-sponse to surfaces on which 2–18% of the polymer sidechainswere functionalized with the RGD peptide or the RDG peptide.To examine RGD dose–response, peptide surface concentrationwas varied between 0 and 6.4 pmol/cm2. In addition, cellswere cultured on uncoated Ti or Ti coated with PLL-g-PEG orPLL-g-PEG/PEG-RGD at an RGD surface concentration of 0.7pmol/cm2, and free RGDS was added to the media to blockintegrin binding. Analyses were performed 24 h after cultureshad achieved confluence on the tissue culture plastic surface.Cell number was reduced on smooth Ti compared to plastic or
To improve the biologic acceptance of intraosseousimplants and reduce healing time, investigators havemodified the surface design. Surfaces with rough mor-phologies exhibit greater pullout strength than thosewith smooth surfaces,1 due in part to the sensitivity ofbone-forming osteoblasts to microtopographical fea-
Correspondence to: B. D. Boyan, Wallace H. Coulter Depart-ment of Biomedical Engineering, Georgia Institute of Technol-ogy, 315 Ferst Drive NW, Atlanta, GA 30332; e-mail:[email protected]
tures of their substrate.2–5 Alternative approacheshave coupled bioligands, such as peptides, adhesionproteins, enzymes or growth factors, to the surface.6–8
Specific amino acid sequences present in many extra-cellular matrix proteins are efficient bioligands for cellattachment to surfaces.7,9,10 These relatively short pep-tides can be produced cost effectively and are lesssusceptible to denaturation in comparison with adhe-sion proteins and growth factors. The most commoncell-binding domain used in this manner is the threeamino acid sequence arginine-glycine-aspartic acidArg-Gly-Asp; (RGD).11–13 This motif is found in celladhesive proteins such as fibronectin or vitronectin,which are important in mediating cell adhesion ofosteoblasts and other cells to synthetic material sur-faces.14,15 Cells bind to RGD through integrin recep-tors present in cell membranes,16,17 thereby initiatingintegrin-mediated signal transduction.18 Binding spec-ificity and the subsequent cellular response depend onthe amino acid sequence of the regions of the proteinthat flank the RGD motif19 and on the types of inte-grins expressed by the cell.20 The amino acid sequenceGRGDSP derived from the type III10 repeat of humanfibronectin21 is one of the frequently used sequencesshowing consistently high activity in binding to inte-grin receptors.22–25
Osteoblast interactions with peptide-functionalizedsurfaces have been studied for the most part usingtissue culture plastic as the model surface.13,23 Boththe type of peptide and its surface density affect cell–surface interactions. For example, integrin #5!3-medi-ated fibroblast spreading occurs at a mean RGD pep-tide spacing of 440 nm (corresponding to 0.001 pmol/cm2), while focal contact formation and extensivef-actin stress fiber formation require a mean spacingof $ 140 nm (corresponding to 0.01 pmol/cm2).26
Osteoblast adhesion and spreading are enhanced athigher RGD peptide surface densities (% 0.6 pmol/cm2) compared with lower densities (0.01 pmol/cm2).27–29 Other studies indicate that higher concen-trations of surface-bound RGD may be needed fordifferentiation and mineralization of osteoblast cul-tures grown on peptide- and polyethylene-glycol-modified titanium surfaces based on interpenetratingpolymer networks.30
Material surface chemistry can have profound ef-fects on how cells respond.2 This is due in part todifferential adsorption of serum components and re-sulting conformational changes that affect integrinbinding and function.3 Use of serum-free media is oneway to limit the adsorption of serum components,31
but serum-free conditions do not exist in vivo. Alter-natively, nonspecific adsorption of serum proteins canbe controlled by using protein-resistant surface coat-ings.8,26,30,32
Poly(ethylene glycol) (PEG) has been in particularsuccessful for use as a protein-nonadhesive back-ground onto which specific peptides may be attached.
The protein-repellent character of PEG has been attrib-uted to steric repulsion and excluded-volume effectsbetween proteins in solution and the PEG-modifiedsurface.33,34 Direct PEGylation of surfaces, the use ofPEG-grafted interpenetration networks,30 and poly-ionic PEG-grafted copolymers such as poly(l-lysine)-g-poly(ethylene glycol) (PLL-g-PEG) 32,35–38 have beenreported to render metal oxide surfaces very nonin-teractive with the biologic environment.
The multifunctional macromolecule PLL-g-PEG isin particular attractive for surface modifications, hav-ing a number of advantages compared to other surfacefunctionalization schemes. PLL-g-PEG spontaneouslyadsorbs from aqueous solution onto negativelycharged surfaces via attachment through the posi-tively charged PLL backbone. Thus, a simple dip-and-rinse process, forming a densely packed PEG layer,can coat metal oxide surfaces such as TiO2. The result-ing surface shows an excellent resistance to proteinadsorption, typically less than 2–3 ng/cm2, as well asto cell adhesion.36,37 The PEG chains in the polymericmolecule can be functionalized at the terminal posi-tion with a biologically active moiety, resulting in apolymer with a controlled density of bioligands.39 Us-ing this method, peptide surface density can be easilyvaried and perfectly controlled over orders of magni-tude through assembly from an aqueous solution thatcontains bioligand-modified and nonmodified poly-meric molecules in a defined ratio.40
Topographically structured or otherwise rough sur-faces and 3D devices can be coated straightforwardlyby this simple, spontaneous molecular assembly tech-nique, in contrast to many other techniques that oftenrequire flat/smooth surfaces for achieving controlled,reproducible results. Tissue culture plastic surfaces donot exhibit topographical features typical of the natu-ral bone surface, or of biomaterial surfaces typicallyused for implants, and osteoblast response to tissueculture plastic is altered from cell response to bone ordentin surfaces.41,42 Indeed, resorption of bone wafersby osteoclasts yields surfaces that support phenotypicexpression of osteoblasts more like that seen on Tisurfaces with rough microtopographies.42
This indicates that accurate in vitro assessment offunctionalized surfaces intended for use in vivo re-quires that substrates with relevant surface morphol-ogies should be used. In this article we used PLL-g-PEG without and with functionalization by an RGDpeptide (GCRGYGRGDSPG),37 covalently linked to afraction of the PEG side-chains in the polymer (abbre-viated as PLL-g-PEG/PEG-RGD). The peptideGCRGYGRDGSPG does not bind integrin receptors;therefore, the corresponding polymer PLL-g-PEG/PEG-RDG served as negative control. The polymerswere assembled onto two types of titanium metalsurfaces: a titanium film evaporated onto smooth sil-icon wafers and a rough titanium surface fabricated byalumina blasting followed by chemical etching. Both
RGD-PEPTIDE INHIBITS OSTEOBLAST DIFFERENTIATION ON PEG 459
titanium surfaces are covered by a native oxide film ofa few nanometer thickness.43 The combination of de-fined substrate topography and biologically designedsurface chemistry allows us to independently investi-gate the effect of microtopography and biochemicalcomposition on osteoblast proliferation, differentia-tion, and expression of factors. Such perfectly inde-pendent control of surface topography and chemicalproperties has so far not been possible or been difficultusing conventional biomolecule immobilization tech-niques.
MATERIALS AND METHODS
Substrates
Two types of Ti surfaces were used. Smooth surfaces wereprepared by depositing 100-nm-thick Ti films onto Si wafersusing physical vapor deposition (PVD, reactive magnetronsputtering, Paul Scherrer Institute, Villigen, Switzerland).Before coating, the Si wafer substrates were ultrasonicallycleaned in toluene (Puriss, Merck, Dietikon, Switzerland) for15 min and dried under N2. Control surfaces consisted oftissue culture polystyrene (TCPS, Nunclon, Nalge Nunc Int.)and tissue culture glass (Labteck II, Nalge Nunc Int.). Theroughness parameters of the Ti-coated Si wafers and controlsurfaces were determined by atomic force microscopy basedon a 10-&m scan length. The arithmetic average, Ra, and theroot mean square, Rq, roughness parameters were found tobe 1.7 ' 0.1 nm and 1.3 ' 0.0 nm (Ti surface),43 2.3 ' 0.2 nmand 2.8 ' 0.2 nm (TCPS), and 2.2 ' 0.3 nm and 2.7 ' 0.4 nm(tissue culture glass), respectively.
To produce rough microtopographies, commercially puretitanium (CP Ti) discs, 15 mm in diameter and 1 mm inthickness, were grit blasted with alumina beads (averageparticle size, 250 &m) under low-impact-energy, industrialparticle-blasting conditions and subsequently etched in ahot solution of HCl/H2SO4 (Institut Straumann AG, Wal-denburg, Switzerland). This surface, known commercially as“SLA,” is characterized by a duplex surface topographywith roughness contributions in the range of typically 20–50&m (originating from the blasting process) and 0.5–2 &m(from the chemical etching process). The average roughnessvalues were measured with the laser noncontact profilome-ter and evaluated following the standard Norm DIN4768,resulting in Ra and Rq values of 5.50 ' 0.30 &m and 6.91 '0.39 &m, respectively,44 confirming measurements made bycontact profilometry indicating Ra values of 4–5 &m.45 Im-pedance spectroscopy indicated that the specific surface areawas 6.5 cm2/cm2 (effective surface area/geometric surfacearea). This means that in comparison to a smooth surface theSLA surface has a specific surface that is higher by a factorof approximately 6.5.46
Synthesis of PLL-g-PEG and PLL-g-PEG/PEG-RGD
To assess the role of RGD surface density on osteoblastphenotypic expression, four different PLL-g-PEG classes of
copolymers were synthesized: the nonfunctionalized poly-mer (PLL-g-PEG), two polymers with different amounts ofRGD peptide covalently linked to the PEG chains (PLL-g-PEG/PEG-RGD), and a polymer containing a scrambledpeptide sequence as a control (PLL-g-PEG/PEG-RDG). Pre-viously published methods37 were used to synthesize thePLL-g-PEG/PEG-peptide polymers. Poly(l-lysine) (Mw
15–30 kDa, Sigma, St. Louis, MO) was reacted with NHS-PEG-methoxy (2 kDa) and NHS-PEG-VS (3.4 kDa), bothactivated polymers from Shearwater, Inc.(Los Altos, CA), ina theoretical ratio of 1 PEG chain per 3.5 lysine monomers.Different mole ratios of NHS-PEG-methoxy and NHS-PEG-VS were used depending on the requested final peptideconcentration. The reaction between PLL and PEG takesplace between the NHS-active ester on the PEG and the(-amines of the lysine residues, the reaction between thepeptide and the PEG between the VS-group on the PEG andthe thiol of the cysteine in the peptide. After 30 min at pH8.4, the cysteine-containing peptide was added in a twofoldexcess when compared with the concentration of NHS-PEG-VS molecules. The following morning, 2-mercaptoetha-nol was added to quench the remaining free vinyl sulfones.The salts and 2-mercaptoethanol were dialyzed awayagainst deionized water over a 3-day period and the sampleswere subsequently lyophilized.
Two peptide sequences were used: N-acetyl-GCRGYGRGD-SPG-amide (RGD) and N-acetyl-GCRGYGRDGSPG-amide(RDG).7,37 The peptide design was based on the integrin-binding sequence (-Arg-Gly-Asp-Ser-Pro-, RGDSP) of fi-bronectin. The –Gly-Cys-Arg- (GCR) sequence was includedfor chemical coupling of the thiol group of the cysteine to thevinyl sulfone functional group of the copolymer, while the–Gly-Tyr-Gly- (GYG) sequence was used as a spacer as wellas a marker to quantify the degree of peptide grafting basedon the specific1 H-NMR signals of the tyrosine at 6.87 and7.16 ppm. The RDG sequence was used in the control pep-tide GCRGYGRDGSPG- to test for the specificity of the RGDpeptide–integrin receptor interactions. Apart from theRGDSP, none of the other amino acid sequences in thepeptides chosen is known to show biospecific interactions(PROSITE database, http://www.expasy.org/). Peptidesynthesis was performed on a Pioneer peptide synthesizerusing standard Fmoc chemistry.47 Peptide molecular masswas confirmed by matrix-assisted laser desorption and time-of-flight mass spectrometry (MALDI-ToF).
The general architecture of the polymers used in this workis based on a PLL backbone of approximately 120 l-lysineunits (average value in view of the polydispersity of thepolymer), a PEG or PEG-vinyl sulfone side-chain of approx-imately 47 ethylene glycol units (PEG Mw ) 2 kDa) orapproximately 80 ethylene glycol units (PEG-VS Mw ) 3.4kDa), respectively, and a grafting ratio g, expressed as thenumber of lysine monomers per PEG side-chain, between3.3 and 4.5. PLL-g-PEG polymers with optimized architec-tures have consistently shown the capacity of forming densePEG-brush adlayers on negatively charged surfaces such asTiO2, rendering such surfaces highly resistant to proteinadsorption when exposed to full serum36 or whole heparin-ized plasma (typically less than 3–5 ng/cm2 adsorbedmass).39
For each polymer batch the grafting ratio g and the degreeof peptide functionalization were determined by NMR asdescribed elsewhere,39 while the protein resistance of the
460 TOSATTI ET AL.
polymeric adlayers upon serum exposure was monitored insitu with the help of optical waveguide lightmode spectros-copy (OWLS).35,36,47 All polymer batches used in this workproved to render the TiO2 surfaces highly resistant to serumprotein adsorption ($ 5 ng/cm2 adsorbed mass from fullserum). Molecular weight and grafting ratio g of the poly-mers as well as the resulting peptide surface concentrationsare summarized in Table I.
Coating of the titanium surfaces with polymers
Polymer (1 mg/mL) was dissolved in phosphate-bufferedsaline (PBS), pH 7.4, filter sterilized (Millex-GW, Millipore,Switzerland), aliquoted, and stored at *20°C. Prior to sur-face modification, the Ti samples were ultrasonicated for 5min in 2-propanol (UVASOL, Merck, Dietikon, Switzerland)to remove adventitious macroscopic contamination anddried under N2. After cleaning and sterilizing in an oxygenplasma (Harrick Plasma Cleaner/Sterilizer PDC-32G, Ossin-ing, NY) for 3 min, the surfaces were immediately placed incell culture wells of 24-well plates. The polymer solutionswere pipetted onto the sterile surfaces so that the surfaceswere just covered by liquid. The samples were then gentlyshaken for 15 min and washed three times with sterile PBS.Prior to cell seeding, the final PBS wash solution was re-moved.
The surface concentration of peptide (in pmol/cm2) wascalculated based on the quantitative NMR data of the poly-mers and the average adsorbed polymer mass (110 ng/cm2,from OWLS measurements). To describe the different sur-faces with adsorbed peptide-functionalized PLL-g-PEG, thefollowing code is used: PLL-g-PEG/PEG-peptide(X) whereX indicates the peptide surface concentration in pmol/cm2
(Table I), referred to as the geometric surface area. Becausethe effective surface area of the SLA surface was higher by afactor of 6.5, the total peptide concentration per unit geo-
metric surface area was also correspondingly higher. For acomplex topography such as the SLA, it is not feasible topredict what fraction of the surface peptide the cells recog-nize; the peptide densities are therefore always specified perunit effective surface area. However, for calculating the totalamount of surface peptides per well of the SLA samples(needed for designing the competition assays; see below) thepeptide surface density per effective area was multipliedwith the specific surface factor 6.5.
Initial experiments used surfaces coated with PLL-g-PEG/PEG-RGD(1.0) and PLL-g-PEG/PEG-RDG(1.7). The copoly-mer solutions for the dose–response experiments were pre-pared by 1:4 dilution of equimolar solutions of PLL-g-PEGand PLL-g-PEG/PEG-RGD(6.4), leading to surfaces withtheoretical concentrations of 6.4, 1.3, 0.26, and 0.05 pmol/cm2 RGD peptide. Previous work demonstrated that thesurface composition of mixed PLL-g-PEG adlayers quantita-tively reflects the solution composition, that is, no preferen-tial adsorption has been observed.40
Cell response to surface treatment
Cell culture model
MG63 osteoblast-like cells, originally isolated from a hu-man osteosarcoma,48 were obtained from the AmericanType Culture Collection (Rockville, MD).
These cells have been well characterized and exhibit nu-merous osteoblastic traits, including increased levels of al-kaline phosphatase activity and osteocalcin synthesis in re-sponse to 1,25(OH)2D3,49,50 as well as enhanceddifferentiation on surfaces with rough microtopogra-phies.51–53
The cells were cultured in Dulbecco’s modified Eagle me-dium (DMEM) containing 10% fetal bovine serum (FBS) and
TABLE IMolecular Weight (Mw) and Grafting Ratio g of the Polymers Used in This Work and Peptide Surface Concentration
of the Polymer-Modified TiO2 Surfaces Calculated from the Polymer Architecture (NMR) and theMass of Adsorbed Polymer (OWLS)
(% of total PEG chains) — 1.7 4.0 1.6 2.5 18MwPolymer
(g/mol)† 96K 72K 81K 76K 93K 86KPeptide surface
concentration on smoothTi(metal) surface(pmol/cm2) 0 0.7 1.7 0.7 1.0 6.4
*PLL-HBr is the bromide salt of poly(l-lysine) used as educt in the synthesis of the copolymer.†Calculated according to the formula published by Tosatti et al.39
RGD-PEPTIDE INHIBITS OSTEOBLAST DIFFERENTIATION ON PEG 461
1% penicillin and streptomycin at 37°C in an atmosphere of5% CO2 and 100% humidity. When the cells reached conflu-ence, they were subcultured onto the test surfaces at aninitial plating density of 9000 cells/cm2. Media werechanged 24 h after plating and on day 4. All assays wereconducted at 7 days postplating, based on previous studiesshowing that cultures on plastic were confluent at thistime.51,54
Experimental design
Initial studies were conducted on smooth Ti surfaces toassess if the PLL-g-PEG coating had an effect on cell re-sponse independent of surface microtopography. For eachexperiment, the following surfaces were used: (1) untreatedTCPS, (2) uncoated tissue culture glass disks, (3) plasma-cleaned Si wafers that were sputter coated with Ti(metal), (4)Ti-coated Si wafers with PLL-g-PEG alone, (5) Ti-coated Siwafers with PLL-g-PEG/PEG-RDG(1.7), and (6) Ti-coated Siwafers with PLL-g-PEG/PEG-RGD(1.0).
To determine whether the effect of the surface coatingswas sensitive to Ti microtopography, Ti SLA disks (rough Tisurface) were compared to smooth Ti-coated Si wafers. Un-treated TCPS and uncoated tissue culture glass wafers wereincluded as controls in addition to uncoated Ti SLA disks.Experimental design was as described above for the smoothTi-coated silicon wafers.
Dose-dependent effects of the RGD peptide on the re-sponse of MG63 cells to PLL-g-PEG were examined usingthe structured Ti SLA disks. Each experiment included: (1)TCPS, (2) plasma-cleaned Ti SLA, (3) Ti SLA coated withPLL-g-PEG, (4) Ti SLA coated with PLL-g-PEG/PEG-RGD(0.05), (5) Ti SLA coated with PLL-g-PEG/PGE-RGD(0.26), (6) Ti SLA coated with PLL-g-PEG/PEG-RGD(1.28), and (7) Ti SLA coated with PLL-g-PEG/PEG-RGD(6.4). The total amount of RGD peptide inside each wellwas 0, 0.44, 2.2, 11.1, and 55.5 pmol/well for surfaces 3–7,respectively.
To determine if the effect of the RGD peptide on theresponse of MG63 cells to PLL-g-PEG was due to integrinbinding to the surface-immobilized ligand, the peptideRGDS (Calbiochem, San Diego, CA) was added to the me-dia. In the first set of studies, RGDS was dissolved in DMEMat a concentration of 45 pmol/mL. MG63 cells were incu-bated in DMEM containing 45 pmol/mL RGDS for 10 minand then transferred to SLA surfaces coated with PLL-g-PEG/PEG-RGD(0.7) at a plating density of 9000 cells/cm2.At each culture medium change, fresh RGDS was againadded at the same concentration as used in the startingmedium (45 pmol/mL). Cell responses were compared tothose on TCPS. Uncoated SLA disks, SLA disks coated withPLL-g-PEG, and SLA disks with PLL-g-PEG/PEG-RGD(0.7)were examined.
In addition, the effect of soluble RGDS on the response ofthe cells to nonfunctionalized PLL-g-PEG was examined byculturing the cells on SLA surfaces coated with PLL-g-PEGin the presence of RGDS in the medium at total concentra-tions corresponding to the surface concentrations in thedose-dependent PLL-g-PEG/PEG-RGD(x) experiments de-scribed above. In these experiments, RGDS was present inthe culture media at concentrations of 0, 0.36, 1.8, 9, and 45
pmol/mL. The cells were incubated in RGDS-containingmedia for 10 min and plated on the SLA at a plating densityof 9000 cells/cm2. PLL-g-PEG-coated SLA surfaces served ascontrol.
Cell response
To determine the number of cells after 7 days of cultureMG63 cells were released from the surfaces by two sequen-tial incubations with 0.25% trypsin in Hank’s balanced saltsolution (HBSS) containing 1 mM ethylenediamine tetraace-tic acid (EDTA) for 10 min at 37°C, followed by the additionof DMEM containing 10% FBS to stop the reaction.55 Thetwo cell suspensions were combined and centrifuged at500 " g for 10 min. Cell pellets were washed with PBS andresuspended in PBS. Cell number was determined using aCoulter Counter (Coulter Electronics, Inc., Hialeah, FL).Cells harvested in this manner exhibited % 95% viabilitybased on trypan blue exclusion.
Osteocalcin levels in the conditioned media were mea-sured using a commercially available radioimmunoassay kit(Human Osteocalcin RIA Kit, Biomedical Technologies,Stoughton, MA) as described previously.56 Alkaline phos-phatase-specific activity was assayed in lysates of isolatedcells as well as in cell layer lysates following the method ofHale et al.57 as previously described.54,56 The first methodassesses the enzyme activity in the cell whereas the secondmethod also includes the matrix vesicles present in theextracellular matrix. Matrix vesicles are enriched in alkalinephosphatase activity immediately prior to the initiation ofcalcification. Therefore, the combined results of these twoassays provide information on the state of differentiation ofthe cell. Higher enzyme activity in cell layers versus cells isindicative of a more differentiated cell. If enzyme activity isincreased but osteocalcin is unaffected, early differentiationis underway. If both parameters are increased differentiationis at a midlevel and if alkaline phosphatase activity is re-duced but osteocalcin is elevated differentiation is complete.Protein content was determined using a commercially avail-able kit (Micro/Macro BCA, Pierce Chemical Co., Rockford,IL). Alkaline phosphatase activity was assayed at 37°C as therelease of p-nitrophenol from p-nitrophenylphosphate at pH10.2.58
To measure the level of total TGF-!1 in the conditionedmedia at time of harvest, a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Promega Corp.,Madison, WI) was used as described previously.51,59 Imme-diately prior to assay, conditioned media were acidified toactivate latent TGF-!1. The amount of TGF-!1 in the celllayer was not examined because of difficulties associatedwith quantitatively extracting this growth factor from thematrix when using samples of this size. Approximately 20%of total TGF-!1 produced by the MG63 cells on plastic is inthe cell layer. When MG63 cells are cultured on Ti surfaces,this increases with increasing roughness (smooth Ti, 30%;SLA, 42%).59 Whether this is true for MG63 cells cultured onPLL-g-PEG-coated surfaces is not known, however. PGE2 inthe conditional media was assessed at harvest using a com-mercially available competitive binding radioimmunoassaykit (NEN Research Products, Boston, MA) as described pre-viously.60
462 TOSATTI ET AL.
Statistical analysis
Because of small but significant variations in reagents,including different batches of coated disks, each separateexperiment had six independent cultures per variable and,as a result, the data set in any one experiment had sufficientpower for analysis. Data were first analyzed by analysis ofvariance; Bonferroni’s modification of Student’s t-test wasused for posthoc testing. p values less than 0.05 were con-sidered significant. In all cases, experiments were performedtwice to ensure the validity of the results. The data presentedare from one of two experiments for each parameter. Bothexperiments yielded comparable results, although absolutebaseline values varied between experiments.
RESULTS
Cell number was regulated by surface chemistry[Fig. 1(a)]. Compared to cultures grown on tissue cul-ture plastic or glass (silicon wafers), MG63 cell num-
ber was reduced on smooth Ti surfaces. Coating thesurface with PLL-g-PEG caused a 50% decrease in cellnumber. This effect was not altered by inclusion of theRDG peptide but the RGD peptide restored cell num-ber to the same levels as seen on tissue culture plastic.
Osteocalcin levels in the conditioned media weremodulated in an opposite manner to cell number [Fig.1(b)]. When compared to tissue culture plastic, osteo-calcin was increased in cultures grown on the siliconwafers and wafers coated with Ti. When cells werecultured on smooth Ti surfaces that were coated withPLL-g-PEG, osteocalcin levels were increased by 100%over the Ti surface alone. The RDG peptide did notalter the PLL-g-PEG effect but the RGD peptide re-duced osteocalcin levels to those seen on the siliconwafers alone.
Alkaline phosphatase-specific activity was also af-fected. Activity in isolated cells [Fig. 1(c)] was de-creased by 50% on surfaces coated with PLL-g-PEG.Inclusion of the RDG peptide had no effect but theRGD peptide partially restored activity to levels seen
Figure 1. Effect of PLL-g-PEG adlayers on osteoblast response to titanium. MG63 osteoblast-like cells were cultured for 7days on functionalized and nonfunctionalized PLL-g-PEG adlayers adsorbed on Ti(metal) coated Si wafers. Experimentalgroups: tissue culture polystyrene (plastic); tissue culture glass (glass); clean Ti(metal) surfaces (Ti); PLL-g-PEG (PEG);PLL-g-PEG/PEG-RDG(1.7) [(PEG-RDG(1.7)]; PLL-g-PEG/PEG-RGD(1.0) [PEG-RGD(1.0)]. The number in brackets followingthe peptide code denotes the peptide surface density in pmol/cm2. (A) Number of MG63 cells released from cultured surfacesby trypsinization. (B) Osteocalcin levels in conditioned media collected at harvest. (C)Alkaline phosphatase-specific activityof isolated MG63 cells. (D) Alkaline phosphatase-specific activity of MG63 cell layer lysates. Values are the mean ' SEM ofsix independent cultures. Data are from one of two separate experiments, both with comparable results. (*)p + 0.05 vs. plastic;(#) p + 0.05 vs. glass; (■) p + 0.05 vs. titanium; (!) p + 0.05 vs. PEG.
RGD-PEPTIDE INHIBITS OSTEOBLAST DIFFERENTIATION ON PEG 463
in cells cultured on PLL-g-PEG alone. The effect ofPLL-g-PEG was also evident when matrix vesicle en-zyme activity was included [Fig. 1(d)]. Cell layer al-kaline phosphatase was reduced by 30% in culturesgrown on smooth Ti and there was a further 50%decrease on smooth Ti surfaces coated with PLL-g-PEG. The RDG peptide had no effect but inclusion ofthe RGD peptide restore enzyme activity to the levelsseen in cell layers on Ti alone.
PLL-g-PEG coatings modulated levels of growthfactors and other mediators in the conditioned media.Growth on smooth Ti had no effect on TGF-!1 com-pared to plastic or glass but there was a sixfold in-crease when MG63 cells were cultured on PLL-g-PEG[Fig. 2(a)]. The RDG peptide did not alter the effect ofPLL-g-PEG but the RGD peptide restored TGF-!1 tolevels observed on Ti alone. The surface effect on PGE2levels was even more pronounced [Fig. 2(b)]. Com-pared to plastic, PGE2 was increased on glass by 260%and on Ti by 490%. However, in cultures grown onPLL-g-PEG PGE2 was elevated 91-fold. The RDG pep-tide reduced the PLL-g-PEG effect by 84%, but levelswere still 13-fold greater than on Ti alone. These sur-face effects were even greater when the cells werecultured on rough Ti SLA samples. Growth on SLAalone reduced cell number; SLA coated with PLL-g-PEG caused a further reduction, which was unaffectedby the RDG peptide [Fig. 3(a)]. In contrast, the RGDpeptide restored cell number to the level observed onplastic. This effect was concentration dependent [Fig.3(b)]. As little as 0.05 pmol/cm2 RGD peptide restoredcell number to approximately 30% of the level on SLAalone. The ability of the RGD peptide to increase cellnumber was partially blocked by inclusion of RGDS inthe culture media of cells grown on PLL-g-PEG/PEG-RGD(0.7) [Fig. 3(c)]. RGDS acted by competitively in-hibiting binding to PEG-RGD because RGDS alone didnot alter the response to PLL-g-PEG at any of theconcentrations tested (Table II).
The stimulatory effect of PLL-g-PEG on osteocalcinlevels was enhanced by surface roughness. PLL-g-PEGcoating caused a four- to fivefold increase in osteocal-cin levels over Ti alone [Fig. 4(a)]. This effect wasreduced by 30% when the RDG peptide was includedin the coating but it was abolished by inclusion of theRGD peptide. The effect of the RGD peptide wasconcentration dependent [Fig. 4(b)]. As little as 0.05pmol/cm2 RGD peptide caused a significant reductionof the PLL-g-PEG-stimulated increase. Inclusion ofRGDS in the culture media partially blocked the in-hibitory effect of RGD peptide on the response ofMG63 cells to PLL-g-PEG [Fig. 4(c)]. RGDS in themedia had no effect on osteocalcin in cultures grownon PLL-g-PEG alone (Table II).
Alkaline phosphatase-specific activity was regu-lated in a similar manner. Cellular enzyme activitywas increased on SLA and further increased by PLL-g-PEG [Fig. 5(a)]. This was unaffected by the RDG
peptide but the RGD peptide reduced activity to thesame level as seen in cells cultured on plastic. Theeffect of RGD peptide was concentration dependent[Fig. 5(b)] and could be partially blocked by inclusionof RGDS in the culture media of cells grown on PLL-g-PEG/PEG-RGD(0.7) [Fig. 5(c)]. Media RGDS alonedid not affect enzyme activity in cells isolated fromcultures grown on PLL-g-PEG alone (Table II). Theeffects of surface chemistry on cell layer alkaline phos-phatase were comparable to those on the cells (datanot shown).
Growth on SLA enhanced the stimulatory effect ofPLL-g-PEG on TGF-!1 levels in the conditioned media[Fig. 6(a)]. PLL-g-PEG increased TGF-!1 by seven- to
Figure 2. Effect of PLL-g-PEG adlayers on TGF-!1 andPGE2 levels in the conditioned media of MG63 cells culturedon smooth Ti surfaces. MG63 osteoblast-like cells were cul-tured for 7 days on functionalized and nonfunctionalizedPLL-g-PEG adlayers adsorbed on Ti(metal)-coated Si wafers.Experimental groups: tissue culture polystyrene (plastic);tissue culture glass (glass); clean Ti(metal) surfaces (Ti);PLL-g-PEG (PEG); PLL-g-PEG/PEG-RDG(1.7) [PEG-RDG(1.7)]; PLL-g-PEG/PEG-RGD(1.0) [PEG-RGD(1.0)]. Thenumber in brackets following the peptide code denotes thepeptide surface density in pmol/cm2. (A) TGF-!1 was as-sayed by ELISA after acid activation. (B) PGE2 productionwas measured using a radioimmunoassay kit. Values are themean ' SEM of six independent cultures. Data are from oneof two separate experiments, both with comparable results.(*)p + 0.05 vs. TCPS; (#)p + 0.05 vs. glass; (■) p + 0.05 vs.titanium; (!) p + 0.05 vs. PEG.
464 TOSATTI ET AL.
eightfold and this was unchanged by the RDG pep-tide. In contrast, it was completely abolished by theRGD peptide. The RGD peptide effect was concentra-tion dependent [Fig. 6(b)]. As little as 0.05 pmol/cm2
RGD peptide reduced the stimulatory effect of PLL-g-PEG by more than 50%. Inclusion of RGDS in themedia partially reversed the effect of the RGD-peptide[Fig. 6(c)]. However, media RGDS alone did not alterthe PLL-g-PEG-induced increase (Table II).
PGE2 was also modulated by PLL-g-PEG but theeffect was less robust in cells cultured on SLA than onsmooth Ti. Whereas the RDG peptide did not alter theincrease in PGE2 due to PLL-g-PEG, the RGD peptidereduced PGE2 to levels like those seen on glass [Fig.7(a)]. The RGD peptide effect was concentration de-pendent [Fig. 7(b)] and sensitive to RGDS blockade[Fig. 7(c)]. In contrast, media RGDS alone did not alterthe PLL-g-PEG effect (Table II).
DISCUSSION
These studies demonstrate for the first time thatPLL-g-PEG coatings have direct effects on osteoblastphysiology. Cell numbers were significantly reducedin cultures grown on Ti surfaces coated with PLL-g-PEG. One possibility is that the coatings were toxic.However, trypan blue staining of the harvested cellsindicated that all cells surviving the 7-day cultureperiod were viable. It is more likely that fewer cellswere able to attach to the PLL-g-PEG-coated surfacesor that adhesion was affected. It is not known if therate of proliferation was sensitive to the surface coat-ing. The number of cells at time of harvest was greaterthan the number of cells inoculated onto the surface,but the time course of the study precluded an assess-ment of whether attachment was reduced and the cellsthat attached proliferated at a normal rate or if the ratewas increased or decreased.
This article demonstrated that inclusion of the RGDpeptide increased cell number, likely by increasingintegrin-mediated signaling because the integrins thatare expressed by osteoblasts on tissue culture plas-tic14,20,62 recognize the RGD motif used in this article.Although all other aspects of the RDG peptide wereidentical to those in the RGD peptide, the RDG pep-tide had no effect. The RDG peptide is not recognizedby the integrins that bind the RGD ligand; thus, theeffect of PLL-g-PEG on cell number was dominant.Together with the results of the competition assays(see below) where addition of soluble RGDS to theculture medium resulted in an important reduction ofthe effect of the surface-bound RGD peptides, theseresults are clear evidence for direct RGD peptide–cellinteraction, even in the presence of serum.
PLL-g-PEG caused marked increases in osteocalcinlevels in the cells grown on smooth Ti, indicating
Figure 3. Effect of PLL-g-PEG adlayers on MG63 cell numberwhen cultured on grit-blasted/acid-etched (SLA CP Ti) sur-faces. (A) Cells were cultured on tissue culture plastic (plastic);tissue culture glass (glass); clean Ti(metal) SLA (SLA CP Ti);SLA coated with PLL-g-PEG (PEG); PLL-g-PEG/PEG-RDG(1.7)[PEG-RDG(1.7)]; PLL-g-PEG/PEG-RGD(1.0) [PEG-RGD(1.0)].The number in brackets following the peptide code denotes thecorresponding peptide surface density in pmol/cm2. (B) Cellswere cultured on plastic; SLA CP Ti; and SLA coated withPLL-g-PEG (PEG) that was either not functionalized [PEG-RGD(0.00)] or functionalized with RGD [PEG-RGD(X)], result-ing in RGD surface densities X of 0.05, 0.26, 1.26, or 6.4 pmol/cm2. (C) Cells were cultured on plastic, SLA CP Ti, or SLA CPTi coated with PEG or PEG-RGD(0.7). In an additional set ofPEG-RGD(0.7) samples, cells were cultured in the presence of45 nM RGDS in solution [PEG-RGD(0.7) , 45 nM RGDS].Values are the mean ' SEM of six independent cultures. Dataare one of two separate experiments, both with comparableresults. (*) p + 0.05 vs. plastic; (!) p + 0.05 vs. glass; (#) p + 0.05vs. SLA; (■) p + 0.05 vs. PEG.
RGD-PEPTIDE INHIBITS OSTEOBLAST DIFFERENTIATION ON PEG 465
expression of a mature osteoblastic phenotype.63 Thisexemplifies the inverse relationship between prolifer-ation and differentiation noted by others with respectto osteogenesis in vitro.64 This interpretation is sup-ported by the change in alkaline phosphatase-specificactivity elicited by the coating. Activity of this enzymeincreases during early stages of differentiation andthen decreases as mineral deposition occurs, coincid-ing with an increase in osteocalcin levels.65,66 Depend-ing on where in the time course the snapshot is taken,alkaline phosphatase and osteocalcin may appear toincrease in parallel, alkaline phosphatase may de-crease as osteocalcin increases, or alkaline phospha-tase may return to constitutive levels and osteocalcinmay be elevated. This article suggests that PLL-g-PEGcaused a significant and rapid increase in differentia-tion and at harvest alkaline-phosphatase specific ac-tivity had already returned to basal levels in thesecultures. The fact that this effect was greater in the celllayer than in the isolated cells supports this interpre-tation as the cell layer would contain any alkalinephosphatase-enriched matrix vesicles released by thecells prior to mineral formation. The hypothesis thatPLL-g-PEG specifically inhibited alkaline phosphataseis unlikely because enzyme activity was stimulated onthe rougher Ti surfaces as discussed below.
PLL-g-PEG also increased the levels of TGF-!1 inthe conditioned media, suggesting that it acted on thecells through mechanisms involving release of growthfactors. It is unlikely that the TGF-!1 was primarilypresent in active form because cell number was low,although it is possible that active TGF-!1 levels werein excess of the concentrations that mediate cell pro-liferation. Arguing against this is the observation thatosteocalcin was increased and TGF-!1 is known toinhibit terminal differentiation of osteoblasts.67
Osteoblast differentiation is associated withPGE2,69,70 suggesting that part of the stimulatory effectof PLL-g-PEG on osteocalcin levels was via a PGE2-dependent mechanism. MG63 cells generate very lowlevels of PGE2 when cultured under conventional con-ditions.53 Enhanced differentiation of osteoblasts onsurfaces with rougher microtopographies, like the TiSLA used in this article, is associated with increased
PGE2 production and inhibition of prostaglandin pro-duction blocks the stimulatory effect of the microto-pography. Our results suggest that a similar mecha-nism is involved. When the cells were cultured onPLL-g-PEG/PEG-RGD, greater integrin binding waspotentiated, PGE2 levels were restored to those seenon smooth uncoated surfaces, and differentiation ofthe cells was also typical of that seen on the smoothuncoated surfaces.
The effect of PLL-g-PEG was sensitive to surfacemorphology, both in terms of rate and magnitude ofcellular response. Growth of MG63 cells on roughermicrotopographies enhances differentiation in itself.54
Thus, the differentiation elicited by PLL-g-PEG waspotentiated by the fact that the responding cells werealready at a further state of maturation in the osteo-blast lineage cascade. The reduction in cell numberwas greater and the increase in osteocalcin levels wasgreater. While PLL-g-PEG decreased alkaline phos-phatase-specific activity in MG63 cells grown onsmooth surfaces, on rough surfaces enzyme activitywas increased to a comparable extent in the cells andin the cell layer. This supports our previous observa-tions that alkaline phosphatase and osteocalcin areindependently regulated.71 Moreover, on the roughersurface PLL-g-PEG induced a greater increase inTGF-!1 and a less robust increase in PGE2. It is im-portant to note, however, that the absolute amounts ofboth local mediators were greater in cultures grownon SLA than on smooth Ti. Whether this was due tothe higher exposure to PLL-g-PEG as a consequence ofthe increased surface area on the rougher surface isnot known.
Regardless of surface morphology, incorporation ofthe RGD peptide restored the behavior of the cells tothat seen on smooth surfaces, in some cases abolishingthe effect of Ti in comparison with tissue culture plas-tic or glass. Much of our understanding of the mech-anism of RGD action is derived from experiments thatare conducted using conventional cell culture meth-odology. Clearly, the concentration of the peptide andits spacing is important in determining how cells at-tach, spread, and differentiate in culture.23,26,72,73 Theresults of this article support this. As the concentration
TABLE IIEffect of RGDS on Phenotypic Expression of MG63 Osteoblast-Like Osteosarcoma Cells Cultured on
*Results are statistically relevant (p + 0.05) in comparison to PLL-g-PEG.
466 TOSATTI ET AL.
Figure 4. Effect of PLL-g-PEG adlayers on osteocalcin levelsin the conditioned media of MG63 cells cultured on grit-blast-ed/acid-etched (SLA CP Ti) surfaces. (A) Cells were culturedon tissue culture plastic (plastic); tissue culture glass (glass);clean Ti(metal) SLA (SLA CP Ti); SLA coated with PLL-g-PEG(PEG); PLL-g-PEG/PEG-RDG(1.7) [PEG-RDG(1.7)]; PLL-g-PEG/PEG-RGD(1.0) [PEG-RGD(1.0)]. The number in bracketsfollowing the peptide code denotes the corresponding peptidesurface density in pmol/cm2. (B) Cells were cultured on plas-tic; SLA CP Ti; and SLA coated with PLL-g-PEG (PEG) that waseither not functionalized [PEG-RGD(0.00)] or functionalizedwith RGD [PEG-RGD(X)], resulting in RGD surface densities Xof 0.05, 0.26, 1.26, or 6.4 pmol/cm2. (C) Cells were cultured onplastic, SLA CP Ti, or SLA CP Ti coated with PEG or PEG-RGD(0.7). In an additional set of PEG-RGD(0.7) samples, cellswere cultured in the presence of 45 nM RGDS in solution[PEG-RGD(0.7) , 45 nM RGDS]. Values are the mean ' SEMof six independent cultures. Data are one of two separateexperiments, both with comparable results. (*)p + 0.05 vs.plastic; (!) p + 0.05 vs. glass; (#) p + 0.05 vs. SLA; (■) p + 0.05vs. PEG.
Figure 5. Effect of PLL-g-PEG adlayers on alkaline phos-phatase-specific activity in homogenates of MG63 cells isolatedfrom cultures grown on grit-blasted/acid-etched (SLA CP Ti)surfaces. (A) Cells were cultured on tissue culture plastic (plas-tic); tissue culture glass (glass); clean Ti(metal) SLA (SLA CPTi); SLA coated with PLL-g-PEG (PEG); PLL-g-PEG/PEG-RDG(1.7) [PEG-RDG(1.7)]; PLL-g-PEG/PEG-RGD(1.0) [PEG-RGD(1.0)]. The number in brackets following the peptide codedenotes the corresponding peptide surface density in pmol/cm2. (B) Cells were cultured on plastic; SLA CP Ti; and SLAcoated with PLL-g-PEG (PEG) that was either not functional-ized [PEG-RGD(0.00)] or functionalized with RGD [PEG-RGD(X)], resulting in RGD surface densities X of 0.05, 0.26,1.26, or 6.4 pmol/cm2. (C) Cells were cultured on plastic, SLACP Ti, or SLA CP Ti coated with PEG or PEG-RGD(0.7). In anadditional set of PEG-RGD(0.7) samples, cells were cultured inthe presence of 45 nM RGDS in solution [PEG-RGD(0.7) , 45nM RGDS]. Values are the mean ' SEM of six independentcultures. Data are one of two separate experiments, both withcomparable results. (*)p + 0.05 vs. plastic; (!) p + 0.05 vs. glass;(#)p + 0.05 vs. SLA; (■) p + 0.05 vs. PEG.
RGD-PEPTIDE INHIBITS OSTEOBLAST DIFFERENTIATION ON PEG 467
Figure 6. Effect of PLL-g-PEG adlayers on TGF-!1 levels inthe conditioned media of MG63 cells cultured on grit-blasted/acid-etched (SLA CP Ti) surfaces. (A) Cells were cultured ontissue culture plastic (plastic); tissue culture glass (glass); cleanTi(metal) SLA (SLA CP Ti); SLA coated with PLL-g-PEG (PEG);PLL-g-PEG/PEG-RDG(1.7) [PEG-RDG(1.7)]; PLL-g-PEG/PEG-RGD(1.0) [PEG-RGD(1.0)]. The number in brackets followingthe peptide code denotes the corresponding peptide surfacedensity in pmol/cm2. (B) Cells were cultured on plastic; SLACP Ti; and SLA coated with PLL-g-PEG [PEG] that was eithernot functionalized [PEG-RGD(0.00)] or functionalized withRGD [PEG-RGD(X)], resulting in RGD surface densities X of0.05, 0.26, 1.26, or 6.4 pmol/cm2. (C) Cells were cultured onplastic, SLA CP Ti, or SLA CP Ti coated with PEG or PEG-RGD(0.7). In an additional set of PEG-RGD(0.7) samples, cellswere cultured in the presence of 45 nM RGDS in solution[PEG-RGD(0.7) , 45 nM RGDS]. Values are the mean ' SEMof six independent cultures. Data are one of two separateexperiments, both with comparable results. (*)p + 0.05 vs.plastic; (!) p + 0.05 vs. glass; (#) p + 0.05 vs. SLA; (■) p + 0.05vs. PEG.
Figure 7. Effect of PLL-g-PEG adlayers on PGE2 levels in theconditioned media of MG63 cells cultured on grit-blasted/acid-etched (SLA CP Ti) surfaces. (A) Cells were cultured ontissue culture plastic (plastic); tissue culture glass (glass); cleanTi(metal) SLA (SLA CP Ti); SLA coated with PLL-g-PEG (PEG);PLL-g-PEG/PEG-RDG(1.7) [PEG-RDG(1.7)]; PLL-g-PEG/PEG-RGD(1.0) [PEG-RGD(1.0)]. The number in brackets followingthe peptide code denotes the corresponding peptide surfacedensity in pmol/cm2. (B) Cells were cultured on plastic; SLACP Ti; and SLA coated with PLL-g-PEG [PEG] that was eithernot functionalized [PEG-RGD(0.00)] or functionalized withRGD [PEG-RGD(X)], resulting in RGD surface densities X of0.05, 0.26, 1.26, or 6.4 pmol/cm2. (C) Cells were cultured onplastic, SLA CP Ti, or SLA CP Ti coated with PEG or PEG-RGD(0.7). In an additional set of PEG-RGD(0.7) samples, cellswere cultured in the presence of 45 nM RGDS in solution[PEG-RGD(0.7) , 45 nM RGDS]. Values are the mean ' SEMof six independent cultures. Data are one of two separateexperiments, both with comparable results. (*)p + 0.05 vs.plastic; (!) p + 0.05 vs. glass; (#)p + 0.05 vs. SLA; (■) p + 0.05vs. PEG.
468 TOSATTI ET AL.
of RGD peptide was increased the effect of PLL-g-PEGwas reduced. For most parameters, low levels of thepeptide were sufficient to restore cell adhesion andspreading, accomplish a complete reduction in theeffect of PLL-g-PEG, and in some instances abolish theeffect of surface roughness.
The form of the peptide may be an important vari-able. The peptide used in this article was designedbased on several experimental requirements. It in-cluded a covalent linker to the PLL-g-PEG and alonger PEG chain as a spacer to ensure that the cellhad access to the RGD motif. Some of the effect of thepeptide was independent of RGD per se as the RDGpeptide caused a partial reduction of the MG63 cellresponse to PLL-g-PEG-coated SLA surfaces, at leastwith respect to osteocalcin, TGF-!1, and PGE2.
In the competitive inhibition assays not all of theeffect of the RGD peptide was blocked by the inclu-sion of RGDS in the media at fivefold excess over theamount of RGD peptide on the surface (correspondingto a maximum concentration of 45 pmol/ml or 45 nMRGDS in solution). This observation is likely to be aconsequence of the soluble peptide concentration cho-sen, which may not have been high enough to com-pletely inhibit binding of cell receptors to surface-immobilized RGD. High soluble RGD peptideconcentration in the media (e.g., 100 &M to 1 mM) is ingeneral necessary to largely inhibit the effect of sur-face-bound peptides.74,75 Even 1 mM GRGDS onlyinhibited 60% of MG63 cell adhesion to recombinanthuman fibronectin peptide (hFNIII9-10).76 It is alsopossible that the integrin binding RGDS does not me-diate some of the inhibition of the response to PLL-g-PEG caused by the RGD peptide.
Previous studies showed that PLL-g-PEG coatingsare protein resistant, preventing adsorption of 99.5%of serum proteins even after 10 days of incubation.37,77
This may explain why fewer cells were present on thePLL-g-PEG-coated surfaces and suggests that cell at-tachment to the RGD peptide-modified surface wasvia an RGD-dependent mechanism. Some protein (upto 5 ng/cm2) does adsorb on PLL-g-PEG-coated TiO2,however, and other serum and medium componentsmay be present on the surfaces that are involved in cellattachment, proliferation, and differentiation of thecells. The small amount of serum components embed-ded in the PLL-g-PEG structure may contribute toosteoblast attachment in the absence of synthetic RGDpeptide at the surface. The PLL-g-PEG-adsorbed massis known to be somewhat lower on TiO2 in compari-son to SiO2 and Nb2O5 substrates,35,36 resulting in amore open PEG brush structure in the former case.Such “open” PEG brush structures have been shownnot only to adsorb small amounts of proteins but alsoto better preserve their active conformation, prevent-ing unfolding and denaturing of the proteinaceousmolecules. Spatially separated, native proteins areprobably interacting with receptors in cell membranes
in a particularly efficient way, attaching cells withoutinducing spreading. For example, affinity of surface-immobilized, active fibronectin for integrin receptorshas been found to be 10–100 times stronger comparedto short, linear peptides containing the RGD se-quence.78 For the sample series with increasing surfacedensities of grafted RGD peptides investigated here,this “osteogenic” effect may be more and more over-ridden by the effect of RGD resulting in increasingdegrees of spreading as has been observed for a num-ber of cell types,8,79,80 including osteoblasts and chon-drocytes.76,81
The changes we see in osteoblast behavior from noRGD peptide grafting characterized by low cell num-ber but high phenotype expression to a high RGDsurface concentration with high cell number but re-duced phenotypic expression is not entirely new. Sim-ilar trends have been reported for chondrocytes onRGD peptide-grafted polymer surfaces74 and forsmooth muscle cells and fibroblasts on RGD peptide-coated glass surfaces,82 where increasing RGD peptideconcentration resulted in accelerated cell adhesion andspreading while at the same time matrix productionwas reduced. Similarly, in a study of RGD peptide(0.2–20 pmol/cm2)-modified titanium surfaces Barberet al.29 concluded that maximum spreading of osteo-blasts and high proliferation rate do not seem topresent the optimum condition for early differentia-tion and phenotype expression. It was speculated thata discrete peptide ligand density might exist that bestsupports cell development and improves bone-mate-rial interfacial strength and integration in osseous tis-sue.
Because increase in RGD peptide surface concentra-tion results in a change of cell shape from essentiallyspherical at low to highly spread at high RGD peptidesurface densities,26,81 an interesting question iswhether the changing regime in proliferation/pheno-type expression has to do with cell shape only. Cellstudies on patterned surfaces, where RGD peptidesurface densities and cell shape can be tailored inde-pendently, are one way to approach this question. Thework by Chen et al.81 provides some strong argumentsthat indeed cell shape is a major factor governing cellgrowth, viability, or death. PLL-g-PEG does not ap-pear to exert its effect through a pathway initiated byRGD binding. RGDS had no effect on the response ofMG63 cells to PLL-g-PEG-coated surfaces. The resultsof this article support the hypothesis that RGD bind-ing may reduce differentiation of osteoblasts to anextent conducive to proliferation rather than stimulat-ing differentiation.
The authors thank Wanda Whitfield for her help in thepreparation of the article. This research was financially sup-ported by the International Team for Oral Implantology(ITI), Waldenburg, Switzerland, the Swiss National ScienceFoundation/National Research Program NRP 47 “Supramo-
RGD-PEPTIDE INHIBITS OSTEOBLAST DIFFERENTIATION ON PEG 469
lecular Functional Materials,” ETH Zurich, the GeorgiaTech/Emory Center for the Engineering of Living Tissues(Atlanta, GA), and the Georgia Research Alliance.
References
1. Buser D. Titanium for dental applications (II). In: Brunette DM,Tengvall P, Textor M, Thomsen P, editors. Titanium in medi-cine. Berlin: Springer-Verlag; 2001. p 875–888.
2. Lincks J, Boyan BD, Blanchard CR, Lohmann Ch, Liu Y, Coch-ran DL, Dean DD, Schwartz Z. Response of MG63 osteoblast-like cells to titanium and titanium alloy is dependent on sur-face roughness and composition. Biomaterials 1998;19:2219–2232.
3. Sinha RK, Morris F, Shah SA, Tuan RS. Surface composition oforthopaedic implant metals regulates cell attachment, spread-ing, and cytoskeletal organization of primary human osteo-blasts in vitro. Clin Orthop Rel Res 1994;305:258–272.
4. Puleo DA, Bizios R. Formation of focal contacts by osteoblastscultured on orthopedic biomaterials. J Biomed Mater Res 1992;26:291–301.
5. Windeler AS, Bonewald LF, Khare AG, Boyan BD, Mundy GR.The influence of sputtered bone substitutes on cell growth andphenotypic expression. In: Davies JE, editor. The bone–bioma-terial interface. Toronto: Toronto Press; 1991. p 205–13.
6. Hayman EG, Pierschbacher MD, Ohgren Y, Ruoslahti E. Serumspreading factor (vitronectin) is present at the cell surface andin tissues. Proc Natl Acad Sci USA 1983;80:4003–4007.
7. Pierschbacher MD, Ruoslahti E. Cell attachment activity offibronectin can be duplicated by small synthetic fragments ofthe molecule. Nature 1984;309:30–33.
9. Massia SP, Hubbell JA. Human endothelial cell interactionswith surface-coupled adhesion peptides on a nonadhesiveglass substrate and two polymeric biomaterials. J Biomed Ma-ter Res 1991;25:223–242.
10. Dee KC, Andersen TT, Bizios R. Design and function of novelosteoblast-adhesive peptides for chemical modification of bio-materials. J Biomed Mater Res 1998;40:371–377.
11. Ruoslahti E, Pierschbacher MD. New perspectives in cell ad-hesion: RGD and integrins. Science 1987;238:491–497.
12. Hubbell JA. Biomaterials in tissue engineering. Biotechnology2003;13:565–576.
13. Healy KE. Molecular engineering of materials for bioreactivity.Curr Opin Solid State Mater Sci 1999;4:381–387.
14. Sinha RK, Tuan RS. Regulation of human osteoblast integrinexpression by orthopaedic implant materials. Bone 1996;18:451–457.
15. Pearson BS, Klebe RJ, Boyan BD, Moskowicz D. Comments onthe clinical application of fibronectin in dentistry. J Dent Res1988;67:515–517.
16. Eble JA, Kuhn K. Integrin–ligand interaction. New York:Springer; 1997.
17. Harmenberg J, Hoglund M, Hellstrom-Lindberg E. G- andGM-CSF in oncology and oncological haematology. Eur JHaematol 1994;55(suppl):1–28.
18. Albelda SM, Buck CA. Integrins and other cell adhesion mol-ecules. FASEB J 1990;4:2868–2880.
19. Akiyama SK. Integrins in cell adhesion and signaling. HumanCell 1996;9:181–186.
20. Hughes DE, Salter DM, Dedhar S, Simpson R. Integrin expres-sion in human bone. J Bone Mineral Res 1993;8:527–533.
21. Hautanen A, Gailit J, Mann DM, Ruoslahti E. Effects of mod-ifications of the RGD sequence and its context on recognitionby the fibronectin receptor. J Biol Chem 1989;264:1437–442.
22. Ochsenhirt SE, Tirrell M, Fields GB, Dillow AK. Use of sup-ported thin films of peptide amphiphiles as model systems ofthe ECM to study the effect of structure–function phenomenaon cell adhesion. In: Dillow AK, Lowman AM, editors. Biomi-metic materials and design. New York: Marcel-Dekker; 2002. p1–28.
23. Garcia AJ, Collard DM, Keselowsky BG, Cutler SM, GallantND, Byers BA, Stephansson SN. Engineering of integrin-spe-cific biomimetic surfaces to control cell adhesion and function.In: Dillow AK, Lowman AM, editors. Biomimetic materialsand design. New York: Marcel-Dekker; 2002. p. 29–55.
24. Irvine DJ, Ruzette AVG, Mayes AM, Griffith LA. Nanoscaleclustering of RGD peptides at surfaces using comb polymers. 2.Surface segragation of comb polymers in polylactide. Biomac-romolecules 2001;2:545–556.
26. Massia SP, Hubbell JA. An RGD spacing of 440 nm is sufficientfor integrin alpha V beta 3-mediated fibroblast spreading and140 nm for focal contact and stress fiber formation. J Cell Biol1991;114:1089–1100.
27. Rezania A, Johnson R, Leftkow AR, Healy KE. Bioactivation ofmetal oxide surfaces. Langmuir 1999;15:6931–6939.
28. Rezania A, Healy KE. The effect of peptide surface density onmineralization of a matrix deposited by osteogenic cells.J Biomed Mater Res 2000;52:595–600.
29. Barber TA, Golledge SL, Castner DG, Healy KE. Peptide-mod-ified IPNs grafted to bulk titanium modulated osteoblast be-havior in vitro. J Biomed Mater Res 2003;64:38–47.
33. Jeon SI, Lee JH, Andrade JD, Degennes PG. protein surfaceinteractions in the presence of polyethylene oxide .1. Simplifiedtheory. J Colloid and Interface Sci 1991;142:149–158.
34. Bjorling M. Interaction between surfaces with attached poly-(ethylene oxide) chains. Macromolecules 1992;25:3956–3970.
35. Kenausis GL, Voros J, Elbert D, Huang NP, Hofer R, Ruiz-Taylor L, Textor M, Hubbell JA, Spencer ND. Poly(l-lysine)-g-poly(ethylene glycol) layers on metal oxide surfaces: Attach-ment mechanism and effects of polymer architecture onresistance to protein adsorption. J Phys Chem B 2000;104:3298–3309.
36. Huang NP, Michel R, Voros J, Textor M, Hofer R, Rossi A,Elbert DL, Hubbell JA, Spencer ND. Poly(l-lysine)-g-poly(eth-ylene glycol) layers on metal oxide surfaces: Surface-analyticalcharacterization and resistance to serum and fibrinogen ad-sorption. Langmuir 2001;17:489–498.
38. Textor M, Tosatti S, Wieland M, Brunette DM. Use of molec-ular assembly techniques for tayloring the chemical propertieson smooth and rough surfaces. Boca Raton, FL: CRC Press;2003.
39. Tosatti S, De Paul SM, Askendal A, Vande Vondele S, HubbellJA, Tengvall P, Textor M. Peptide functionalized poly(l-lysine-
470 TOSATTI ET AL.
g-poly (ethylene glycol) on titanium: resistance to protein ad-sorption in full heparinized human blood plasma. Biomaterials2003;24:4949–4958.
40. Voros J, Ramsden JJ, Csucs G, Szendro I, Textor M, SpencerND. Biotin-derivatized poly(l-lysine)-g-poly(ethylene glycol):A novel polymeric interface for bioaffinity sensing. Langmuir2003;18:220–230.
41. Schwartz Z, Lohmann CH, Wieland M, Cochran DL, Dean DD,Textor M, Bonewald LF, Boyan BD. Osteoblast proliferationand differentiation on dentin slices are modulated by pretreat-ment of the surface with tetracycline or osteoclasts. J Periodon-tol 2000;71:622–633.
42. Boyan BD, Schwartz Z, Lohmann CH, Sylvia VL, Cochran DL,Dean DD, Puzas JE. Pretreatment of bone with osteoclastsaffects osteoblast phenotypic expression. J Orthoped Res 2003;21:638–647.
43. Tosatti S, Michel R, Textor M, Spencer ND. Self-assembledmonolayers of dodecyl and hydroxy-dodecyl phosphates onboth smooth and rough titanium and titanium oxide surfaces.Langmuir 2002;18:3537–3548.
44. Wieland M, Hanggi P, Hotz W, Textor M, Keller BA, SpencerND. Wavelength-dependent measurement and evaluation ofsurface topographies: Application of a new concept of windowroughness and surface transfer function. Wear 2000;237:231–252.
45. Lohmann CH, Tandy EM, Sylvia VL, Hell-Vocke AK, CochranDL, Dean DD, Boyan BD, Schwartz Z. Response of normalfemale human osteoblasts (NHOst) to 17!-estradiol is modu-lated by implant surface morphology. J Biomed Mater Res2002;62:204–213.
46. Contu F, Elsener B, Bohni H. Characterization of implant ma-terials in fetal bovine serum and sodium sulfate by electro-chemical impedance spectrospcopy. J Biomed Mater Res 2002;62:412–421
47. Kurrat R, Textor M, Ramsden JJ, Boni P, Spencer ND. Instru-mental improvements in optical waveguide light mode spec-troscopy for the study of biomolecule adsorption. Rev SciInstrum 1997;67:2172–2176.
48. Franceschi RT, James WM, Zerlauth G. 1#,25-dihydroxyvita-min D3 specific regulation of growth, morphology, and fi-bronectin in a human osteosarcoma cell line. J Cell Physiol1985;123:401–409.
49. Boyan BD, Schwartz Z, Bonewald LF, Swain LD. Localizationof 1,25-(OH)2D3 responsive alkaline phosphatase in osteoblast-like cells (ROS 17/2.8, MG 63, and MC 3T3) and growthcartilage cells in culture. J Biol Chem 1989;264:11879–11886.
50. Geivelis M, Turner DW, Pederson ED, Lamberts BL. Measure-ments of interleukin-6 in gingival crevicular fluid from adultswith destructive periodontal disease. J Periodontol 1993;64:980–983.
51. Kieswetter K, Schwartz Z, Hummert TW, Cochran DL, Simp-son J, Dean DD, Boyan BD. Surface roughness modulates thelocal production of growth factors and cytokines by osteoblast-like MG63 cells. J Biomed Mater Res 1996;32:55–63.
52. Boyan BD, Lohmann CH, Sisk M, Liu Y, Sylvia VL, CochranDL, Dean DD, Schwartz Z. Both cyclooxygenase-1 and cyclo-oxygenase-2 mediate osteoblast response to titanium surfaceroughness. J Biomed Mater Res 2001;55:350–359.
53. Bonewald LF, Kester MB, Schwartz Z, Swain LD, Khare AG,Johnson TL, Leach RJ, Boyan BD. Effects of combining trans-forming growth factor ! and 1,25-dihydroxyvitamin D3 ondifferentiation of a human osteosarcoma (MG-63). J Biol Chem1992;267:8943–8949.
54. Martin JY, Schwartz Z, Hummert TW, Schraub DM, Simpson J,Lankford J, Dean DD, Cochran DL, Boyan BD. Effect of tita-nium surface roughness on proliferation, differentiation, andprotein synthesis of human osteoblast-like cells (MG63).J Biomed Mater Res 1995;29:389–401.
55. Martin JY, Dean DD, Cochran DL, Simpson J, Boyan BD,Schwartz Z. Proliferation, differentiation, and protein synthe-sis of human osteoblast-like cells (MG63) cultured on previ-ously used titanium surfaces. Clin Oral Implants Res 1996;7:27–37.
56. Boyan BD, Batzer R, Kieswetter K, Liu Y, Cochran DL, Szmuck-ler-Moncler S, Dean DD, Schwartz Z. Titanium surface rough-ness alters responsiveness of MG63 osteoblast-like cells to1#,25-(OH)2D3. J Biomed Mater Res 1998;39:77–85.
57. Hale LV, Kemick ML, Wuthier RE. Effect of vitamin D metab-olites on the expression of alkaline phosphatase activity byepiphyseal hypertrophic chondrocytes in primary cell culture.J Bone Miner Res 1986;1:489–495.
58. Bretaudiere JP, Spillman T. Alkaline phosphatases. In: Berg-meyer HU, editor. Methods of enzymatic analysis. Weinheim,Germany: Verlag Chemica; 1984. p 75–92.
59. Sisk M, Lohmann CH, Cochran DL, Sylvia VL, Simpson JP,Dean DD, Boyan BD, Schwartz Z. Inhibition of cyclooxygenaseby indomethacin modulates osteoblast response to titaniumsurface roughness in a time-dependent manner. Clin Oral Im-plants Res 2001;12:52–61.
60. Schwartz Z, Dennis R, Bonewald LF, Swain LD, Gomez R,Boyan BD. Differential regulation of prostaglandin E2 synthe-sis and phospholipase A2 activity by 1,25-(OH)2D3 in threeosteoblast-like cell lines (MC-3T3-E1, ROS 17/2.8, and MG-63).Bone 1992;13:51–58.
61. Saito T, Albelda SM, Brighton CT. Identification of integrinreceptors on cultured human bone cells. J Orthoped Res 1994;12:384–394.
62. Stein GS, Lian JB, Owen TA. Relationship of cell growth to theregulation of tissue-specific gene expression during osteoblastdifferentiation. FASEB J 1990;4:3111–3123.
63. Lian JB, Stein GS. Concepts of osteoblast growth and differen-tiation: Basis for modulation of bone cell development andtissue formation. Crit Rev Oral Biol Med 1992;3:269–305.
64. Bellows CG, Aubin JE, Heersche JN, Antosz ME. Mineralizedbone nodules formed in vitro from enzymatically released ratcalvaria cell populations. Calcif Tissue Int 1986;38:143–154.
65. Harris SE, Bonewald LF, Harris MA, Sabatini M, Dallas SL,Feng JQ, Ghosh-Choudhury N, Wozney JM, Mundy GR. Ef-fects of transforming growth factor ! on bone nodule forma-tion and expression of bone morphogenetic protein 2, osteo-calcin, osteopontin, alkaline phosphatase, and type I collagenmRNA in long-term cultures of fetal rat calvarial osteoblasts.J Bone Miner Res 1994;9:855–863.
66. Bonewald LF, Mundy GR. Role of transforming growth factor-beta in bone remodeling. Clin Orthoped Rel Res 1990;250:261–276.
67. Dziak RM. Prostaglandins as mediators of bone cell metabo-lism. In: Rubin RP, Weiss GB, Putney JW, editors. Calcium inbiological systems. New York: Plenum Publishing; 1985. p533–539.
68. Raisz LG, Fall PM. Biphasic effects of prostaglandin E2 on boneformation in cultured fetal rat calvariae: Interaction with cor-tisol. Endocrinology 1998;126:1654–1659.
69. Lohmann CH, Bonewald LF, Sisk MA, Sylvia VL, CochranDL, Dean DD, Boyan BD, Schwartz Z. Maturation statedetermines the response of osteogenic cells to surface rough-ness and 1,25-dihydroxyvitamin D3. J Bone Miner Res 2000;15:1169 –1180.
70. Maheshwari G, Brown G, Lauffenburger DA, Wells A, GriffithLG. Cell adhesion and motility depend on nanoscale RGDclustering. J Cell Sci 2000;113:1677–1686.
71. Healy KE, Harbers GM, Barber TA, Sumner DR. Osteoblastinteractions with engineered surfaces. In: Davies JE, editor.Bone engineering. Toronto: em squared inc.; 2000. p 268–281.
72. Jeschke B, Meyer J, Jonczyk A, Kessler H, Adamietz P, MeenenNM, Kantlehner M, Goepfert C, Niles B. RGD-peptides for
RGD-PEPTIDE INHIBITS OSTEOBLAST DIFFERENTIATION ON PEG 471
tissue engineering of articular cartilage. Biomaterials 2002;23:3455–3463.
73. Danilov YN, Juliano RL. (Arg-Gly-Asp)n-albumin conjugatesas a model substratum for integrin-mediated cell-adhesion.Exp Cell Res 1989;182:186–196.
74. Jang JH, Ku Y, Chung CP, Heo SJ. Enhanced fibronectin-mediated cell adhesion of human osteoblast by fibroblastgrowth factor, FGF-2. Biotech Lett 2002;24:1659–1663.
75. Michel R, Lussi JW, Csucs G, Reviakine I, Danuser G, KettererB, Hubbell JA, Textor M, Spencer ND. Selective molecularassembly patterning: A new approach to micro- and nano-chemical patterning of surfaces for biological applications.Langmuir 2002;18:3281–3287.
76. Koo LY, Irvine DJ, Mayes AM, Lauffenburger DA, Griffith LG.Co-regulation of cell adhesion by nanoscale RGD organizationand mechanical stimulus. J Cell Sci 2002;115:1423–1433.
77. Olbrich KC, Andersen TT, Blumenstock FA, Bizios R. Surfaces
modified with covalently-immobilized adhesive peptides affectfibroblast population motility. Biomaterials 1996;17:759–764.
78. Cook AD, Hrkach JS, Gao NN, Johnson IM, Pajvani UB, Can-nizzaro SM, Langer R. Characterization and development ofRGD-peptide-modified poly(lactic acid-co-lysine) as an inter-active, resorbable biomaterial. J Biomed Mater Res 1997;35:513–523.
79. Harbers GM, Barber TA, Stile RA, Sumner DR, Healy KE.Mimetic peptide-modified materials for control of cell differ-entiation. In: Dillow AK, Lowman AM, editors. Biomimeticmaterials and design. New York: Marcel Dekker; 2002. p 55–90.
80. Mann BK, Tsai AT, Scott-Burden T, West JL. Modification ofsurfaces with cell adhesion peptides alters extracellular matrixdeposition. Biomaterials 1999;20:2281–2286.
81. Chen CS, Mrksich M, Huang S, Whitesides GM, Ingber DE.Geometric control of cell life and death. Science 1997;276:1425–1428.
![Acoustical investigation of poly(dA).poly(dT), poly[d(A-T)].poly[d(A-T)], poly(A) . poly(U) and DNA hydration in dilute aqueous solutions](https://static.documents.pub/doc/80x56/63455bae03a48733920b53d8/acoustical-investigation-of-polydapolydt-polyda-tpolyda-t-polya.jpg)
