at SciVerse ScienceDirect

Biomaterials 34 (2013) 2947e2959

Contents lists available

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

A nano-hydroxyapatite e Pullulan/dextran polysaccharide compositemacroporous material for bone tissue engineering

Jean Christophe Fricain a,1, Silke Schlaubitz b,1, Catherine Le Visage c,1, Isabelle Arnault a,Sidi Mohammed Derkaoui c, Robin Siadous a, Sylvain Catros a, Charlotte Lalande a,b, Reine Bareille a,Martine Renard b, Thierry Fabre a, Sandro Cornet b, Marlène Durand b, Alain Léonard d,Nouredine Sahraoui d, Didier Letourneur c,1, Joëlle Amédée a,*,1

a Inserm U1026, University Bordeaux Segalen, Tissue Bioengineering, F-33076 Bordeaux, FrancebClinical Research Center e Technological Innovation, Inserm, Bordeaux University Hospital, Pessac 33600, Francec Inserm U698, Cardiovascular Bioengineering, Univ Paris 13, Sorbonne Paris Cité, 75877 Paris Cedex 18, Univ Paris Diderot, CHU X. Bichat, 75018 Paris, Franced Teknimed, L’Union 31240, France

a r t i c l e i n f o

Article history:Received 27 November 2012Accepted 9 January 2013Available online 30 January 2013

Keywords:PolysaccharidesNano-hydroxyapatiteOsteogenic differentiationBone defectMineralizationBone formation

* Corresponding author. Tel.: þ33 5 57 57 17 37; faE-mail address: joelle.amedee@inserm.fr (J. Améd

1 Authors with equal contributions.

0142-9612/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.biomaterials.2013.01.049

a b s t r a c t

Research in bone tissue engineering is focused on the development of alternatives to allogenic andautologous bone grafts that can stimulate bone healing. Here, we present scaffolds composed of thenatural hydrophilic polysaccharides pullulan and dextran, supplemented or not with nanocrystallinehydroxyapatite particles (nHA). In vitro studies revealed that these matrices induced the formation ofmulticellular aggregates and expression of early and late bone specific markers with human bonemarrow stromal cells in medium deprived of osteoinductive factors. In absence of any seeded cells,heterotopic implantation in mice and goat, revealed that only the composite macroporous scaffold(Matrix þ nHA) (i) retained subcutaneously local growth factors, including Bone Morphogenetic Protein2 (BMP2) and VEGF165, (ii) induced the deposition of a biological apatite layer, (iii) favored the formationof a dense mineralized tissue subcutaneously in mice, as well osteoid tissue after intramuscular im-plantation in goat. The composite scaffold was thereafter implanted in orthotopic preclinical models ofcritical size defects, in small and large animals, in three different bony sites, i.e. the femoral condyle ofrat, a transversal mandibular defect and a tibial osteotomy in goat. The Matrix þ nHA induced a highlymineralized tissue in the three models whatever the site of implantation, as well as osteoid tissue andbone tissue regeneration in direct contact to the matrix. We therefore propose this composite matrix asa material for stimulating bone cell differentiation of host mesenchymal stem cells and bone formationfor orthopedic and maxillofacial surgical applications.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

During the past decade, the importance of tissue engineering toaddress limitations in tissue grafting has become increasingly clearfor a wide variety of diseases, including osteoarticular pathologies.The goal is to transplant biofactors (cells, genes and/or proteins)within a suitable biodegradable scaffold, and to combine mechan-ical function with tissue regeneration [1,2]. However, bone tissueengineering has not become routine clinical practice to date, asshown by the limited number of clinical trials of tissue-engineered

x: þ33 5 56 90 05 17.ée).

All rights reserved.

constructs for bone reconstruction [3,4]. Clinicians are still lookingfor a ready-to-use biomaterial, deprived of growth factors or livingcells, that can be produced rapidly for different clinical applicationsin different shapes and sizes, and able to generate bone in critically-sized defects. Therefore, we postulate that amatrix for regeneratinglarge bone defects should promote osteogenic differentiation ofhost mesenchymal stem cells thanks to its own intrinsic chemicaland structural properties. This matrix should also promote thegrowth of a dense mineralized bone tissue and bone formationwithin the defect after its implantation.

To address this issue, materials with high hydrophilic propertiesappear suitable for mimicking the aqueous in vivo environment. Forthis reason, hydrogels have been used extensively as three-dimensional (3D) matrices [5]. They represent promising systemsfor the healing and regeneration of damaged tissues since they are

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e29592948

highly permeable and facilitate the transport of nutrients andmetabolites [5].

One of the first polymers used for bone tissue engineering wasbased on a hydrolytically copolymer of polylactic-co-glycolic acid(PLGA) but its use for large bone defect regeneration was con-troversial as inflammatory events were observed [6]. Numerousstudies have focused on the use of other polymers such as collagen[7], chitosan [8], alginate [9] and porous poly(epsilon-caprolactone-co-L-lactide) sponges [10]. For bone tissue regeneration, thesenatural or synthetic polymers are often associated with calciumphosphate [11], mainly for promoting osteoblast adhesion[12] and differentiation in the context of tissue-engineeredconstructs. In vitro, their ability to activate bone cell differentia-tion of mesenchymal stem cells has always been demonstrated inan osteoinductive medium containing b-glycerophosphate anddexamethasone. In vivo, their ability to generate bone tissue wasalso evidenced when these composite materials are preliminarycombined with mesenchymal stem cells before their orthotopicimplantation.

We used in this study macroporous matrices composed of thenatural hydrophilic polysaccharides pullulan and dextran that hasalready been used for cell therapy [13e16]. Here, these matriceswere supplemented or not with nanocrystalline hydroxyapatite(nHA) particles for bone tissue regeneration. We first evaluatein vitro, the cell responses to these macroporous matrices, theability of these scaffolds to drive Human BoneMarrow Stromal cells(HBMSCs) to osteogenic lineage in a medium totally deprived ofosteoinductive factors. Then, we studied the properties of bothscaffolds, in absence of cell seeding, firstly in ectopic sites, i.e.subcutaneously in mice and intramuscularly in goat and thenorthotopically, using three different experimental models in small(rat) and large (goat) animals.

2. Materials and methods

2.1. Synthesis of nanocrystalline hydroxyapatite nHA

nHAwas synthesized by wet chemical precipitation at room temperature. Fifty-nanometer shaped nHA crystals were obtained and characterized by using trans-mission electron microscopy, Fourier-transformed infrared spectroscopy and X-raydiffraction [17].

2.2. Synthesis and characterization of polysaccharide-based macroporous matrices

Macroporous scaffolds were synthesized using a blend of pullulan/dextran75:25 (pullulan, MW 200,000, Hayashibara Inc.; dextran, MW 500,000, Sigma)prepared by dissolving 9 g of pullulan and 3 g of dextran into 40mL of distilled watercontaining 14 g of NaCl. This scaffold is referred in the text as Matrix scaffold. For thepatented preparation of macroporous scaffolds with nanocrystalline hydroxyapatitenHA (Matrix þ nHA) [18], 13 mL of distilled water were replaced by 13 mL of nano-hydroxyapatite suspension (nHA, 6.36% w/v). Chemical cross-link was carried outusing trisodium trimetaphosphate. After incubation at 50 �C for 15 min, resultingscaffolds were cut into the desired shape, soaked in PBS, then washed extensivelywith a 0.025% NaCl solution. After freeze-drying, scaffolds were stored at roomtemperature until use. To measure hydroxyapatite content in dried scaffolds (weight%), scaffolds were enzymatically digested as previously described [19] and resultinghydroxyapatite suspensions were centrifuged, rinsed 3 times with distilled water,dried at 60 �C for 24 h, then were weighed. Scaffolds (Matrix and Matrix þ nHA)were observed using Environmental Scanning ElectronMicroscopy (ESEM) and BackScattered Electron Microscopy (BSEM) on a JEOL 6700F electron microscope.

2.3. In vitro evaluation of polysaccharide-based macroporous matrices withHBMSCs

2.3.1. Isolation of human bone marrow stromal cellsCells isolated from human bone marrow were aspirated from the femoral dia-

physis or iliac bone after obtaining consent from patients (age 30e80 years) un-dergoing hip prosthesis surgery after trauma. The human bone marrow was thensequentially filteredwith syringes fittedwith 16-,18-, and 21-gauge needles. Culturemedium was Dulbecco’s Modified Eagle’s Medium (DMEM, Gibco) supplementedwith 10% (v/v) Fetal Bovine Serum (FBS, Gibco) and cells were maintained in thisbasal medium for in vitro experiments with scaffolds.

2.3.2. HBMSCs behavior within the 3D scaffoldsThe viability of cells cultured within the two matrices (Matrix and

Matrix þ nHA) was investigated at 7 and 15 days of culture in basal medium usinga L3224 LIVE/DEAD� viability/cytotoxicity kit according to the manufacturer’s pro-tocol (Molecular Probes). The 3D cultured cells were stained with calcein AM andethidium homodimer-1. Membrane-permeant calcein AM is cleaved by esterase inlive cells to yield cytoplasmic green fluorescence and membrane-impermeantethidium homodimer-1 labels membrane-compromised cells with red fluo-rescence. Images were recorded with a fluorescence microscope (Nikon Diaphot 300inverted microscope).

Cell growth was quantified by measuring the metabolic cell activity using theMTT (3-(4,5-dimethylthiazolyl-2)-2,5-diphenyltetrazolium bromide) assay. Theyellow tetrazolium MTT is reduced by metabolically active cells into an insolubleformazan salt, and the resulting intracellular purple formazan is solubilized andquantified by spectrophotometry. Briefly, disks (6 mm diameter � 3 mm thickness)of matrices (Matrix and Matrix þ nHA) were seeded with a density of 250,000HBMSCs per scaffold in DMEM containing 10% (v/v) FCS without osteogenic addi-tives. Cultures were incubated at 37 �C in a humidified atmosphere for 1, 7 and 15days. The culture medium was replenished every 3 days. For each time-point ofculture, after 3 h of incubation of MTT at 37 �C, the MTT solution was removed, theformazan crystals were dissolved in dimethylsulfoxide, and 100 mL were aspiratedand then poured into another 96-well plate for absorbance measurement at 540 nm.Cell proliferation on Tissue Culture Polystyrene (TCPS) of plastic culture dishes wasused as a control. All cell assays described were performed with at least 4 replicatesfor each condition tested.

Staining of actin fibers of paraformaldehyde fixed and permeabilized cells cul-tured for 7 days within the matrices in basal medium was performed by incubationwith Alexa 488-conjugated phalloidin (Molecular Probes, 1:40). Nuclei were coun-terstained with 10 mg/mL Hoechst-33342 (Invitrogen). Observations were made ona Leica TCS SP5 confocal microscope (Leica Microsystems, Wetzlar, Germany).

Intracellular Alkaline Phosphatase (ALP) activity was detected in HBMSCs cul-tured in the scaffolds, using the conversion of a colorless p-nitrophenyl phosphate toa colored p-nitrophenol (Sigma diagnostic kit (85L-2), Aldrich). Cells were culturedafter 7 days into the two matrices (Matrix and Matrix þ nHA) in basal medium, andfixed with paraformaldehyde 4% (v/w) for 40 min at 4 �C. Cells were then stainedwith alkaline dye (Fast blue RR salt supplemented with Naphthol AS-MX phosphatealkaline solution 0.25%, Sigma, Aldrich) for 30 min. Samples were observed with anoptical microscope (Zeiss, Axiovert 25).

Immunostaining of osteocalcin (OCN) was performed in HBMSCs cultured for 7days inside the two types of matrices in basal medium. Cells were fixed in 4% (w/v)paraformaldehyde for 40min at 4 �C and permeabilized in Triton X 100 0.1% (v/v) for30 min at 4 �C. Fixed cells were incubated for 1 h in PBS 0.1 M pH 7.4 (Gibco) con-taining 1% (w/v) BSA (bovine serum albumin), then overnight at 4 �C with primarymouse monoclonal antibodies against human osteocalcin (TAKARA, OCG2). Subse-quently, cells were washed in PBS 0.1 M pH 7.4 (Gibco) and incubated with Alexafluor 488-conjugated rabbit anti-mouse IgG 1:4000 (Molecular probes) for 2 h at37 �C. Cells were thereafter observed with a fluorescence microscope (Nikon Dia-phot 300 inverted microscope).

2.3.3. Environmental Scanning Electron Microscopy (ESEM)HBMSCs were cultured for 7 days within the two scaffolds in BM culture con-

ditions, and fixed with 2.5% (w/w) glutaraldehyde/cacodylate for 18 min at roomtemperature for ESEM analysis. Fixed materials were washed twice into cacodylate0.1 M buffer for 5 min. Samples were observed on a QUANTA 200 (FEI Company)scanning electron microscope with controlled pressure. Environmental conditionsin ESEM modality were 7 torr, 7 Kv, 6 mmWD. For ESEM, a specific detector “GSED”(gas secondary electron detector) and the Peltier platine at 4 �C were used duringthe observations. Image acquisition was done using SIS Scandium software.

2.3.4. Quantitative Real Time Polymerase Chain Reaction (Q-PCR)Total RNA was extracted from two-dimensional HBMSCs cultures (plastic

culture dishes as control) and from cells seeded onto the two types of matrices(Matrix and Matrix þ nHA) (500,000 cells/disk of 6 mm diameter � 3 mmthickness) using the Nucleospin� RNA kit (Macherey-Nagel, Düren, Germany). Onemg was used as a template for single-strand cDNA synthesis with the Superscriptpreamplification system (Gibco, Paisley, UK) in a 20 ml final volume containing20 mM TriseHCl pH 8.4, 50 mM KCl, 2.5 mM MgCl2, 0.1 mg/mL BSA, 10 mM DTT,0.5 mM of each dATP, dCTP, dGTP, dTTP, 0.5 mg oligo (dT), and 200 U of reversetranscriptase. After incubation at 42 �C for 50 min, the reaction was stopped at70 �C for 15 min. Five mL of cDNA diluted at a 1:80 ratio were mixed with MESA-green (2�, Fisher Scientific, Pittsburgh, PA, USA) and 10 mM each of forward andreverse primers (Table 1) in a 25 mL final volume and loaded onto a 96-well plate.Amplification was performed using the thermocycler (ICycler, Biorad, Hercules,CA, USA). Data were analyzed with the iCycler IQ� software and compared by theDDCt method. Each Q-PCR was performed in duplicate. Briefly, the mean Ct valueof the target gene was normalized to its averaged Ct values of the housekeepinggene P0 to give a DCt value, which was then normalized to control samples (plasticcultures dishes) to obtain a DDCt value. Three different experiments wereperformed.

Table 1Primer sequences used for PCR amplification. ALP: Alkaline Phosphatase; OCN:Osteocalcin; P0: Ribosomic protein as housekeeping gene.

Humangene

GenBank Primer sequences TM (�C)

ALP BC021289 Forward 50 AGC CCT TCA CTG CCA TCC TGT 30

Reverse 50 ATT CTC TCG TTC ACC GCC CAC 3065

OCN NM_199173 Forward 50 ACC ACA TCG GCT TTC AGG AGG30

Reverse 50 GGG CAA GGG CAA GGG GAA GAG3065

P0 BC015690 Forward 50 ATG CCC AGG GAA GAC AGG GC 30

Reverse 50 CCA TCA GCA CCA CAG CCT TC 3065

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e2959 2949

2.4. Implantation of polysaccharide-based macroporous matrices in small and largeanimals

The procedures and treatment of rats and goats were based on the principles ofLaboratory Animal Care formulated by the National Society for Medical Research andapproved by the Animal Care and Experiment Committee of University of BordeauxSegalen. Experiments were carried out in accredited animal facilities according to Eu-ropean recommendations for laboratoryanimal care (directive86/609CEEof24/11/86).

2.4.1. Non-osseous implantation of matrices in mice (subcutaneously) and goat(intramuscularly)

Inmice,Matrix andMatrixþnHA (cylinders of 4mmdiameter� 4mmthickness)were inserted into subcutaneous pockets in the dorsum of 12-week-old Balb/c miceweighing 25e30 g. Different methods were used for analyzing the newly formedsubcutaneous tissue. Six samples of Matrix and Matrix þ nHA were analyzed bymicro-CT (see Materials and methods 2.5) and histologically (see Materials andmethods 2.6) after 15, 30 and 60 days of implantation. Two samples of Matrix andMatrixþnHAwere analyzedbyX-raydiffraction before the implantation (Day 0) after15, 30 and 60 days of implantation (see Materials and methods 2.7). Six samples ofMatrix and Matrix þ nHA were analyzed by Enzyme Immunoassay (EIA) for quanti-fication of BMP2 and VEGF165 after 2, 15, 30 and 60 days of implantation (seeMaterials and methods 2.8).

In goats,Matrix andMatrixþ nHA (cylinders of 10mmdiameter� 10mmdepth)were inserted into intramuscular sites in seven 4-year-old adult goats with anaverage weight of 70 � 15 kg. Goats were anesthetized by intravenous injection ofthiopental (12 mg/kg) for induction and anesthesia was maintained by inhalation ofhalothane/oxygen. Separate facial incisions were created in the paraspinal muscles(L1eL3). Using blunt dissection, intramuscular pockets were created and filled withtwo samples of Matrix or Matrix þ nHA for each side of the animal. After 1 month ofimplantation, three animals were euthanized by an overdose of pentobarbital andpotassium chloride, and six samples of Matrix and Matrix þ nHA were treated formicro-CTandhistological analysis. After 6months of implantation, four animalswereeuthanized by an overdose of pentobarbital and potassium chloride and six samplesof Matrix and Matrix þ nHA were treated for micro-CT and histological analysis.

2.4.2. Osseous implantation of matrices in the femoral condyle of ratsMedial defects of 5 mm diameter and 6 mm depth were created in left and right

femoral condyles of Wistar rats weighing 350e400 g by using a dental microdrill.Matrix and Matrix þ nHA were implanted into each bone defect. A control withoutscaffold (Empty group) was also performed. Implants were analyzed 15, 30 and 90days after surgery with micro-CT and histological techniques. At each time, sixsamples were used for Matrix and Matrix þ nHA groups and two samples for theempty group. Three separate experiments were performed. Samples were firstanalyzed by micro-CT and then histologically. Each sample was then divided intotwo parts, one for non-decalcified section and the other for decalcification, dehy-dration and paraffin inclusion.

2.4.3. Osseous implantation of Matrix þ nHA in goat mandibleBased on the data obtained in rats, implantations of the Matrix þ nHA versus the

empty group were performed in goat. Matrix þ nHA matrices were inserted into themandibular site in seven 4-year-old adult goats with an average weight of 70 � 15 kg.Seven goats were anesthetized with induction by intravenous injection of thiopental(12 mg/kg) and maintained by inhalation of halothane/oxygen. Both sides of the man-dible were shaved and scrubbed with antiseptic solution. An incision was made alongthe inferior border of the angle of the mandible. Soft tissue and periosteum werereflected down to bone. A bicortical bone defect was made using surgical drills from2 mm to 10 mm diameter. Defects were filled with Matrix þ nHA or left empty. Threeanimalswereeuthanizedafter1month (M1)and fouranimals after6months (M6)byanoverdose of pentobarbital and potassiumchloride. AtM1, four samples ofMatrixþ nHAand one empty sample were treated for micro-CT and histology. At M6, six samples ofMatrix þ nHA and one empty sample were treated for micro-CT and histology.

2.4.4. Osseous implantation of Matrix þ nHA in a tibial osteotomy model in goatImplants of Matrix þ nHA (12 mmwide � 4 cm long) were inserted into a tibial

osteotomy model in seven 4-year-old adult goats with an average weight of

70�15kg. Tibiawas shaved and scrubbedwith antiseptic solution. An incisionof 5 cmwas made along the tibia 8 cm from the epiphysis. Soft tissue and periosteum werereflected down to bone. A bone corner defect of 12mmwide and 4 cm longwasmadeusing a surgical saw. As for the mandibular model, two groups of implantation wereperformed: defects were filled with Matrix þ nHA or left empty. Three animals wereeuthanized after 1 month (M1) and four animals after 6 months (M6) by an overdoseof pentobarbital andpotassiumchloride. AtM1, four samples ofMatrixþnHAandoneempty sample were treated for micro-CT and histology. At M6, six samples ofMatrix þ nHA and one empty sample were treated for micro-CT and histology.

2.5. Micro-computed tomography (micro-CT)

Micro-CT was performed on Explore Locus SP X-ray mCT devices (GeneralElectric, Milwaukee, WI) ex vivowith a source voltage of 80 kV and a current of 60 mAto obtain a 15 mm resolution from 900 X-ray radiographs with an exposure time of3000 ms. After scanning, cross-sectional slices were reconstructed and three-dimensional analyses were performed using eXplore MicroView� software (Gen-eral Electric Healthcare, Milwaukee, WI). Reconstruction of the region of interestwas performed after correction of the center of rotation and calibration of mineraldensity. Each scan was reconstructed using the same calibration system to dis-tinguish bone and air. Mineral Content (MC) andMineral Density (MD) volumeweremeasured for each group. All the samples were treated by micro-CT before histo-logical analysis. After scanning, cross-sectional slices were reconstructed and three-dimensional analyses were performed using Microview� software.

2.6. Histological procedure

One part of the fixed samples was decalcified (Decalcifiant DC3, Labonord,France), dehydrated and paraffin-embedded. Transversal sections (4e5 mm inthickness) were prepared and treated with Masson Trichrome for mineralized bone(blue) and osteoid staining (red). The other part of the samples was dehydrated ina graded series of ethanol and then embedded with methylmethacrylate, which wassubsequently polymerized. Four transversal sections of 10e15 mm were obtainedfrom each sample using a modified diamond blade microtome (Leica MicrosystemsSP1600, France). These sections were stained by Goldner’s Trichrome to assess newbone formation or with Von Kossa to evaluate mineralization. The sections wereanalyzed with an Eclipse 80i light microscope (Nikon, Japan). Pictures werecaptured using a DXM 1200 C (Nikon, Japan) CCD camera.

2.7. X-ray diffraction analysis

Two subcutaneous implants of Matrix and Matrix þ nHAwere retrieved after 15(D15), 30 (D30) and 60 days (D60) of implantation in mice. Two samples of Matrixand Matrix þ nHA before implantation (D0) were used as controls. To obtain a finepowder without any organic tissues, they were treated with bleach for 2 h at roomtemperature, rinsed in PBS and then centrifuged to obtain the pellet. Structuralproperties were explored by diffraction (XRD) using a PANalytical X’pert MPD dif-fractometer (Bragg Brentano tet geometry) equipped with a secondary mono-chromator using copper radiation (mean l ¼ 1.5418 Å).

2.8. Protein extraction from subcutaneous implants and Enzyme Immunoassay (EIA)analysis of VEGF165 and BMP2 within the implants

The amounts of VEGF165 and BMP2 retained within the two types of matrices(Matrix, Matrix þ nHA) implanted subcutaneously were quantified at 2, 15, 30 and60 days after implantation with the VEGF165 immunoassay kit (MMV00,Quantikine�, R&D systems) and BMP2 immunoassay kit (DBP200, Quantikine�, R&Dsystems), according to the manufacturer’s instructions. Data are expressed in pg ofgrowth factor per mg of proteins extracted from the implants and quantified usingbicinchoninic acid protein assay kit (Thermoscientific) (n¼ 6 for each sample and foreach time of implantation). The concentration of growth factors was also quantifiedin muscle and in skin samples, as controls, treated as described above.

2.9. Statistical analysis

All data were expressed as means � standard deviation (SD). They were ana-lyzed using standard analysis of Student’s t-test. Differences were considered sig-nificant when p < 0.05 (*). The symbol ** indicates a significant difference withp < 0.01. ***Indicates a significant difference with p < 0.001. n.s. indicates non-significant difference.

3. Results

3.1. The macroporous composite polysaccharide-based scaffold

Polysaccharide-based porous scaffolds were prepared bychemical cross-linking of dextran and pullulan with sodium

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e29592950

trimetaphosphate, and sodium chloride as porogen [19]. Pre-liminary in vitro and in vivo degradation experiments evidenced thematrix biodegradation, which can be controlled by the cross-linking conditions and porosity [19] [16]. Composite scaffolds(Matrix þ nHA) were prepared by dispersing nHA nanoparticles inthe aqueous polysaccharide solution before the cross-linking step[18]. Macroscopic images of 5 mm diameter freeze-dried scaffoldsare shown in Fig. 1A1 and A2. Scaffold porosity, which was assessedusing ESEM, was around 37% for both types of matrices (Fig. 1A3and A4). In this study, Matrix þ nHA scaffolds contained 2.8 � 0.1%(w/w) of hydroxyapatite. Dispersion of nHA within theMatrix þ nHA was evaluated by BSEM (Fig. 1B) and images withhigh magnification (Fig. 1B2) revealed that these nanoparticuleswere homogeneously dispersed within the composite scaffold.

3.2. HBMSCs behavior within the polysaccharide-based scaffold

We first examined the behavior and morphology of HBMSCscultured within the 3D porous polysaccharide-based scaffolds in

Fig. 1. Macroporous composite polysaccharide-based scaffolds. Macroporous scaffolds werehydroxyapatite suspension (6.36% w/v) was added to the polysaccharide solution. ChemicalMatrix (A1), or supplemented with nHA (Matrix þ nHA) (A2) was identical as observed mScanning Electron Microscopy (A3: Matrix; A4: Matrix þ nHA). (B): Scanning Electron MicrnHA dispersion within the structure (B1, B2).

the culture medium deprived of osteogenic factors (BM). Micro-scopic observations revealed that the two macroporous matrices(Matrix, Matrix þ nHA) induced the formation of multicellularaggregates (Fig. 2A1 and A2). These cellular aggregates were viablein the twomatrices, as revealed by the LIVEDEAD� assay performedafter 15 days of culture (Fig. 2A3 and A4). Their size ranged from 20to 200 mm and they are distributed within the pores of the twomatrices. Actin/phalloidin and DAPI staining indicated that theaggregates formed in both matrices were composed of numerouscells exhibiting actin fibers (Fig. 2B1). ESEM performed after 7 daysof culture confirmed that these cell clusters were integrated withinthe porous structure of the two scaffolds (Fig. 2B2) and surroundedby an extracellular matrix.

Data fromMTTassays indicated that HBMSCs proliferate slightlyfrom day 1 to day 15 within the Matrix (Fig. 2C). The supplemen-tation with nHA (Matrix þ nHA) was beneficial by increasing cellproliferation, reaching a maximum value at day 7, as for Matrixscaffolds. At 7 and 15 days, HBMSCs inMatrixþnHAhad the highestproliferation values, compared to HBMSCs cultured in Matrix.

synthesized using a blend of pullulan/dextran 75:25 into water containing NaCl. Nano-cross-link was carried out using trisodium trimetaphosphate. (A): The morphology ofacroscopically. Porous scaffolds hydrated in PBS were observed with Environmental

oscopy of Matrix þ nHA with backscattered electrons at two magnifications evidenced

Fig. 2. In vitro responses of Human Bone Marrow Stromal Cells (HBMSCs) to the macroporous scaffolds. (A): Optical microscopy shows that HBMSCs formed multicellular aggregateswithin the pores of the Matrix (A1) and the Matrix þ nHA (A2), with viable cells observed by LIVE/DEAD� assay after 15 days of culture in Matrix (A3) and Matrix þ nHA (A4). (B):Multicellular aggregates analysis. (B1): Actin in multicellular aggregates formed within Matrix þ nHA after 7 days of culture was visualized using phalloidin-Alexa 488 (green) andnuclei were stainedwith DAPI (blue). (B2): ESEM ofmulticellular aggregates formedwithinMatrixþ nHA after 7 days of culture. (C): Quantitative analysis of theMTTassays performedfrom day 1 to day 15 of culture in basal medium inMatrix andMatrixþ nHA. (D): mRNA levels of Alkaline Phosphatase (ALP) (D1) and osteocalcin (OCN) (D2) quantified by Real TimePolymerase Chain Reaction after 1 (Day 1) and 7 days (Day 7) of HBMSCs in basal medium within the Matrix ( ) and the Matrix þ nHA (-). Data are expressed as relative geneexpression normalized to P0, as housekeeping gene and to mRNA levels in two-dimensional cultures (,). Three independent experiments were performed; *p � 0.05; ***p � 0.001,N.S: not significant. (D1a): Cytochemistry of alkaline phosphatase activity in HBMSCs cultured for 7 days in the Matrix. (D1b): Cytochemistry of alkaline phosphatase activity inHBMSCs cultured for 7 days inMatrixþ nHA. (D2a): Immunostaining of osteocalcin inHBMSCs cultured for 7 days inMatrix. (D2b): Immunostaining of osteocalcin in HBMSCs culturedfor 7 days in Matrix þ nHA. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e2959 2951

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e29592952

The mRNA levels of an early osteoblastic marker (ALP)(Fig. 2D1), and a late osteoblastic differentiation marker, i.e.osteocalcin (OCN) (Fig. 2D2) were quantified in cells cultured ina Basal Medium (BM) within the Matrix, Matrix þ nHA, or ontoplastic culture dishes as control in the same medium (Fig. 2D).Quantitative PCR showed that the HBMSCswithinMatrix expressedhigh ALP and OCN gene levels after 1 and 7 days, when compared tothe two-dimensional (2D) cultures (Fig. 2D1 and D2). The genelevels were significantly over-expressed compared with HBMSCscultured within the Matrix þ nHA at all culture time-points. Wecould also noticed that the ratio of stimulation for OCN mRNA levelat day 7, was at least 10-fold higher in HBMSCs cultured in the twomatrices compared with 2D cultures. Regarding the proteinexpression after 7 days of culture in BM, multicellular aggregatesdistributed within the pores of the two matrices exhibit the ALPenzymatic activity in Matrix (Fig. 2D1a) and Matrix þ nHA(Fig. 2D1b), as well osteocalcin protein (Fig. 2D2a: Matrix; 2D2b:Matrix þ nHA) revealed here by immunostaining.

Fig. 3. Subcutaneous implantation of matrices in mice: micro-CT, histological and X-ray diffand 4 mm thickness) were inserted into subcutaneous pockets in the dorsum of the 12-wsubcutaneous implantation of Matrix þ nHA on right side (indicated by an arrow) and Matrix(B): Mineral Content (MC) and Mineral Density (MD) were measured from reconstructeMatrix þ nHA). Data are presented as means � standard deviation for n ¼ 8 samples forsignificant difference compared to the other group with p < 0.01. (C): Representative histologimplanted subcutaneously in mice after 15 days (D15) and 60 days (D60). (D): X-ray Diffractdays (D15) after subcutaneous implantation in mice. Diffraction peaks corresponding to Hhistological decalcified sections and Masson Trichrome staining of Matrix þ nHA samples, 60The black arrows indicate osteoblast-like cells adjacent to the Matrix þ nHA (M) (E2). The

3.3. Subcutaneous implantation of the composite polysaccharide-based scaffold in mice

Since these two matrices promote in vitro osteoblastic differen-tiation of HBMSCs, they were first implanted without any cells,subcutaneously in mice. Strikingly, only for the Matrix þ nHAsamples, micro-CT images (Fig. 3A) revealed high degree of miner-alization, beginning at D15 from the periphery of the Matrix þ nHAand increasing significantly over time, as shown after quantificationof MC (Mineral Content) and MD (Mineral Density) at each timepoint (Fig. 3B). For all implantation series, Matrix þ nHA led toa densely calcified tissue after 30 and 60 days. In contrast, no min-eralization occurred with Matrix alone that was implanted on theleft side of the same mouse, whatever the time of implantation(Fig. 3A and B). Von Kossa staining performed on undecalcifiedsections (Fig. 3C) confirmed that Matrix alone did not stimulate theformation of mineralized tissue, whatever the time of implantation(D15 and D60). The porous composite scaffold Matrix þ nHA

raction analysis of the implants. Matrix and Matrix þ nHA (cylinders of 4 mm diametereek-old Balb/c mice weighing 25e30 g. (A): Representative micro-CT images of theon left side (indicated by *), after 15 (D15), 30 (D30) and 60 days (D60) of implantation.d three-dimensional micro-CT images obtained with both matrices ( : Matrix; -:each group and for each time of implantation. The symbol ** indicates a statisticallyical undecalcified sections and Von Kossa staining of Matrix and Matrix þ nHA samplesion (XRD) analysis of powders of Matrix and Matrix þ nHA implants before (D0) and 15ydroxyapatite (HA) are plotted. Halite peaks are indicated by H. (E): Representativedays after implantation at different magnifications. M indicate the Matrix þ nHA (E1).white arrows show numerous vessels formed within the tissue (E3).

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e2959 2953

induced a high and progressive increase of Von Kossa staining fromD15 to D60 (Fig. 3C).

The XRD patterns (Fig. 3D) of powders of Matrix alone before(D0) and 15 days (D15) after subcutaneous implantation in mice,showed only peaks of Halite (H), due to the preparation of thesamples prior XRD analysis, whatever the time of implantation. Inopposite, the phase spectrum of Matrix þ nHA samples at D15evidenced specific carbonated hydroxyapatite peaks (Fig. 3D),while no carbonated HA peak was observed in Matrix þ nHAsamples before implantation (D0). Similar observations wereobserved after 30 and 60 days of implantation.

Masson Trichrome staining performed at D60 on decalcifiedsections of the Matrix þ nHA scaffold implants (Fig. 3E) revealeda dense collagen mineralized tissue (Fig. 3E1), mainly observedaround the implant, exhibiting cuboid cells and lining cells alongthe interface with the scaffold (Fig. 3E2), as well as numerousvessels within this dense collagen network (Fig. 3E3). No inflam-matory responsewas visible with these two scaffolds, whatever thetime of implantation.

3.4. VEGF165andBMP2quantificationwithin thematrices implantedsubcutaneously

The VEGF165 and BMP2 concentrations retained by the twomatrices were quantified during the initial phase (two days aftersurgery), and between 15 and 60 days after subcutaneous im-plantation in mice. The amounts of VEGF165 and BMP2 in skin andmuscle were also measured as controls (Table 2). The BMP2 con-centration retained in the tissue formed within the Matrix þ nHAwas significantly higher than in the Matrix group, whatever thetime of implantation, and was much higher than that observed inmuscle or skin samples. This was especially true at the initial phaseafter two days of implantation, where the amount of BMP2 retainedin the Matrix þ nHA implant was 10-fold higher than in the Matrixalone. On the other hand, the VEGF165 concentration retained inMatrix þ nHA implants was significantly higher than in the Matrixalone at D30 and D60, owing to the involvement of angiogenesiswith blood vessel formation during the regeneration phase.

3.5. Intramuscular implantation of the composite polysaccharide-based scaffold in goat

To complete the ectopic implantation data, the two matrices(Matrix andMatrixþ nHA) were then implanted intramuscularly ingoat. Mineralization of the implants was analyzed after 1 (M1) and6 months (M6) by micro-CT (Fig. 4A). Mineralization of theMatrix þ nHA started from the periphery of the samples at onemonth and was strongly marked after 6 months of implantation,

Table 2Enzyme Immunoassay (EIA) analysis of VEGF165 and BMP2 retained within Matrixand Matrix þ nHA implants after 2 days (D2), 15 days (D15), 30 days (D30) and 60days (D60) of subcutaneous implantation in mice. Muscle and skin were used ascontrols. Data are presented as means � standard deviation for n ¼ 6 samples foreach group and for each time of implantation. n.s. indicates non-significant differ-ence. *Indicates a significant difference with p < 0.05. **Indicates p < 0.01.

VEGF165 (pg/mg protein) BMP2 (pg/mg protein)

Matrix Matrix þ nHA Matrix Matrix þ nHA

D2 58 � 2 68 � 11 ns 123 � 25 1412 � 153**D15 24 � 2 22 � 12 ns 219 � 35 420 � 86*D30 9 � 1 43 � 4* 47 � 3 132 � 13*D60 4 � 1 58 � 3* 52 � 6 286 � 45*

Skin 9 � 2 82 � 1Muscle 11 � 1 8 � 1

compared toMatrix alone for which no specific signal was observedat any time of implantation.

Histology on undecalcified sections with Von Kossa staining andon decalcified sections with Masson Trichrome staining is shown inFig. 4B and C, respectively. Von Kossa staining of non-decalcifiedsections (Fig. 4B) confirmed the presence of a mineralized tissueformed only in the Matrix þ nHA samples at M1 and which washighly visible at M6. Masson Trichrome staining (Fig. 4C) revealednew osteoid tissue in the Matrix þ nHA group at M1 and wellevidenced osteoid tissue at M6 in direct contact with the scaffold.After one month of implantation (M1), around 60% of the samplesshowed new bone formation (4 samples/6 samples analyzed) andaround 90% after 6 months (M6) (5 samples/6 samples analyzed).Masson Trichrome staining revealed the presence of trabeculaestained red with lining osteoblasts and embedded osteocytes.

3.6. Implantation of the composite polysaccharide-based scaffold inthe femoral condyle of rats

Disks of Matrix and Matrix þ nHA were implanted in bone de-fects created in both the left and right side of femoral condyles(Fig. 5A and B). Micro-CT images showed that Matrixþ nHA formeda dense mineralized tissue in the bone defect compared to Matrixalone or the empty group (Fig. 6A). The mineralization increased involume fromD15 to D90, as shown byMC andMD analysis (Fig. 6B).Both micro-CT images and quantification analysis showed thatwhile Matrix alone promoted an increase in MC and MD from D30to D60 compared to the empty group, both MD and MC remainedsignificantly lower than that observed with Matrix þ nHA for eachtime of implantation. The MD of the Matrix þ nHA group reached843 � 87 mg/cm3 after 90 days of implantation, a value similar tothat of trabecular bone in a femoral condyle (around 800 mg/cm3).

Histological analysis on non-decalcified sections confirmed themineralization of the tissue formed within the defect, as evidencedby the increased intensity of Von Kossa staining in Matrix þ nHAsamples from D15 to D90 (Fig. 6C) compared to the empty andMatrix groups at the same time of implantation. Masson Trichromestaining performed at D90 (Fig. 6D) revealed new bone formation,in direct contact to thematerial, mainly in theMatrixþ nHA groupscompared to the Matrix alone and empty groups.

3.7. Implantation of the composite polysaccharide-based scaffold ina mandibular defect in goat

From data obtained in small animals, the Matrix þ nHA wasthereafter implanted in bone defects in large animals. Firstly, theporous composite scaffold was designed (Fig. 5D) for filling a newtransversal mandibular defect of 1 cm in diameter � 8 mm depth(Fig. 5C). Samples were implanted for 1 and 6 months (M1 and M6).Micro-CT analysis showed that Matrix þ nHA induced formation ofa dense mineralized tissue (Fig. 7A) that increased over time. Miner-alization of the bone defect started at 1 month and was clearly evi-denced at 6 months. The mineralized tissue completely filled thecritical size bone defect after 6 months of implantation. Micro-CTimages of the empty group demonstrated that this bone defectcould not heal spontaneously in view of the small amount of miner-alized tissues observed in the empty group after 6 months of im-plantation (Fig. 7A). Von Kossa staining (Fig. 7B) confirmed themicro-CT data. In the empty group, Von Kossa staining showed fewmineralized tissue at 1monthwithin the defect thathad still notfilledit after 6 months of implantation. In opposite, Von Kossa staining inMatrix þ nHA implants increased with time of implantation. InMatrixþ nHA implants, Masson Trichrome staining (Fig. 7C) showednewosteoid tissue in direct contactwith the remainingMatrixþ nHAmatrix samples that filled the bone defect after 6 months of

Fig. 4. Intramuscular implantation of matrices in goat: micro-CT and histological analysis of the implants. Matrix and Matrix þ nHA (cylinders of 10 mm diameter and 10 mmdepth) were inserted into intramuscular sites in seven 4-year-old adult goats with an average weight of 70� 15 kg. Goats were anesthetized by induction by intravenous injection ofthiopental (12 mg/kg) and maintained by inhalation of halothane/oxygen. Separate facial incisions were created in the paraspinal muscles (L1eL3). Using blunt dissection,intramuscular pockets were created and filled with Matrix or Matrix þ nHA samples. One or 6 months after implantation, each animal was euthanized and samples were retrievedfor micro-CT and histological analysis. (A): Representative micro-CT images of the intramuscular implantation of Matrix and Matrix þ nHA, one (M1) and six months (M6) afterimplantation. (B): Representative histological undecalcified sections and Von Kossa staining of Matrix and Matrix þ nHA samples after one and six months. (C): Representativehistological decalcified sections and Masson Trichrome staining of Matrix and Matrix þ nHA samples after one and six months.

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e29592954

implantation. In contrast, Masson Trichrome staining showed anempty defect at both time points in the empty group.

3.8. Implantation of the composite polysaccharide-based scaffold ina tibial osteotomy model in goat

A model of tibial osteotomy was developed by creating a defect12 mm wide and 4 cm deep (Fig. 5E) and Matrix þ nHA wasdesigned for filling this bone defect (Fig. 5F). Samples wereretrieved after 1 and 6 months of implantation. In this model,micro-CT analysis showed that Matrix þ nHA induced the forma-tion of a mineralized tissue within the bone defect that increasedwith time of implantation, and the regeneration of the cortical boneat 6 months (Fig. 8A). In the empty group, the cortical bone was notregenerated and no evidence of mineralizationwas observed in thecenter of the defect, even after 6 months of implantation. Histo-logical analysis confirmed the micro-CT data. Von Kossa stainingevidenced the presence of calcified tissue with Matrix þ nHA(Fig. 8B), and Masson Trichrome staining revealed organizedlamellar bone after 6 months of implantation (Fig. 8C) as well asregeneration of the cortical bone. In contrast, histological analysisdemonstrated that the model was not able to heal spontaneously inthe empty group, although some fibrous tissue could be seen alongthe host bone, but large areas of the bone defect remained empty(Fig. 8C).

4. Discussion

Porous composite scaffolds composed of polymers supple-mented with hydroxyapatite are promising substrates for bone

tissue engineering because of their ability to mimic the naturalnanostructure of the bone tissue. We have developed here porouspolysaccharide-based scaffolds composed of pullulan, dextran andnano-hydroxyapatite particles. This 3D structure was prepared bycross-linking the biopolymers in the absence of organic solvents.Pullulan and dextran are hydrophilic and their biochemical simi-larity with the extracellular matrix is the rationale for their use asbiodegradable scaffolds for tissue engineering [5]. Pullulan isa neutral, linear, non-immunogenic polysaccharide produced bythe fungus Aureobasidium pullulans [20], and is an attractive bio-material thanks to its good mechanical properties and bio-compatibility [21]. Dextran is synthesized from sucrose by bacteriaand was used as a blood substitute. A cross-linking process carriedout in aqueous conditions provided a simple method to obtaina polysaccharide-based scaffold [22] that can be easily manufac-tured into a variety of shapes and sizes for different clinical appli-cations. Investigations combining cross-linking with a porogenagent and a salt-leaching technique allowed preparation of porousscaffolds of controlled porosity that could be colonized by variouscell types including mesenchymal stem cells [16,23]. The porositydetermined by ESEM was around 37%, but could be increased tomore than 50% to improve tissue infiltration, if needed [18]. Inter-estingly, the chemical versatility of these scaffolds makes it possibleto include other components in the 3D structure. Here, nanoscalehydroxyapatite was associated with polysaccharides for orthopedicapplications and to mimic the nano- and the microstructure of thebone tissue. Nano-hydroxyapatite powders can be synthesized bya variety of methods to provide a stoichiometry and a crystallizedproduct, but these methods often require high temperature, longheat treatment times and wet chemical methods that can alter the

Fig. 5. Experimental models performed in rat and goat and corresponding matrices used for bone filling. For osseous implantation in the femoral condyles in rats (A), osseousimplantation in mandibular site in goat (C), osseous implantation in tibial site in goat (E), 5 mm diameter scaffolds and 6 mm depth (B), 1 cm diameter cylinders and 8 mm depth(D), and 12 mm width and 4 cm long corner scaffolds (F) were prepared, respectively.

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e2959 2955

bioactivity of the hydroxyapatite. Here, nano-sized hydroxyapatite(nHA) was synthesized by wet precipitation [24] at room temper-ature. Fifty-nanometer-long nHA crystals have already beenobtained and physico-chemically characterized [17]. Specialattention was paid to the preparation of composites with nano-particules of hydroxyapatite in solution within the polymer [25].The interfacial strength between nHA and the polymer is a veryimportant factor because lack of adhesion between the two phaseswill result in early failure [26]. The nHA solution was mixed withthe aqueous solution of polysaccharides [18], a simple step in thedesign of this nanocomposite, in order to overcome the problem ofagglomeration of nHA particles [27]. Nanoscale-organized HA inthis polymer could offer an increase of specific surface for cell andtissue colonization.

The biological behavior of these macroporous polysaccharide-based scaffolds was first assessed in vitro. Regarding the choice ofthe cell culturemedium to assess the direct effect of the 3Dmatriceson the HBMSCs behavior, we chose a basal medium deprived ofosteogenic factors (b-glycerophosphate, ascorbic acid, dex-amethasone) that could masked the effect of the scaffold on cellproliferation and differentiation. In vitro studies performed withHBMSc demonstrated that both Matrix and Matrix þ nHA offer

a suitable macroporous architecture, allowing the formation ofmulticellular aggregates distributed homogenouslywithin the poresof the scaffolds (Fig. 2). Cells proliferatewithin the two scaffolds, butHBMSCs in Matrix þ nHA achieved higher proliferation rates com-pared with Matrix, during the two weeks of culture. Strikingly,Matrix alone, compared with the two-dimensional cultures onplastic culture dishes or with the Matrix þ nHA, stimulates theexpression of early (ALP) and late osteoblastic (OCN) markers inmesenchymal stem cells, over time of culture. HBMSCs inMatrix þ nHA exhibited comparable values for ALP gene level withthat on 2D cultures but not for OCN expression. As cell-seedingdensity was similar in cultures in Matrix and Matrix þ nHA, theproliferation rates of HBMSCs were distinct inside the twomatrices,suggesting a different sequence of expression of the bone specificmarkers in the two types of scaffolds during thefirstweek of culture.Moreover, the supplementation of nHA within the matrix mayaccelerate in vitro the kinetic of mineralization of the extracellularmatrix, leading to an up-regulation of osteocalcin, a bone specificmarker involved in the control of mineralization. Bone cell differ-entiation of HBMSCs observed in these two scaffolds might berelated to the specific distribution of these cells in multicellularaggregates. The formation of the cell aggregates inside a 3D structure

Fig. 6. Osseous implantation of matrices in the femoral condyles in rats: micro-CT and histological analysis of the implants. Medial defects of 5 mm diameter and 6 mm depth werecreated in both left and right femoral condyles of Wistar rats weighing 350e400 g using a dental microdrill. Matrix and Matrix þ nHA were implanted into each bone defect. Acontrol without scaffold (Empty) was also conducted. Implants were analyzed 15, 30 and 90 days after surgery and treated for micro-CT and histological analysis. At each time, sixsamples were used for Matrix and Matrix þ nHA groups and two samples were used for the empty group. (A): Representative micro-CT images of the femoral condyle of rats, 15days (D15), 30 days (D30) and 90 days (D90) without scaffold (Empty) and after implantation of Matrix or Matrix þ nHA. (B): Mineral Content (MC) and Mineral Density (MD) weremeasured from reconstructed three-dimensional micro-CT images obtained for Empty (,), Matrix ( ), and Matrix þ nHA (-). Data are presented as means � standard deviationfor n ¼ 6 samples for Matrix and Matrix þ nHA groups, and n ¼ 2 for the Empty group. *Indicates a statistically significant difference compared to the other group with p < 0.05. (C):Representative histological undecalcified sections and Von Kossa staining of Empty group, Matrix, and Matrix þ nHA samples at Day 15 (D15) and Day 90 (D90). (D): Representativehistological decalcified sections and Masson Trichrome staining of Empty group, Matrix, and Matrix þ nHA samples at Day 90 (D90).

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e29592956

for tissue regeneration is of particular importance. Cellular com-munication and cell interactions were identified as key factors forcontrolling cell differentiation and extracellular matrix production[28]. The spheroid configuration might promote interactions be-tween cells by way of cellular junctional proteins (cadherins andconnexins), thereby promoting osteogenic differentiation andstimulating the production of a mineralized matrix. This wasobserved with mesenchymal stem cells [28] and embryonic stemcells [29].

Our in vitro data suggested that after implantation in osseoussite of matrices without seeded cells, they may activate osteogenicdifferentiation of the host mesenchymal stem cells. With regards tothese in vitro data, Matrix and Matrix þ nHA were implantedectopically. Two different animal models (mice and goat) and twodifferent ectopic sites (subcutaneous, intramuscular) were used toassess mineralization and osteoid formation. The calcification wasdemonstrated here by micro-CT and Von Kossa staining whileosteoid tissue was evidenced by Masson Trichrome staining ofdecalcified sections. The scaffold without nHA did not induce nei-ther mineralization, nor osteoid formation, whatever the exper-imental models used. In opposite, striking results were obtained

with Matrix þ nHA for the level of mineralization of the newlyformed tissues found in non-osseous sites in small and large ani-mals. We evidenced mineralization of the Matrix þ nHA implantsas early as 15 days after subcutaneous implantation in mice, butwithout osteoid tissue formation. Osteoid formationwas evidencedin Matrix þ nHA implanted intramuscularly in goat after one andsix months of implantation. We observed bone trabeculae liningthe pores of the material including lining osteoblasts and embed-ded osteocytes. This in vivo response appears dependent on theassociation of the two components within the matrix i.e. poly-saccharides and nHA, since the polysaccharides alone (Matrixgroup) do not elicit the same effect as Matrix þ nHA on the calci-fication of tissue after ectopic implantation and not induce osteoidformation intramuscularly in goats.

As mentioned in numerous studies, the physicochemical prop-erties (chemical composition, micro and nanostructure) of bio-materials influence their bioactive potential. Recent reviews andstudies [30e32] highlight at least two mechanisms whereby boneinduction is initiated. Micro and nanostructure could increase thespecific area and promote interactions with growth factors includ-ingBMPs triggeringundifferentiated inducible osteoprogenitor cells

Fig. 7. Osseous implantation of Matrix þ nHA in the mandible of goat: micro-CT and histological analysis of the implants. Matrix þ nHA matrices were inserted into the mandibularsite in seven 4-year-old adult goats with an average weight of 70 � 15 kg. An incision was made along the inferior border of the angle of the mandible and a bicortical bone defectwas made using surgical drills from 2 mm to 10 mm diameter. Defects were filled with Matrix þ nHA or left empty. Animals were euthanized after 1 month and 6 months andsamples were treated for micro-CT. (A): Representative micro-CT images of mandibular implantation of Matrix þ nHA are shown at one month (M1) (n ¼ 4) and six months (M6)(n ¼ 6) after implantation. (B): Representative histological undecalcified sections and Von Kossa staining of Empty group and Matrix þ nHA samples after one and six months. (C):Representative histological decalcified sections and Masson Trichrome staining of Empty group and Matrix þ nHA samples after one and six months. «M» indicates the residualmatrices remaining in the defect.

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e2959 2957

to the osteogenic lineage [33]. In the case of calcium phosphate-based materials, the release of inorganic ions may also help theformation of a biological carbonated apatite layer that may alsocontrol stem cell differentiation. In this work, XRD analysis con-firmed that a biological carbonated apatite layer already appeared15 days after subcutaneous implantation in the tissue formed withMatrix þ nHA only. This is consistent with the hypothesis thatrelease of Ca2þ, PO4

3�, HPO42� from the composite material (here,

from nHA and phosphate links between macromolecular chains)into the surrounding tissuesmay increase the local saturation of thebiologic fluid, causing precipitation of carbonated apatite that in-corporates calcium, phosphate and other ions (Mg2þ, Naþ, CO3

2�),as well as proteins [34].

In addition, we demonstrate in the present work massiveretention of BMP2 in theMatrixþ nHA that was clearly evident twodays after implantation in the ectopic site. The amount of BMP2retained and/or produced by the host cells remained substantiallyhigher in the tissue formed within the macroporous compositescaffold than within the Matrix alone. These data support thefindings of Maire and colleague, showing the ability of dextran-derived hydrogels to retain in vitro osteogenic growth factorsincluding BMP2 [35,36]. Moreover, VEGF165 retained and/or pro-duced in the tissue formed inside the composite scaffold could alsoplay a role in the vascularization of the tissue and the coupling withosteogenesis.

From the in vivo data obtained ectopically, we next implantedthese composite scaffolds in an osseous site, and produced themwith customized shapes and sizes. The ability of fabricating a pre-designed scaffold with the patient-specific anatomy requirements,

is an additional value for this matrix [37]. Clinically relevant in vivomodels are needed to address the challenging construct designdecisions for bone tissue regeneration [38,39]. Bone defect modelsshould be critically-sized, such that they will not heal if left empty.Moreover, in vivo assays in small animals should be made as chal-lenging as possible to allow discrimination of the effects of differentconstruct designs. The rat calvarial defect model can be madecritically-sized and is commonly used to test bone repair materials[40]. However, this experimental model cannot be considered asa physiological bone repair model since no constraint applied to thenewly formed bone. Here, we used a model of bone defect in thefemoral condyles of rats that has been set up for mice [41,42]. A3 mm diameter defect in femoral condyle of rat will heal in somecases when left empty without scaffold (data not shown), whereasa 5 mm diameter defect used in this work will consistently fail toheal. Here, theMatrix without nHA induces less mineralized tissuesthan Matrix þ nHA as demonstrated by micro-CT analysis and byhistology.

One can take into consideration the differences in tissueregeneration kinetics between small and large animals or humans.Nevertheless, it allowed an alternative for screening and opti-mizing biomaterials dedicated for bone regeneration. We thentested the ability of the Matrix þ nHA to generate bone in criticalsize defect in large animals. While some large animal models mayclosely represent the mechanical and physiological human clinicalsituation, it should be pointed out that there are only models, witheach animal model having unique advantages and drawbacks.Currently, there are numerous large animal models for testingimplant materials in vivo (pig, sheep.) [30,39]. More specifically,

Fig. 8. Osseous implantation of Matrix þ nHA in a tibial osteotomy model in goat: micro-CT analysis and histological analysis of the implants. Implants of Matrix þ nHA (corner of12 mm wide and 4 cm length; Fig. 2e and f) were inserted into a tibial osteotomy model in seven 4-year-old adult goats with an average weight of 70 � 15 kg. An incision of 5 cmwas made along the tibia 8 cm from the epiphysis. A bone corner defect of 12 mmwide and 4 cm long was made using a surgical saw. Defects were filled with Matrix þ nHA or leftempty. Goats were euthanized after 1 month and 6 months. (A): Representative micro-CT images of the implantation of Matrix þ nHA in a tibial osteotomy model one month (M1)(n ¼ 4) and six months (M6) (n ¼ 6) after implantation. (B): Representative histological undecalcified sections and Von Kossa staining of Empty group and Matrix þ nHA samplesafter one and six months. (C): Representative histological decalcified sections and Masson Trichrome staining of Empty group and Matrix þ nHA samples after one and six months.«M» indicates the residual matrices remaining in the defect.

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e29592958

for studies investigating boneeimplant interactions, an under-standing of the bone characteristics of the species, such as bonemicrostructure and composition, as well as bone modeling andremodeling properties are important when later extrapolating theresults to the human situation. In another words, while no speciesfulfill all of the requirements of an ideal model, an understandingof the differences in bone architecture and remodeling will assistin the selection of a suitable model for a defined clinical appli-cation. Finally, the size of the animal must be considered to ensurethat it is appropriate for the number and size of implants chosen[43]. Regarding the choice of the large animal models, the pigdemonstrates a good likeness with human bone. However, diffi-culties may arise in relation to their increasing size with time andease of handling. In this respect, sheeps and goats show morepromise as animal models for the testing of bone implant mate-rials. There is little information about the utility of goats versussheep for implant-related studies. The literature reports that thegoat is a suitable animal model for testing human implants andmaterials as they are considered to have a metabolic rate and boneremodeling rate similar to that of humans [44]. Dai et al. [45] alsosupported the use of goats for studies related to bone healing dueto their comparable bone healing capacity and tibial blood supplywith that of humans. In the same way, Lamerigts et al. [46] foundthat the goat is a suitable model to study bone graft incorporation,as the sequence of events occurring during incorporation of bonegrafts is similar in humans and goats.

With respect to these requirements, the ability to Matrix þ nHAto regenerate osteoid formation and the proof of its osteo-conductivity were demonstrated by treating critical size defectsperformed in long bone (osteotomy model) or in the mandibule. In

these two preclinical models in goat, Matrix þ nHA promoteda densemineralized tissue evidenced as well in the periphery of thedefect as in the central part of the defects. In the mandibular model,the defect was completely filled by a mineralized tissue. In theosteotomy model, although micro-CT analysis revealed that thebone defects treated with Matrix þ nHA appeared not totallycompletely filled after 6 months, histological data showed newosteoid formation in the whole defect, as well in the center of thebone defect. In addition, histological data evidenced osteoid tissueapposed to the matrix that remained after 6 months. Although, weobserved a biodegradability of the matrices, we still evidencedmaterials after 6 months of implantation. The biodegradability ofthese composite scaffolds could be improved by either decreasingthe crosslinker content [47] and/or increasing the porosity of thescaffold [19]. For both defects (mandibular site and osteotomy),we observed a dense lamellar collagen network as well as a layeredarrangement of osteocytes. In addition, the tibial osteotomymodel revealed that the Matrix þ nHA supports the regenerationof a dense cortical bone as showed by micro-CT and histologicalstudy.

5. Conclusion

The composite based-polysaccharide scaffold (Matrix þ nHA)offers some strong evidence that a mineral phase incorporatedwithin these natural polysaccharides leads to a chemical andstructural architecture, activates early calcification and osteoidtissue formation in non-osseous and osseous sites in several pre-clinical models from small to large animals.

J.C. Fricain et al. / Biomaterials 34 (2013) 2947e2959 2959

Acknowledgments

This work was supported by grants from Inserm (NationalInstitute for Health and Medical Research), from University Bor-deaux Segalen, Universities of Paris 7 and Paris 13, and by twogrants from the French Research National Agency (ANR-09-EBIO-001 3D Cell; ANR-10-EMMA-009-01 MATRIþ). Thanks to M. Girardand the Laboratory Animal Care Facility. Thanks to TechnologyPlatform for Biomedical Innovation, Bordeaux Segalen University,Bordeaux, France for allowing the micro-CT analysis. Thanks also toE. Lebraud for XRD observations, to P. Guitton for technical supportand to R. Cooke for editing the manuscript.

References

[1] Langer R. The evolution of biomaterials. Nat Mater 2009;8:444e5.[2] Bueno EM, Glowacki J. Cell-free and cell-based approaches for bone regen-

eration. Nat Rev Rheumatol 2009;5:685e97.[3] Chatterjea A, Meijer G, van Blitterswijk C, de Boer J. Clinical application of

human mesenchymal stromal cells for bone tissue engineering. Stem Cells Int2010;2010:215625.

[4] Hollister SJ, Murphy WL. Scaffold translation: barriers between concept andclinic. Tissue Eng Part B Rev 2011;17:459e74.

[5] Gloria A, De Santis R, Ambrosio L. Polymer-based composite scaffolds fortissue engineering. J Appl Biomater Biomech 2010;8:57e67.

[6] Chang NJ, Lin CC, Li CF, Wang DA, Issariyaku N, Yeh ML. The combined effectsof continuous passive motion treatment and acellular PLGA implants onosteochondral regeneration in the rabbit. Biomaterials 2012;33:3153e63.

[7] Kim HW, Gu HJ, Lee HH. Microspheres of collagen-apatite nanocompositeswith osteogenic potential for tissue engineering. Tissue Eng 2007;13:965e73.

[8] Costa-Pinto AR, Correlo VM, Sol PC, Bhattacharya M, Srouji S, Livne E, et al.Chitosan-poly(butylene succinate) scaffolds and human bone marrow stromalcells induce bone repair in a mouse calvaria model. J Tissue Eng Regen Med2012;6:21e8.

[9] Ghahramanpoor MK, Hassani Najafabadi SA, Abdouss M, Bagheri F, BaghabanEslaminejad M. A hydrophobically-modified alginate gel system: utility in therepair of articular cartilage defects. J Mater Sci Mater Med 2011;22:2365e75.

[10] Holmbom J, Sodergard A, Ekholm E, Martson M, Kuusilehto A, Saukko P, et al.Long-term evaluation of porous poly(epsilon-caprolactone-co-L-lactide) asa bone-filling material. J Biomed Mater Res A 2005;75:308e15.

[11] Thein-Han WW, Misra RD. Biomimetic chitosan-nanohydroxyapatite com-posite scaffolds for bone tissue engineering. Acta Biomater 2009;5:1182e97.

[12] Balasundaram G. Using hydroxyapatite nanoparticles and decreased crystal-linity to promote osteoblast adhesion similar to functionalizing with RGD.Biomaterials 2006;27:2798e805.

[13] Abed A, Assoul N, Ba M, Derkaoui SM, Portes P, Louedec L, et al. Influence ofpolysaccharide composition on the biocompatibility of pullulan/dextran-based hydrogels. J Biomed Mater Res A 2011;96:535e42.

[14] Le Visage C, Chaubet F, Autissier A, Letourneur D. Method for preparingporous scaffold for tissue engineering. Patent WO 2009/047346, 2009.

[15] Le Visage C, Letourneur D. Method for preparing porous scaffold for tissueengineering, cell culture and cell delivery. Patent WO2009/047347, 2009.

[16] Le Visage C, Gournay O, Benguirat N, Hamidi S, Chaussumier L, Mougenot N,et al. Mesenchymal stem cell delivery into rat infarcted myocardium usinga porous polysaccharide-based scaffold: a quantitative comparison withendocardial injection. Tissue Eng Part A 2012;18:35e44.

[17] Catros S, Fricain JC, Guillotin B, Pippenger B, Bareille R, Remy M, et al. Laser-assisted bioprinting for creating on-demand patterns of human osteoproge-nitor cells and nano-hydroxyapatite. Biofabrication 2011;3:025001.

[18] Derkaoui S, Le Visage C, Letourneur D, Fricain J, Catros S, Amédée J. Porouspolysaccharide scaffold comprising nano-hydroxyapatite and use for boneformation. Patent PCT/EP2011/064924, 2010.

[19] Autissier A, Le Visage C, Pouzet C, Chaubet F, Letourneur D. Fabrication ofporous polysaccharide-based scaffolds using a combined freeze-drying/cross-linking process. Acta Biomater 2010;6:3640e8.

[20] Leathers TD. Biotechnological production and applications of pullulan. ApplMicrobiol Biotechnol 2003;62:468e73.

[21] Shingel KI. Current knowledge on biosynthesis, biological activity, andchemical modification of the exopolysaccharide, pullulan. Carbohydr Res2004;339:447e60.

[22] Lack S, Dulong V, Picton L, Le Cerf D, Condamine E. High-resolution nuclearmagnetic resonance spectroscopy studies of polysaccharides crosslinked by

sodium trimetaphosphate: a proposal for the reaction mechanism. CarbohydrRes 2007;342:943e53.

[23] Lalande C, Miraux S, Derkaoui SM, Mornet S, Bareille R, Fricain JC, et al.Magnetic resonance imaging tracking of human adipose derived stromal cellswithin three-dimensional scaffolds for bone tissue engineering. Eur Cell Mater2011;21:341e54.

[24] Afshar A, Ghorbani M, Ehsani N, Saeri MR, Sorell CC. Some important factorsin the wet precipitation process of hydroxyapatite. Mater Des. 2003;24:197e203.

[25] Sun F, Zhou H, Lee J. Various preparation methods of highly porous hy-droxyapatite/polymer nanoscale biocomposites for bone regeneration. ActaBiomater 2011;7:3813e28.

[26] Supova M. Problem of hydroxyapatite dispersion in polymer matrices: a re-view. J Mater Sci Mater Med 2009;20:1201e13.

[27] Boissard CI, Bourban PE, Tami AE, Alini M, Eglin D. Nanohydroxyapatite/pol-y(ester urethane) scaffold for bone tissue engineering. Acta Biomater 2009;5:3316e27.

[28] Burns JS, Rasmussen PL, Larsen KH, Schroder HD, Kassem M. Parameters inthree-dimensional osteospheroids of telomerized human mesenchymal(stromal) stem cells grown on osteoconductive scaffolds that predict in vivobone-forming potential. Tissue Eng Part A 2010;16:2331e42.

[29] Gothard D, Roberts SJ, Shakesheff KM, Buttery LD. Engineering embryonicstem-cell aggregation allows an enhanced osteogenic differentiation in vitro.Tissue Eng Part C Methods 2010;16:583e95.

[30] Barradas AM, Yuan H, van Blitterswijk CA, Habibovic P. Osteoinductive bio-materials: current knowledge of properties, experimental models and bio-logical mechanisms. Eur Cell Mater 2011;21:407e29.

[31] Li J, Habibovic P, Yuan H, van den Doel M, Wilson CE, de Wijn JR, et al. Bio-logical performance in goats of a porous titanium alloy-biphasic calciumphosphate composite. Biomaterials 2007;28:4209e18.

[32] Miron RJ, Zhang YF. Osteoinduction: a review of old concepts with newstandards. J Dent Res 2012;91:736e44.

[33] Ripamonti U. Smart biomaterials with intrinsic osteoinductivity: geometriccontrol of bone differentiation. In: Davies JM, editor. Bone engineering. Tor-onto, Canada: Em Squared Corporation; 2000. p. 215e21.

[34] Habibovic P, Yuan H, van der Valk CM, Meijer G, van Blitterswijk CA, deGroot K. 3D microenvironment as essential element for osteoinduction bybiomaterials. Biomaterials 2005;26:3565e75.

[35] Maire M, Chaubet F, Mary P, Blanchat C, Meunier A, Logeart-Avramoglou D.Bovine BMP osteoinductive potential enhanced by functionalized dextran-derived hydrogels. Biomaterials 2005;26:5085e92.

[36] Maire M, Logeart-Avramoglou D, Degat MC, Chaubet F. Retention of trans-forming growth factor beta1 using functionalized dextran-based hydrogels.Biomaterials 2005;26:1771e80.

[37] Billiet T, Vandenhaute M, Schelfhout J, Van Vlierberghe S, Dubruel P. A reviewof trends and limitations in hydrogel-rapid prototyping for tissue engineering.Biomaterials 2012;33:6020e41.

[38] Cancedda R, Giannoni P, Mastrogiacomo M. A tissue engineering approach tobone repair in large animal models and in clinical practice. Biomaterials 2007;28:4240e50.

[39] Pearce AI, Richards RG, Milz S, Schneider E, Pearce SG. Animal models forimplant biomaterial research in bone: a review. Eur Cell Mater 2007;13:1e10.

[40] Schlegel KA, Lang FJ, Donath K, Kulow JT, Wiltfang J. The monocortical criticalsize bone defect as an alternative experimental model in testing bone sub-stitute materials. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2006;102:7e13.

[41] Grellier M, Granja PL, Fricain JC, Bidarra SJ, Renard M, Bareille R, et al. Theeffect of the co-immobilization of human osteoprogenitors and endothelialcells within alginate microspheres on mineralization in a bone defect. Bio-materials 2009;30:3271e8.

[42] Malaval L, Monfoulet L, Fabre T, Pothuaud L, Bareille R, Miraux S, et al.Absence of bone sialoprotein (BSP) impairs cortical defect repair in mouselong bone. Bone 2009;45:853e61.

[43] Schimandle JH, Boden SD. Spine update. Animal use in spinal research. Spine1994;19:2474e7.

[44] Anderson ML, Dhert WJ, de Bruijn JD, Dalmeijer RA, Leenders H, vanBlitterswijk CA, et al. Critical size defect in the goat’s os ilium. A model toevaluate bone grafts and substitutes. Clin Orthop Relat Res 1999:231e9.

[45] Dai KR, Xu XL, Tang TT, Zhu ZA, Yu CF, Lou JR, et al. Repairing of goat tibialbone defects with BMP-2 gene-modified tissue-engineered bone. Calcif TissueInt 2005;77:55e61.

[46] Lamerigts NM, Buma P, Huiskes R, Schreurs W, Gardeniers J, Slooff TJ. Incor-poration of morsellized bone graft under controlled loading conditions. A newanimal model in the goat. Biomaterials 2000;21:741e7.

[47] Gliko-Kabir I, Yagen B, Baluom M, Rubinstein A. Phosphated crosslinked guarfor colon-specific drug delivery. II. In vitro and in vivo evaluation in the rat.J Control Release 2000;63:129e34.