ARTICLE IN PRESS

0142-9612/$ - se

doi:10.1016/j.bi

�Correspond3990.

E-mail addr1Equally con

Biomaterials 26 (2005) 2479–2489

www.elsevier.com/locate/biomaterials

Effects of scaffold composition and architecture on human nasalchondrocyte redifferentiation and cartilaginous matrix deposition

Sylvie Miota,1, Tim Woodfieldb,c,1, Alma U. Danielsd, Rosemarie Suetterline,Iman Peterschmittf, Michael Heberera, Clemens A. van Blitterswijkc, Jens Riesleb,

Ivan Martina,�

aDepartments of Surgery and of Research, University Hospital Basel, Hebelstrasse 20, 4031 Basel, SwitzerlandbIsoTis S.A., PO Box 98, 3720 AB, Bilthoven, The Netherlands

cInstitute for Biomedical Technology, University of Twente, Enschede, The NetherlandsdLaboratory for Orthopaedic Biomechanics, Felix Platter Hospital, Basel, Switzerland

eM.E. Muller Institute at the Biozentrum, University of Basel, SwitzerlandfUFR Pluridisciplinaire Enseignement Professionnalise Superieur, Universite Haute Alsace, Colmar, France

Received 23 March 2004; accepted 25 June 2004

Available online 10 August 2004

Abstract

We investigated whether the post-expansion redifferentiation and cartilage tissue formation capacity of adult human nasal

chondrocytes can be regulated by controlled modifications of scaffold composition and architecture. As a model system, we used

poly(ethylene glycol)-terephthalate–poly(butylene)-terephthalate block copolymer scaffolds from two compositions (low or high

PEG content, resulting in different wettability) and two architectures (generated by compression molding or three-dimensional (3D)

fiber deposition) with similar porosity and mechanical properties, but different interconnecting pore architectures. Scaffolds were

seeded with expanded human chondrocytes and the resulting constructs assessed immunohistochemically, biochemically and at the

mRNA expression level following up to 4 weeks of static culture.

For a given 3D architecture, the more hydrophilic scaffold enhanced cell redifferentiation and cartilaginous tissue formation after

4 weeks culture, as assessed by higher mRNA expression of collagen type II, increased deposition of glycosaminoglycan (GAG) and

predominance of type II over type I collagen immunostain. The fiber-deposited scaffolds, with a more accessible pore volume and

larger interconnecting pores, supported increased GAG deposition, but only if a more hydrophilic composition was used.

By applying controlled and selective modifications of chemico-physical scaffold parameters, we demonstrate that both scaffold

composition and architecture are instructive for expanded human chondrocytes in the generation of 3D cartilaginous tissues. The

observed effects of composition and architecture were likely to have been mediated, respectively, by differential serum protein

adsorption and efficiency of nutrient/waste exchange.

r 2004 Elsevier Ltd. All rights reserved.

Keywords: Cartilage tissue engineering; Nasal chondrocytes; Extracellular matrix; Wettability; Porosity; Scaffold

e front matter r 2004 Elsevier Ltd. All rights reserved.

omaterials.2004.06.048

ing author. Tel: +41-61-265-2384; Fax: +41-61-265-

ess: imartin@uhbs.ch (I. Martin).

tributing authors.

1. Introduction

One of the challenges in tissue engineering is thedesign and fabrication of biodegradable scaffolds whichare ‘instructive’ for specific cellular functions, and maythus regulate cell adhesion, proliferation, expression of aspecific phenotype and extracellular matrix deposition ina predictable and controlled fashion. It has long been

ARTICLE IN PRESS

x

O

O

C

O

(O-CH2-CH2)ny

O (CH2)4-O C

O

C

O

poly(ethylene glycol) terephthalate poly(butylene) terephthalate

CM 300/55/45 3DF 300/55/45

CM 1000/70/30 3DF 1000/70/30

C

(A)

(B)

Fig. 1. (A) Chemical structure of PEGT–PBT block copolymers. The

MW of PEG (a) and the weight ratio between PEGT and PBT (b/c)

define the specific composition obtained (a/b/c). (B) 3D reconstruction

by mCT scans of different PEGT/PBT scaffold architectures (generated

by CM or 3DF deposition techniques) and compositions (300/55/45 or

100/70/30).

S. Miot et al. / Biomaterials 26 (2005) 2479–24892480

known that cell behavior on synthetic polymers isrelated to both the physical and chemical properties ofthe substratum [1]. Scaffold physical structure maycontrol cell function by regulating the diffusion ofnutrients, waste products and cell–cell interactions [2],whereas scaffold surface chemistry indirectly affects celladhesion, morphology and subsequent cellular activityby controlling adsorption of ions, proteins and othermolecules from the culture medium [1,3]. It is then thismolecularly populated surface that the cells sense andrespond to biochemically by means of specific receptors(e.g., integrins) [4].

Cartilage cells are an ideal model for the study ofcell–substrate interactions, due to the tight relationshipswhich have been established between chondrocytemorphology and function [5]. When chondrocytes arereleased from their extracellular matrix and culturedunder conditions promoting a spread morphology, suchas on two-dimensional (2D) tissue culture plastic, theyprogressively lose their original phenotype (i.e., expres-sion of collagen type II and aggrecan) and displayfibroblastic, pre-chondrogenic features (i.e., expressionof collagen type I and versican), a process typicallydescribed as dedifferentiation [6,7]. Chondrocyte ded-ifferentiation can be reversed by culturing cells insubstrates supporting a spherical morphology, such aspolymer gels [6,8] or porous polymer scaffolds [9,10].The influence of specific surface properties of variousbiomaterials on chondrocyte behavior has been so farmostly investigated using 2D films [11–15]. However, thegrowing appreciation that cell behavior in 2D culturesdoes not necessarily translate into 3D systems [16,17]prompts for extended studies in 3D scaffolds where theintroduction of an additional parameter, 3D architec-ture, could modulate the effects of changes in thesubstrate composition.

As a model scaffold system, we selected a segmentedblock copolymer consisting of soft, hydrophilic poly(-ethylene glycol)-terephthalate (PEGT) segments andhard, hydrophobic poly(butylene)-terephthalate (PBT)segments, denoted as a/b/c, where a represents the PEGmolecular weight (MW, g/mol), and b and c representthe weight percentage of PEGT and PBT blocks,respectively. By varying the PEGT MW and the weightpercentage ratio between PEGT and PBT, it is possibleto modulate the wettability, protein adsorption, swellingand mechanical properties of the substrate [18,19].Furthermore, by changing the fabrication technique(i.e., compression molding (CM) or 3D fiber (3DF)deposition) for a given PEGT/PBT composition, it ispossible to fabricate 3D scaffolds with the same bulkcomposition and overall porosity but different inter-connecting pore architectures [20].

Previously, Papadaki et al. [15] reported that primarybovine chondrocytes seeded on 2D PEGT/PBT filmsbest retained their chondrocytic phenotype when a 1000/

70/30 composition was used. More recently, Mahmoodet al. [18] confirmed that 2D PEGT/PBT films producedfrom 1000/70/30 compositions better maintained thephenotype of primary adult human articular chondro-cytes as compared to 300/55/45 compositions. The lattercomposition, however, allowed redifferentiation ofexpanded human nasal chondrocytes in 3D scaffoldsproduced by CM [21].

In the present study, we produced 3D scaffolds in thetwo above-described PEGT/PBT compositions either byCM or 3DF deposition. We then used the generatedscaffolds to investigate whether the post-expansionredifferentiation and cartilage tissue formation capacityof adult human nasal chondrocytes can be regulated bycontrolled changes in PEGT/PBT composition andarchitecture. The selected cell source was chondrocytesfrom the human nasal septum, which can de- and re-differentiate [21,22], generate cartilage over a wide rangeof patient ages [23] and, as compared to human articularchondrocytes, proliferate faster and exhibit a higherchondrogenic capacity after expansion [24].

2. Material and methods

2.1. Scaffold fabrication

PEGT/PBT copolymers were obtained from IsoTisS.A. (Bilthoven, The Netherlands). 3D scaffolds madefrom 300/55/45 and 1000/70/30 compositions (Fig. 1A)were produced using either a CM and particle leachingtechnique [25] or a 3DF deposition technique [19,20].

ARTICLE IN PRESSS. Miot et al. / Biomaterials 26 (2005) 2479–2489 2481

Briefly, CM scaffolds were prepared by mixing PEGT/PBT granules with sodium chloride grains (75% byvolume), sieved to obtain particles ranging in size from400 to 600 mm. After CM under heat and pressure, thepolymerized PEGT/PBT block was then immersed indemineralized water for 48 h to remove the sodiumchloride, and dried under reduced pressure in a vacuumoven. Porous 3DF scaffolds were constructed bysuccessively layering a 0–901 pattern of molten PEGT/PBT fibers from a +250 mm nozzle onto a computer-controlled x2y2z table. Fibers within each layer werespaced 1.0mm apart, but were offset (staggered) by0.5mm between layers to optimize the ‘‘visible’’ surfaceavailable for cell seeding (Fig. 1B). Square blocks of30� 30� 4mm3 were produced with a consistent poresize and 100% interconnecting pore volume (PV) (Fig.1B). Cylindrical scaffolds, +5� 4mm2, were coredfrom the bulk porous CM and 3DF blocks prior to cellseeding. Previous thermal analysis studies have demon-strated that CM and 3DF processing techniques do notresult in changes of PEG MW or PEGT/PBT composi-tion [19], and therefore, any differences seen in thisstudy between architectures of CM and 3DF scaffoldscould not be related to differences in scaffoldcomposition.

2.2. Scaffold characterization

Scaffold architecture was analyzed using micro-computed tomography (mCT) using a desktop MicroCTmachine (mCT-40, Scanco Medical, Bassersdorf, Swit-zerland) as described previously [20]. Briefly, 200 2Dslices (2048� 2048 pixels, 12 mm resolution) werescanned of every sample covering a height of 2.4mm.The resulting gray-scale images were segmented and athreshold applied to extract the polymer phase and theninverted to extract the pore phase. This allowed 3Dreconstruction of the scaffold volume (SV) and PV fromstacked 2D images (Fig. 1B), which in turn could beanalyzed to give information on SV percent (vol%)porosity, surface area to volume ratio, pore size andaccessible PV using previously described techniques[2,20,26–28]. In addition to mCT, scaffold porosity wasalso determined using mass/volume techniques [20].

To characterize the mechanical properties of scaffoldswith varying composition and architecture, unconfinedstatic and dynamic compression tests were carried outon hydrated scaffolds to determine the equilibriummodulus (20% strain) and dynamic stiffness (0.1Hz) forcomparison with native articular cartilage tissue, asexplained previously [19,20].

2.3. Chondrocyte isolation and expansion

The source of hyaline cartilage was the human nasalseptum, collected in the University Hospital of Basel

after informed consent. Independent experiments wereperformed with cells from two patients (31 and 28 yearsold). Chondrocytes were isolated upon 22-hour incuba-tion at 37 1C in 0.15% type II collagenase and re-suspended in Dulbecco’s modified Eagle’s medium(DMEM) containing 10% fetal bovine serum, 4.5mg/ml D-glucose, 0.1mM non-essential amino acids, 1mM

sodium pyruvate, 100mM HEPES buffer, 100U/mlpenicillin, 100 mg/ml streptomycin and 0.29mg/ml glu-tamine (complete medium). Chondrocytes were platedin plastic dishes at a density of 104 cells/cm2 in acomplete medium further supplemented with 1 ng/mltransforming growth factor-b1 (TGF-b1), 5 ng/ml fibro-blast growth factor-2 (FGF-2) and 10 ng/ml platelet-derived growth factor-bb (PDGF-bb) (all from R&DSystems, Minneapolis, MN). This specific combinationof growth factors has previously been shown to enhancehuman nasal chondrocyte proliferation and post-expan-sion differentiation ability [22]. When cells weresubconfluent, they were detached by treatment with0.3% type II collagenase, followed by 0.05% trypsin/0.53mM EDTA, and re-plated at 5� 103 cells/cm2.When cells again approached confluence, second pas-sage cells were detached and seeded on scaffolds asdescribed below.

2.4. Chondrocyte seeding and culture on three-

dimensional scaffolds

Cylindrical scaffolds (+5� 4mm2 thick) were pre-wet with 70% ethanol, extensively rinsed in completemedium and blotted dry. Cells (5� 106/scaffold, re-suspended in 54 ml of complete medium) were staticallyseeded on the scaffolds placed in dishes coated with athin film of 1% agarose. Constructs were staticallycultured for 2 and 4 weeks in complete mediumsupplemented with 10 mg/ml insulin and 0.1mM ascorbicacid, with culture medium completely replaced twice aweek. Duplicate specimens were generated and assessedfor each experimental group (e.g., type of scaffold andtime point). After harvesting, each construct was dividedinto three parts which were independently processed forhistological and immunohistochemical, biochemical orgene expression analysis.

2.5. Histological and immunohistochemical analysis

Tissue constructs were fixed in 4% formalin for 24 hat 4 1C, dehydrated, embedded in paraffin and cross-sectioned (7 mm thick). For histological analysis, sec-tions were stained with Safranin-O for sulfated glyco-saminoglycans (GAGs). Immunohistochemical analysiswas performed as described by Baschong et al. [29].Sections were treated with 0.25% ammonia (NH3) in70% ethanol for 1 h during deparaffinization in order toeliminate autofluorescence. Sections were then sequen-

ARTICLE IN PRESSS. Miot et al. / Biomaterials 26 (2005) 2479–24892482

tially treated with 10mg/ml sodium borohydride(NaBH4) for 40min at 4 1C and 1mg/ml testicularhyaluronidase for 30min at 37 1C. Non-specific siteswere blocked by incubation with swine serum diluted1:10 in PBS for 10min at room temperature (RT).Sections were double labeled by incubation with apolyclonal antibody against collagen type I (T40111R,Biodesign International, Saco, ME) followed by a goatanti-rabbit secondary antibody conjugated with Alexa488 (Molecular Probes) and with a monoclonal anti-body against collagen type II (II-II6B3, DevelopmentalStudies Hybridoma Bank, Iowa City, IA) followed by adonkey anti-mouse secondary antibody labeled withCy3 (Jackson Immuno Research Lab). Other sectionswere labeled with a monoclonal antibody againstsmooth muscle a-actin (SMA, clone 1A4, Sigma)followed by a donkey anti-mouse secondary antibodylabeled with Cy3. All incubations with the antibodieswere performed for 60min at RT. All sections weremounted with Mowial-1188 (Hoechst, Frankfurt, Ger-many), dried overnight and stored in the dark untilmicroscopic imaging using a confocal laser scanningmicroscope (TCS 4-D CLSM, Leica AG, Heidelberg).The image stacks were generated as 0.5 mm opticalsections in parallel in the 488 nm (green) and 568 nm(red) channels. As a control for tissue quality, a series ofdifferential interference contrast (DIC) images wasacquired in a subsequent scan of the same field.

2.6. Biochemical analysis

Tissue constructs were digested in 1ml of proteinase-K solution (1mg/ml proteinase-K in 50mM Tris with1mM EDTA, 1mM iodoacetamide and 10 mg/ml pep-statin-A) for 15 h at 56 1C. GAG contents weremeasured spectrophotometrically after reaction withdimethylmethylene blue dye [30] with chondroitinsulfate as a standard. The GAG content was normalizedto the amount of DNA which was measured using aCyQUANT cell proliferation assay kit (MolecularProbes), with calf thymus DNA as a standard.

2.7. Gene expression analysis

RNA was extracted from tissue constructs using250 ml Trizol (Life Technologies, Basel, CH) accordingto the manufacturer’s instructions. cDNA was generatedfrom total RNA using reverse-transcriptase Stratascript(Stratagene) in the presence of dNTP and DTT. Real-time PCR reactions were performed and monitoredusing the ABI prism 7700 Sequence Detection Systemand the Sequence Detector V program (Perkin-ElmerApplied Biosystems). The PCR master mix was basedon AmpliTaq Gold DNA polymerase (Perkin-ElmerApplied Biosystems). In the same reaction, cDNAsamples were analyzed both for collagen type II and

for the housekeeping gene (18S ribosomal RNA), usingpreviously described sequences of primers and probes[31]. Each cDNA sample was assessed at least induplicate and collagen type II mRNA expression levelsnormalized to the corresponding 18S rRNA levels.

2.8. Statistical analysis

Data are presented as mean 7standard deviation.The normality of the populations to be analyzed wastested by Shapiro–Wilk tests. Means were compared byStudent’s t-tests in case of normal population or byMann–Whitney tests in case of failure of normality.Statistical analyses were performed using the Sigma Statsoftware (SPSS Inc., Version 7.5), with po0.05 as thecriteria for statistical significance.

3. Results

3.1. Scaffold characterization

Two processing techniques (CM and 3DF) were usedto fabricate porous scaffolds from two different PEGT/PBT copolymer compositions, resulting in reproducibleand comparable volume percent porosities, as deter-mined by mass–volume ðm=V Þ and 3D mCT techniques(Table 1). The scaffold surface area per volume ratiowas approximately three times greater in CM scaffoldsthan in 3DF scaffolds.

Fig. 2 shows the accessible PV distribution and poresize distribution for both 300/55/45 and 1000/70/30compositions of CM and 3DF scaffolds represented as apercentage of total PV. The accessible PVs in both CMand 3DF scaffolds were approximately 100% intercon-nected. However, 3DF scaffolds had a considerablygreater accessibility cut-off value of �360 mm (taken asthe pore size guaranteeing at least 90% of the PV wasaccessible) compared with �100 mm in CM scaffolds(Fig. 2A). Indeed, for the pore range investigated in thisstudy (excluding the pore size extremes of 6 and 1000 mmpores), a greater percentage of the PV was accessible in3DF scaffolds as compared with CM scaffolds, andtypically corresponded to between 300 and 400 mmgreater accessible pore size for a given percentage PV.Furthermore, as shown in Fig. 2B, 300/55/45 and 1000/70/30 3DF scaffolds had considerably greater averagepore sizes (388 and 380 mm, respectively) than CMscaffolds (182 and 160 mm, respectively). The leftwardshift in the average pore size distribution observed inCM and 3DF scaffolds of different compositions (Fig.2B) was probably due to the documented swellingcharacteristics of the more hydrophilic 1000/70/30compositions [19]. To give an indirect measure ofscaffold tortuosity or complexity, 3D mCT analysis alsoallowed determination of a connectivity density (CD)

ARTICLE IN PRESS

0%

20%

40%

60%

80%

100%

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

Pore Size (mm)

Acc

essi

ble

Po

re V

olu

me

as a

% o

f T

ota

l PV

CM 300/55/45

CM 1000/70/30

3DF 300/55/45

3DF 1000/70/30

90%

(A)

0%

2%

4%

6%

8%

10%

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Pore Size (mm)

Fre

qu

ency

as

% o

f T

ota

l PV

CM 300/55/45

CM 1000/70/30

3DF 300/55/45

3DF 1000/70/30

(B)

Fig. 2. Accessible PV distribution (A) and pore size distribution (B)

for both 300/55/45 and 1000/70/30 compositions of CM and 3DF

scaffolds represented as a percentage of total PV. The dotted line in

(A) represents the 90% cut-off value, used to determine the pore size at

which 90% of the PV was no longer accessible. Arrows in (A) indicate

direction of pore accessibility measurement, from the scaffold

periphery toward the scaffold interior.

Table 1

Structural characterization of scaffold architecture

Sample Measured vol% porosity (%) Surface area/volume

(mm�1)

Average pore size

(mm)

Pore size range

(mm)

Pore size at X90%

accessible pore

volume (mm)

m=V mCT mCT mCT mCT mCT

CM 300/55/45 75.671.9 81.8 10.0 182 6–450 98

CM 1000/70/30 76.072.1 77.8 12.2 160 6–444 95

3DF 300/55/45 70.271.2 74.4 3.8 380 132–450 380

3DF 1000/70/30 71.571.8 71.2 4.0 388 156–600 360

S. Miot et al. / Biomaterials 26 (2005) 2479–2489 2483

index, defined as the maximal number of pore walls thatmay be cut without completely separating the scaffoldstructure [32]. 3DF scaffolds displayed a very low CD(ranging between 5.1 and 7.3) and a PV that was 90%accessible at pore sizes greater than 360 mm, demonstrat-ing a simple, slightly tortuous pore architecture. Incontrast, CM scaffolds exhibited a higher CD (rangingbetween 246.2 and 407.7) and a PV that was 90%accessible at pore sizes greater than 100 mm, indicative ofa more complex, tortuous pore architecture.

Similar equilibrium modulus and dynamic stiffnessproperties were obtained for both CM and 3DF

scaffolds. 300/55/45 scaffolds yielded higher equilibriummodulus and dynamic stiffness values than 1000/70/30scaffolds of the same geometry (Fig. 3). This wasprobably due to the lower PEG MW and greaterweight percentage ratio of ‘‘stiff’’ PBT blocks to‘‘soft’’ PEGT blocks present in the 300/55/45 scaffolds,and limited water uptake. Furthermore, 300/55/45scaffolds yielded equilibrium modulus valuesapproximately 200–300% higher than native immaturebovine and human articular cartilage, whereas dynamicstiffness values were approximately 50% of nativetissues.

3.2. Histological and immunohistochemical analysis

After expansion in the monolayer for two passages,nasal chondrocytes were seeded and cultured in CM and3DF PEGT/PBT scaffolds based on the composition300/55/45 or 1000/70/30.

After 4 weeks of cultivation in constructs based on300/55/45 scaffolds, most cells appeared elongated,indicating a dedifferentiated phenotype, and the Safra-nin-O staining intensity was relatively faint (Fig. 4A andC). In constructs based on 1000/70/30 scaffolds, cellsexhibited a rounded morphology, characteristic of amore differentiated phenotype, and the extracellularmatrix was intensely stained for sulfated GAG (Fig. 4Band D). A similar trend was observed after 2 weeks ofculture, although GAG staining intensity was generallyweaker than at 4 weeks (data not shown). For eachcomposition of PEGT/PBT, the Safranin-O staining wasrelatively more intense in the constructs based on 3DFscaffolds (Fig. 4C and D) than in those based on CMscaffolds (Fig. 4A and B). After 4 weeks of culture, in allexperimental groups, staining for Safranin-O waspreferential at the periphery of the constructs, whilethe center was generally only faintly stained. Thispattern of GAG distribution is consistent with previousobservations reported for tissue-engineered cartilageconstructs of similar dimensions cultured under staticconditions [33].

After 4 weeks of culture, immunohistochemicalstaining for collagen type I was positive in constructs

ARTICLE IN PRESS

Fig. 3. Equilibrium modulus (20% strain) (A) and dynamic stiffness (0.1Hz) (B) of hydrated CM and 3DF scaffolds compared with natural articular

cartilage: # taken from [19] and * taken from [47].

Fig. 4. Representative cross-sectional fields of constructs generated by 4 weeks culture of human nasal chondrocytes on CM 300/55/45 (A), CM

1000/70/30 (B), 3DF 300/55/45 (C) or 3DF 1000/70/30 (D) scaffolds. Histological sections were stained by Safranin-O. Scale bar=50mm.

S. Miot et al. / Biomaterials 26 (2005) 2479–24892484

from all experimental groups (Fig. 5A, C, E and G),whereas staining for collagen type II was not detected inconstructs based on CM 300/55/45 scaffolds (Fig. 5B).Collagen type II staining was predominant to that forcollagen type I in constructs based on 1000/70/30scaffolds, both in the CM (Fig. 5F) and 3DF (Fig.5H) architectures, whilst it was of similar intensity tocollagen type I staining in constructs based on 3DF 300/55/45 scaffolds (Fig. 5D).

Constructs were also immunohistochemically stainedfor SMA, a contractile actin isoform expressed byfibroblastic cells [34] and dedifferentiated chondrocytes[10,35]. After 2 and 4 weeks of culture, cells were faintlystained in the constructs based on 300/55/45 scaffolds,independently of their architecture, whilst no SMAstaining was detected in constructs based on 1000/70/30scaffolds (data not shown).

3.3. Biochemical analysis

The amounts of accumulated GAG normalized to theDNA content (GAG/DNA) significantly increased from2 to 4 weeks of culture in all experimental groups (Fig.6), indicating a temporal development of cartilaginoustissue, in agreement with the histological observations.After 4 weeks of culture, constructs based on 1000/70/30scaffolds contained significantly higher GAG/DNAratios than those based on 300/55/45 scaffolds, bothfor CM (p ¼ 0:021) and 3DF (p ¼ 0:021) architectures.Constructs based on 3DF scaffolds contained signifi-cantly higher GAG/DNA ratios than those based onCM scaffolds, but only if the 1000/70/30 compositionwas used (p ¼ 0:021) (Fig. 6).

The amounts of DNA were not significantly differentin the constructs based on the four types of scaffolds and

ARTICLE IN PRESS

Fig. 5. Collagen type I (left column) and type II (right column) immunostaining of chondrocytes cultured for 4 weeks in the scaffolds CM 300/55/45

(A, B), 3DF 300/55/45 (C, D), CM 1000/70/30 (E, F) or 3DF 1000/70/30 (G, H). Scale bar=200mm.

S. Miot et al. / Biomaterials 26 (2005) 2479–2489 2485

ARTICLE IN PRESS

Glycosaminoglycans/DNA

0

5

10

15

20

25

30

CM300/55/45

CM1000/70/30

3DF 300/55/45

3DF 1000/70/30

GA

G/D

NA

(m

g/m

g) 2 weeks

4 weeks

o*+

oo

o*

Fig. 6. GAG content normalized to the DNA amount of constructs

generated by chondrocytes cultured for 2 (white bars) or 4 weeks (gray

bars) in the scaffolds CM 300/55/45, CM 1000/70/30, 3DF 300/55/45

and 3DF 1000/70/30. Statistically significant differences (po0:05) areindicated as follows: * different from composition 300/55/45 for the

same architecture; + different from architecture CM for the same

composition; O different from 2 weeks of culture for the same

architecture and composition.

CII expression

0.01

0.10

1.00

10.00

CM300/55/45

CM1000/70/30

3DF 300/55/45

3DF1000/70/30

CII

mR

NA

leve

l

2 weeks

4 weeks

o*+

*

+

o

Fig. 7. Collagen type II (CII) mRNA expression level normalized to

18S rRNA (arbitrary units) by chondrocytes cultured for 2 (white bars)

or 4 weeks (gray bars) in the scaffolds CM 300/55/45, CM 1000/70/30,

3DF 300/55/45 and 3DF 1000/70/30. Statistically significant differ-

ences (po0:05) are indicated as follows: * different from composition

300/55/45 for the same architecture; + different from architecture CM

for the same composition; O different from 2 weeks of culture for the

same architecture and composition.

S. Miot et al. / Biomaterials 26 (2005) 2479–24892486

cultured for 2 or 4 weeks (data not shown), probablyexcluding the possibility that chondrogenesiswas indirectly regulated through differential cellproliferation.

3.4. Collagen type II gene expression

From 2 to 4 weeks of culture, the collagen type IImRNA expression levels increased significantly in theconstructs based on 3DF 1000/70/30 scaffolds(p ¼ 0:021), and decreased significantly in those basedon CM 300/55/45 scaffolds (p ¼ 0:019). No statisticallysignificant temporal change was detected in the otherexperimental groups (Fig. 7).

After 4 weeks of culture, chondrocytes in 1000/70/30scaffolds expressed significantly higher collagentype II mRNA levels than in 300/55/45 scaffolds,both for CM (p ¼ 0:020) and 3DF (p ¼ 0:020)architectures. Chondrocytes cultured in 3DF scaffoldsexpressed significantly higher collagen type IImRNA levels than in CM scaffolds, both for 1000/70/30 (p ¼ 0:020) and 300/55/45 (p ¼ 0:019) composi-tions.

Collagen type II mRNA expression levels as afunction of scaffold composition and architecturewere in agreement with collagen type II proteindeposition, assessed immunohistochemically. Inparticular, after 4 weeks of culture, constructs basedon CM 300/55/45 scaffolds, where collagen type IIprotein was not detected (Fig. 5B), displayed thelowest collagen type II mRNA expression levels(Fig. 7).

4. Discussion

4.1. Summary

In the present study, we demonstrated that a PEGT/PBT composition previously reported to support main-tenance of the chondrocytic phenotype in 2D films (i.e.,1000/70/30) [18] is also capable of promoting chondro-cyte redifferentiation and cartilaginous matrix accumu-lation in 3D porous scaffolds. This result was obtainedfor both of the 3D architectures studied, although ahighly interconnecting and accessible pore architecture(i.e., 3DF scaffolds) further improved chondrogenesis in1000/70/30 scaffolds.

4.2. Cell/biomaterial interactions

The behavior of chondrocytes cultured on a specificsubstrate is determined by sequential events startingfrom integrin-mediated interactions with the proteinsadsorbed on the biomaterial [36], subsequent adhesionand morphological changes, finally regulating thedifferentiation stage and thus the quality and quantityof ECM deposition. As compared to the 300/55/45substrate, the higher wettability of the 1000/70/30material (contact angle of 39711 versus 48731[20,37]), combined with a larger size and density of thehydrophilic PEG chains [38,39], is expected to generate agreater volume of moving water molecules at thescaffold surface, which is known to reduce unspecificadsorption of serum proteins, including fibronectin[14,18]. A reduced adsorption of fibronectin, proteinwhich is generally associated with cell spreading [18,40],is consistent with our result that chondrocytes displayed

ARTICLE IN PRESSS. Miot et al. / Biomaterials 26 (2005) 2479–2489 2487

a more round shape and expressed reduced amounts ofSMA in the hydrophilic 1000/70/30 scaffolds. Based onknown structure–function relationships in chondrocytebiology [40], these morphological and cytoskeletalfeatures would then explain the greater extent of cellredifferentiation, assessed by the higher mRNA expres-sion of collagen type II, and the generation of superiorcartilaginous tissues, assessed by the increased deposi-tion of GAG and the predominance of type II over typeI collagen. The observation that a more hydrophilicsubstrate supports differentiated cellular functions isconsistent with previous studies performed using fetalbovine and human chondrocytes [18,41], rat marrowstromal cells [39], rat hepatocytes [42] and humanhepatoblastoma cells [43]. One of the most innovativeaspects of our work is the indication that the composi-tion of a porous 3D scaffold is instructive for thedevelopment of engineered cartilage in the samedirection predicted by studies of chondrocytedifferentiation performed in 2D membranes [18].Further studies should investigate the behavior of cellscultured in the different substrate compositions usingserum-free media, which would be highly desirable inview of future clinical applications due to the concernsregarding the unpredictable biological effects ofserum and the potential contaminations by xenogeneicpathogens.

4.3. Scaffold architecture

As compared to cells cultured in 1000/70/30 CMscaffolds, those cultured in 1000/70/30 3DF scaffoldsdisplayed an increased extent of redifferentiation andgenerated tissues which were closer to natural cartilage,histologically and biochemically. Since the two scaffoldsshare an identical composition and are characterized bysimilar percent porosity and mechanical properties, theresult was probably due to the higher accessibility cut-off value, average pore size and lower CD of the 3DFscaffolds. This conclusion, which relies on the uniqueopportunity of applying controlled modifications ofselective physical parameters in the scaffolds investi-gated, underlines the importance of designing 3Darchitectures with highly accessible and interconnectingPVs, thereby facilitating the diffusion rate of nutrientsand gases as well as waste removal [44]. The change inthe scaffold architecture from CM to 3DF did not resultin increased GAG fractions if a 300/55/45 compositionwas used. This observation is consistent with a previousstudy in which GAG accumulation by primary bovinechondrocytes cultured for 2 weeks in scaffolds based on300/55/45 composition was similar in both 3DF and CMarchitectures; although when implanted subcutaneouslyin nude mice, 3DF constructs showed significantlyhigher GAG deposition than CM constructs [20]. Takentogether, these findings suggest that, in a static in vitro

environment, scaffold architectures with a highlyaccessible and interconnecting PV can better modulatethe generation of engineered cartilage if certainpre-requisites necessary to promote chondrogenesis(e.g., a supportive substrate composition) are met.Since ECM was found to be preferentially depositedat the construct periphery in all experimentalgroups, further studies should investigate the possibilityof enhancing the uniformity of the generated tissuesby using bioreactors that increase mass transferrates by the application of various regimes of fluid flow[45,46].

5. Conclusions

This study advances the identification of criteria fordesign or selection of scaffolds with properties which areinstructive for the redifferentiation of human expandedchondrocytes and supportive for the generation of 3Dcartilaginous tissues. Specifically, we demonstrated theimportance of a hydrophilic composition containinglarge and densely distributed PEG chains (e.g., 1000/70/30), combined with a highly interconnecting andaccessible pore architecture (e.g., 3DF). Issues that havenot been addressed in this study are related to themechanical properties and degradation rate of thescaffolds. Indeed, in contrast to 300/55/45 scaffolds,those based on the 1000/70/30 composition poorlymatched the equilibrium modulus and dynamicstiffness of native cartilage and are expected tofurther decrease their mechanical properties in ashort period of time due to a relatively fastdegradation rate [37]. In this regard, it has to beunderlined that the choice of the scaffold composi-tion could be related to the cartilage repair strategyselected (e.g., graft of a cell-free scaffold, of afreshly cell-seeded construct or of a more developedengineered tissue). For example, if constructs arepre-cultured for a sufficient period of time, thepresence of deposited ECM could compensate forintrinsically lower mechanical properties of thescaffold, which are on the other hand essentialfor a direct implantation approach. While a highlyinterconnecting and accessible pore architecturewhich also provides suitable mechanical propertiesmay always be recommended (e.g., 3DF), for a strategyrelying on grafting of immature cartilage constructs, onecould envision a composite composition where a back-bone with appropriate mechanical properties (e.g., 300/55/45) is coated with another substrate capable ofpromoting chondrocyte redifferentiation and cartilagi-nous matrix deposition (e.g., 1000/70/30). Further workis required to verify these principles in appropriateanimal models.

ARTICLE IN PRESSS. Miot et al. / Biomaterials 26 (2005) 2479–24892488

Acknowledgments

This work was supported by the Swiss Federal Officefor Education and Science (B.B.W.) under the FifthEuropean Framework Growth Program (FP5 projectSCAFCART, Grant No. 99.0291). We are grateful tothe Ear-Nose-Throat Department, University Hospitalof Basel (head of clinic: Prof. Dr. R. Probst) forproviding cartilage biopsies.

References

[1] Lydon MJ, Minett TW, Tighe BJ. Cellular interactions with

synthetic polymer surfaces in culture. Biomaterials

1985;6:396–402.

[2] Zeltinger J, Sherwood JK, Graham DA, Mueller R, Griffith LG.

Effect of pore size and void fraction on cellular adhesion,

proliferation, and matrix deposition. Tissue Eng 2001;7:557–72.

[3] Singhvi R, Kumar A, Lopez GP, Stephanopoulos GN, Wang DI,

Whitesides GM, Ingber DE. Engineering cell shape and function.

Science 1994;264:696–8.

[4] Loeser RF. Integrin-mediated attachment of articular chondro-

cytes to extracellular matrix proteins. Arthritis Rheum

1993;36:1103–10.

[5] Loty S, Forest N, Boulekbache H, Sautier JM. Cytochalasin D

induces changes in cell shape and promotes in vitro chondrogen-

esis: a morphological study. Biol Cell 1995;83:149–61.

[6] Benya PD, Shaffer JD. Dedifferentiated chondrocytes reexpress

the differentiated collagen phenotype when cultured in agarose

gels. Cell 1982;30:215–24.

[7] Binette F, McQuaid DP, Haudenschild DR, Yaeger PC,

McPherson JM, Tubo R. Expression of a stable articular cartilage

phenotype without evidence of hypertrophy by adult human

articular chondrocytes in vitro. J Orthop Res 1998;16:207–16.

[8] Bonaventure J, Kadhom N, Cohen-Solal L, Ng KH, Bour-

guignon J, Lasselin C, Freisinger P. Reexpression of cartilage-

specific genes by dedifferentiated human articular chondrocytes

cultured in alginate beads. Exp Cell Res 1994;212:97–104.

[9] Grigolo B, Lisignoli G, Piacentini A, Fiorini M, Gobbi P,

Mazzotti G, Duca M, Pavesio A, Facchini A. Evidence for

redifferentiation of human chondrocytes grown on a hyaluronan-

based biomaterial (HYAff 11): molecular, immunohistochemical

and ultrastructural analysis. Biomaterials 2002;23:1187–95.

[10] Martin I, Suetterlin R, Baschong W, Heberer M, Vunjak-

Novakovic G, Freed LE. Enhanced cartilage tissue engineering

by sequential exposure of chondrocytes to FGF-2 during 2D

expansion and BMP-2 during 3D cultivation. J Cell Biochem

2001;83:121–8.

[11] Hambleton J, Schwartz Z, Khare A, Windeler SW, Luna M,

Brooks BP, Dean DD, Boyan BD. Culture surfaces coated with

various implant materials affect chondrocyte growth and meta-

bolism. J Orthop Res 1994;12:542–52.

[12] Ishaug-Riley SL, Okun LE, Prado G, Applegate MA, Ratcliffe A.

Human articular chondrocyte adhesion and proliferation on

synthetic biodegradable polymer films. Biomaterials

1999;20:2245–56.

[13] Lahiji A, Sohrabi A, Hungerford DS, Frondoza CG. Chitosan

supports the expression of extracellular matrix proteins in human

osteoblasts and chondrocytes. J Biomed Mater Res

2000;51:586–95.

[14] Lee JH, Kopecek J, Andrade JD. Protein-resistant surfaces

prepared by PEO-containing block copolymer surfactants. J

Biomed Mater Res 1989;23:351–68.

[15] Papadaki M, Mahmood T, Gupta P, Claase MB, Grijpma DW,

Riesle J, van Blitterswijk CA, Langer R. The different behaviors

of skeletal muscle cells and chondrocytes on PEGT/PBT block

copolymers are related to the surface properties of the substrate. J

Biomed Mater Res 2001;54:47–58.

[16] Cukierman E, Pankov R, Stevens DR, Yamada KM. Taking

cell–matrix adhesions to the third dimension. Science

2001;294:1708–12.

[17] Abbott A. Cell culture: biology’s new dimension. Nature

2003;424:870–2.

[18] Mahmood TA, Miot S, Frank O, van Blitterswijk CA, Langer R,

Martin I, Riesle J. Differential protein adsorption on polymer

biomaterials regulates adhesion and phenotype of human

articular chondrocytes. Submitted for publication.

[19] Woodfield TBF, Malda J, de Wijn J, Peters F, Riesle J, van

Blitterswiijk CA. Design of porous scaffolds for cartilage tissue

engineering using a 3-dimensional fiber-deposition technique.

Biomaterials 2004;25:4149–61.

[20] Malda J, Woodfield T, van der Vloodt F, Wilson C, Martens DE,

Tramper J, van Blitterswiijk CA, Riesle J. The effect of PEGT/

PBT scaffold architecture on the composition of tissue engineered

cartilage. Biomaterials, in press.

[21] Malda J, Kreijveld E, Temenoff JS, van Blitterswijk CA, Riesle J.

Expansion of human nasal chondrocytes on macroporous

microcarriers enhances redifferentiation. Biomaterials

2003;24:5153–61.

[22] Tay A, Farhadi J, Suetterlin R, Pierer G, Heberer M, Martin I.

Cell yield, proliferation and post expansion differentiation

capacity of human ear, nasal and rib chondrocytes. Tissue Eng

2004;10:762–70.

[23] Rotter N, Bonassar LJ, Tobias G, Lebl M, Roy AK, Vacanti CA.

Age dependence of biochemical and biomechanical properties of

tissue-engineered human septal cartilage. Biomaterials

2002;23:3087–94.

[24] Kafienah W, Jakob M, Demarteau O, Frazer A, Barker MD,

Martin I, Hollander AP. Three-dimensional tissue engineering of

hyaline cartilage: comparison of adult nasal and articular

chondrocytes. Tissue Eng 2002;8:817–26.

[25] Du C, Meijer GJ, van de Valk C, Haan RE, Bezemer JM,

Hesseling SC, Cui FZ, de Groot K, Layrolle P. Bone growth in

biomimetic apatite coated porous polyactive 1000PEGT70PBT30

implants. Biomaterials 2002;23:4649–56.

[26] Ruegsegger P, Koller B, Muller R. A microtomographic system

for the nondestructive evaluation of bone architecture. Calcif

Tissue Int 1996;58:24–9.

[27] Muller R, Ruegsegger P. Micro-tomographic imaging for the

nondestructive evaluation of trabecular bone architecture. Stud

Health Technol Inform 1997;40:61–79.

[28] Laib A, Barou O, Vico L, Lafage-Proust MH, Alexandre C,

Rugsegger P. 3D micro-computed tomography of trabecular and

cortical bone architecture with application to a rat model of

immobilisation osteoporosis. Med Biol Eng Comput

2000;38:326–32.

[29] Baschong W, Suetterlin R, Laeng RH. Control of autofluores-

cence of archival formaldehyde-fixed, paraffin-embedded tissue in

confocal laser scanning microscopy (CLSM). J Histochem

Cytochem 2001;49:1565–72.

[30] Farndale RW, Buttle DJ, Barrett AJ. Improved quantitation and

discrimination of sulphated glycosaminoglycans by use of

dimethylmethylene blue. Biochim Biophys Acta 1986;883:

173–7.

[31] Barbero A, Ploegert S, Heberer M, Martin I. Plasticity of clonal

populations of dedifferentiated adult human articular chondro-

cytes. Arthritis Rheum 2003;48:1315–25.

[32] Odgaard A. Three-dimensional methods for quantification of

cancellous bone architecture. Bone 1997;20:315–28.

ARTICLE IN PRESSS. Miot et al. / Biomaterials 26 (2005) 2479–2489 2489

[33] Obradovic B, Meldon J, Freed L, Vunjak-Novakovic G.

Glycosaminoglycans deposition in engineered cartilage: experi-

ments and mathematical model. AIChE J 2000;46:1860–71.

[34] Premdas J, Tang JB, Warner JP, Murray MM, Spector M. The

presence of smooth muscle actin in fibroblasts in the torn human

rotator cuff. J Orthop Res 2001;19:221–8.

[35] Kinner B, Spector M. Smooth muscle actin expression by human

articular chondrocytes and their contraction of a collagen–glyco-

saminoglycan matrix in vitro. J Orthop Res 2001;19:233–41.

[36] Horbett TA. Principles underlying the role of adsorbed plasma

proteins in blood interactions with foreign materials. Cardiovasc

Pathol 1993;2:137S–48S.

[37] Deschamps AA, Claase MB, Sleijster WJ, de Bruijn JD, Grijpma

DW, Feijen J. Design of segmented poly(ether ester) materials and

structures for the tissue engineering of bone. J Control Release

2002;78:175–86.

[38] Gref R, Minamitake Y, Peracchia MT, Trubetskoy V, Torchilin

V, Langer R. Biodegradable long-circulating polymeric nano-

spheres. Science 1994;263:1600–3.

[39] Gopferich A, Peter SJ, Lucke A, Lu L, Mikos AG. Modulation of

marrow stromal cell function using poly(D,L-lactic acid)-block-

poly(ethylene glycol)-monomethyl ether surfaces. J Biomed Mater

Res 1999;46:390–8.

[40] Zanetti NC, Solursh M. Effect of cell shape on cartilage

differentiation. In: Stein WD, Bronner F, editors. Cell shape:

determinants, regulation and regulatory role. New York: Aca-

demic Press; 1989. p. 291–327.

[41] LiVecchi AB, Tombes RM, LaBerge M. In vitro chondrocyte

collagen deposition within porous HDPE: substrate microstruc-

ture and wettability effects. J Biomed Mater Res 1994;28:

839–50.

[42] De Bartolo L, Morelli S, Bader A, Drioli E. Evaluation of cell

behaviour related to physico-chemical properties of polymeric

membranes to be used in bioartificial organs. Biomaterials

2002;23:2485–97.

[43] Krasteva N, Harms U, Albrecht W, Seifert B, Hopp M, Altankov

G, Groth T. Membranes for biohybrid liver support systems—

investigations on hepatocyte attachment, morphology and

growth. Biomaterials 2002;23:2467–78.

[44] Malda J, Woodfield T, van der Vloodt F, et al. The

effect of PEGT/PBT scaffold architecture on oxygen gradients

in tissue engineered cartilaginous constructs. Biomaterials, in

press.

[45] Martin I, Obradovic B, Treppo S, Grodzinsky AJ, Langer R,

Freed LE, Vunjak-Novakovic G. Modulation of the mechanical

properties of tissue engineered cartilage. Biorheology 2000;37:

141–7.

[46] Davisson T, Sah RL, Ratcliffe A. Perfusion increases cell content

and matrix synthesis in chondrocyte three-dimensional cultures.

Tissue Eng 2002;8:807–16.

[47] Treppo S, Koepp H, Quan EC, Cole AA, Kuettner KE,

Grodzinsky AJ. Comparison of biomechanical and biochemical

properties of cartilage from human knee and ankle pairs. J

Orthop Res 2000;18:739–48.