lable at ScienceDirect

Biomaterials 29 (2008) 3662–3670

Contents lists avai

Biomaterials

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

Polylactic acid fibre-reinforced polycaprolactone scaffolds for bonetissue engineering

Vincenzo Guarino a,*, Filippo Causa b, Paola Taddei c, Michele di Foggia c, Gabriela Ciapetti d,Desiree Martini e, Concezio Fagnano c, Nicola Baldini d, Luigi Ambrosio a

a Institute of Composite and Biomedical Materials (IMCB-CNR), P.le Tecchio 80, 80125 Naples, Italyb Department of Experimental and Clinical Medicine, University of Magna Graecia, Germaneto, 88100 Catanzaro, Italyc Department of Biochemsitry ‘‘G. Moruzzi’’, Alma Mater Studiorum, University of Bologna, Via Belmeloro 8/2, 40126 Bologna, Italyd Laboratory for Pathophysiology of Orthopaedic Implants, Istituti Ortopedici Rizzoli, Via di Barbiano 1/10, 40136 Bologna, Italye Department of Human Anatomy, Alma Mater Studiorum, University of Bologna, Via Irnerio 48, 40126 Bologna, Italy

a r t i c l e i n f o

Article history:Received 26 March 2008Accepted 19 May 2008Available online 10 June 2008

Keywords:Fibrous compositeScaffoldsDegradationProgenitor cellsBone tissue engineering

* Corresponding author. Tel.: þ39 81 768 2513; faxE-mail address: vguarino@unina.it (V. Guarino).

0142-9612/$ – see front matter � 2008 Elsevier Ltd.doi:10.1016/j.biomaterials.2008.05.024

a b s t r a c t

The employment of composite scaffolds with a well-organized architecture and multi-scale porositycertainly represents a valuable approach for achieving a tissue engineered construct to reproduce themiddle and long-term behaviour of hierarchically complex tissues such as spongy bone. In this paper,fibre-reinforced composites scaffold for bone tissue engineering applications is described. These arecomposed of poly-L-lactide acid (PLLA) fibres embedded in a porous poly(3-caprolactone) matrix, andwere obtained by synergistic use of phase inversion/particulate leaching technique and filament windingtechnology. Porosity degree as high as 79.7% was achieved, the bimodal pore size distribution showingpeaks at ca 10 and 200 mm diameter, respectively, accounting for 53.7% and 46.3% of the total porosity. Invitro degradation was carried out in PBS and SBF without significant degradation of the scaffold after 35days, while in NaOH solution, a linear increase of weight lost was observed with preferential degradationof PLLA component. Subsequently, marrow stromal cells (MSC) and human osteoblasts (HOB) reacheda plateau at 3 weeks, while at 5 weeks the number of cells was almost the same. Human marrow stromalcell and trabecular osteoblasts rapidly proliferate on the scaffold up to 3 weeks, promoting an orientedmigration of bone cells along the fibre arrangement. Moreover, the role of seeded HOB and MSC oncomposite degradation mechanism was assessed by demonstrating a more relevant contribution to PLLAdegradation of MSC when compared to HOB. The novel PCL/PLLA composite scaffolds thus showedpromise whenever tuneable porosity, controlled degradability and guided cell–material interaction aresimultaneously requested.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

The biomimicry of the extracellular matrix (ECM) to promotecell/tissue integration represents a widespread approach in tissueengineering. The peculiar ECM fibrillar structure, characterized bycollagen bundles well-interpenetrated in a proteoglycan-basedsubstrate, provides the basic guidance for cell organization, survivaland function. For these reasons, an approach based on the bio-mimicry-inspired design of micro- and nano-structured materialsfor tissue engineered scaffolds certainly moves towards the in-creasing demand to recreate a 3D substrate, ECM-like, for a morephysiological growth of cells and tissue regeneration [1,2].

: þ39 81 768 2404.

All rights reserved.

Over the last two decades, biodegradable polymers, such aspoly(lactic acid) PLA, poly(glycolic acid) PGA and poly(3-capro-lactone) PCL, belonging to the family of poly(a-hydroxyesters), haveemerged as a class of biomaterials of growing interest for applica-tion in surgery, drug delivery and tissue engineering (i.e. sutures forwound healing, devices for bone fracture internal fixation, carriersfor delivery of bioactive molecules, scaffolds for the regeneration oftissues or organs [3]). As a general rule, they exhibit predictable andreproducible mechanical and physical properties, such as tensilestrength, elastic modulus and degradation rate under controlledconditions [4]. Because of their chemical and structural similarities,they degrade in similar fashion through ester hydrolysis processesand decarboxylation from the chain ends, assuring their completeremoval by natural pathways. However, different degradationkinetics may occur, depending on specific interatomic and in-termolecular bonding, which induces a different susceptibility tohydrolytic attack and breakdown [5] with different effects on the

V. Guarino et al. / Biomaterials 29 (2008) 3662–3670 3663

final degradation behaviour. For instance, PCL is very attractive intissue engineering because of its good biocompatibility and proc-essability, but its high hydrophobicity and low degradability in vivomake it less suitable for long-term applications [6,7]. On the con-trary, PLA and PGA show degradation rates, physical and mechan-ical properties adjustable over a wide range by using variousmolecular weights and copolymers, but, their degradation processbased on bulk erosion mechanism may promote the undesiredrelease of acidic products which can cause a strong inflammatoryresponse up to the premature scaffold failure [8].

A key consideration in scaffold design is that the complexcomposition and structural organization of hard mineralized tis-sues, such as bone, cannot be replicated using a single materialoffering a limited range of properties. In particular, scaffolds forbone regeneration have to provide a highly interconnected porousstructure to guide cell ingrowth for tissue formation, whereasmaintaining a sufficient mechanical strength to supply the struc-tural requirement of the substituted tissue [9].

In this context, the development of composite scaffolds bycombining two or more types of materials with selected propertiesmay represent an appropriate and advantageous solution, satisfy-ing all the mechanical and physiological demands of the host tissue[10]. To date, polymer matrices reinforced by ceramic fillers such ashydroxyapatite and tricalcium phosphates represent a promisingmaterial, able to mimic the collagen/hydroxyapatite micro-/macro-morphology of ‘‘native bone material’’ [11–13]. However, thevarying success of these approaches may be attributed to a non-homogeneous spatial distribution of the ceramic particles into thepolymer matrix with poor mechanical response and bioactivity[14].

In this work, an alternative biomimetic approach is proposed,founded on the use of polymer-based composite scaffold withcontrolled pore spatial distribution and programmed degradationkinetics obtained by integrating polymer continuous fibres. Indeed,continuous fibres integrated within the polymeric matrix are ableto mimic the collagen fibres spatial disposition of the ECM, to guidethe preferential cell growth [15–18] and to support the stresstransfer from the polymer matrix improving the mechanicalproperties of the scaffold.

In detail, a composite scaffold obtained by incorporating poly(L-lactide) (PLLA) continuous fibres within a PCL matrix through thesynergic combination of phase inversion/salt leaching techniqueand the filament winding technology has been proposed. Thecombination of phase inversion with salt leaching, per se, assureshighly porous structures with a pore volume fraction exceeding90% of total volume and a bimodal pore distribution [19]. Filamentwinding technology is a powerful instrument to produce compositepolymers with improved mechanical response in compression [20]and tailored degradation kinetics.

The long-term degradation of PCL matrix following a two-stepmechanism – i.e. random hydrolytic ester cleavage and weight lossthrough the diffusion of oligomeric species from the bulk – guar-antees a substrate with mechanical properties and support functionuntil tissue regeneration has been accomplished [21]. Meanwhile,PLLA fibre bundles, which are strictly ordered by helicoidal ar-rangement inside the scaffold due to their degrading faster than thePCL matrix, may promote the formation of preferential channels toguide cell migration.

In the current work, a morphological investigation involving theporosity assessment was performed by using scanning electronmicroscopy, gravimetric measurements and mercury intrusionporosimetry, whereas the in vitro degradation kinetics in differentaqueous media was investigated by the combined use of vibrationalmicro-Raman spectroscopy and thermogravimetry (TGA), alreadyproved suitable for this purpose [22–26]. Finally, the interactions ofhuman cells, marrow stromal cells (MSC) and osteoblasts (HOB),

respectively, with the scaffold have been studied to assess the roleof chemistry and scaffold morphology on bone cell behaviour.

2. Materials and methods

2.1. Materials

Poly(3-caprolactone) (PCL, MW 65 kDa) and PLLA fibres (75 D-tex) were pur-chased, respectively, from Sigma–Aldrich and Sofradim companies. N,N-dimethyl-acetamide (J.T. Baker) and ethanol (Merck, >99.9%) with reagent grade were usedwithout further purification. Sodium chloride crystals (Fluka, AT> 99.9%) weresieved in a specific size range (300–500 mm) by sifting for 3 h.

2.2. Scaffolds preparation

Poly(3-caprolactone) pellets were dissolved in N,N-dimethylacetamide 20% w/wby stirring for about 3 h at 58 �C. NaCl 300–500 mm-sized particles were added tothe PCL matrix (weight ratio: 9/1) to get a homogeneous mixture. Fibre-reinforcedcomposite structures were obtained by the integration of PLLA fibres prepared byfilament winding technique (Fig. 1). The filament winding machine (mod. AS LAB101 – T.EL.MEC, Italy) consists of a rotating mandrel and a fibre dispensing headmoving along the length of the mandrel. The head motion which is synchronizedwith the mandrel rotation enables to define the angle of the fibre with respect to themandrel axis, so that helical plies optimized to handle expected loads can be laiddown. Before the winding on the mandrel, the single fibre has to go across a resinbath: the matrix viscosity and the fibre shear rate play a main role on the fibre-to-matrix adhesion. Moreover, the machine parameters (i.e. winding pitch, windingangle) may be set in order to get different winding patterns as required. In thespecific case, PLLA fibres impregnated with the PCL-based mixture were rotated byusing a winding angle of 45� and a winding pitch of 500 mm, respectively, on Tefloncoated steel mandrel for five fibre-cycle. The solvent was extracted from the matrixby ethanol dipping (35 ml for gram of composite) for 24 h. Then, samples weretotally immersed in bi-distilled water for 7 days to leach out salt and any othercontaminants. Finally, the porous material was dried under hood at room temper-ature for 6 h to obtain the final sample ready to use.

2.3. Scaffold morphological characterization

2.3.1. Porosity: qualitative evaluationOuter and inner pore morphologies of the composite scaffolds were studied by

using Leica Cambridge scanning electron microscope (Stereoscan S440) with a 20 kVacceleration voltage. Briefly, raw composite specimens were fractured parallel andperpendicular to the surface, using a razor blade. The resulting transverse andlongitudinal sections were sputter coated with gold under vacuum (EMSCOPESC500). Representative areas of coated surfaces were observed at different magni-fications. Structural parameters such as the pore content, pore size and inter-connectivity were visualized by SEM, and macropore as well as micropore networks,together forming the total void volume of the scaffold, were measured.

2.3.2. Porosity: quantitative analysisIn order to obtain a quantitative estimation of the porosity degree of composite

scaffolds, two different methods were used – gravimetric measurements andmercury intrusion porosimetry. The former allowed to determine the density andthe porosity degree by calculating, in triplicate, the volume and the scaffold mass.The material density was calculated starting from the ones of each componentnormalized by their volumetric fraction. A rough estimation of the specimenporosity was carried out by using the following equation:

Porosityð%Þ ¼ 1� ½ðweight=volumeÞ=r polymer� � 100

Meanwhile, an accurate analysis of the porosity features was performed by mer-cury intrusion porosimetry (Thermo Electron mod. Pascal 140-240 and ultra macro-pore kit). Through the mercury intrusion within the pores induced by an appliedpressure gradient, it has been possible to perform an accurate evaluation of the totalpore volume fraction, the average pore diameter, and the pore size distribution intothe 3D scaffold. As mercury intrusion occurs, the cumulative volume of the intrudedmercury at each pressure step is recorded in an intrusion curve, while pore sizes andpore volume distribution by pore size are calculated on the basis of the Washburnequation with the assumption of a cubic pore shape [14]. Afterward, bulk and absolutedensity values were calculated in agreement with the mercury volume measure-ments, and compared with those just calculated by gravimetric measurements. Indetail, the bulk density was obtained by the effective mercury volume intruded intothe structure which is referred to scaffold material with pore spaces. Instead, absolutedensity relative to the scaffold strut material is derived by the volume measurement atambient pressure when the sample is only enveloped by the intruding fluid.

2.4. In vitro degradation

The composites were weighed and immersed in three different aqueous media,at 37 �C:

Fig. 1. Scheme of the filament winding procedure used to prepare PLLA fibre-reinforced scaffolds.

V. Guarino et al. / Biomaterials 29 (2008) 3662–36703664

(a) Phosphate-buffered saline (PBS) at pH 7.5 (KH2PO4 0.0087 M, Na2HPO4

0.0304 M and NaCl 0.154 M).(b) Simulated body fluid (SBF) buffered at pH 7.5 (Tris 50 mM, HCl 45 mM) con-taining Naþ (142 mM), Kþ (4 mM), Ca2þ (2.5 mM), Cl� (148.8 mM), HCO3

� (4.2 mM),HPO4

2� (1 mM) [27].(c) 0.01 M of NaOH solution.

Each degradation test was performed in triplicate under sterile conditions. Tothis purpose, the samples were kept in ethanol overnight before immersion in thedegradation media. The NaOH, PBS and SBF solutions were renewed every week.Once a week up to 35 days, the samples were collected, washed with distilled water,vacuum-dried at room temperature and weighed. The % weight loss of the sampleswas calculated according to the equation

% weight loss ¼ 100ðw0 �wtÞw0

(1)

where wt is the dry weight remaining at a given degradation time t and w0 is theinitial weight.

2.4.1. Micro-Raman spectroscopyMicro-Raman spectra of the untreated, in vitro degraded and cell-seeded sam-

ples (see below) were obtained using a Jasco NRS-2000C instrument with a micro-scope with 100� magnification. The spectra were recorded in backscatteringconditions with 5 cm�1 spectral resolution using the 488 nm line (Innova Coherent70) with a power of about 10 mW. The detector was a 160 K frozen digital CCD (Spec-10: 100B, Roper Scientific Inc.). The instrument was equipped with a camera forvisual examination of the specimen surface at the laser focusing spot, to recordseparately the spectra of PCL and PLLA. Due to the intrinsic orientation of fibresamples, micro-Raman polarized spectra of PLLA were recorded; the rotation of thepolarization plane of the incident radiation was obtained with a half-wave plate.According to Porto’s nomenclature, the scattering geometry can be described bya four-letter notation: A(BC)D, where A, B, C, and D stand for one of the axis of theexternal reference frame, X, Y, or Z. A is the direction of the laser beam, B the po-larization of the laser, C the direction of the polarization analyzer and D the directionof view (the direction for the collection of the scattered intensities). To gain insightinto the degree of PLLA chain orientation, the depolarization ratio R of the 875 cm�1

band was calculated as the ratio of the intensity measured in the two X(ZY)X and

X(YY)X optical configurations (indicated for simplicity as ZY and YY), according to theequation

R ¼ IZY

IYY(2)

where IZY and IYY are the intensities measured (as peak heights) in the ZY and YYconfigurations, respectively.

2.4.2. Thermogravimetric analysisTGA measurements of the untreated, in vitro degraded and cell-incubated

samples were performed with a Mettler TA-STAR, TGA/SDTA 851e thermo-balance,in air with a heating rate of 2 �C/min, from 25 to 600 �C. By the analysis of the firstderivative curves [28], the weight percentages of PLLA and PCL components wereobtained from the 200–330 �C and 330–420 �C regions, respectively.

2.5. In vitro cell cultures

2.5.1. Human trabecular bone cellsHuman osteoblasts were obtained from patients (aet. 58–79) undergoing

treatment for hip joint replacement. The study was approved by the IOR institutionalreview board. Briefly, bone marrow containing trabecular bone fragments was col-lected during surgery following reaming of the femoral canal prior to insertion of theimplant. Bone chips were isolated using a cell strainer with several washes inphosphate-buffered saline (PBS) solution, then grown in Dulbecco’s modified Eaglemedium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM L-glut-amine, and 1% penicillin–streptomycin at 37 �C in a humidified 5% CO2 atmosphere.Cells began to migrate from the chips within 10–14 days; at confluence, the osteo-blastic phenotype of outgrowing cells was confirmed by cytochemical staining foralkaline phosphatase. Harvesting and sub-culturing of cells were obtained bytreatment with a mixture of 0.05% trypsin and 0.02% ethylenediamine tetra-aceticacid (EDTA) in Mg2þ- and Ca2þ-free phosphate-buffered saline (PBS). Passage 2–3cells were seeded on the scaffolds.

2.5.2. Human marrow stromal cellsMarrow stromal cells (MSC) were obtained from the same source as outlined for

osteoblasts. Following separation of bone chips using a cell strainer the bonemarrow was layered onto Ficoll gradient and centrifuged at 400� g for 30 min.

V. Guarino et al. / Biomaterials 29 (2008) 3662–3670 3665

Mononuclear cells were collected, washed twice in PBS, and grown in alpha modi-fication of Dulbecco’s Eagle medium (a-MEM) supplemented with 10% fetal bovineserum (FBS), 2 mM L-glutamine, 1% penicillin–streptomycin, and 50 mg/ml ascorbicacid 2-phosphate at 37 �C in a humidified 5% CO2 atmosphere. At confluence, MSCwere detached with trypsin and characterized for osteogenic ability by colonyforming assay at 14 days, as described by Basso et al. [26]. Passage 2–3 cells wereseeded on the scaffolds.

2.5.3. Cell seedingPolymer samples were treated with 75% ethanol for 2 h. Then, samples were

washed with PBS, dipped in antibiotic/antimycotic 1:100 in PBS for 2 h, and pre-wetted in a-MEM for 1 h. The cell culture was performed in 96-well tissue-cultureplates with change of medium twice a week. Cells from passages 2–3 were seeded at20,000 per sample in 50 mL of medium, and the samples incubated for 2 h at 37 �C toallow cell attachment to the composite surface; then the wells were filled with150 mL of culture medium. Cells on tissue-culture polystyrene (TCPS) provided thecontrols. Culture medium was added with 3 mM b-glycerophosphate for in vitromineralization.

2.5.4. Cellular assays: viability and alkaline phosphataseCell viability was measured, by adding the Alamar blue dye (Biosource Int.) 10%

v/v to the culture well. After 4 h, 100 mL were transferred to a 96-well plate and readat 490 nm using a plate reader (Cytofluor 2350, Millipore Corp.). Results wereexpressed as Relative Fluorescence Units. Then, alkaline phosphatase (ALP) releasedin the medium after 1, 2, 3, 4, and 5 weeks post-seeding and in the lysed cells at5 weeks was assayed. The reaction is based on the conversion of 5 mM colorlessp-nitrophenyl phosphate to coloured p-nitrophenol according to the manufacturer’sprotocol (Sigma, St. Louis, MO). The color change was measured by spectropho-tometer at 405 nm and the amount of enzyme released by the cells was quantifiedby comparison with a standard curve. At the last time endpoint alkaline phosphatasein cell lysate (1% Triton X-100) was measured using the same assay.

Concerning the SEM observation of the cell-seeded scaffolds, they were fixedwith Karnovsky’s fixative, then post-fixed in 1% osmium tetroxide, dehydrated ingraded alcohols, dried in hexamethyldisilazane, sputter coated with palladium–goldand finally analysed in a Philips 515 scanning electron microscope.

Fig. 2. SEM analysis of fibre-reinforced composite scaffolds: (A) bimodal pore sizedistribution and (B) organization of PLLA fibres inside porous PCL matrix.

3. Results

The preliminary investigation of the structural architecture ofcomposite scaffolds was made by SEM (Fig. 2A, B). In Fig. 2A, abimodal porosity characterized by two different pore size rangewas shown. In detail, a microporosity with small pore sizes ranging1–10 mm and a macro-porosity with 100–400 mm pore size havebeen distinguished. By the cross-section image (Fig. 2B) at highermagnification, the structural organization of the composite scaffoldwas evidenced: PLLA fibres were well integrated into the PCLmatrix without any significant alteration of pore morphology. Morespecifically, the arrangement of single pore between the fibres wasguaranteed by the adequate selection of the fibre step during thewinding process which must be larger than the NaCl crystal size, inorder to maintain the needful spaces for developing an inter-connected pore network. The presence of single pores, interspersedamong the fibres and larger than the NaCl crystals, so as to maintainthe spaces needed for interconnectivity, was ensured by adjust-ment of the winding pitch during the winding process.

As support of the qualitative investigation on scaffoldmorphology, porosity was measured by two non-invasive meth-odologies, i.e. the gravimetric method and mercury intrusionporosimetry. In Table 1, the porosity degree obtained by the twodifferent methods is reported. The composite scaffold showeda porosity degree ranging from 76.8� 2.6% (gravimetric method) to79.7% (mercury intrusion porosimetry). Both values are lower thanthe theoretical value defined by the volumetric fraction of sodiumchloride crystals used during the processing (82% in volume). In

Table 1Porosity and characteristic density data obtained on the PCL/PLLA composite scaffolds b

Method Total porosity (%) Relative pore volume fraction (%)

Micropores (0.1–10 mm) Macropores (10–300 mm

GM 76.8 – –MIP 79.7 46.3 53.7

particular, the measurement performed by the gravimetric methodis rough (total porosity %: 76.8� 2.6%) and may be affected bya significant error of the calculated scaffold volume. Even if closedpores are not included, the mercury intrusion method proves themore reliable method for evaluating the interconnected porosity.Two characteristic densities of the composite scaffold have beencalculated, i.e. the bulk density and the absolute density. The formeris related to the porous scaffold density including all voids (closedand open pores, external voids or roughness) according to the ASTMD5004 standard, whereas the absolute density is related only to thesolid material excluding the pore volume contribution. All calcu-lated data were in agreement with the porosity values previouslycalculated. Finally, mercury intrusion porosimetry enabled to esti-mate the bimodal distribution, confirming the full interconnectionwithin the pore network (Fig. 3). According to many literature re-lated to scaffold design pores with sizes smaller than 10 mm havebeen referred as microporosity, while larger ones as macro-porosity[29]. The corresponding data are reported in Table 1 where relativemicropore and macropore volume fractions are shown. Indeed,macropores represent the 53.7% of the total scaffold porosity withan average pore size of about 166 mm, while the remaining 46.3%are micropores with an average pore size of 2.24 mm.

y the gravimetric method (GM) and mercury intrusion porosimetry (MIP)

Average pore radius (mm) Density (g/cm3)

) Micropores (0.1–10 mm) Macropores (10–300 mm) Bulk Absolute

– – 0.269 1.1622.24 166.41 0.255 1.259

Fig. 3. Pore size distribution in PCL/PLLA composite scaffolds.

V. Guarino et al. / Biomaterials 29 (2008) 3662–36703666

Regarding the degradation properties, the weight loss of thePCL/PLLA samples as a function of degradation time in the differentmedia is reported in Fig. 4. No significant weight changes weredetected during degradation in PBS. In detail, the plateau observedin the weight loss profile during the first 35 days in PBS clearlyindicates a slow degradation rate of the PCL/PLLA system, almostunchanged during the degradation period. In the case of treatmentwith SBF media, a slight but progressive weight gain up to 0.8% after28 days was observed. On the contrary, the in vitro degradation inNaOH solution showed a nearly linear increase of weight loss up to14% after 35 days, which suggests a more ‘‘degradative activity’’ ofthe medium used. This result is not surprising, since the catalyticeffect of the OH� ion on the degradation of aliphatic polyesters hasbeen described [22,25,30].

Meanwhile, further investigation by TGA allowed to evaluate thecomposition modifications induced by the degradation treatmentin different media. TGA measurements demonstrated that un-treated samples contain 85% w/w of PCL and 15% of PLLA, inagreement with the theoretical proportions used during thescaffold preparation. No significant changes in composition wereobserved upon degradation in PBS and SBF, while an increase of PCLweight fraction up to 91% w/w was evidenced during the treatmentin NaOH. The analysis of the cell-seeded samples did not allow anaccurate estimation of composition, due to the residual presence ofthe cell culture media.

In this context, micro-Raman analysis was performed on cell-seeded samples to overcome the limitation of the TGA analysis.Therefore, this technique allowed to gain more insight into thedegradation mechanism and kinetics of the composite and the cellrole in degradation.

To illustrate this, the spectra corresponding to the mostdegraded samples – i.e. those treated with the NaOH solution for35 days – have been reported here.

Fig. 4. Trend of the % weight loss of PCL/PLLA samples as a function of degradationtime in the different media: (-) NaOH solution; (:) SPB; and (C) SBF. The data re-ported are mean values of duplicate tests.

As regards the PCL component, Raman spectra did not revealsignificant changes (Fig. 5A) upon degradation. In previous papers[24,25], the I1305/I1285 intensity ratio between the bands at 1305and 1285 cm�1 was identified as marker of crystallinity. As can beseen from Fig. 5B, the I1305/I1285 intensity ratio remained constantafter 35 days of immersion in the different media as well asincubation with MSC and HOB.

In Fig. 5C, D the Raman spectra of PLLA fibres before (a) and afterdegradation for 35 days in NaOH (b) are shown. In agreement withthe measured weight loss, the spectra of the NaOH-treated samplesshowed the most significant changes, such as an increase of the923 cm�1 band intensity, the appearance of the 523 cm�1 band andthe increase of the splitting of the 412–398 and 315–297 cm�1

doublets (YY scattering geometry, Fig. 5C). All these bands have beenreported as the most sensitive to changes in structural order [31–35]. In particular, the area of the 412 cm�1 band, normalized to thearea of the 875 cm�1 band (insensitive to sample crystallinity), hasbeen reported to increase with PLLA crystallinity [33]. The A412/A875

ratio (expressed as peak area) increased from 0.8 to 1.2 upon deg-radation in NaOH for 35 days, while remained nearly constant in theother media. As can be seen in Fig. 5D (ZY scattering geometry), the710 cm�1 band increased in intensity with respect to the 736 cm�1

component: these two bands have been reported as good indicatorsof the 103 and 31 helices [32], the most likely structures for PLLA[34]. Several other bands showed changes in intensity (Fig. 5D),which can be ascribed to structural rearrangements as well. The I875/I923 intensity ratio was identified as marker of degradation; it wasfound to increase upon degradation in NaOH, as shown in Fig. 5E.The increase was significantly less pronounced upon degradation inPBS, SBF and cell culture medium (CTRL samples), while the samplesincubated with MSC and HOB were characterized by intermediateI875/I923 values (being higher for MSC than HOB, Fig. 5E). Thepolarization ratio of the band at 875 cm�1 (R875, Fig. 5F) showed ananalogous trend. The I875/I1459 intensity ratio was identified asspectroscopic marker of the polymeric chain length [22,23]. Thespecimens degraded for 35 days in NaOH solution showed thehighest I875/I1459 decrease (from 1.64 to 1.10), indicating the mostsignificant chain length decrease. The samples treated for the sametime period with PBS, SBF and cell culture medium did not displaysignificant I875/I1459 changes, while for the samples incubated withHOB and MSC, the I875/I1459 ratio reached 1.52 and 1.40, respectively,confirming the above reported trends.

In the assessment of osteoconductivity, marrow stromal cellsand osteoblasts derived from patients were found to adhere to thecomposite and proliferate to some extent. As shown in Fig. 6A, B thegrowth of both MSC and HOB reached a plateau at 3 weeks; at 5weeks the number of cells is approximately the same and conflu-ence was not attained. Alkaline phosphatase (ALP) was released at1–3 weeks of culture to decrease thereafter (Fig. 7A, B). MSCproduced more ALP than HOB, as shown in Fig. 8. As expected,TCPS-seeded cells recorded high values, as the cells easily grow toconfluence on plastic bottom well. The structure of the compositeafter the cell seeding is shown in Fig. 9 by SEM at low magnifica-tion: fibres run parallel in bundles within the PCL matrix withhomogeneous spatial distribution (Fig. 9A). The attachment andspreading of cells are observed preferentially on the matrix to formsome clusters of cells (Fig. 9B), whereas on PLLA fibres the cellsalign themselves with the long axis parallel to the filaments, andextend their filopodia from each other.

4. Discussion

In devising an ideal scaffold for bone tissue regeneration, themain goal remains the accurate control of pore size and porosity, aswell as degradation kinetics, with the aim to elicit a ‘‘desired’’behaviour of cells during bone morphogenesis.

Fig. 5. Raman spectra of the PCL matrix (A) before (a) and after (b) degradation for 35 days in NaOH solution. (B) I1305/I1285 intensity ratios obtained from the Raman spectra of PCLbefore and after treatment for 35 days. Raman spectra of a PLLA fibre before (a) and after (b) degradation for 35 days in NaOH solution ((C): YY scattering geometry; (D): ZYscattering geometry). (E) I875/I923 intensity ratios obtained from the Raman spectra of PLLA recorded in the ZY scattering geometry before and after treatment for 35 days. (F) R875

depolarization ratio obtained from the Raman spectra of PLLA before and after treatment for 35 days. CTRL¼ control samples for cell cultures.

V. Guarino et al. / Biomaterials 29 (2008) 3662–3670 3667

In the current work the morphological features, the degradationproperties and the cellular response of a new polymer-basedcomposite were investigated. Indeed, we started from the obser-vation that the presence of structured fibrillar organization in thetissue structures supporting cells, such as collagen, actin, etc.,provides the rationale for introducing micro- and nano-fibres in thebiomimetic scaffolds. The variety of biomedical applicationsemploying polymer fibres both on micrometric [36] and nano-metric scale [37–39] has been recently reviewed, demonstratinga good penetration of bone marrow stromal cells into compositescaffold systems.

In this context, an effective solution to obtain scaffolds withspecific chemico-physical properties and controlled morphologymay be a PCL/PLLA composite scaffold (fibres of PLLA in PCL matrix),prepared through the synergic use of phase inversion and salt

leaching technique, and the filament winding technology. Amongbiomedical polymers, PCL is characterized by an extended re-sorption time (up to 2 years as a bulk), while PLLA fibres aredegraded at a higher rate: both have good compatibility [3,40]. Thecombination of polymers with different degradation kinetics insuch a way may be an interesting approach to obtain scaffolds withcontrolled degradation rates according to the specific biomedicalapplication. Furthermore, the large pore size of 250–400 mm wasshown to be more effective for MSC and HOB culture than thesmaller pore size as just discussed elsewhere [6,41].

The PCL/PLLA composite scaffolds, examined with conventionalanalytical methods, show very interesting pore morphology interms of porosity degree and pore size distribution. The phase in-version and salt leaching techniques generate a bimodal porosity[42] with micropores, resulting from the solvent removal, which

Fig. 6. (A) Viability of human marrow stromal cells (MSC) seeded on PCL/PLLA scaf-folds (mean� sd, n¼ 4) and (B) viability of human osteoblasts (HOB) seeded on PCL/PLLA scaffolds (mean� sd, n¼ 4).

Fig. 7. (A) ALP release from human marrow stromal cells (MSC) seeded on PCL/PLLAscaffolds (mean� sd, n¼ 4) and (B) ALP release from human osteoblasts (HOB) seededon PCL/PLLA scaffolds (mean� sd, n¼ 4).

Fig. 8. ALP content in MSC and HOB seeded on PCL/PLLA scaffolds for 5 weeks(mean� sd, n¼ 4).

V. Guarino et al. / Biomaterials 29 (2008) 3662–36703668

allow the nutrient supply and waste removal, while macropores(about 300 mm), due to leaching of sodium chloride crystals allowfor an optimal ingrowth of the healing bone tissue. Furthermore,the combined effect of the reinforcing action of PLLA fibres and theslow degradation rate of the single components in media with a pHvalue similar to the body fluids may positively contribute to themechanical performance under compression of the compositesystem. Then, changes in fibre and matrix features, such as di-ameter and fibre chemistry, during degradation could providea guide for the bone remodelling process by achieving a compositestructure with morphological and structural properties which thenevolve, during degradation, towards a softer and more porousmaterial needed in the long term.

A polymer-based scaffold material has to degrade and resorb ata controlled rate at the same time as the bone tissue regenerates. Inorder to study the in vitro degradation, traditional assays withbuffered solutions such as SBF and PBS showed negligible degra-dation of scaffolds after 35 days. This could be explained by theethanol interaction occurring in the preparation phases, which isprobably responsible for the higher poly(a-hydroxy acids) crystal-linity with increase of their degradation times [43]. In the work ofother authors [30], the NaOH solution is chosen as a model toaccelerate the degradation of the scaffold. Indeed, it is well knownthat the degradation products of aliphatic polyesters determine anacid environment. However, in this context it must be stressed thatesters and polyesters undergo hydrolysis reactions catalyzed bybases as well as acids [23].

A near linear increase of weight loss was observed during thefirst 35 days of degradation in NaOH solution. This has also beenfound by Cam et al. [30], who reported that PLLA of different mo-lecular weights lost their weight almost linearly with time up to100 days; after then the degradation rate noticeably decreased. Thisobservation could be explained by the progressive formation ofchannels, which could guide the PLLA degradation products toimmediately leave the porous PCL matrix after their formation,therefore providing the linear rate of weight loss (Fig. 4). The

composition and morphology of the polymeric scaffolds appearunaffected by degradation in PBS and SBF, whereas in NaOH solu-tion the progressive enrichment in PCL component, as revealed byTGA, suggested a preferential degradation of the more hydrophilicPLLA component.

The morphology of PCL appears unaffected after immersion for35 days in the different media as well as incubation with HOB andMSC. As regards the PLLA component, micro-Raman spectroscopyshowed that degradation is conditioned by the medium composi-tion and cell activity.

Among the five different environmental conditions of degra-dation investigated, the NaOH solution induced the most relevantstructural rearrangements (Fig. 5E, F). In this medium, the PLLAcomponent underwent a significant increase in crystallinity andpolymer chain shortening, as revealed by the A412/A875 and I875/

Fig. 9. SEM image of cell seeded on PCL/PLLA for 5 weeks: (A) MSC aligned by ‘contact guidance’ along PLLA fibres (bar¼ 100 mm) and (B) clusters of HOB cells on the matrix andPLLA fibres (bar¼ 100 mm).

V. Guarino et al. / Biomaterials 29 (2008) 3662–3670 3669

I1459 ratios, respectively. The increased crystallinity of PLLA is notsurprising, as reported by previous studies [16,22–25]. In semi-crystalline polymers degradation starts in the amorphous part,since water can enter more easily into these domains [44]increasing the proportion of crystallised regions. Furthermore, thehydrolytic attack on the ester backbone decreases the polymericchain length with the formation of short oligomers, more mobile,which tend to recrystallize [45].

In PBS, SBF and cell culture medium, structural rearrangementsappeared irrelevant (Fig. 5E, F). The presence of seeded HOB andMSC also plays a role on the degradation mechanism, showing anintermediate behaviour with a more relevant contribution of MSCthan HOB. These results were confirmed also by the trend of theI875/I1459 intensity ratio. Even if in vitro degradation assays repre-sent a model for scaffold performance in vivo, providing importantinsights into the hydrolytic degradation profile, the effective in vivodegradation is also affected by enzymes, cellular activity and pHchanges.

Focusing on cell–material interactions, human-derived boneforming cells were shown to adhere to the scaffolds and growwithin them, but the confluence was not reached at 5 weeks ofculture. It is recognized that culture conditions affect the quality ofthe engineered tissue [46], and deep penetration of cells andnutrients within scaffolds is hampered in static cultures at longtime endpoints. Moreover, human primary cells derived frompatients, as used in our study, show a reduced proliferation rate incomparison to rat or rabbit MSC and cell lines, often used for testingscaffold colonization [47,48].

Moreover, it is increasingly shown that human cells exhibitsignificant responses to surface changes, and use adhesion mech-anisms different from those used by osteoblast-like cells [49,50].

In our study human marrow stromal cells and trabecular osteo-blasts rapidly proliferate on the 3D scaffold up to 3 weeks, andrelease ALP, to entering a stationary state thereafter. In addition tothe interconnected porosity of the scaffold, which provides thesuitable environment for cell attachment and proliferation, thefibre arrangement leads to an oriented migration of bone cells,known as ‘contact guidance’. As a consequence, the varyingarrangement of cells within the 3D structure may better withstandmechanical forces applied, with a potentially improved tissue or-ganization. These results further confirm the main contribution ofthe scaffold morphology on the cell–material interaction: the highinternal surface of the scaffolds due to the macro-porosity and thecurvature of fibre on a micrometric scale positively influence theattachment and proliferation of MSC [36,39]. Apparently, both MSCand HOB show the same reaction to the PCL/PLLA topography.According to the degradation study, MSC are able to induce somechanges in the composite structure when seeded for 5 weeks on itssurface, whereas HOB are less active. The finding that human cellsin vitro induce some changes in polymer fibres, as shown in Fig. 5E,

F, is confirming their potential active role in vivo, where cells,enzymes, MMP, and other factors could act coordinately to degradethe polymers. However, it has to be recognized that the persistenceof the scaffold in a complex medium such as the serum-addeda-MEM at 37 �C for 5 weeks, might be able per se to change thepolymeric structure to some extent, therefore changing proteinadsorption and potential osteoblast adhesion [51].

5. Conclusions

A new type of composite material for bone regeneration, com-posed of a porous PCL matrix reinforced with continuous PLLAfibres, was fabricated by the synergic use of filament winding andphase inversion/salt leaching technique. The chemistry and thedegrees of micro- and macro-porosity attained with such tech-niques determine the structural and biological performances of thecomposite. Concerning the degradation properties, over 5 weeks,the treatment with NaOH solution (pH 12) induced the most sig-nificant changes. These involved only the more hydrophilic PLLAcomponent, which underwent significant structural rearrange-ments, increase in crystallinity and decrease in polymeric chainlength. Furthermore, encouraging results in cellular studies arereached in the case of HOB and MSC: the latter were found moreactive on the composite, due either to a slightly higher number ofcells compared to HOB, or to a higher activity of the cells.

Acknowledgement

The authors wish to thank Dr. Gennaro Tafuri and Dr. FernandoSarracino for valuable advice in filament winding technology. Thisstudy partially supported from the Ministero dell’Universita e dellaRicerca (FIRB – RBAU01N79B) and Rete nazionale di ricercaTISSUENET n. RBPR05RSM2.

References

[1] Datta N, Holtorf HL, Sikavitsas VI, Jansen JA, Mikos AG. Effect of bone extra-cellular matrix synthesized in vitro on the osteoblastic differentiation ofmarrow stromal cells. Biomaterials 2005;26(9):971–7.

[2] Currey JD, Butler G. The mechanical properties of bone tissue in children.J Bone Joint Surg Am 1975;57:810–4.

[3] Middleton JC, Tipton AJ. Synthetic biodegradable polymers as orthopaedicdevices. Biomaterials 2000;21(23):2335–46.

[4] Freed LE, Vunjak-Novakovic G, Biron RJ, Eagles DB, Lesnoy DC, Barlow SK, et al.Biodegradable polymer scaffolds for tissue engineering. Nat Biotech 1994;12:689–93.

[5] Sivalingam G, Madras G. Thermal degradation of binary physical mixtures andcopolymers of poly(3-caprolactone), poly(D, L-lactide), poly(glycolide). PolymDegrad Stab 2004;84:393–8.

[6] Hutmacher DW. Scaffolds in tissue engineering bone and cartilage. Bio-materials 2000;21(24):2529–43.

[7] Sung HJ, Meredith C, Johnson C, Galis ZS. The effect of scaffold degradationrate on three-dimensional cell growth and angiogenesis. Biomaterials 2004;25(26):5735–42.

V. Guarino et al. / Biomaterials 29 (2008) 3662–36703670

[8] Martin C, Winet H, Bao JY. Acidity near eroding polylactidepolyglycolide invitro and in vivo in rabbit tibial bone chambers. Biomaterials 1996;17(24):2373–80.

[9] Hubbell JA. Materials as morphogenetic guides in tissue engineering. CurrOpin Biotechnol 2003;14(5):551–8.

[10] Heller J, Barr J, Shah DT, Ng SY, Shen HR, Baxter BC. Poly(ortho esters). In:Ma PX, Elisseeff J, editors. Scaffolding in tissue engineering. Boca Raton, FL:CRC Press; 2005. p. 91–110.

[11] Kweon HY, Yoo MK, Park IK, Kim TH, Lee HC, Lee HS, et al. A novel degradablepolycaprolactone networks for tissue engineering. Biomaterials 2003;24:801–8.

[12] Manjubala I, Woesz A, Pilz C, Rumpler M, Fratzl-Zelman N, Roschger P, et al.Biomimetic mineral-organic composite scaffolds with controlled internalarchitecture. J Mater Sci Mater Med 2005;16:1111–9.

[13] Kim HW, Knowles JC, Kim HE. Hydroxyapatite and gelatin composite foamsprocessed via novel freeze-drying and crosslinking for use as temporary hardtissue scaffolds. J Biomed Mater Res 2005;72A:136–45.

[14] Guarino V, Causa F, Netti PA, Ciapetti G, Pagani S, Martini D, et al. The role ofhydroxyapatite as solid signal on performance of PCL porous scaffolds for bonetissue regeneration. J Biomed Mater Res B Appl Biomater 11 Mar 2008. doi:10.1002/jbm.b.31055.

[15] Cooper JA, Lu HH, Ko FK, Freeman JW, Laurencin CT. Fiber-based tissue-engineered scaffold for ligament replacement: design considerations andin vitro evaluation. Biomaterials 2005;26(13):1523–32.

[16] Causa F, Sarracino F, De Santis R, Netti PA, Ambrosio L, Nicolais L. Basicstructural parameters for the design of composite structures as ligamentaugmentation devices. J Appl Biomater Biomech 2006;4:21–30.

[17] Xong X, Bien H, Chung CY, Yin L, Fang D, Hsiao BS, et al. Electrospun fine-textured scaffolds for heart tissue constructs. Biomaterials 2005;26:5330–8.

[18] Cronin EM, Thurmond FA, Bassel-Duby R, Sanders Williams R, Wright WE,Nelson KD, et al. Protein-coated poly(L-lactic acid) fibers provide a substratefor differentiation of human skeletal muscle cells. J Biomed Mater Res 2004;69A:373–81.

[19] Guarino V, Causa F, Ambrosio L. Porosity and mechanical properties re-lationship in PCL based scaffolds. J Appl Biomater Biomech 2007;5(3):149–57.

[20] Causa F, Netti PA, Ambrosio L, Ciapetti G, Baldini N, Pagani S, et al. Poly-3-caprolactone/hydroxyapatite composites for bone regeneration: in vitrocharacterization and human osteoblast response. J Biomed Mater Res 2006;76A:151–62.

[21] Griffith L. Polymeric biomaterials. Acta Mater 2000;48:263–77.[22] Taddei P, Tinti A, Fini G. Vibrational spectroscopy of polymeric biomaterials.

J Raman Spectrosc 2001;32:619–29.[23] Taddei P, Monti P, Simoni R. Vibrational and thermal study on the in vitro and

in vivo degradation of a poly(lactic acid)-based bioabsorbable periodontalmembrane. J Mater Sci Mater Med 2002;13:469–75.

[24] Taddei P, Tinti A, Reggiani M, Fagnano C. In vitro mineralization of bioresorbablepoly(3-caprolactone)/apatite composites for bone tissue engineering: a vibra-tional and thermal investigation. J Mol Struct 2005;744–747:135–43.

[25] Taddei P, Di Foggia M, Causa F, Ambrosio L, Fagnano C. In vitro bioactivity ofpoly(epsilon-caprolactone)-apatite (PCL-AP) scaffolds for bone tissue engi-neering: the influence of the PCL/AP ratio. Int J Artif Organs 2006;13:719–25.

[26] Basso N, Bellows CG, Heersche JNM. Effect of simulated weightlessness onosteoprogenitors cell number and proliferation in young and adult rats. Bone2005;36:173–83.

[27] Kokubo T, Kushitani H, Sakka S, Kitsugi T, Yamamuro T. Solutions able toreproduce in vivo surface-structure changes in bioactive glass–ceramic A-W.J Biomed Mater Res 1990;24:721–34.

[28] Sivalingam G, Vijayalakshmi SP, Madras G. Enzymatic and thermal degrada-tion of poly(3-caprolactone), poly (D,L-lactide), and their blends. Ind Eng ChemRes 2004;43:7702–9.

[29] Ginebra MP, Delgado JA, Harr I, Almirall A, DelValle S, Planell JA. Factorsaffecting the structure and properties of an injectable self-setting calciumphosphate foam. J Biomed Mater Res 2007;80A:351–61.

[30] Cam D, Hyon SH, Ikada Y. Degradation of high molecular weight poly-(L-lactide) in alkaline medium. Biomaterials 1995;16:833–43.

[31] Pitt CG, Chasalow FI, Hibionada YM, Klimas DM, Schindler A. Aliphatic poly-esters. I. The degradation of poly(3-caprolactone) in vivo. J Appl Polym Sci1981;26:3779.

[32] Kang S, Hsu SL, Stidham HD, Smith PB, Leugers MA, Yang X. A spectroscopicanalysis of poly(lactic acid) structure. Macromolecules 2001;34:4542–8.

[33] Smith PB, Leugers A, Kang S, Hsu SL, Yang X. An analysis of the correlationbetween structural anisotropy and dimensional stability for drawn poly(lacticacid) films. J Appl Polym Sci 2001;82:2497–505.

[34] Kister G, Cassanas G, Vert M, Pauvert B, Terol A. Vibrational analysis of poly-(L-lactic acid). J Raman Spectrosc 1995;26:307–11.

[35] Brizzolara D, Cantow HJ, Diederichs K, Keller E, Domb AJ. Mechanism of thestereocomplex formation between enantiomeric poly(lactide)s. Macromole-cules 1996;29:191–7.

[36] Yasuda K, Inoue S, Tabata Y. Influence of culture method on the proliferationand osteogenic differentiation of human adipo-stromal cells in nonwovenfabrics. Tissue Eng 2004;10:1587–96.

[37] Zhang Y, Lim CT, Ramakrishna S, Huang ZM. Recent development of polymernanofibers for biomedical and biotechnological applications. J Mater Sci MaterMed 2005;16:933–46.

[38] Zhang YZ, Ouyang HW, Lim CT, Ramakrishna S, Huang ZM. Electrospinning ofgelatine fibers and gelatin/PCL composite fibrous scaffolds. J Biomed MaterRes Appl Biomater 2005;72B:156–65.

[39] Christenson EM, Anseth KS, Van der Beucken JP, Chan CK, Ercan B, Jansen JA,et al. Nanobiomaterial applications in orthopaedics. J Orthop Res 2007;27:11–22.

[40] Holmbom J, Sodergard A, Ekholm F, Martson M, Kuusilehto A, Saukko P, et al.Long-term evaluation of porous poly(espilon-caprolactone-co-L-lactide) asa bone-filling material. J Biomed Mater Res A 2005;75:308–15.

[41] Savarino L, Baldini N, Greco M, Capitani O, Pinna S, Valentini S, et al. Theperformance of poly-3-caprolactone scaffolds in a rabbit femur model withand without autologous stromal cells and BMP4. Biomaterials 2007;28:3101–9.

[42] Guarino V, Causa F, Ambrosio L. Bioactive scaffolds for connective tissueregeneration. Expert Rev Med Devices 2007 May;4(3):405–18.

[43] Blasi P, D’sousa SS, Selmin F, De Luca P. Plasticizing effect of water on poly-(lactide-co-glycolide). J Controlled Release 2005;108(1):1–9.

[44] Fischer EW, Sterzel HJ, Wegner G. Investigation of the structure of solutiongrown crystals of lactide copolymers by means of chemical reactions. Kolloid-Z u Z-Polymere 1973;251:980–90.

[45] Landes C, Ballon A, Roth C. In patient versus in vitro degradation of P(L/DL)LAand PLGA. J Biomed Mater Res B Appl Biomater 2006;76(2):403–11.

[46] Yoshimoto H, Shin YM, Terai H, Vacanti JP. A biodegradable nanofiber scaffoldby electrospinning and its potential for bone tissue engineering. Biomaterials2003;24:2077–82.

[47] Shao XX, Hutmacher DW, Ho ST, Goh JCH, Lee EH. Evaluation of a hybridscaffold/cell construct in repair of high-load-bearing osteochondral defects inrabbits. Biomaterials 2006;27:1071–80.

[48] Koh YH, Bae CJ, Sun JJ, Jun IK, Kim HE. Macrochanneled poly(3-caprolactone)/hydroxyapatite scaffold by combination of bi-axial machining and lamination.J Mater Sci Mater Med 2006;17:773–8.

[49] Mwale F, Wang HT, Nelea V, Luo L, Antoniou J, Wertheimer MR. The effect ofglow discharge plasma surface modification of polymers on the osteogenicdifferentiation of committed human mesenchymal stem cells. Biomaterials2006;27:2258–64.

[50] Kilpadi K, Sawyer AA, Prince CW, Chang PL, Bellis SL. Primary human marrowstromal cells and Saos-2 osteosarcoma cells use different mechanisms toadhere to hydroxyapatite. J Biomed Mater Res 2004;68A:273–85.

[51] Tang ZG, Hunt JA. The effect of doping of polycaprolactone films on the controlof osteoblast adhesion and proliferation in vitro. Biomaterials 2006;27:4409–18.