ARTICLE IN PRESS

0142-9612/$ - se

doi:10.1016/j.bi

�CorrespondEngineering, N

Singapore. Tel.

E-mail addr

(F. Chen), g010

(M.A. Woodru

(D.W. Hutmac

Biomaterials 28 (2007) 814–824

www.elsevier.com/locate/biomaterials

Combined marrow stromal cell-sheet techniques and high-strengthbiodegradable composite scaffolds for engineered functional bone grafts

Yefang Zhoua, Fulin Chenb, Saey Tuan Hob, Maria Ann Woodruffb,Tit Meng Lima, Dietmar W. Hutmacherb,c,�

aDepartment of Biological Sciences, Faculty of Science, National University of Singapore, Singapore 119260, SingaporebDivision of Bioengineering, Faculty of Engineering, National University of Singapore, Singapore 119260, Singapore

cDepartment of Orthopaedic Surgery, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 119260, Singapore

Received 21 June 2006; accepted 20 September 2006

Available online 11 October 2006

Abstract

In this study, cell sheets comprising multilayered porcine bone marrow stromal cells (BMSC) were assembled with fully interconnected

scaffolds made from medical-grade polycaprolactone–calcium phosphate (mPCL–CaP), for the engineering of structural and functional

bone grafts. The BMSC sheets were harvested from culture flasks and wrapped around pre-seeded composite scaffolds. The layered cell

sheets integrated well with the scaffold/cell construct and remained viable, with mineralized nodules visible both inside and outside the

scaffold for up to 8 weeks culture. Cells within the constructs underwent classical in vitro osteogenic differentiation with the associated

elevation of alkaline phosphatase activity and bone-related protein expression. In vivo, two sets of cell-sheet-scaffold/cell constructs were

transplanted under the skin of nude rats. The first set of constructs (5� 5� 4mm3) were assembled with BMSC sheets and cultured for 8

weeks before implantation. The second set of constructs (10� 10� 4mm3) was implanted immediately after assembly with BMSC sheets,

with no further in vitro culture. For both groups, neo cortical and well-vascularised cancellous bone were formed within the constructs

with up to 40% bone volume. Histological and immunohistochemical examination revealed that neo bone tissue formed from the pool of

seeded BMSC and the bone formation followed predominantly an endochondral pathway, with woven bone matrix subsequently

maturing into fully mineralized compact bone; exhibiting the histological markers of native bone. These findings demonstrate that large

bone tissues similar to native bone can be regenerated utilizing BMSC sheet techniques in conjunction with composite scaffolds whose

structures are optimized from a mechanical, nutrient transport and vascularization perspective.

r 2006 Elsevier Ltd. All rights reserved.

Keywords: Bone engineering; Bone marrow stromal cells; Mesenchymal stem cell; Composites; Biodegradable polymers; Functional tissue engineering

1. Introduction

Tissue formation within the body, as part of a develop-ment, healing and/or repair process, is a complex event inwhich cell populations in combination with extra cellularmatrix, self-assemble into functional units and ultimatelyinto tissues and organs. There is intense academic and

e front matter r 2006 Elsevier Ltd. All rights reserved.

omaterials.2006.09.032

ing author. Division of Bioengineering, Faculty of

ational University of Singapore, Singapore 119260,

: +65 68741036; fax: +65 67773537.

esses: scip1129@nus.edu.sg (Y. Zhou), cflfx@tom.com

29965@nus.edu.sg (S.T. Ho), biewma@nus.edu.sg

ff), dbsltm@nus.edu.sg (T.M. Lim), biedwh@nus.edu.sg

her).

commercial interest in finding methods to stimulate andcontrol these events, and eventually to replicate these eventsoutside the body as closely as possible [1]. This interest hasresulted in tissue engineering (TE) emerging as a well-recognized research area, in the arena of regenerativemedicine. The most common concept underlying TE is tocombine living cells, biologically active molecules and astructural scaffold to form a ‘‘tissue-engineering construct’’(TEC) to promote the repair and regeneration of tissues [2].In the bone TE literature [3–5], an increasing trend is

seen towards the fabrication of polymer-based compositematerials to obtain matrices with better osteoconductiveproperties. The main approach is to develop composite andbiomimetic scaffolds with nanocrystallites of inorganic

ARTICLE IN PRESSY. Zhou et al. / Biomaterials 28 (2007) 814–824 815

biological compounds such as calcium phosphate (CaP)salts dispersed within polymer matrices. Composites ofpolymer and ceramics are favored by bone engineers due totheir favorable mechanical properties, including strengthvia the ceramic phase, toughness and plasticity via thepolymer phase, and greatly improved mechanical stiffnessof the matrix [5]. Biological advantages of these compo-sites, such as improved biocompatibility arise from anincreased initial flash spread of serum proteins in the invitro phase compared to the more hydrophobic polymersurface. When used in vivo, ceramic particles bothembedded and on the surface of the composites allow forimproved tissue integration by buffering pH change andproviding a suitable microenvironment that mimics thehost tissue’s inorganic phase [6].

In the past, we have successfully reported strategies forbone engineering that are based on immobilizing bonemarrow stromal cells (BMSC) within fibrin glue that is theninjected into medical-grade polycaprolactone (mPCL)-based matrices [7]. However, despite fibrin glue achievinghigher seeding efficiencies, its cell-specific degradationkinetics and cost might mitigate its widespread use. Hencewe have developed our second-generation scaffolds frommPCL–CaP composites whose mechanical properties are inthe range of cancellous bone and that have improved cellattachment and subsequent cell proliferation, in theabsence of fibrin glue [8]. Despite improved attachmentto PCL–CaP composites in the absence of fibrin, approxi-mately 30–40% of the cells fail to attach due to the lowsurface to volume ratio of these scaffolds.

To address this issue, this study investigated the use ofcell-sheet technology [9] in conjunction with pre-seeding asan approach for optimizing the aforementioned techniques.Whilst confluent cultured cells are usually harvested byenzymatic digestion before seeding on a scaffold, thistechnology allowed us to harvest an intact cell sheet.Okano’s group developed the use of single sheets of culturedcorneal epithelial cells and multilayered cardiomyocytessheets for engineering transplantable cornea and myocardialtissues using a smart cell culture plate technology [10,11].Other groups [12] utilized dermal fibroblasts sheets for skinand L’heureux and colleagues reported the engineering ofhuman blood vessels by endothelial and smooth muscle cellsheets [13] using standard tissue culture plates.

Our first hypothesis was that partially mineralized andstrong cell sheets could be grown on normal tissue cultureplates under appropriate cell culture conditions. We furtherhypothesized that not only structural, but also functionalbone could be engineered in vivo by using mineralized cellsheets in combination with novel composite scaffolds.

2. Methods

2.1. Scaffold fabrication and characterization

Flakes of mPCL (Birmingham Polymers, Alabama, USA) and tri-

calcium phosphate (TCP) (Biobone Inc, China) (mPCL–TCP) (80:20%)

were prepared into +1.7070.10mm monofilaments via a filament

extrusion process using an extruder built in-house. A fused deposition

modeling (FDM) 3000 rapid prototyping system (Stratasys Inc, USA) was

used to fabricate composite scaffolds with a bulk dimension of

50� 50� 4mm3 (Length, breadth and height, respectively), a lay down

pattern of 01/601/1201 and a porosity of 50%. Details of scaffold

fabrication and characterization can be found elsewhere [14].

mPCL–TCP scaffolds were cut into 5� 5� 4mm3 (length, breadth and

height) or 10� 10� 4mm3 blocks. Five molar NaOH treatments were

used to increase surface roughness and hydrophilicity. Scaffolds were

washed and sterilized using 70% ethanol prior to use. The scaffold

porosity, interconnectivity, distribution of CaP within the mPCL matrix,

compressive modulus and surface/volume ratio were measured as

previously reported [15]. Mechanical compression tests were conducted

using an Instron 4302 Material Testing Sytsem in accordance with the

ASTM D695-96 guidelines. The specimens were compressed at a rate of

1mm/min up to a strain level of approximately 60% at room temperature

under 75% humidity. The stress–strain curves were obtained to evaluate

the compressive stiffness (Young’s modulus) and compressive strength of

scaffolds (n ¼ 4). Phase contrast imaging and tomography (PCIT) were

used to provide a 3D reconstruction of the matrix of a mPCL–CaP

scaffold at the Singapore Synchrotron Light Source. One thousand

projections were collected during the experiment with 2 s per projection.

Each projection was a 2D image with 1392� 1040 pixels and 12-bit

accuracy and the image comprised 976� 976� 728mm3 and a spatial

resolution of �1mm was achieved.

2.2. Cell isolation and culture

Porcine BMSC were isolated and cultured as reported previously [16].

Pigs were obtained from the animal holding unit of the National

University of Singapore (NUS) and samples of bone marrow were

harvested in accordance with the IACUC approval of NUS. Briefly,

BMSC were aspirated from the bone marrow, gradient centrifuged and

plated into flasks containing Dulbecco’s Modified Eagle’s medium

(DMEM) low glucose (GIBCO, CA, USA) containing 10% fetal bovine

serum (GIBCO, USA) and 2% antibiotics (200mg/ml penicillium and

200mg/ml streptomycin). Only passage two to four cultures were used for

all the experiments. At confluence, culture media was changed to

osteogenic media consisting of standard media plus L-ascorbic acid-

2-phosphate (50mg/ml), b-glycerophosphate (10mM) and dexamethasone

(100 nM) (Sigma, USA). Control cultures (uninduction group) were

maintained in standard media. Media was changed every 3 days.

2.3. Fabrication of sheet–scaffold constructs

Confluent cells in flasks with density of 50,000 cells/cm2 were cultured

in osteogenic medium for 1–2 weeks, until BMSC sheets were formed and

could be detached intact from the substratum using a cell scraper (Fig.

1A). The sheets were then wrapped around the pre-seeded scaffolds (1

million cells per scaffold and 25 cm2 sheet for 5� 5� 4mm3 scaffolds, or 4

million cells per scaffolds and 150 cm2 sheet for 10� 10� 4mm3

scaffolds), (Fig. 1B–C). The pre-seeded scaffolds were prepared by

delivering cells in 20–100ml volume and the scaffolds containing the cell

suspensions were kept in an incubator for 2 h before the media was topped

up. The pre-seeded constructs were cultured in DMEM media with 10%

FBS. Cell sheets were wrapped on 5� 5� 4mm3 and 10� 10� 4mm3

scaffolds, respectively. The assembled sheet-scaffold construct was held in

culture medium until implantation (Fig. 1D).

2.4. Construct characterization in vitro

Cell attachment on scaffolds was performed by phalloidin–propidium

iodide (PI) staining according to the manufacturers’ protocol (Molecular

Probes Inc., Oregon, USA). Qualitative cell viability was assessed using

fluorescein diacetate–propidium iodide (FDA–PI) staining. To observe cell

ARTICLE IN PRESS

Fig. 1. Experimental procedures for the assembly of scaffold–cell constructs with cell-sheet construction and implantation. (A) Porcine bone marrow

stromal cell (BMSC) sheets were harvested using cell scraper. (B) The BMSC sheet was wrapped around the pre-seeded scaffold. (C) The constructs after

assembly demonstrated dense sheet coverage over the pre-seeded scaffolds (Right) compared with the unassembled one (Left). (D) Constructs were

implanted into subcutaneous pockets of 6-month-old nude rats and the implants were accepted by the host tissue with no detectable foreign body reaction

over the entire study period.

Y. Zhou et al. / Biomaterials 28 (2007) 814–824816

growth inside the scaffolds, specimens were cut at the middle and the depth

of visualized specimen was 200–300mm. The specimens were viewed using

an Olympus IX70-HLSH100 confocal laser-scanning microscope [17].

For SEM, samples were fixed in paraformaldehyde, dehydrated in a

gradient series of ethanol and sputter-coated with gold. Images were taken

by XL30SEM (FEI Inc, OR, USA) at 15Kv. Von Kossa staining was

utilized to assess the degree of mineralization throughout the scaffold–cell

construct, as referred to in previous reports [18].

2.5. Alkaline phosphatase (ALP) activity

Cellular ALP activity was determined using a kinetic assay, based on

measuring the rate of p-nitrophenol formation from p-nitrophenyl

phosphate (kit 104-LL, Sigma). Briefly, cell lysates and culture super-

natant were combined with 50ml of ALP reagent and the activity measured

in a 96-well plate following 30min incubation at 37 1C. ALP activity was

read at 405 nm (Bio-Rad, USA) as per the manufactures instructions. ALP

activity in the lysates was expressed as nanomoles of p-nitrophenol

produced per minute per microgram of protein [17]. The ALP activity in

the culture media was expressed as unit/L in accordance with manufac-

turer’s protocol since total proteins in the media were mainly from serum

in media and could not be used to normalize all samples.

2.6. RNA isolation and RT-PCR

Total cellular RNA was extracted weekly using Trizol reagent

(Invitrogen Corp., Carlsbad, California, USA) and the cDNA synthesis

was performed from 2mg total RNA using Superscript II and Oligo dT

(Invitrogen) according to the manufacturer’s instructions. The expression

of bone matrix proteins was semi-quantitated by PCR using specific

primers as follows: (1) GAPDH, glyceraldehyde-3-phosphate dehydro-

genase, NM_002046, 192 bp sense 50-GTTCCAATATGATTCCACCC-30,

and anti-sense 50-TGAGTCCTTCCACGATACC-30; (2) Collagen type I

(Col I), NM-000088.2, 501 bp, sense 50-CTGGCAAAGAAGGCGGC-

AAA-30 and anti-sense 50-CTCACCACGATCACCACTCT-30; (3) Runx2,

AY406695, 381 bp, sense 50-CCAGATGGGACTGTGGTTAC-30 and anti-

sense 50-ACTTGGTGCAGAGTTCAGGG-30; (4) Osteopontin (OPN),

AF052124, 347 bp, 50-CCAAGTAAGTCCAACGAAAG-30 and anti-sense

50-GGTGATGTCCTCGTCTGTA-30; (5) Osteocalcin (OCN), X53698.1,

315 bp, sense 50-CATGAGAGCCCTCACACTC-30 and anti-sense 50-

AGAGCGACACCCTAGACCG-30; (6) Osterix (Osx), AF477981, 566 bp,

sense 50-TGGGGTATCTCTTGATTAGG-30 and anti-sense 50-TTGGGG-

CCTTGAGACAGCA-30. Target gene expression values were semi-

quantified by densitometry and standardized against GAPDH. The

reaction products were also cloned into pGEM-TEasy vector (Promega)

and sequenced for confirmation.

2.7. Western blot

Cell lysates were prepared using ice-cold lysis buffer (5mM MgCl2,

150mM NaCl, 1% triton-100, pH 7.5) containing protease inhibitors

(Calbiochem, UK). The protein concentrations were determined using a

Protein Assay Kit (Bio-Rad). Cell lysates (40mg) were resolved by 6–12%

SDS-PAGE (polyacrylamide gel electrophoresis) gels and the proteins

were transferred to nitrocellulose membranes (Amersham, UK). After

blocking with 5% low fat milk in tris-buffered saline (TBS), membranes

were then incubated with either mouse anti-OCN (Biodesign, ME, USA),

OPN (DSHB*, IA, USA) or –actin (Santa Cruz, CA, USA). The primary

antibody was diluted 1:1000 in TBS with 0.1% Tween (TBST) overnight at

4 1C, then incubated for 1 h with secondary antibodies diluted 1:1000 in

TBST. The specific signals were visualized by chemiluminesence (Super-

signal west pico kit, Pierce, USA) and the intensity of bands was assessed

using densitometry.

2.8. Implantation

The animal research protocol was reviewed and approved by the

IACUC of National University of Singapore (NUS) (NIDCR 00–113).

Nude rats originally obtained from Harlan Sprague Dawley (Indianapolis,

USA) were bred and maintained in pathogen-free conditions. Twelve

constructs consisting of cell-sheets assembled with scaffolds measuring

5� 5� 4mm3, were cultured in osteogenic medium for 8 weeks and then

transplanted subcutaneously into the dorsal pockets of 4-month-old nude

rats weighing between 110 and 130 g. Four constructs with cell-sheets

assembled with scaffolds measuring 10� 10� 4mm3 were also implanted

immediately without further culture. For each group, two mPCL–CaP

scaffolds without cells and cell sheets were implanted as a control. Prior to

implantation, cells in two constructs per group were labeled with 2 mM of

the cFDA-SE (carboxyfluorescein diacetate succinimidyl ester) (Molecular

Probes). Transplants were recovered 4, 8 and 12-weeks post-transplanta-

tion, fixed in 4% formalin and decalcified in 10% formic acid for paraffin

embedding. Paraffin sections (10mm) were deparaffinized, hydrated, and

stained with hematoxylin and eosin (H&E), safranin-O/fast green (S/F).

Immunohistochemical analysis was performed by using monoclonal

antibodies against OCN (Sigma) and Col I (Sigma), Col II (Chondrex),

which had been biotinylated through Animal research kits (DAKO, USA).

The signals were visualized using Envision System Horseradish Peroxidase

kits (DAKO) and samples were counterstained with hematoxylin.

2.9. Micro CT (mCT) scan and X-ray analysis

A mCT scanner (Skyscan, Belgium) was used to determine bone growth

occurring in the engineered constructs. Specimens were scanned with a

resolution of 35 mm and approximately 500 scan slices were taken and

reconstructed according to the manufactures recommendations (Skyscan,

Belgium). The output was reconstructed into 3D stacks using Mimics 7.3.

The thresholds used in this study were 68–1732 Housefield units (HU) for

cortical bone and �70–67 HU for cancellous bone based on the threshold

calculations for samples of porcine femur bone. With conventional X-ray

analysis, samples were also analyzed using a Mammomat 3000 (Siemens)

X-ray machine at the condition of 25KV and 2.8Mas. After mCT

ARTICLE IN PRESSY. Zhou et al. / Biomaterials 28 (2007) 814–824 817

scanning and X-ray analysis, the specimens underwent compression

testing as described in Section 2.1 (n ¼ 4).

2.10. Statistics analysis

All values were presented as means7standard deviations. All data was

subjected to one-way analysis of variance (ANOVA) of post-hoc testing

(SPSS Version 11.02). Significance levels were set at po0:05. Data were

the average of three replicates performed under identical conditions.

3. Results

3.1. In vitro characterization of tissue-engineered constructs

BMSC sheet-scaffold constructs were cultured for up to8 weeks and were evaluated based on morphologicalappearance, cell adhesion and viability (Fig. 2). Thecomposite scaffolds (mPCL–CaP, 80:20) with a lay-downpattern of 01/601/1201 presented a fully interconnectedhoney-comb-like pore morphology with a porosity of 50%(Fig. 2A). The surface/volume ratio was 8.5mm2/mm3 andthe scaffolds had a compression modulus of 124MPa andcompressive strength of 4.2MPa. Phase Contrast Imagingand Tomography (PCIT) demonstrated that the CaPparticles with sizes in the mm3 range were uniformlyembedded within PCL matrix (Fig. 2A inset). During invitro expansion, cells were well attached on the surface ofscaffolds and exhibited mesenchymal cell-like morphologywith actin alignment on cell bodies (Fig. 2B,C). After 1week in culture, viable cells (�90%) formed bridges overthe scaffold and only a relatively small proportion of dead,PI-stained cells (Fig. 2D) were observed. After 3–5 weeks in

Fig. 2. BMSC–scaffold constructs as observed by SEM and confocal microscop

image of the matrix with visible particles of CaP (white) in the mm3 range withi

were aligned with the cell body after 3 days. (C) Strong actin fiber staining was

center at day 21 (Phalloidin-PI). (D) BMSCs formed bridges in scaffolds after

scaffold after 35 days, with limited cells stained red. (F) BMSCs in cell sheets w

PI staining). (G) The top view of the constructs shows BMSCs sheets covering

the pores with an even distribution. (I) Distinct mineral-like nodules formed

showing nodules formed at 35 days under induction. Scale bar: (A) 500mm, (

1000mm, (I) 50mm and (J) 100mm.

culture, bars of the interior constructs where covered bylive cells (�70%) that had spread evenly on the surface ofscaffolds and throughout its pores (Fig. 2E). When cell-sheets were wrapped on the pre-seeded mPCL-TCPscaffolds, a high proportion of cells remained viable forup to 8 weeks (Fig. 2F). Assessment of the cell-sheetconstructs after 3 weeks culture, via SEM, showed auniform covering of neo-tissue inside and around the pre-seeded mPCL-TCP scaffolds (Fig. 2G, H) and the calciumdeposition was confirmed by the presence of mineralizednodules (Fig. 2I, J).

3.2. ALP activity

Measurements of ALP activity from cell lysates andmedium following osteogenic media supplementation wereundertaken to assess the ability of the constructs toaugment in vitro osteogenesis. Following induction,cellular ALP activity significantly increased to day 21(Po0:05), and then maintained a plateau to day 56(Fig. 3A). In comparison, cultures from the uninducedgroup expressed minimal ALP activity and failed to showan increase over time. Similarly, ALP activity in themedium following induction was significantly higher thanthe uninduced group reaching a peak at day 49 (Po0:05)that was 4-fold higher than the uninduced group (Fig. 3B).

3.3. Expression of osteogenic biomarkers

To confirm the constructs were able to support in vitroosteogenic differentiation, RNA from cells cultured on the

y. (A) mPCL–CaP scaffold architecture. Inset shows the reconstructed 3D

n the PCL (yellow). (B) Actin fibers visualized using Phalloidin-PI staining

observed at the BMSC–scaffold interface while comparatively weak in the

7 days (FDA-PI staining). (E) The BMSCs maintained viability inside the

rapped onto scaffolds remained viable after 56 days in vitro culture (FDA-

the scaffolds. (H) Dense BMSCs grew inside the scaffolds and nearly filled

on the BMSCs sheet after osteogenic induction. (J) Von kossa staining

B) 50 mm, (C) 80 mm, (D) 200mm, (E) 80 mm, (F) 40 mm, (G) 500mm, (H)

ARTICLE IN PRESS

0

1

2

3

4

5

56

49

4942

35

3528

21

2114

7

Construct non-induction

Construct under induction

Construct non-induction

Construct under induction

7

1

1

0

5

10

15

ALP

act

ivity

(U

nit/L

)

20

25

30

Days

Days

(nm

ol/u

g/m

in)

p-ni

trio

phen

yl p

hosp

hate

6

7

8

9

(B)

(A) 10

Fig. 3. ALP activity of BMSC sheet–scaffold constructs both in cell

lysates (A) and the medium (B). (A) Lysates of constructs demonstrated

significantly elevated ALP activity after osteogenic induction. ALP levels

remained high over the whole culture period. (B) ALP activity in the

medium. Constructs under induction demonstrated higher extracellular

ALP activity than constructs without induction over the whole culture

period (po0:05).

Y. Zhou et al. / Biomaterials 28 (2007) 814–824818

constructs was extracted and RT-PCR was performed tomonitor the temporal expression of the osteo-relatedmRNA transcripts, Runx2, Osx, OPN, Col I and OCN.Fig. 4 shows that Osx and OCN expression levels weresignificantly up-regulated at least 10 and 5 times, respec-tively after induction and maintained this elevated levelthroughout the culture period. OPN expression levels werealso up regulated, and the levels of Runx2 and Col I wereslightly increased in induced constructs. To further confirmosteogenesis on a protein level, OCN and OPN synthesiswere also measured using western blots. As shown inFig. 5, OCN was specifically expressed with inducedconstructs and its expression remained stable over 7 weeksculture. OPN expression increased around 3–4 times atweek 3 and then decreased slightly.

3.4. In vivo bone formation

To assess the in vivo bone-forming properties of theengineered constructs, samples were implanted under theskin flank of nude rats.

For assembled constructs which had undergone 8 weeksin vitro culture, bone formation without fibrous tissuecapsule formation was observed after 4 weeks of implanta-tion with the mineralized tissue observed around theconstruct (Fig. 6A, B). Strong mineralization was verifiedby X-ray radiography and mCT with average bone volume

of 49mm3 (Fig. 6B–C). In contrast, control specimens(mPCL–CaP scaffolds without cells) showed mild fibrousencapsulation with structured and vascularised fibroustissue throughout the scaffold architecture (Fig. 6D). In allcell sheet construct specimens, stages of neo-bone forma-tion appeared to mainly follow a path akin to endochon-dral ossification, as observed by the presence ofhypertrophic chondrocytes within the constructs (Fig. 6E,F). Calcified cartilage tissue was detected at the interface ofneo-bone and neo-cartilage with weak glycosaminoglycan(GAG) staining and eosin staining were also observed(Fig. 6F, G). Intramembranous trabecular bone formationwas also observed with the typical osteocyte embedded indense matrix accompanied by both blood vessel andmarrow cavities next to scaffold bars (Fig. 6H, I). Bone-related extracellular matrix biomarkers were detected.Specifically, the neo-bone stained positive for OCN andCol I and negative for Col II and safranin-O (Fig. 6J–L).The neo cartilage-like tissue, which was detected mainly inthe interior part of construct, stained positive for Col IIand safranin-O but very weakly for Col I and OCN(Fig. 6I, J–L). These implants, with 8 weeks in vitro cultureperiods, demonstrated improved compressive modulusover the control implants but there was no statisticallysignificant difference between them (data not shown).For the constructs implanted immediately after assem-

bly, dense neo tissue was formed around the constructswith a well vascularised and non-fibrotic gross appearance,while the control demonstrated only membrane-like fibroustissue formation (Fig. 7A). Extensive bone formation wasobserved after 4, 8 and 12 weeks in constructs with strongradiograph signal similar to fully mineralized compactbone (Fig. 7B–E). mCT showed that a mineral densitysimilar to cortical and cancellous bone was observed at theperiphery and the interior of the constructs, respectively.(Fig. 7F, G). The volume of neo bone accounted for about40% of the whole construct for all time points with amaximal volume of 403mm3 and the cells within the neo-bone demonstrated positive cFDA staining, previouslylabeled before implantation (Fig. 7H, I). No scaffolddegradation was detected for up to 12 weeks of implanta-tion (data not shown). The bone formation followed bothintramembranous ossification and endochondral ossifica-tion while only limited safranin-O staining and sparselydistributed chondrocytes were observed in constructs(Fig. 7J–M). Most neo-bone tissues demonstrated lamellarbone characteristics, resulting from the rapid manner inwhich bone was laid down (Fig. 7J, K). The alignedosteoblasts were observed on the surface of the mPCL-CaPstruts with osteocytes inside the pore architecture accom-panied by blood vessel and marrow cavities (Fig. 7L).Specifically, the neo-bone stained extensively for OCN andCol I with limited signals detected for Col II in somechondrocyte locations (Fig. 7N–Q). Compression testingrevealed that the constructs had modulus of 184MPa andcompressive strength of 5.2MPa, which was 30% higherthan the control scaffolds (without cells).

ARTICLE IN PRESS

Non-induction(a)

(b)

Days

OCN

OPN

Col I

Runx2

GAPDH

14

12

Rel

ativ

e ex

pres

sion

leve

l

10∗

∗

∗∗

8

6

4

2

Osterix Osteocalcin Osteopontin

Day 1 Day 7 Day 14

Day 28 Day 42 Day 49

0

Osx

Induction

1 7 14 28 42 49 1 7 14 28 42 49

Fig. 4. In vitro RT-PCR profiles to analyze expression of osteogenic marker genes. (A) Representative gel electrophoresis figures showing different gene

expression levels at different time points. OCN, Osx and OPN expression level were up-regulated while Runx2 and Col I levels did not show distinct

variation after osteogenic induction. (B) Intensity levels showing Osx, OCN and OPN mRNA expression levels in sheet-scaffolds constructs significantly

increased after induction (po0:05).

Y. Zhou et al. / Biomaterials 28 (2007) 814–824 819

4. Discussion

In this study we have demonstrated the feasibility ofengineering large (400mm3) structural and functional bonetissue grafts by combining pre-seeded novel mPCL–CaPcomposites and partially mineralized cell-sheets. Theengineered constructs could be candidates for mediumload-bearing applications, or, if used in combination withfixation devices could be utilized in high load-bearingapplications. The scaffold design is based on a fullyinterconnected large pore network which supports goodnutrient support and fast vascularization, slow degradationkinetics and can sustain higher mechanical loadingthan those PCL-based constructs used in low load-bearingareas [19].

The composite scaffolds in this study showed mechanicalproperties in the range of cancellous bone and havedemonstrated good support of BMSC attachment andproliferation as well as differentiation and subsequently invitro and in vivo bone formation. Design and fabricationof composite scaffolds are currently an area of intenseinterest in bone TE, and the main idea is to producescaffolds with inorganic biological compounds such as CaP

salts dispersed within polymer matrices [20,21]. Compositescan confer favorable mechanical properties, includingstrength via the ceramic phase, toughness and plasticityvia the polymer phase, as well as a increased mechanicalstiffness [22]. Most importantly, composite surfaces pro-vide a bioactive environment that leads to improved cellattachment, elevation of certain osteogenic biomarkerexpression levels and better integration with host tissue,when compared to contemporary polymer surfaces. BMSCdemonstrated a classical osteogenic differentiation processon the composites with a clear increase in ALP activity,followed by an increase in OCN expression, which was inline with results from various other studies [23,24].Interestingly, the two key transcriptional factors (Osxand Runx2) revealed different kinetic patterns at thetranscription level. Osx expression was highly upregulatedby induction while Runx2 did not change significantlyduring the entire culture period. The results confirmed thatRunx2 is involved in osteogenesis by modulation of itsactivity not by quantificational change in gene or proteinexpression [25,26].This study demonstrated that cell-sheets with high cell

density could be combined with scaffolds that possess good

ARTICLE IN PRESS

Non-induction Induction(A)

(B)

Days

OCN

OPN

Actin

4.5

4 1 14 28 42

∗

∗

49

3.5

3

2.5

Fol

d ch

ange

2

1.5

1

0.5

0

Osteocalcin Osteopontin

1 14 42 49 14 28 42 49

Fig. 5. Western blot assay for OCN and OPN expression. (A) OCN was specifically detected in induced constructs and OPN expression significantly up-

regulated after induction. (B) Analysis of intensity of bands indicated that OCN expression from constructs under induction peaked at day 28 while OPN

peaked at 14 days. The non-induction samples show no OCN expression and no statistical difference in OPN expression.

Y. Zhou et al. / Biomaterials 28 (2007) 814–824820

mechanical properties and could potentially be placed inbone defects that undergo significant load bearing. Anyscaffold used in bone regeneration is generally required tohave biological, biomechanical and biomaterial functions.These include appropriate porous morphology for thediffusion of nutrients and the invasion of vascularity froma surrounding tissue, appropriate material surface chem-istry and biocompatibility to allow cells to adhere andexpress their normal phenotypes, sufficient mechanicalproperties as a load-bearing construct during the regenera-tion process, and adequate biodegradability after boneremodeling in the porous part has occurred. Thus, ascaffold with controlled design of the macro and micro-structure as well as biological and biomaterial considera-tions should balance these complex functions.

From the viewpoint of biomechanical and functional TEfor bone regeneration, the mechanical function representedby stiffness and strength, and capabilities for transporta-tion and exchange of nutrients and wastes could be apossible design target to be optimized [27]. Since thestiffness of scaffolds is proportional to Fp

2, where Fp is thevolume fraction of the polymer, it is the design goal toincorporate high-modulus materials. This generally in-volves a hard reinforcement phase within a ductile matrix.Despite the incorporation of ceramics resulting in a highermodulus the major problem faced is the interfacialadhesion between the reinforcement and the polymermatrix. Unless effective stress transfer takes place acrossinterfaces, the compression and shear strength are not

expected to be significantly improved. Certainly, a certaindegree of compromise may be unavoidable to produce acomposite system with acceptable scaffold performance inhigh load-bearing areas. Hence, in our developed compo-site system it is a balance of properties between strengthand porosity.Reviewing the bone-engineering literature, it can be

concluded that scaffolds should not only have a pore sizebigger than 300 mm but also possess pore interconnectivityof a similar size to facilitate good in vitro nutrient/wastetransport and in vivo vascularization [28]. However, thedrawback for scaffolds optimized using these designparameters, like the ones used in this study, include acompromised surface/volume ratio, which leads to sig-nificant loss of cells during seeding. The application ofhydrogels such as fibrin glue [14,29] allows the achievementof up to 100% seeding efficiency. However, this eleganttechnology has associated drawbacks such as cell- andpatient-dependent degradation kinetics, and high cost.Hence, our results show that using the cell-sheet technol-ogy circumvents these disadvantages associated withoptimizing the scaffold morphology from a nutrient/wasteand vascularization point of view.The cell-sheet technique has proven effective for the

engineering of soft tissues, including skin [30], blood vessels[13], corneal [10], and myochardium [11] owing to thesheets effectively preserving the cell–cell contact and ECM.To our knowledge this is the first report detailing the use ofcell sheets for hard TE. Our results show the layered cell

ARTICLE IN PRESS

Fig. 6. Cell sheet–scaffold constructs after in-vitro culture for 49 days were implanted into nude rat and harvested after 28 days. (A) Gross appearance of

implants showing the implants were well vascularized. Inset shows that the implants were not encapsulated by rat fibrous tissue. (B) X-ray images. Distinct

bone formation was observed in the constructs. The inset shows the top view of X-ray image. (C) Micro CT images show the neo tissue in the constructs

had similar mineral density to native bone. (D) Hematoxylin and eosin (H&E) staining of cell free scaffold group. Fibrous tissue was mainly found in

constructs. (E) Growth plate like tissue formed in constructs as shown by H&E staining. (F) High-magnification image indicates the osteocyte (black

arrow) formation in neo bone while chondrocytes like cells (white arrow) were located in calcified cartilage. (G) Safranin-O/fast green staining indicated

endochondral ossification. Cartilage-like tissue composed of chondrocytes (white arrow) was observed in the interior part of constructs with positive

safranin-O staining, and next to neo bone tissue embedded with osteocytes (black arrow). (H) Distinct bone tissue formed inside the pores and scaffold

surface. (I) H&E staining indicated typical cancellous bone tissue formation. Marrow cavities (Ma) were surrounded by neo bone tissue with distinct

osteocytes embedded. Blood vessels (BV) penetrated into tissue with red blood cells stained, indicating vascularization. (J) Osteocalcin staining. Strong

signals (black arrow head) were mainly detected at the periphery of constructs while weak signals were detected in the interior of the calcified cartilage. (K)

Collagen type-I staining. Extensive staining (black arrow head) was observed with weak and no staining observed on chondrocyte-like cells. (L) Collagen

type-II staining. Only limited signals (black arrow head) were detected on some chondrocyte-like cells. BV: blood vessel+red blood cells; Ma: marrow; Bo:

bone; OC: osteocyte; CY: chondrocyte; CC: calcified cartilage. Scale bar: (D, F, G, I–L) 50mm (E,H) 200mm.

Y. Zhou et al. / Biomaterials 28 (2007) 814–824 821

sheets after wrapping on composites contributed to theextensive in vivo bone formation and this indicated that thecell-sheet techniques could have promising potential inhard TE. In addition to the obvious advantages of usingcell-sheet techniques to achieve a highly efficient celldelivery system over the traditional cell suspension systems[9,31], the layered and condensed cell sheets might mimicthe in vivo deposition of bone matrix formation processwhere osteoblasts are attached on the mineralized sheet likeextracelluar matrix [32]. Furthermore, it is postulated thatthe cell sheets underwent significant contraction after beingwrapped around the composites scaffolds. It is reported

that cell-generated contraction forces stimulate a conden-sation-like process between the sheets and these biomecha-nical-related forces are thought to play a critical role in thespecific type of tissue that developed [33,34].The neo-bone formed in constructs demonstrated

endochondral ossification characteristics with the observa-tion of hypertrophy chondrocytes, which may be related tothe specific architecture of the constructs—the interior partof the constructs experienced hypoxia and had limitedmedium diffusion. The assembled constructs had a size ofover 4mm, which is the limitation of oxygen and nutrientdiffusion in vivo [21]. A dense mineralized BMSC sheet was

ARTICLE IN PRESS

Fig. 7. Constructs with BMSC sheet-scaffolds were implanted into nude rat, harvested after 28, 56 and 84 days. (A) Gross appearance of BMSC sheet-

scaffolds constructs with induction (Right) and non-induction control constructs (Left) after 4 weeks. (B) X-ray detected bone like tissue formation in 28

day (C) 56 day (D) 84 day implants. (E) X-ray image of constructs without cell seeding (Control). (F) Micro CT demonstrated the overall highly

mineralized tissue similar to cortical (golden) and cancellous (red) bone in implanted constructs after 28 day. (G) Micro CT images disclosed the hard

tissue formation within constructs. The red color represented mineralized tissue with similar density with cancellous bone while the golden color

represented cortical bone. (H) Fluorescence was detected on the formed bone tissue after 28 days; the fluorescence came from the cFDA labeled BMSC. (I)

Micro CT quantification of implant tissue compositions depicted substantial bone formation in the induction group, accounting for 40% total volume for

the 28, 56 and 84 day implantation, while the control group formed only connective tissue. (J) H&E staining shows lamellar bone-like tissue formed in

both outer part and interior of constructs after 28 days. (K) High-magnification image shows wel-organized lamellar bone like tissue (Bo) with distinct

osteocytes located within bone tissue. (L) The typical osteoblasts (OB, black arrow) located on the surface of neo-mineralized tissue with marrow cavities

and blood vessels in 56 days implants. (M) Safranin-O staining demonstrated that hypertrophic chondrocytes (white arrow) were observed in 56 day

implants in very low numbers. High magnification shows the chondrocytes (white arrow) with weak safranin-O staining surrounded by the osteocyte

(black arrow). (N) OCN staining. Strong signals (arrow head) were detected on neo mineralized tissue while very weak to no-signals were detected on

chondrocyte-like cells (O). (P) Collagen type-I staining. Extensive staining was detected on the neo-mineralized tissue in constructs. (Q) Collagen type-II

staining. Limited signals (arrow head) were detected on chondrocyte like cells while no staining for the mineralized tissues. BV: blood vessel+red blood

cells; Ma: marrow; OB: osteoblast; Bo: bone; OC: osteocyte; CY: chondrocyte. Scale bar: (J) 200mm; (H–Q) 50mm.

Y. Zhou et al. / Biomaterials 28 (2007) 814–824822

observed in the periphery of the constructs, which mighthave hindered nutrient diffusion and the early establish-ment of blood vessels into the middle of constructs.Consequently a hypoxia environment, promoting carti-lage-like tissue and chondrocyte development was created[35,36]. Intramembranous bone formation was also ob-served at the outer side of neo bone, demonstrating typicaltrabecular bone features as observed in skeletal formationin a developing embryo and during fracture repair; when

osteogenic cells in the presence of a good blood supply andadjacent to a local deposit of calcium salts will differentiateinto bone cells [37]. This is reasonable, since the cells indense mineralized sheets were well contacted with newblood vessel in vivo.The implanted constructs demonstrated improved me-

chanical properties with compression strength of 184MPa,which is at the lower range of cortical bone. This impliesthat the engineered bone tissue integrated well within the

ARTICLE IN PRESSY. Zhou et al. / Biomaterials 28 (2007) 814–824 823

composites. On the other hand, the mechanical propertiesof constructs were still less than native cortical bone sincethe implants were ectopically implanted, and did notexperience biomechanical stimulation comparable with aorthotopic sites [14]. Given a more dynamic environment,the constructs may exhibit superior mechanical strength.Another observation was that constructs with long invitro culture time (8 weeks) demonstrated lower inferiormechanical properties and less bone volume formationthan the immediate implanted constructs without pre-culture. The results might have arisen owing to: (1) Densersheets in the immediate group (9 layers) than in the in vitrogroup (3 layers) and this meant that the immediate groupexperienced stronger condensation processes due to cell-sheet self-contraction [3,34]; (2) The long in vitro culturetime may also have reduced the ability of cells in constructsto synthesize compact mineralized tissue [38].

5. Conclusion

In conclusion, large bone grafts with a volume of up to400mm3 were engineered through combination of BMSCsheets and composite scaffolds. The results presented inthis study show the potential to use the sheet-scaffoldconstructs in medium and high load-bearing defects.Several studies applying the presented technology in load-bearing orthotopic models in immunocompetent animalshave been planned and are currently being undertaken inour laboratory.

Acknowledgements

We thank Mr. Tan Kim Cheng (Temasek Polytechnic,Singapore) for support in the scaffold fabrication andDr. Marian Cholewa (Singapore Synchrotron LightSource) for support with the Phase Contrast Imaging andTomography (PCIT). This work was supported in part bythe Biomedical Research Council (BMRC), Grant numberR-397-000-005-305.

Reference

[1] Langer R, Vacanti JP. Tissue engineering. Science 1993;260:920–6.

[2] Langer RS, Vacanti JP. Tissue engineering: the challenges ahead. Sci

Am 1999;280:86–9.

[3] Davies JE. Bone engineering. em squared incorporated. Toronto,

Canada, 2000.

[4] Hutmacher DW. Scaffold design and fabrication technologies

for engineering tissues—state of the art and future perspectives.

J Biomater Sci Polym Ed 2001;12:107–24.

[5] Vacanti JP, Langer R. Tissue engineering: the design and fabrication

of living replacement devices for surgical reconstruction and

transplantation. Lancet 1999;354(Suppl. 1):SI32–4.

[6] Rezwan K, Chen QZ, Blaker JJ, Boccaccini AR. Biodegradable and

bioactive porous polymer/inorganic composite scaffolds for bone

tissue engineering. Biomaterials 2006;27:3413–31.

[7] Schantz JT, Teoh SH, Lim TC, Endres M, Lam CX, Hutmacher DW.

Repair of calvarial defects with customized tissue-engineered bone

grafts I. Evaluation of osteogenesis in a three-dimensional culture

system. Tissue Eng 2003;9(Suppl. 1):S113–26.

[8] Endres M, Hutmacher DW, Salgado AJ, Kaps C, Ringe J, Reis RL,

et al. Osteogenic induction of human bone marrow-derived mesench-

ymal progenitor cells in novel synthetic polymer–hydrogel matrices.

Tissue Eng 2003;9:689–702.

[9] Yang J, Yamato M, Kohno C, Nishimoto A, Sekine H, Fukai F,

et al. Cell sheet engineering: recreating tissues without biodegradable

scaffolds. Biomaterials 2005;26:6415–22.

[10] Nishida K, Yamato M, Hayashida Y, Watanabe K, Yamamoto K,

Adachi E, et al. Corneal reconstruction with tissue-engineered cell

sheets composed of autologous oral mucosal epithelium. N Engl J

Med 2004;351:1187–96.

[11] Shimizu T, Sekine H, Yang J, Isoi Y, Yamato M, Kikuchi A, et al.

Polysurgery of cell sheet grafts overcomes diffusion limits to produce

thick, vascularized myocardial tissues. FASEB J 2006.

[12] Pouliot R, Larouche D, Auger FA, Juhasz J, Xu W, Li H, et al.

Reconstructed human skin produced in vitro and grafted on athymic

mice. Transplantation 2002;73:1751–7.

[13] L’Heureux N, Paquet S, Labbe R, Germain L, Auger FA. A

completely biological tissue-engineered human blood vessel. FASEB

J 1998;12:47–56.

[14] Shao XX, Hutmacher DW, Ho ST, Goh JC, Lee EH. Evaluation of a

hybrid scaffold/cell construct in repair of high-load-bearing osteo-

chondral defects in rabbits. Biomaterials 2006;27:1071–80.

[15] Hutmacher DW, Schantz T, Zein I, Ng KW, Teoh SH, Tan KC.

Mechanical properties and cell cultural response of polycaprolactone

scaffolds designed and fabricated via fused deposition modeling. J

Biomed Mater Res 2001;55:203–16.

[16] Chen F, Chen S, Tao K, Feng X, Liu Y, Lei D, et al. Marrow-derived

osteoblasts seeded into porous natural coral to prefabricate a

vascularised bone graft in the shape of a human mandibular ramus:

experimental study in rabbits. Br J Oral Maxillofac Surg 2004;

42:532–7.

[17] Zhou Y, Hutmacher DW, Sae-Lim V, Lim TM. Effect of collagen-I

modified composites on proliferation and differentiation of human

alveolar osteoblasts. Aust J Chem 2006;59:571–8.

[18] Zhou YF, Sae-Lim V, Chou AM, Hutmacher DW, Lim TM. Does

seeding density affect in vitro mineral nodules formation in novel

composite scaffolds? J Biomed Mater Res A 2006;78:183–93.

[19] Schantz JT, Lim TC, Ning C, Teoh SH, Tan KC, Wang SC, et al.

Cranioplasty after trephination using a novel biodegradable burr hole

cover: technical case report. Neurosurgery 2006;58 ONS-E176.

[20] Griffith LG, Naughton G. Tissue engineering—current challenges

and expanding opportunities. Science 2002;295:1009–14.

[21] Muschler GF, Nakamoto C, Griffith LG. Engineering principles of

clinical cell-based tissue engineering. J Bone Joint Surg Am 2004;86-

A:1541–58.

[22] El-Ghannam A. Bone reconstruction: from bioceramics to tissue

engineering. Expert Rev Med Devices 2005;2:87–101.

[23] Owen TA, Aronow M, Shalhoub V, Barone LM, Wilming L,

Tassinari MS, et al. Progressive development of the rat osteoblast

phenotype in vitro: reciprocal relationships in expression of genes

associated with osteoblast proliferation and differentiation during

formation of the bone extracellular matrix. J Cell Physiol 1990;

143:420–30.

[24] Stringa E, Filanti C, Giunciuglio D, Albini A, Manduca

P. Osteoblastic cells from rat long bone. I. Characterization of their

differentiation in culture. Bone 1995;16:663–70.

[25] Prince M, Banerjee C, Javed A, Green J, Lian JB, Stein GS, et al.

Expression and regulation of Runx2/Cbfa1 and osteoblast pheno-

typic markers during the growth and differentiation of human

osteoblasts. J Cell Biochem 2001;80:424–40.

[26] Shui C, Spelsberg TC, Riggs BL, Khosla S. Changes in Runx2/Cbfa1

expression and activity during osteoblastic differentiation of human

bone marrow stromal cells. J Bone Miner Res 2003;18:213–21.

[27] Adachi T, Osako Y, Tanaka M, Hojo M, Hollister SJ. Framework

for optimal design of porous scaffold microstructure by computa-

tional simulation of bone regeneration. Biomaterials 2006;27:

3964–72.

ARTICLE IN PRESSY. Zhou et al. / Biomaterials 28 (2007) 814–824824

[28] Karageorgiou V, Kaplan D. Porosity of 3D biomaterial scaffolds and

osteogenesis. Biomaterials 2005;26:5474–91.

[29] Risbud M, Ringe J, Bhonde R, Sittinger M. In vitro expression of

cartilage-specific markers by chondrocytes on a biocompatible

hydrogel: implications for engineering cartilage tissue. Cell Trans-

plant 2001;10:755–63.

[30] Cooper ML, Andree C, Hansbrough JF, Zapata-Sirvent RL,

Spielvogel RL. Direct comparison of a cultured composite skin

substitute containing human keratinocytes and fibroblasts to an

epidermal sheet graft containing human keratinocytes on athymic

mice. J Invest Dermatol 1993;101:811–9.

[31] Ouyang HW, Toh SL, Goh J, Tay TE, Moe K. Assembly of bone

marrow stromal cell sheets with knitted poly (L-lactide) scaffold for

engineering ligament analogs. J Biomed Mater Res B Appl Biomater

2005;75:264–71.

[32] Goldstein AS. Effect of seeding osteoprogenitor cells as dense

clusters on cell growth and differentiation. Tissue Eng 2001;7:

817–27.

[33] Stopak D, Harris AK. Connective tissue morphogenesis by fibroblast

traction. I. Tissue culture observations. Dev Biol 1982;90:383–98.

[34] Tacchetti C, Tavella S, Dozin B, Quarto R, Robino G, Cancedda

R. Cell condensation in chondrogenic differentiation. Exp Cell Res

1992;200:26–33.

[35] Pfander D, Cramer T, Schipani E, Johnson RS. HIF-1 alpha controls

extracellular matrix synthesis by epiphyseal chondrocytes. J Cell Sci

2003;116:1819–26.

[36] Robins JC, Akeno N, Mukherjee A, Dalal RR, Aronow BJ,

Koopman P, et al. Hypoxia induces chondrocyte-specific gene

expression in mesenchymal cells in association with transcriptional

activation of Sox9. Bone 2005;37:313–22.

[37] Stevens MM, Marini RP, Schaefer D, Aronson J, Langer R, Shastri

VP. In vivo engineering of organs: the bone bioreactor. PNAS

2005;102:11450–5.

[38] Van den Dolder J, Vehof JW, Spauwen PH, Jansen JA. Bone

formation by rat bone marrow cells cultured on titanium fiber mesh:

effect of in vitro culture time. J Biomed Mater Res 2002;62:350–8.