Influence of collagen-fibril-based coatings containing decorin and biglycan on osteoblast behavior

Influence of collagen-fibril-based coatings containingdecorin and biglycan on osteoblast behavior

Timothy Douglas,1 Ute Hempel,2 Carolin Mietrach,2 Manuela Viola,3 Davide Vigetti,3

Sascha Heinemann,1 Susanne Bierbaum,1 Dieter Scharnweber,1 Hartmut Worch1

1Institute of Material Science, Max Bergmann Center of Biomaterials, Technische Universitat Dresden,Budapester Strasse 27, 01069 Dresden, Germany2Institute of Physiological Chemistry, Medical Faculty Carl Gustav Carus, Technische Universitat Dresden,Fiedlerstrasse 42, 01307 Dresden, Germany3Department of Experimental and Clinical Biomedical Sciences, Universita dell’Insubria, Via H.J. Dunant 5,21100 Varese, Italy

Received 7 November 2006; revised 27 January 2007; accepted 3 April 2007Published online 17 July 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/jbm.a.31501

Abstract: Collagen is used as a scaffold material for tissueengineering as well as a coating material for implants with aview to enhancing osseointegration by mimicry of the boneextracellular matrix in vivo. The biomimicry strategy can betaken further by incorporating the small leucine-rich proteo-glycans (SLRPs) decorin and biglycan, which are expressedin bone. Both bind to fibrils during fibrillogenesis in vitro. Inthis study, the ability of collagen types I, II, and III to binddecorin and biglycan was compared. Collagen type II boundsignificantly more SLRPs in fibrils than collagen I and III,with more biglycan than decorin bound by all three collagentypes. Therefore, type II fibrils with bound decorin or bigly-can or neither were used to coat titanium surfaces. Bioavail-ability of SLRPs was confirmed by direct ELISA after SLRPbiotinilation. The in vitro behavior of osteoblasts from ratcalvaria (rOs) and human knee (hOs) cultured on different

surfaces was compared. Proliferation and collagen synthesiswere determined. Also, the influence of SLRPs on the for-mation of focal adhesions by rO was investigated. Biglycanenhanced the formation of focal adhesions after 2 and 24 h.Decorin and biglycan affected rO and hO proliferation andcollagen synthesis differently. Biglycan stimulated hO pro-liferation significantly but had no effect on rO proliferation,and also inhibited rO collagen synthesis significantly whilenot affecting hO collagen synthesis. Decorin promoted hOproliferation slightly but did not influence rO proliferation.The results could be relevant when designing implant coat-ings or tissue engineering scaffolds. � 2007 Wiley Periodi-cals, Inc. J Biomed Mater Res 84A: 805–816, 2008

Key words: collagen type II; decorin; biglycan; osteoblast;titanium

INTRODUCTION

Titanium and its alloys are widely used as ortho-pedic materials due to their high biocompatibility,but require stable fixation in the surrounding boneto be successful. Surface modifications which mimicthe extracellular matrix (ECM) of bone by coatingimplants with collagen I, its main organic compo-nent, have improved osteoblast attachment, prolifer-ation, and differentiation in vitro, while increasingearly bone remodeling, bone contact, and formationof surrounding bone in vivo.1–8 Collagen types I, II,and III, whose molecules consist of an uninterrupted

triple helix of � 300 nm in length and 1.5 nm in di-ameter, form fibrils by self-assembly from collagenmolecules and demonstrate a cross-striated, D-peri-odic banding pattern, where D ¼ 67 nm.9 Collagentype I is the most abundant type in mammals, and isthe type present in mineralized bone, but duringbone development and repair, collagen types II andIII are also expressed.10,11

The strategy of mimicking bone ECM can be takena step further by incorporating the small leucine-richproteoglycans (SLRPs) decorin and biglycan whichare found in bone12,13 and are made up of a proteincore substituted with glycosaminoglycan (GAG)chains consisting of either chondroitin sulphate(CS6 and CS4, also known as CS A and CS C, respec-tively) or dermatan sulphate (DS, also known as CSB); decorin possesses one chain and biglycan two.14

In bone, these chains have been reported to consistentirely of CS4,15 whereas in bovine articular

Correspondence to: T. Douglas; e-mail: [email protected] grant sponsor: German Ministry for Education

and Research (BMBF)

' 2007 Wiley Periodicals, Inc.

cartilage, a mixture of CS4 and DS is present.16 Bothdecorin and biglycan have been expressed in bonecell culture in vivo17 and both are believed to playroles in mineralization.18 Knock-out mice defectivein biglycan and decorin show a severe reduction inbone mass.19

Surfaces coated with collagen type I fibrils withbound decorin showed significantly accelerated andenhanced formation of focal adhesions by osteoblastscompared with those without decorin.20 In addition,studies have reported that decorin and biglycan areable to bind to and modulate the activity of thegrowth factors TGF-beta 1, which has been reportedto influence osteoblast proliferation, induce the syn-thesis of collagen and other ECM proteins, and causeosteoblast chemotaxis, and BMP-4, which enhancesosteoblast differentiation.21–26 Therefore, as well asinfluencing cell behavior directly, decorin and bigly-can could potentially function as growth factor reser-voirs or modulators.

The aims of this study were (i) to select one of thecollagen types I, II, and III on the basis of the abilityof fibrils to bind decorin and biglycan as determinedby colormetric assay and to confirm the bioavailabil-ity of fibril-bound decorin and biglycan; (ii) to studyand compare the proliferation and collagen synthesisby osteoblasts from rat calvaria (rOs) and humanknee (hOs) cultured on titanium surfaces coatedwith collagen fibrils of the selected collagen typewith bound decorin and biglycan. In addition, GAGchains of both decorin and biglycan were character-ized using high-performance liquid chromatography(HPLC). It was considered advantageous to bind asmuch decorin and biglycan as possible to fibrils inorder to increase the extent of a possible osteoblastreaction to decorin and/or biglycan. Collagen II waschosen and four surface types were compared: baretitanium, titanium coated with collagen II fibrils, tita-nium coated with collagen II fibrils with bound bi-glycan, and titanium coated with collagen II fibrilswith bound decorin. To our best knowledge, no-oneelse has yet studied osteoblast behavior on coatingsof collagen II, nor investigated the effect of decorinand biglycan as a component of a coating on theparameters listed earlier.

MATERIALS AND METHODS

Materials

Chemicals specific for the HPLC analysis of GAG chainsof decorin and biglycan were obtained as described byViola et al.27 All other chemicals, including the pepsin-treated collagen types I (bovine skin), II (bovine trachealcartilage), and III (human placenta), as well as decorin(bovine articular cartilage, molecular weight (MW) �100 kDa,

of which about 40 kDa is core protein and about 60 kDaGAG chain) and biglycan (bovine articular cartilage, MW200–350 kDa of which about 45 kDa is core protein andabout 155–295 kDa GAG chains), were obtained fromSigma-Aldrich Chemie GmbH, Germany, unless otherwisestated. All chemical reagents used were reagent grade andall solutions were prepared with high purity deionizedwater. Titanium (grade II) was obtained from GoodfellowGmbH, Germany.

HPLC analysis of GAG chains ofdecorin and biglycan

The composition of decorin and biglycan GAG chainswas determined by HPLC as described by Viola et al. withminor modifications.27 In brief, derivatization of D-disac-charides from standard CS/DS, decorin, and biglycan wasperformed as described by Calabro et al.,28 using 2 nmolof each standard disaccharide or the disaccharide derivedfrom decorin digestions (10 g of UA). A 40 lL volume of12.5 mM AMAC solution in glacial acetic acid/DMSO(3:17, v/v) was added, and samples were incubated for10–15 min at room temperature. A 40 lL volume of afreshly prepared solution of 1.25M NaBH3CN in waterwas added to each sample, followed by an overnight incu-bation at 378C. An appropriate dilution of these samples in0.1M ammonium acetate at pH 7.0 was used for the HPLCanalysis.

Separation and analysis of AMAC-derivatives of D-di-saccharides were performed as described by Karousouet al.29 with a Jasco-Borwin chromatograph system with afluorophore detector (Jasco FP-920, kex ¼ 442 nm and kem¼ 520 nm). Chromatography was carried out using areversed phase column (C-18, 4.6 mm 3 150 mm, particlesize of 3 lm; Bischoff) at room temperature, equilibratedwith 0.1M ammonium acetate buffer at pH 7.0, filteredthrough a 0.22-lm membrane filter. A gradient elution wasperformed using a binary solvent system composed of0.1M ammonium acetate buffer, pH 7.0 (eluent A), andacetonitrile (eluent B). The flow rate was 1 mL/min, andthe following program was used: pre-run of column with100% eluent A for 20 min, isocratic elution with 100% elu-ent A for 5 min, gradient elution to 15% eluent B for 15min, to 30% eluent B for 35 min, and from 30 to 50% for 5min. Sample peaks were identified and quantified compar-ing the fluorescence spectra with standard disaccharides,using Jasco-Borwin software. A 10 lL volume of the deriv-atized samples (1/10 of the total) was analyzed andcompared.

Preparation of fibrils

Fibrils from collagen types I, II, and III were preparedaccording to the method of Williams et al.30 Briefly, colla-gen was dissolved at 1 mg/mL in 10 mM acetic acid over-night at 48C. Aliquots were mixed on ice with equal vol-umes of double concentrated fibrillogenesis buffer (50 mMsodium dihydrogenphosphate and 10 mM potassium dihy-drogenphosphate at pH 7.4 to yield 60 mM phosphate intotal) in 1.5-mL microcentrifuge tubes (Brandt, Germany).

806 DOUGLAS ET AL.

Journal of Biomedical Materials Research Part A DOI 10.1002/jbm.a

SLRPs were added to the reaction solution before the startof fibrillogenesis at a SLRP:collagen ratio of 1:7 by mass.Fibril formation took place at 378C overnight. Fibrils weresubsequently separated by centrifugation for 15 min at10,000g. The supernatant was retained for analysis. Thepellet was retained either for analysis or to coat titaniumsurfaces.

Protein determination in supernatant and pellet

Collagen concentration in supernatants was determinedusing a protocol based on the method of Lowry et al.31

Two-hundred microliter supernatant was mixed with 1 mLof a freshly prepared solution formed by diluting 2% (w/w) Na2CO3 in 0.1M NaOH with 0.5% CuSO4�5H2O (w/w)in 1% (w/w) Na3C6H5O7�2H2O by a factor of 50. Fifteenminutes after mixing, 100 lL Folin and Ciocalteu’s Phenolreagent was added and the mixture shaken vigorously.After subsequent incubation in the dark for 90 min,absorbance was measured at 700 nm. A calibration curvederived from solutions of the respective tropocollagen typein 10 mM acetic acid ranging between 0 and 500 lg/mLsubjected to the same procedure enabled determination ofcollagen concentration. Spectroscopic measurements weremade using a Tecan Spectrafluor Plus spectrometer.

In experiments to determine SLRP content of fibrils,samples were prepared in triplicate. The mass of collagenin the pellet was calculated by subtracting the mass of col-lagen detected in the supernatant from the initial mass ofcollagen before fibrillogenesis.

SLRP determination by dimethylmethyleneblue assay

SLRPs were quantified using the protocol of Chouet al.32 Briefly, pellets in microcentrifuge tubes were resus-pended in 500 lL of a 0.1 mg/mL Papain solution inHank’s balanced salt solution (HBSS) using pulses from anultrasound horn (UP 100H; Dr. Hieschler GmbH, Germany)at cycle ¼ 1 and 100% amplitude for 3 s. Thereafter digestiontook place at 608C for 24 h. After digestion, 40 lL of the solu-tion was transferred to a 96-well Nunc microplate andreacted with 250 lL of 1,9-dimethymethylene blue (DMMB)solution composed of 21 mg of DMMB, 5 mL absolute etha-nol, and 2 mg sodium formate per 1 L, with pH adjusted to1.5 using 6M HCl. The amount of SLRPs was determined bymeasuring the absorbance at 590 nm and comparing it to acalibration curve consisting of SLRP solutions in HBSS withconcentrations ranging from 0 to 100 lg/mL. Spectroscopicmeasurements were made using a Tecan Spectrafluor Plusspectrometer.

Preparation of collagen-coated titanium surfaces

Titanium discs of diameter 9.5 mm were roughened byblasting with corundum of diameter 250 lm (Finox,Germany). Blasted surfaces were then cleaned with 1%Triton X-100, acetone, and 96% ethanol, rinsed with dis-tilled water, and air-dried. Fibrils formed in the step

described above were resuspended in buffer solution usingpulses from an ultrasound horn (UP 100H; Dr. HieschlerGmbH) to form a suspension with a ‘‘concentration’’ of 1mg fibrils/mL buffer solution. Hundred microliter of thissuspension was added dropwise onto the surface of eachtitanium disc. After being coated with fibrils by adsorptionfor 30 min, surfaces were washed three times with distilledwater and air-dried. Coatings were crosslinked overnightat room temperature by immersion in a solution of 19.17mg/mL N-(3-dimethylaminopropyl-N0-ethyl-carbodiimidehydrochloride (EDC) and 5.75 mg/mL N-hydroxysuccini-mide (NHS) in 60% 0.1M Na2HPO4 at pH 5.5 (vol %) and40% absolute ethanol (vol %). After removal of the cross-linking solution, surfaces were immersed three times for30 min in 0.1M Na2HPO4 at pH 9.1 and then twice for 30min in 4M NaCl, and finally rinsed five times with dou-ble-distilled water and air-dried. Sterilization was carriedout using ethylene oxide at 458C. To verify that treatmentat 458C would not denature collagen fibrils, titanium surfa-ces coated with uncrosslinked fibrils were heated at 458Cfor 24 h and their morphology examined using atomicforce microscopy, as described previously.33 No differencein fibril morphology could be seen after treatment at 458C(data not shown).

Bioavailability of decorin and biglycan

Decorin and biglycan were biotinilated via the core pro-tein using EZ-LinkTM Sulpho-NHS-Biotin (Perbio ScienceDeutschland GmbH, Germany) according to the manufac-turer’s protocols and incorporated into fibrils during fibril-logenesis as described earlier. Titanium surfaces werecoated with fibrils containing biotinilated SLRPs asdescribed earlier but without sterilization. BiotinilatedSLRPs were detected by direct enzyme-linked immunosor-bent assay (ELISA): coated titanium surfaces and 24-wellplates (Nunc GmbH, Germany) were blocked for 2 h with3% (w/w) bovine serum albumin (BSA) in Phosphate Buf-fered Saline (PBS). After washing three times with distilledwater, surfaces were immersed in 500 lL of 1 lg/mLStreptavidin-conjugated alkaline phosphatase (Dianova,Germany) in PBS in the blocked wells for 1 h. After wash-ing three times with distilled water, 500 lL of 1 mg/mL 4-nitrophenyl phosphate disodium salt hexahydrate in alka-line phosphatase buffer (0.1M glycine, 1 mM MgCl2, 1 mMZnCl2, pH 10.4) was added. The reaction was stopped byaddition of 200 lL of 3M NaOH. Absorbance was meas-ured at 405 nm.

Cell preparation and culture

Throughout this study, NIH guidelines for the care anduse of laboratory animals (NIH Publication No. 85-23 Rev.1985) were observed. Rat calvarial osteoblasts were iso-lated according to Geissler et al.3 Briefly, cells wereobtained from the calvariae of newborn Wistar Kyoto ratsby sequential digestion with 3.56 U collagenase P/mL(Roche) and 15 U trypsin/mL (Roche) in phosphate buf-fered saline (PBS) and subcultured in Dulbecco’s ModifiedEagle’s Medium DMEM (Biochrom) containing 10% fetal

INFLUENCE OF DECORIN AND BIGLYCAN ON OSTEOBLASTS 807


calf serum (FCS). The osteogenic differentiation and osteo-blast phenotype of these cells (rO) was confirmed by deter-mination of alkaline phosphatase activity, collagen type Isynthesis, osteocalcin mRNA and formation of calciumphosphate deposits. After three to eight passages, cellswere used for the experiments. Adult human osteoblasts(hO) from the knee of a 76-year-old female donor werepurchased from PromoCell (Heidelberg, Germany). Thesecells were cultured and expanded for 7 days in Osteoblastgrowth medium (distributed by PromoCell, Heidelberg,Germany).

For cell culture experiments, osteoblasts were plated ina small volume (200 lL for Ø 16-mm samples) at 12,500cells/cm2 (rO) and 7000 cells/cm2 (hO) on the top of thetitanium disks. After 2-h incubation at 378C the mediumwas filled up to 1 mL. Culture medium was changed twicea week.

Focal adhesion contacts: Immunofluorescencestaining

After incubation of the cells on the different titaniumcoatings fixation was performed with 4% paraformaldehyde(w/v) in PBS for 15 min at 258C. The cells were then perme-abilized with 0.1% Triton X-100 in PBS for 20 min. Followingthis, unspecific binding sites were blocked with 1% BSAwith 0.05% Tween 20 (w/v) in PBS. Primary antibody bind-ing (64 lg mouse anti-vinculin (hVin I clone)-IgG/mL of 1%BSA/0.05% Tween 20 in PBS; Sigma) was allowed for 1 h at258C. After washing with PBS, the cells were exposed to anadditional incubation in blocking buffer for 10 min. Second-ary antibody (66 lg goat anti-mouse TRITC-labeled IgG/mLof 1% BSA with 0.05% Tween 20 in PBS; DAKO) and FITC-phalloidine (1.5 lg/mL of 1% BSA with 0.05% Tween 20 inPBS; Sigma) were used for 1 h at 258C. Subsequently DAPI(0.2 lg of 40,6-diamidino-2-phenylindol/mL of PBS; Sigma)for nuclei staining was applied for 15 min at 258C. Stainedcells on coated titanium plates were then embedded inMowiol 4-88. The staining was visualized using an AxioPhotfluorescence microscope, digital images were acquired withan AxioCam MRm camera (Zeiss) working with the AxioVi-sion software version 4.4. The fluorescence signals weredetected with the following optics: TRITC excitation 546 nm,emission 590 nm; FITC excitation 450–490 nm, emission515–565 nm; DAPI excitation 365 nm, emission 420 nm.

Proliferation rate

Proliferation of osteoblasts was determined by quantify-ing DNA synthesis as described previously.2 Briefly, afterculture for 24 h, medium was supplemented with 1 lCimethyl-[3H]thymidine/mL (80 Ci/mmol; Amersham) andcells were incubated for 24 h. DNA was precipitated with10% trichloroacetic acid. After washing with 96% ethanol,cells were lysed in 100 mM NaOH containing 22 mMNa2CO3 and 3.5 mM SDS. Methyl-[3H]thymidine incorpo-rated into DNA was determined by liquid scintillationcounting on a 1450 TRILUX Microbeta (Perkin ElmerWallac GmbH, Freiburg, Germany).

Collagen synthesis rate

Collagen synthesis was determined according to Scuttet al.34 with some modifications. At day 3 of culture, cellscultured on different surfaces were incubated in mediumsupplemented with 2.5 lCi [3,4-3H]proline/mL (60 Ci/mmol; Amersham), 300 lM ascorbate, and 200 lM b-ami-nopropionitril for 24 h. The extracellular collagen wasextracted by the addition of 10M acetic acid to give a finalacetic acid concentration of 0.5M, followed by incubationat 48C for 24 h. After that, collagen was precipitated by theaddition of 10 lg of rat-tail collagen/mL (Roche) and 4.3MNaCl, both to 0.5M acetic acid, to give a final concentrationof 0.85M NaCl. The samples were thoroughly shaken andincubated at 48C for 1 h. After centrifugation at 20,800g for30 min, the pellet was washed three times with 0.85MNaCl and dissolved in 0.5M acetic acid. Aliquots wereused for scintillation counting. Cell number was assessedby methylene blue staining according to Currie.35

Statistics

For analysis of statistical significance, experiments wererepeated three times. Results are presented as mean 6 SEM(standard error from mean). A one-way ANOVA (analysisof variance) with Bonferoni post test was applied.

RESULTS

Analysis of decorin and biglycan GAG chains

The distribution of disaccharides present in theGAG chains of decorin and biglycan as measured byHPLC are shown in Table I. The sulfation patternsof decorin and biglycan did not differ drastically.Decorin had slightly more sulfated chains than bigly-can, with 0.97 sulfate groups per disaccharide unitcompared with 0.93 for biglycan. The predominantdisaccharide in both decorin and biglycan was Ddi-mono4S, corresponding to chondroitin-4-sulfate(CS4) or dermatan sulfate (DS), which made upabout half of the total GAG (52.95% for decorin,49.70% for biglycan). DDi-mono6S, corresponding tochondroitin-6-sulfate (CS6) was the second mostcommon disaccharide unit in both proteoglycans,with biglycan having a higher proportion thandecorin (42.78–34.23%). Decorin contained moretrisulfated and nonsulfated disaccharide units thanbiglycan.

Binding of decorin and biglycan to collagen typesI–III and selection of collagen type for coating

The mass of decorin and biglycan bound per mgfibrils formed can be seen in Figure 1. All three col-lagen types bound more biglycan than decorin, with

808 DOUGLAS ET AL.


type II binding more than types I and III. It was con-sidered desirable to bind as much decorin and bigly-can as possible to maximize potential decorin/bigly-can–osteoblast interactions. Hence, it was decided touse fibrils of collagen II and not collagen I or III tocoat titanium surfaces for the following investiga-tions. Collagen II fibrils without SLRPs and withbound decorin/biglycan were compared.

Bioavailability of decorin and biglycan

The bioavailability of biglycan and decorin in col-lagen II fibrils adsorbed to titanium surfaces isshown in Figure 2. The adsorbance value obtainedusing direct ELISA is a semiquantitative measure ofthe bioavailability of the protein cores of the twoSLRPs. On coated surfaces, the signal for biotinilated

TABLE IDistribution (%) of Disaccharides Present in the GAG of Examined Decorin and Biglycan (See Text for Details)

Sample Disaccharide unit Decorin Biglycan

Ddi-nonSCS 9.46 7.26

Ddi-mono6S CS6 (CS C) 34.23 42.78

Ddi-mono4S CS4 (CS A) or DS (CS B) 52.95 49.70

Ddi-di(4,6)S 0.00 0.23

Ddi-tri(2,4,6)S 3.36 0.03

Mean number of sulfate groups per disaccharide unit 0.97 0.93

Figure 1. Quantification of decorin and biglycan boundper mg collagen fibrils of collagen types I, II, and III dur-ing fibrillogenesis in 30 mM phosphate buffer by dimethyl-methylene blue (DMMB) assay according to the method ofChou et al.32 Decorin/biglycan:collagen (w/w) ratio in thestarting solution ¼ 1:7. Experiments were performed intriplicate; error bars show standard deviation. [Color figurecan be viewed in the online issue, which is available atwww.interscience.wiley.com.]

Figure 2. Comparison of bioavailability of biotinilateddecorin and biglycan bound to collagen II fibrils coatedonto titanium surfaces. Biotinilated core protein wasdetected by direct ELISA. Ti: uncoated titanium; Ti/II: tita-nium coated with collagen II fibrils; Ti/II/BG: titaniumcoated with collagen II fibrils with bound biglycan; Ti/II/dcn: titanium coated with collagen II fibrils with bounddecorin. Experiments were performed in triplicate; errorbars show standard deviation. [Color figure can be viewedin the online issue, which is available at www.interscience.wiley.com.]



biglycan was stronger than that for biotinilateddecorin. The signals for both SLRPs were muchhigher than those for bare titanium and titaniumcoated with collagen II fibrils without SLRPs.

Focal adhesion formation

The effect of decorin and biglycan on the forma-tion of focal adhesions by rO is shown in Figure 3.The presence of biglycan led to an increase in thenumber of focal adhesions formed after both 2 and24 h. It was not clear if decorin caused an increasein the number of focal adhesions formed. Bare tita-nium surfaces and those coated with collagen IIwithout biglycan or decorin showed fewer focaladhesions than surfaces containing biglycan, bothafter 2 and 24 h.

Proliferation

The effect of decorin and biglycan on the proli-feration of hOs and rOs after 2 days can be seen inFigure 4. The proliferation of rOs was higher thanthat of hOs on all surfaces. Neither decorin nor bi-glycan significantly affected rO proliferation, how-ever, decorin and in particular biglycan significantlypromoted hO proliferation.

Collagen synthesis

The effect of decorin and biglycan on collagen syn-thesis by hOs and rOs after 4 days is presented inFigure 5. Neither decorin nor biglycan significantlyaffected hO collagen synthesis. However, a slightreduction in rO collagen synthesis was observed forthe collagen II coating with no decorin or biglycan,whereas a highly significant reduction was detectedin the presence of biglycan and no significant changein the presence of decorin could be seen.

DISCUSSION

The aims of this study were (i) to select one of thecollagen types I, II, and III on the basis of ability tobind decorin and biglycan to fibrils and to confirmthe bioavailability of the fibril-bound decorin andbiglycan; (ii) to study and compare the behavior ofosteoblasts from rat calvaria (rOs) and human knee(hOs) cultured on titanium surfaces coated withcollagen fibrils of the selected collagen type.

Decorin/biglycan GAG chain(s) show similarsulfation patterns

Biglycan and decorin showed similar sulfation pat-terns (Table I), the only noticeable differences being

that decorin contained more trisulfated disaccharideunits (3.36% compared with 0.03% for biglycan) andless Ddi-mono6S, corresponding to CS6 (34.23% com-pared with 42.78% for biglycan). CS4/DS makes upapproximately half and CS6 more than one third ofthe GAG of both biglycan and decorin, data consist-ent with the literature since biglycan and decorinfrom bovine cartilage have been reported to haveGAG chains which are similar in composition,although sulfated at the C4 position.36

Collagen type II binds more decorin andbiglycan than collagen types I and III andmore biglycan is bound than decorin

More SLRPs were bound by collagen II than colla-gen I and III, with considerably more biglycan thandecorin bound by all three collagen types (Fig. 1).These results correlated well with data from a previ-ous publication of our group.37 Furthermore, stron-ger interactions between collagen II and biglycanthan collagen II and decorin have been reported.38,39

SLRPs are believed to interact with collagen bothvia the GAG chains40 and the core protein, with bi-glycan and decorin core proteins competing foridentical or adjacent binding sites on collagen Ifibrils.41 If it is assumed that identical molaramounts of decorin and biglycan do indeed bind tofibrils via the protein core, the higher mass of bigly-can bound may be at least partly explained by bigly-can’s higher molecular weight (200–350 kDa com-pared with 100 kDa for decorin). GAG-mediatedbinding is believed to be of an ionic nature.42–46 Thecharge densities of the decorin and biglycan used inthis study must be similar in view of the similardegrees of sulfation (Table I), and so differences incharge density must be small and unlikely to exert alarge effect. Approximately half of the decorin andbiglycan GAG chains is composed of Ddi-mono4S,which can be CS4 or DS. Collagen has been reportedto have a higher affinity for DS than CS4.45,47,48 Dif-ferent proportions of CS4 and DS in the GAG chainsof decorin and biglycan could lead to differences inthe strength of their interactions with collagen. Dif-ferences in GAG chain length may also play a role.Greater binding of higher molecular weight DS andCS to tropocollagen has been reported.45,48 The bi-glycan used has longer GAG chains than decorin(155–295 kDa for two biglycan chains comparedwith 60 kDa for one decorin chain), and longerGAGs may be more successful at ‘‘bridging’’ basicregions in neighboring collagen molecules, resultingin a more stable interaction.

Fibrils of collagen II are known to be markedlythinner than those of collagen I,49 which may offer alarger surface area for SLRP binding to collagen,

810 DOUGLAS ET AL.


assuming that SLRPs bind to the outer surface ofcollagen, as has been suggested.50 It may be that col-lagens I, II, and III have different inherent affinitiesfor GAGs and SLRPs under the fibrillogenesis con-ditions used; collagen types I, II, and III tend tobe associated with different GAGs in vivo51 and

collagen II has bound more CS than collagen I,52

whereas a greater affinity of biglycan for collagen IIthan types I and III has been reported.38,39 CollagenII occurs in cartilage, which has a high proteoglycancontent and is the source tissue for the decorin andbiglycan used in this study. Hence, its ability to bind

Figure 3. Effect of decorin and biglycan on focal adhesion formation of rat osteoblasts to coated surfaces as assessed bycytoskeletal staining after 2 h (top) and 24 h (bottom) in the presence of 10% fetal calf serum. Red: vinculin; green: actin;blue: nuclei; yellow: focal adhesions. Ti: uncoated titanium; Ti/coll II: titanium coated with collagen II fibrils; Ti/coll II/BG: titanium coated with collagen II fibrils with bound biglycan; Ti/coll II/dcn: titanium coated with collagen II fibrilswith bound decorin. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]



a larger amount of proteoglycan may be advanta-geous physiologically.

Bioavailability: Adsorbance signal forbiotinilated biglycan is higher than that forbiotinilated decorin

Biotinilated biglycan and decorin could bedetected by direct ELISA, suggesting that the coreproteins of both are bioavailable. Hence, the GAGchains of both decorin and biglycan are present onthe surface and presumably also bioavailable. The

signal for biglycan was stronger than that fordecorin (Fig. 2).

Assuming that an equal number of biotin groupsare attached to decorin and biglycan, whose coreproteins share 55% homology53 and have similar mo-lecular weights, this result would mean that moremoles of biglycan core protein than decorin coreprotein were detected on titanium surfaces, whichwould in turn mean that more GAG in the form ofchains was present on surfaces containing biglycan.

Figure 4. Effect of decorin and biglycan on proliferationof (a) human and (b) rat osteoblasts measured after 2 daysby quantifying DNA synthesis. Ti: uncoated titanium; Ti/coll II: titanium coated with collagen II fibrils; Ti/coll II/BG: titanium coated with collagen II fibrils with bound bi-glycan; Ti/coll II/dcn: titanium coated with collagen IIfibrils with bound decorin. Experiments were performedin triplicate; error bars show standard deviation frommean. Significances: * ¼ <0.05, *** ¼ <0.001. [Color figurecan be viewed in the online issue, which is available atwww.interscience.wiley.com.]

Figure 5. Effect of decorin and biglycan on collagen syn-thesis of (a) human and (b) rat osteoblasts measured after4 days using the method of Scutt et al.34 Ti: uncoated tita-nium; Ti/coll II: titanium coated with collagen II fibrils;Ti/coll II/BG: titanium coated with collagen II fibrils withbound biglycan; Ti/coll II/dcn: titanium coated with colla-gen II fibrils with bound decorin. Experiments were per-formed in triplicate; error bars show standard deviationfrom mean. Significances: * ¼ <0.05, *** ¼ <0.001. [Colorfigure can be viewed in the online issue, which is availableat www.interscience.wiley.com.]

812 DOUGLAS ET AL.


Biglycan stimulates formation of focaladhesions by rO

Biglycan promoted the formation of focal adhe-sions on rough titanium after both 2- and 24-h incu-bation (Fig. 3).

Both CS and DS are negatively charged; the nega-tive charges may bind cations from the medium,which are required for focal adhesion formation.20

The differences observed for decorin and biglycancould be explained by the different masses bound tofibrils (Fig. 1); the higher amount of biglycan thandecorin bound and the stronger signal obtained fromcoated surfaces containing biglycan (Fig. 2) wouldresult in a higher amount of GAG on the surfaceand in turn a higher negative charge, leading to ahigher number of focal adhesions.

Decorin and biglycan stimulate proliferation ofhOs. Biglycan but not decorin inhibits collagensynthesis of rOs

Both decorin and, in particular, biglycan stimu-lated the proliferation of hOs (Fig. 4). Collagen syn-thesis by hOs was not affected by biglycan ordecorin, but rO collagen synthesis was reduced con-siderably by biglycan but not decorin (Fig. 5).

SLRPs and GAGs have affected the proliferationof a number of cell types. The overexpression of bi-glycan in smooth muscle cells led to an increase inproliferation both in vitro and in vivo.54 CS bound tocollagen I has improved proliferation of fibroblasts55

and chondrocytes.56

There are several hypotheses to explain howdecorin and biglycan may affect proliferation andcollagen synthesis.

One possibility is that the negative charges of theGAG chains enhance proliferation. Negative chargeshave enhanced osteoblast-like cell proliferation57 andsubstrate-bound proteoglycans containing CS have,depending on the circumstances, stimulated or in-hibited nerve growth, with the GAG chains media-ting the effects.58 Since more biglycan than decorinis bound to fibrils (Fig. 1) and the signal from bioti-nilated biglycan on titanium surfaces is stronger(Fig. 2), the amount of GAG and hence the negativecharge is expected to be higher, which might havecaused the effect of biglycan on proliferation to bestronger than that of decorin.

It is also conceivable that either the protein coresof decorin and biglycan and/or their GAG chainscan stimulate proliferation or inhibit collagen synthe-sis by binding growth factors either present in themedium or secreted by cells, thereby increasing theirbioavailability to the cells and/or potentiating theireffects positively or negatively. Binding of growth

factors and chemokines to biglycan and decorin coreproteins and CS and DS has been reported.22,23,59–61

The higher amount of biglycan present comparedwith decorin could have led to more proliferation-stimulating growth factors being bound and hence agreater increase in hO proliferation or decrease in rOcollagen synthesis.

Alternatively, the make-up of the protein coresand GAG chains may play a role in growth factorbinding and potentiation of their effects.62–64 Thedecorin and biglycan used in this study exhibiteddifferences in their GAG chains (Table I) and it isknown from literature that decorin and biglycancore proteins show structural differences.53,65 It ispossible that as a result there are differences in theabilities of decorin and biglycan to bind growth fac-tors and modulate their effects.

Changes in fibril thickness caused by decorin andbiglycan may also exert an influence. A previouspublication of our group showed that, at theSLRP:collagen ratios used in this study, decorinreduced the thickness of collagen II fibrils, whereasbiglycan did not affect fibril thickness.33 It is notimpossible that topographical changes on the nano-meter scale caused by differences in fibril thicknesscould influence proliferation and/or collagen synthe-sis. Different surface treatments resulting in varyingsurface roughnesses on the nanometer scale havebeen reported to affect proliferation of osteo-blasts66,67 and collagen synthesis by osteoblast-likecells.68,69

rO proliferation was considerably higher than hOproliferation (Fig. 4) and the same pattern for hOproliferation and collagen synthesis was not seen forrO (Figs. 4 and 5). There are several possible explan-ations for these differences. Studies have shown thatproliferation of osteoblasts depends on donor ageand skeletal site of origin.70–74 Also, differentmediums were used for the culture of rOs and hOs;differences in the composition of the media mayhave caused differences in proliferation and collagensynthesis. In addition, decorin and biglycan maymodulate the effect of growth factors on differentcell types differently. For example, decorin hinderedthe effect of TGF-beta 1 on osteosarcoma cells23 butenhanced its bioactivity in an osteoblast-like cellculture.24

SUMMARY AND OUTLOOK

Fibrils of collagen types I, II, and III bound morebiglycan than decorin. More biglycan and decorinbound to fibrils of collagen type II than to fibrils ofcollagen types I and III. Therefore, to study the effectof decorin and biglycan on primary rat and human



osteoblast reaction in vitro, it was decided to coat ti-tanium surfaces with collagen II fibrils with bounddecorin or biglycan or neither. It has been shownthat SLRPs bound to collagen II fibrils on titaniumsubstrates are able to influence the proliferation ofprimary osteoblasts and their collagen synthesis. Dif-ferences between rat and human osteoblast behaviorwere observed. These results could have implica-tions for the selection of collagen and SLRP type foruse in implant surface modification and tissue engi-neering. It remains to be investigated whether simi-lar results are obtained using collagen types I andIII, or if the presence of decorin or biglycan canmodulate the effect of exogenously added growthfactors.

The authors thank M. Schuhmann, H. Zimmermann, S.Heinemann, and Ch. Kupke for technical assistance.

References

1. Kim HW, Li LH, Lee EJ, Lee SH, Kim HE. Fibrillar assemblyand stability of collagen coating on titanium for improvedosteoblast responses. J Biomed Mater Res A 2005;75:629–638.

2. Becker D, Geissler U, Hempel U, Bierbaum S, ScharnweberD, Worch H, Wenzel KW. Proliferation and differentiation ofrat calvarial osteoblasts on type I collagen-coated titaniumalloy. J Biomed Mater Res 2002;59:516–527.

3. Geissler U, Hempel U, Wolf C, Scharnweber D, Worch H,Wenzel K. Collagen type I-coating of Ti6Al4V promotesadhesion of osteoblasts. J Biomed Mater Res 2000;51:752–760.

4. Roehlecke C, Witt M, Kasper M, Schulze E, Wolf C, Hofer A,Funk RW. Synergistic effect of titanium alloy and collagentype I on cell adhesion, proliferation and differentiation ofosteoblast-like cells. Cells Tissues Organs 2001;168:178–187.

5. Schliephake H, Aref A, Scharnweber D, Bierbaum S, RoesslerS, Sewing A. Effect of immobilized bone morphogenic pro-tein 2 coating of titanium implants on peri-implant bone for-mation. Clin Oral Implants Res 2005;16:563–569.

6. Schliephake H, Scharnweber D, Dard M, Sewing A, Aref A,Roessler S. Functionalization of dental implant surfaces usingadhesion molecules. J Biomed Mater Res B Appl Biomater2005;73:88–96.

7. Rammelt S, Schulze E, Bernhardt R, Hanisch U, ScharnweberD, Worch H, Zwipp H, Biewener A. Coating of titaniumimplants with type-I collagen. J Orthop Res 2004;22:1025–1034.

8. Bernhardt R, van den Dolder J, Bierbaum S, Beutner R,Scharnweber D, Jansen J, Beckmann F, Worch H. Osteocon-ductive modifications of Ti-implants in a goat defect model:Characterization of bone growth with SR muCT and histol-ogy. Biomaterials 2005;26:3009–3019.

9. Kadler KE, Holmes DF, Trotter JA, Chapman JA. Collagenfibril formation. Biochem J 1996;316 (Part 1):1–11.

10. Ballock RT, O’Keefe RJ. The biology of the growth plate.J Bone Joint Surg Am 2003;85:715–726.

11. Lawton DM, Andrew JG, Marsh DR, Hoyland JA, FreemontAJ. Mature osteoblasts in human non-union fractures expresscollagen type III. Mol Pathol 1997;50:194–197.

12. Ingram RT, Clarke BL, Fisher LW, Fitzpatrick LA. Distribu-tion of noncollagenous proteins in the matrix of adult humanbone: Evidence of anatomic and functional heterogeneity.J Bone Miner Res 1993;8:1019–1029.

13. Bianco P, Fisher LW, Young MF, Termine JD, Robey PG.Expression and localization of the two small proteoglycansbiglycan and decorin in developing human skeletal and non-skeletal tissues. J Histochem Cytochem 1990;38:1549–1563.

14. Iozzo RV. The biology of the small leucine-rich proteogly-cans. Functional network of interactive proteins. J Biol Chem1999;274:18843–18846.

15. Fisher LW, Termine JD, Dejter SW Jr, Whitson SW, Yanagish-ita M, Kimura JH, Hascall VC, Kleinman HK, Hassell JR,Nilsson B. Proteoglycans of developing bone. J Biol Chem1983;258:6588–6594.

16. Rosenberg LC, Choi HU, Tang LH, Johnson TL, Pal S, Web-ber C, Reiner A, Poole AR. Isolation of dermatan sulfateproteoglycans from mature bovine articular cartilages. J BiolChem 1985;260:6304–6313.

17. Waddington RJ, Roberts HC, Sugars RV, Schonherr E. Differ-ential roles for small leucine-rich proteoglycans in bone for-mation. Eur Cell Mater 2003;6:12–21; discussion 21.

18. Waddington RJ, Hall RC, Embery G, Lloyd DM. Changingprofiles of proteoglycans in the transition of predentine todentine. Matrix Biol 2003;22:153–161.

19. Ameye L, Young MF. Mice deficient in small leucine-richproteoglycans: Novel in vivo models for osteoporosis, osteoar-thritis, Ehlers-Danlos syndrome, muscular dystrophy, andcorneal diseases. Glycobiology 2002;12:107R–116R.

20. Bierbaum S, Douglas T, Hanke T, Scharnweber D, Tippelt S,Monsees TK, Funk RH, Worch H. Collageneous matrix coat-ings on titanium implants modified with decorin and chon-droitin sulfate: Characterization and influence on osteoblasticcells. J Biomed Mater Res A 2006;77:551–562.

21. Mundy GR. Regulation of bone formation by bone morpho-genetic proteins and other growth factors. Clin Orthop RelatRes 1996:24–28.

22. Hildebrand A, Romaris M, Rasmussen LM, Heinegard D,Twardzik DR, Border WA, Ruoslahti E. Interaction of thesmall interstitial proteoglycans biglycan, decorin and fibromo-dulin with transforming growth factor beta. Biochem J 1994;302 (Pt 2):527–534.

23. Hausser H, Groning A, Hasilik A, Schonherr E, Kresse H.Selective inactivity of TGF-b/decorin complexes. FEBS Lett1994;353:243–245.

24. Takeuchi Y, Kodama Y, Matsumoto T. Bone matrix decorinbinds transforming growth factor-b and enhances its bioactiv-ity. J Biol Chem 1994;269:32634–32638.

25. Chen XD, Fisher LW, Robey PG, Young MF. The small leucine-rich proteoglycan biglycan modulates BMP-4-induced osteo-blast differentiation. FASEB J 2004;18:948–958.

26. Moreno M, Munoz R, Aroca F, Labarca M, Brandan E,Larrain J. Biglycan is a new extracellular component of theChordin-BMP4 signaling pathway. EMBO J 2005;24:1397–1405.

27. Viola M, Karousou EG, Vigetti D, Genasetti A, Pallotti F,Guidetti GF, Tira E, De Luca G, Passi A. Decorin from differ-ent bovine tissues: Study of glycosaminoglycan chain byPAGEFS. J Pharm Biomed Anal 2006;41:36–42.

28. Calabro A, Benavides M, Tammi M, Hascall VC, Midura RJ.Microanalysis of enzyme digests of hyaluronan and chon-droitin/dermatan sulfate by fluorophore-assisted carbo-hydrate electrophoresis (FACE). Glycobiology 2000;10:273–281.

29. Karousou EG, Militsopoulou M, Porta G, De Luca G, HascallVC, Passi A. Polyacrylamide gel electrophoresis of fluoro-phore-labeled hyaluronan and chondroitin sulfate disaccha-rides: Application to the analysis in cells and tissues. Electro-phoresis 2004;25:2919–2925.

30. Williams BR, Gelman RA, Poppke DC, Piez KA. CollagenFibril Formation. J Biol Chem 1978;253:6578–6585.

31. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein mea-surement with the Folin phenol reagent. J Biol Chem 1951;193:265–275.

814 DOUGLAS ET AL.


32. Chou CH, Cheng WT, Lin CC, Chang CH, Tsai CC, Lin FH.TGF-b1 immobilized tri-co-polymer for articular cartilage tis-sue engineering. J Biomed Mater Res B Appl Biomater 2006;77:338–348.

33. Douglas T, Heinemann S, Bierbaum S, Scharnweber D,Worch H. Fibrillogenesis of collagen types I, II, and III withsmall leucine-rich proteoglycans decorin and biglycan. Bio-macromolecules 2006;7:2388–2393.

34. Scutt A, Berg A, Mayer H. A semiautomated, 96-well plateassay for collagen synthesis. Anal Biochem 1992;203:290–294.

35. Currie GA. Platelet-derived growth-factor requirements forin vitro proliferation of normal and malignant mesenchymalcells. Br J Cancer 1981;43:335–343.

36. Cheng F, Heinegard D, Malmstrom A, Schmidtchen A, YoshidaK, Fransson LA. Patterns of uronosyl epimerization and 4-/6-O-sulphation in chondroitin/dermatan sulphate from decorinand biglycan of various bovine tissues. Glycobiology 1994;4:685–696.

37. Douglas T, Heinemann S, Bierbaum S, Scharnweber D,Worch H.Collagen types I, II, III with small leucine-rich pro-teoglycans decorin and biglycan. Biomacromolecules 2006;7:2388–2393.

38. Bidanset DJ, Guidry C, Rosenberg LC, Choi HU, Timpl R,Hook M. Binding of the proteoglycan decorin to collagentype VI. J Biol Chem 1992;267:5250–5256.

39. Vynios DH, Papageorgakopoulou N, Sazakli H, Tsiganos CP.The interactions of cartilage proteoglycans with collagens aredetermined by their structures. Biochimie 2001;83: 899–906.

40. Pogany G, Hernandez DJ, Vogel KG. The in vitro interactionof proteoglycans with type I collagen is modulated by phos-phate. Arch Biochem Biophys 1994;313:102–111.

41. Schonherr E, Witsch-Prehm P, Harrach B, Robenek H, Rauter-berg J, Kresse H. Interaction of biglycan with type I collagen.J Biol Chem 1995;270:2776–2783.

42. Obrink B, Wasteson A. Nature of the interaction of chondroi-tin 4-sulphate and chondroitin sulphate-proteoglycan withcollagen. Biochem J 1971;121:227–233.

43. Mathews MB. The interaction of collagen and acid mucopoly-saccharides. A model for connective tissue. Biochem J 1965;96:710–716.

44. Mathews MB, Decker L. The effect of acid mucopolysacchar-ides and acid mucopolysaccharide-proteins on fibril forma-tion from collagen solutions. Biochem J 1968;109:517–526.

45. Obrink B. A study of the interactions between monomerictropocollagen and glycosaminoglycans. Eur J Biochem 1973;33:387–400.

46. Obrink B, Laurent TC, Carlsson B. The binding of chondroi-tin sulphate to collagen. FEBS Lett 1975;56:166–169.

47. Gelman RA, Blackwell J. Collagen-mucopolysaccharide inte-ractions at acid pH. Biochim Biophys Acta 1974;342:254–261.

48. Obrink B, Sundelof LO. Light scattering in the study of asso-ciating macromolecules. The binding of glycosaminoglycansto collagen. Eur J Biochem 1973;37:226–232.

49. Fertala A, Sieron AL, Hojima Y, Ganguly A, Prockop DJ. Self-assembly into fibrils of collagen II by enzymic cleavage ofrecombinant procollagen II. Lag period, critical concentration,and morphology of fibrils differ from collagen I. J Biol Chem1994;269:11584–11589.

50. Pieper JS, Hafmans T, Veerkamp JH, van Kuppevelt TH.Development of tailor-made collagen-glycosaminoglycan mat-rices: EDC/NHS crosslinking, and ultrastructural aspects.Biomaterials 2000;21:581–593.

51. Junqueira LC, Toledo OM, Montes GS. Correlation of specificsulfated glycosaminoglycans with collagen types I, II, and III.Cell Tissue Res 1981;217:171–175.

52. Negroiu G, Mirancea N, Mirancea D, Oancea A, Moldovan L.Collagen-Chondroitin sulfate sonstrates conditioned as Spongesand Membranes. Rev Roum Biochim 1992;29:23–28.

53. Svensson L, Heinegard D, Oldberg A. Decorin-binding sitesfor collagen type I are mainly located in leucine-rich repeats4–5. J Biol Chem 1995;270:20712–20716.

54. Shimizu-Hirota R, Sasamura H, Kuroda M, Kobayashi E,Hayashi M, Saruta T. Extracellular matrix glycoprotein bigly-can enhances vascular smooth muscle cell proliferation andmigration. Circ Res 2004;94:1067–1074.

55. Zhong S, Teo WE, Zhu X, Beuerman R, Ramakrishna S, YungLY. Formation of collagen-glycosaminoglycan blended nano-fibrous scaffolds and their biological properties. Biomacromo-lecules 2005;6:2998–3004.

56. van Susante JLC, Pieper J, Buma P, van Kuppevelt TH, vanBeuningen H, van Der Kraan PM, Veerkamp JH, van denBerg WB, Veth RPH. Linkage of chondroitin-sulfate to type Icollagen scaffolds stimulates the bioactivity of seeded chon-drocytes in vitro. Biomaterials 2001;22:2359–2369.

57. Ohgaki M, Kizuki T, Katsura M, Yamashita K. Manipulationof selective cell adhesion and growth by surface charges ofelectrically polarized hydroxyapatite. J Biomed Mater Res2001;57:366–373.

58. Erskine L, McCaig CD. Integrated interactions between chon-droitin sulphate proteoglycans and weak dc electric fieldsregulate nerve growth cone guidance in vitro. J Cell Sci 1997;110 (Part 16):1957–1965.

59. Sugahara K, Mikami T, Uyama T, Mizuguchi S, Nomura K,Kitagawa H. Recent advances in the structural biology ofchondroitin sulfate and dermatan sulfate. Curr Opin StructBiol 2003;13:612–620.

60. Trowbridge JM, Gallo RL. Dermatan sulfate: New functionsfrom an old glycosaminoglycan. Glycobiology 2002;12:117R–125R.

61. Kuschert GS, Coulin F, Power CA, Proudfoot AE, HubbardRE, Hoogewerf AJ, Wells TN. Glycosaminoglycans interactselectively with chemokines and modulate receptor bind-ing and cellular responses. Biochemistry 1999;38:12959–12968.

62. Takada T, Katagiri T, Ifuku M, Morimura N, Kobayashi M,Hasegawa K, Ogamo A, Kamijo R. Sulfated polysaccharidesenhance the biological activities of bone morphogenetic pro-teins. J Biol Chem 2003;278:43229–43235.

63. McCaffrey TA, Falcone DJ, Du B. Transforming growthfactor-b 1 is a heparin-binding protein: Identification of puta-tive heparin-binding regions and isolation of heparins withvarying affinity for TGF-beta 1. J Cell Physiol 1992;152:430–440.

64. Lyon M, Rushton G, Gallagher JT. The interaction of thetransforming growth factor-betas with heparin/heparan sul-fate is isoform-specific. J Biol Chem 1997;272:18000–18006.

65. Fisher LW, Termine JD, Young MF. Deduced proteinsequence of bone small proteoglycan I (biglycan) showshomology with proteoglycan II (decorin) and several noncon-nective tissue proteins in a variety of species. J Biol Chem1989;264:4571–4576.

66. Guizzardi S, Galli C, Martini D, Belletti S, Tinti A, RaspantiM, Taddei P, Ruggeri A, Scandroglio R. Different titaniumsurface treatment influences human mandibular osteoblastresponse. J Periodontol 2004;75:273–282.

67. Mustafa K, Wennerberg A, Wroblewski J, Hultenby K, LopezBS, Arvidson K. Determining optimal surface roughness ofTiO blasted titanium implant material for attachment, prolif-eration and differentiation of cells derived from human man-dibular alveolar bone. Clin Oral Implants Res 2001;12:515–525.

68. Eisenbarth E, Velten D, Breme J. Biomimetic implant coat-ings. Biomol Eng 2007;24:27–32.

69. Lincks J, Boyan BD, Blanchard CR, Lohmann CH, Liu Y,Cochran DL, Dean DD, Schwartz Z. Response of MG63 osteo-blast-like cells to titanium and titanium alloy is dependent on



surface roughness and composition. Biomaterials 1998;19:2219–2232.

70. Declercq H, Van den Vreken N, De Maeyer E, Verbeeck R,Schacht E, De Ridder L, Cornelissen M. Isolation, prolifera-tion and differentiation of osteoblastic cells to study cell/bio-material interactions: Comparison of different isolation tech-niques and source. Biomaterials 2004;25:757–768.

71. Egrise D, Vienne A, Martin D, Schoutens A. Trabecular bonecell proliferation ex vivo increases with donor age in the rat:It is correlated with the extent of bone loss and not with his-tomorphometric indices of bone formation. Calcif Tissue Int1996;59:45–50.

72. Kasperk C, Wergedal J, Strong D, Farley J, Wangerin K,Gropp H, Ziegler R, Baylink DJ. Human bone cell pheno-types differ depending on their skeletal site of origin. J ClinEndocrinol Metab 1995;80:2511–2517.

73. Martinez ME, del Campo MT, Medina S, Sanchez M, San-chez-Cabezudo MJ, Esbrit P, Martinez P, Moreno I, RodrigoA, Garces MV, Munvera L. Influence of skeletal site of originand donor age on osteoblastic cell growth and differentiation.Calcif Tissue Int 1999;64: 280–286.

74. Shigeno Y, Ashton BA. Human bone-cell proliferation in vitrodecreases with human donor age. J Bone Joint Surg Br 1995;77:139–142.

816 DOUGLAS ET AL.


Documents

Influence of collagen-fibril-based coatings containing decorin and biglycan on osteoblast behavior