lable at ScienceDirect

Biomaterials 32 (2011) 5756e5764

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomateria ls

The influence of elastin-like recombinant polymer on the self-renewing potentialof a 3D tissue equivalent derived from human lamina propria fibroblasts and oralepithelial cells

Beste Kinikoglu a,b, José Carlos Rodríguez-Cabello c, Odile Damour a, Vasif Hasirci d,e,*aBanque de Tissus et Cellules, Hospices Civils de Lyon, 69437 Lyon, FrancebMETU, Department of Biotechnology, Ankara, TurkeycGIR BIOFORGE, CIBER-BBN, Universidad de Valladolid, Po de Belén s/n 47011 Valladolid, SpaindMETU, Department of Biological Sciences, 06531 Ankara, TurkeyeMETU, BIOMATEN Center of Excellence in Biomaterials and Tissue Engineering, 06531 Ankara, Turkey

a r t i c l e i n f o

Article history:Received 12 March 2011Accepted 20 April 2011Available online 17 May 2011

Keywords:Elastin-like recombinant polymerNanofibrous scaffoldCell proliferation3D tissue equivalent

* Corresponding author. METU, BIOMATEN, DeparBiotechnology Research Unit, 06531 Ankara, Turkefax: þ90 312 210 15 42.

E-mail address: vhasirci@metu.edu.tr (V. Hasirci).

0142-9612/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.biomaterials.2011.04.054

a b s t r a c t

Three-dimensional epithelial tissue equivalents tend to lose their self-renewing potential progressivelyduring culture as their epithelial cells lose their proliferative capacity with time. Even though the tissueengineered construct can mimic the native tissue well, it rapidly degrades after implantation due to theinsufficient number of proliferating cells in the equivalent. In the present study we demonstrate for thefirst time that the use of an elastin-like recombinant polymer (ELR) engineered to contain the celladhesion peptide RGD can result in a 3D tissue equivalent with high self-renewing potential, containingas many proliferative cells as the native tissue itself. The 3D tissue equivalent was reconstructed by thecoculture of human lamina propria fibroblasts and oral epithelial cells in the nanofibrous ELR-collagenscaffold. Histological, immunohistological and transmission electron microscopic analyses of this oralmucosa equivalent demonstrated the expression of markers characteristic of epithelial proliferation(Ki67) and differentiation (keratin 13), and also the presence of a pluristratified epithelium and anultrastructurally well-organized basement membrane expressing laminin 332. The synthesis of newextracellular matrix by the fibroblasts was also demonstrated. The scaffold proposed here presents greatpotential for tissue engineering applications, and also for studies of epithelial proliferation, and epithelialdisorders including carcinogenesis.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Recombinant polymers (which have also been termed ‘Recom-binamers’ in recent publications [1]), aremacromolecules producedusing recombinant DNA technology by introducing a desired geneinto the genetic content of a host organism such asmicroorganisms,plants or other eukaryotic organisms. This way, it is possible tobioengineer protein-based polymers of well-defined and complexstructure [1]. Elastin-like recombinant polymers (ELRs), which forma subclass of protein-based recombinant polymers, are composed ofthe pentapeptide repeat Val-Pro-Gly-Xaa-Gly (VPGXG), whichmimics from the hydrophobic domain of tropoelastin where X

tment of Biological Sciences,y. Tel.: þ90 312 210 51 80;

All rights reserved.

represents any natural or modified amino acid, except proline [2].The first ELR products were simple peptides, to which the cells didnot attach. Soon after, they were enriched with short peptidesequences having specific bioactivity [3] and have been used ascoatings [4],films [5] for improved cell attachment, and as hydrogelsto promote chondrogenesis [6].

The major challenge for tissue engineering in terms of successafter grafting is to secure the survival of the cultured cells [7].Indeed, oral mucosal epithelial cell sheets successfully recon-structed in vitro were reported to degenerate one week aftertransplantation [8]. It was suggested that host subcutaneous tissueswere unable to promote maintenance of stem and progenitor cellsand therefore could not produce long-term survival [8]. This indi-cates the importance of the presence of highly proliferative cellssuch as stem or progenitor cells in the tissue engineered constructto ensure its self-renewal and therefore its post-transplantationviability. The native oral mucosa itself has all its basal cells inproliferative stage, which are responsible of the high turnover and

B. Kinikoglu et al. / Biomaterials 32 (2011) 5756e5764 5757

constant renewal of its epithelium in vivo [9]. Proliferation isconfined to the basal layer of the epithelium and there is a steadyupward migration and progressive differentiation of cells. Undernormal conditions, epithelial cells that detach from the basallamina lose their growth potential, and this process is linked to theonset of differentiation [10]. Even though the basal cells in nativeoral mucosa itself have such high turnover rates, the epithelial cellsin reconstructed oral mucosa lose their proliferative capacityduring culture [11e13]. Time-course analysis of oral epithelialdevelopment in an oral mucosal equivalent showed that thenumber of proliferative epithelial cells in the model peaks at 1week after their seeding on lamina propria equivalents, and thisnumber gradually decreases down to only a few cells at the end ofa 3 week culture [11]. In our previous study with our patentedcollagen-glycosaminoglycan-chitosan scaffold [14], we observedthe same pattern: at the end of 3 weeks of culture, only a fewproliferative cells (Ki67 positive) remained in the reconstructedoral mucosa [15]. Other studies on tissue-engineered oral mucosaalso found that at the end of the culture period (even for cultures ofonly 11 days) either no proliferative cell [16] or only a few remainedin the oral mucosal equivalents [13,17,18], which is much less thanfound in native oral mucosa. These results led the researchers toinvestigate into the mechanisms of various regenerative medicineapproaches and epithelial stem cell biology [8].

The aim of the present study was to take advantage of an ELRbioengineered to contain the cell adhesion sequence RGD by usingit for the first time as a scaffold material to reconstruct a three-dimensional, full-thickness tissue equivalent. The hypothesisbehind our study was that the use of this ELR containing RGD,which is also the active motif of epidermal growth factor (EGF),a known mitogen [19], would enhance cellular adhesion, prolifer-ation and migration, providing a better niche for the proliferativeprogenitor epithelial cells for their long-term culture. The nano-fibrous ELR-collagen scaffold was prepared by electrospinning ofthe proteinmixture onto collagen foamswhich served as support tothe fibers to facilitate their handling. The nanofibrous ELR-collagenscaffold was consecutively seeded with human oral fibroblasts andepithelial cells, isolated from the non-keratinized oral cavity, toproduce an epithelialized, full-thickness, oral mucosal equivalent.This tissue engineered oral mucosa was characterized by histology,immunohistochemistry and transmission electron microscopy.

2. Materials and methods

2.1. H-RGD6 expression and purification

The ELR gene was constructed using standard genetic engineering techniques.The polymer biosynthesis was carried out by using Escherichia coli system ofrecombinant protein production, while its purification was performed with severalcycles of temperature-dependent reversible precipitations as described elsewhere[4,20]. Purified ELR was dialyzed and then lyophilized. The purity and molecularweight of recombinant ELR were verified by sodium dodecyl sulfate polyacrylamidegel electrophoresis (SDS-PAGE) and matrix-assisted laser desorption/ionizationtime-of-flight (MALDI-TOF) mass spectroscopy.

2.2. Scaffold preparation

Collagen type I was isolated from SpragueeDawley rat tails as reported earlier[21]. Collagen (9.3%) and ELR (13.2%) were dissolved in 1,1,1,3,3,3 hexafluoro-2-propanol (HFIP, SigmaeAldrich, USA) and PBS (pH 7.2), respectively; and fourvolumes of collagen solutionweremixedwith one volume of ELR solution so that thefinal total protein concentrationwas 10%. The resulting ELR-collagenmixture (1:3, w/w)was electrospun directly onto collagen foamswhichwould serve as support to thenanofibrous scaffold. Electrospinning was performed by loading the solution ina 10 mL syringe fitted with a 3 cm 18G blunt-end needle (Ayset AS, Turkey) anddispensing at a rate of 20 mL/min. The solution was ejected by applying a voltage of22.5 kV and the fibers were collected on collagen foams taped onto flat, groundedplate placed at a distance of 10 cm from the syringe needle. The resulting scaffoldswere crosslinked by dehydrothermal treatment (145 �C for 48 h) under 27 in.Hgpressure. Electrospuncollagen scaffolds (10%)without ELRwereprepared as controls.

2.3. Scaffold characterization

Morphology of fibrous scaffolds was studied with a FEI Quanta 200F (USA)scanning electron microscope (SEM) after sputter coating (12 nm) with gold-palladium. Fiber diameters were measured using the image processing programImageJ (NIH, USA) using 4 different areas of the image to calculate the averagevalues. The results were expressed as mean � standard error of the mean.

2.4. Origin, isolation, and culture of epithelial cells

Epithelial cells were isolated from normal human oral mucosal biopsiesremoved from the non-keratinized cheek region of the mouth in accordance withFrench Ethical Regulations (Loi de Bioethique 2004e800), and obtained withinformed consent from patients undergoing oral surgery. The biopsies were firstmeasured, and then cut into small pieces in order to increase the efficacy of theenzymes used. The separation of the epithelium from the lamina propria was per-formed with dispase (GIBCO, Invitrogen, USA), 10 mg/mL for 3 h at 4 �C. Afterseparation, epithelium was treated with trypsin 0.5 g/L e EDTA 0.2 g/L (GIBCO,Invitrogen, USA) for 20 min to extract the cells, which were collected every 10 min.Epithelial cells were grown at 8000e10000 cells/cm2 on a feeder layer of irradiatedhuman fibroblasts in the following medium: DMEM-Ham-F12 2.78/1 (GIBCO, Invi-trogen, USA), 10% fetal calf serum (Hyclone, France), 0.4 mg/mL hydrocortisone(Upjohn, USA), 0.12 UI/mL insulin (Umuline, Lilly, France), 0.033 mg/mL selenium(Laboratoire Aguettant, France), 0.4 mg/mL isoprenaline hydrochloride (Isuprel,Sterling Winthrop, USA), 2 � 10�9 M, tri iodo thyronine (Sigma, USA), 10 ng/mLepidermal growth factor (Austral Biologicals, USA), and antibiotics. The cells wereseeded on the scaffolds at passage 3.

2.5. Origin, isolation, and culture of fibroblasts

Fibroblasts were isolated from the same biopsies as the epithelial cells. Afterepithelium-lamina propria separation, isolation was performed with collagenase A(Roche Diagnostics, Switzerland), 0.1 U/mL for 20 min at 37 �C with continuousstirring. The digest was purified by passing it through a 70 mm cell strainer (BDBiosciences, USA). This procedure was repeated 6 times, and then the digest wasimmediately placed in themonolayer culture. Fibroblasts were seeded at a density of10 000 cells/cm2 and cultured in fibroblast medium composed of DMEM (GIBCO,Invitrogen, USA), 10% newborn calf serum (NCS) (Hyclone, France), and antibiotics.The medium was changed every two days until confluence was reached. Atconfluence, cells were resuspended using trypsin 0.5 g/L - EDTA 0.2 g/L (GIBCO,Invitrogen, USA), then amplified over three passages (from P0 to P2) and seeded intothe scaffold at P3.

2.6. Preparation of lamina propria equivalents

Lamina propria equivalents consisted of ELR-collagen nanofibrous scaffolds inwhich human oral mucosal fibroblasts were cultured. Briefly, lamina propriaequivalents were prepared by adding a suspension of 2.5 � 105/cm2 on top of the1 cm2 ELR-collagen scaffold. Equivalents were then cultured for 21 days in a mediumcomposed of DMEM (GIBCO, Invitrogen, USA), 10% fetal calf serum (Hyclone, France),10 ng/mL epidermal growth factor (Austral Biologicals, USA), 50 mg/mL ascorbic acid(Bayer, Germany). Culture mediumwas changed daily until the seeding of epithelialcells.

2.7. Preparation of epithelialized oral mucosa equivalents

Human epithelial cells were plated on lamina propria equivalents at a concen-tration of 2.5 � 105/cm2. Epithelialized oral mucosal substitutes were cultured inepithelial cell medium supplemented with 50 mg/mL ascorbic acid (Bayer, Germany)under submerged conditions for 7 days. They were then elevated at the aireliquidinterface for the remaining 14 days in another mediumwith DMEM-Ham-F12 2.2/1(GIBCO, Invitrogen, USA), 8 mg/mL bovine serum albumin (Sigma, USA), 0.4 mg/mLhydrocortisone (Upjohn, USA), 0.12 UI/mL insulin (Umuline, Lilly, France), 50 mg/mLascorbic acid (Bayer, Germany), and antibiotics.

2.8. Histology

Tissue equivalents were fixed in 4% formaldehyde solution and embedded inparaffin. Sections, 5 mm thick, were cut and stained using hematoxylin-phloxin-saffron (HPS).

2.9. Immunohistochemistry

The primary antibodies used in this study were anti-keratin 13 (dilution 1:75)(Chemicon, USA), anti-laminin 332 (dilution 1:100) (Chemicon, USA), and anti-Ki67(dilution 1:50) (Dako, Denmark). For the detection of keratin 13 and laminin 332tissue equivalents were embedded in optimum cutting temperature compound(OCT) (Tissue-Tek, Sagura, Japan) and frozen at �20 �C. Then, sections of 5 mmthickness were fixed in acetone for 10 min at �20 �C and blocked in phosphate

B. Kinikoglu et al. / Biomaterials 32 (2011) 5756e57645758

buffered saline containing 4% bovine serum albumin (Sigma, USA) and 5% normalgoat serum (Chemicon, USA). Primary antibodies were incubated for 90 min at roomtemperature. The secondary antibody was AlexaFluor 488 IgG (Invitrogen, USA). Forthe detection of the nuclear antigen Ki67, tissue constructs were fixed in formalinand embedded in paraffin. Sections were dewaxed, rehydrated, pretreated at hightemperature for antigen retrieval. They were then incubated with the primaryantibody overnight followed by the secondary antibody coupled to peroxidase. Abrownprecipitate indicating the distribution of the target proteinwas obtained withdiaminobenzidine (DAB) enzyme substrate (Dako, Denmark). Counterstaining wasdone with Harris hematoxylin. Propidium iodide stain (Vector Laboratories, USA)was used to counterstain the cell nuclei. In all immunofluorescence stainings, nativenon-keratinized human oral mucosa was used as positive control. Specimens wereanalyzed with a Nikon Eclipse Fluorescence Microscope.

2.10. Transmission electron microscopy

Tissue constructs were fixed with 2% glutaraldehyde 0.1 M NaCacodylate/HCl, pH7.4 for 2 h and postfixed with 1% osmium tetroxide-0.15 M NaCacodylate/HCl, pH 7.4for 1 h. After dehydration in a growing gradient of ethanol, the samples wereembedded in Epon and finally polymerized at 60 �C for 72 h. The blocks were cutusing an ultramicrotome and sections of 60e80 nm thickness were contrasted withuranyl acetate and lead citrate. Observations were done with a Philips CM 120transmission electron microscope.

2.11. Statistical analysis

Number of Ki67 positive cells in the epithelia of the 2 models (ELR-collagen andcontrol collagen scaffolds) were determined 3 times by 2 experimenters using Ki67immunostaining micrographs of 3 samples for each model and 3 different areas ofeach image. The statistical analysis of the data was done by using the non-parametric two-tailed ManneWhitney U test (Statext v1.41) with 5% significancelevel, a ¼ 0.05. The results were expressed as mean � standard error of the mean.

3. Results and discussion

3.1. H-RGD6 expression and purification

In order to verify the effectiveness of production and purificationmethods, SDS-PAGE and MALDI-TOF mass spectroscopy were per-formed. SDS-PAGE electrophoresis (Fig. 1) of purified ELR sampleconfirmed its purity and monodisperse nature. Production yield ofpurified ELRwas about 200mg/L of bacterial culture. Fig.1 shows alsothe ELR MALDI-TOF mass spectrum, experimental molecular weightfound by this techniquematcheswell the expectedmolecularweight

Fig. 1. Assessment of H-RGD6 purity and molecular weight. The expected mass of the polypeto doubly charged species. (b) Analysis of biopolymer extract by SDS-PAGE.

of the polymer (60661 Da). Complete amino acid sequence of theelastin-like polypeptide containing RGD amino acid sequences wasMESLLP [[(VPGIG)2 (VPGKG)2 (VPGIG)2]2 (AVTGRGDSPASS)[(VPGIG)2 (VPGKG)2 (VPGIG)2]2]6).

3.2. Scaffold characteristics

Scanning electron microscopic analysis showed that the ELR-collagen (1:3 w/w) fibrous scaffold was highly porous withrandomly oriented fibers of an average diameter of 263 � 23 nm(Fig. 2). An advantage of this scaffold was the use of a non-chemicalcrosslinkage approach. It was possible to crosslink the nanofibrousscaffold by theDHTmethod,which does not involve a chemical agentsuch as glutaraldehyde and therefore no release of aldehydes wouldoccur during degradation. In addition, the fiber morphology waspreserved after the DHT, as opposed to other crosslinking methodswhich involve the incubation of the fibrous structure in a solution ofa chemical agent (such as EDC/NHS or genipin). These methodsdisrupt the fibermorphology since both collagen and ELR are solublein aqueous medium. Thus, DHT was found to be a good method toemploy when ELR or collagen is used as the scaffold material.

3.3. Histology

Histological analysis of the oral mucosal equivalent (Fig. 3)showed that the fibroblasts seeded onto the ELR-collagen nano-fibrous scaffold were able to proliferate, migrate within the thick-ness of scaffold, and synthesize new extracellular matrix filling thepores of the scaffold, and forming a lamina propria equivalent (LPE).At the top of this LPE, epithelial cells proliferated during 7 days ofculture under submerged conditions and 14 days of culture at theaireliquid interface, forming a thick non-keratinized, multilayeredepithelium, giving rise to a full-thickness oral mucosal equivalent(OME). The epithelial cells in the superficial layer were seen toretain their nuclei and stratum corneum was absent, as in nativenon-keratinized oral mucosa. The epithelium they formed wasfirmly anchored to the underlying lamina propria equivalent bya continuous and well-organized basement membrane.

ptide is 60611 Da (a) MALDI-TOF of the biopolymer ELR. Signal at 30253 Da is assigned

Fig. 2. Scanning electron microscopic (SEM) analysis of the ELR-collagen (1:3 w/w) fibrous scaffold electrospun from a mixture of 10% total protein concentration, (a) x3000 (b)x24000.

B. Kinikoglu et al. / Biomaterials 32 (2011) 5756e5764 5759

Nanofibrous structures have great potential with theirbiomimetic architecture for promoting cell growth and main-taining cell functions, and it has been demonstrated that a three-dimensional nanofibrous structure similar to that of naturallyoccurring extracellular matrix (ECM) provides better physical andmechanical microenvironment for cell proliferation and differ-entiation [22]. Electrospinning provides a relatively simpleapproach to fabricate macro- to nanoscaled fibers that are withinthe size range of the extracellular matrix (ECM) [23]. However,poor cellular infiltration into electrospun scaffolds was also oftenreported [23e29]. Therefore, several techniques such as selectiveleaching of polyethylene oxide (PEO) or gelatin [23], salt leaching[22,30], gas forming [31] have been proposed to increase theporosity of electrospun scaffolds to allow better cell migration.Micron-sized fibers were also reported to result in enhanced cellinfiltration compared to nanosized fibers [23]. In the presentstudy, the average diameter of the electrospun ELR-collagenfibers was 260 nm, and no additional technique such as salt orporogen leaching has been used to increase the porosity of thescaffold. Nevertheless, the fibroblasts seeded onto our scaffoldwere able to migrate through the whole thickness of the scaffold,proliferate and populate the scaffold and synthesize new extra-cellular matrix such as collagen type I as demonstrated by theTEM analysis.

It has been reported in literature that in electrospinning therewas a positive correlation between solution concentration, fiberdiameter and pore size: the higher was the concentration of the

Fig. 3. Histological analysis of the full-thickness human oral mucosal equivalent basedon the nanofibrous ELR-collagen scaffold. Cell nuclei were stained in blue by hema-toxylin, cytoplasm in pink by phloxine and extracellular matrix of connective tissue inorange/yellow by saffron. Bar ¼ 50 mm.

solution, the thicker the resulting fibers were and the larger thepore sizes in the fibrous mat [23,32]. That is why in the presentstudy, ELR-collagen solution was prepared at the highest possibleconcentration (10% total protein concentration) in order to obtaina relatively large pore sized scaffold. The resulting ELR-collagenfibers were surprisingly in the nanoscale (mean fiber diameter:260 nm) probably due to the possibility that ELR binds to collagenand interferes with incorporation of more collagen molecules intothe fibers [33]. As a result, a 3D scaffold mimicking ECM in itsstructure (small fiber diameters and high matrix porosity) could beproduced. The seeding of epithelial cells on top of fibroblasts in thescaffoldsmight have further enhanced fibroblastmigration throughthe scaffold.

HFIP is a strong organic solvent and its use in electrospinningof proteins such as collagen might raise concerns as to whether itmight damage protein bioactivity. However, studies demon-strated that there were only minor differences between collagenelectrospun from HFIP and from PBS based solutions [34]. Elec-trophoretic analysis showed that collagen fibrils electrospunfrom HFIP exhibited a 67 nm repeat banding pattern that isthought to expose a binding site in the native collagen fibril thatenhances cell adhesion and migration [35]. Results of in vivoexperiments also supported that the use of HFIP during electro-spinning did not harm the bioactivity of the collagen [36]. It wasfound in vivo that constructs composed of collagen electrospunfrom HFIP were fully infiltrated with interstitial cells and werewell integrated with the surrounding muscle tissue of the host.The implant was free from fibrotic encapsulation and there wasa smooth continuum of cells from the host tissue into the elec-trospun collagen. In contrast, electrospun constructs of dena-tured collagen delaminated from the host tissue after 7 daysin vivo. It was revealed that these implants developed fibroticcapsules, were poorly infiltrated with interstitial cells, andappeared to be infiltrated by lymphocytes [35]. In the presentin vitro study the significant impact of ELR that we observed ontissue development indicates that the use of HFIP did not havea significant influence on ELR bioactivity either.

3.4. Immunohistochemistry

Keratin 13 (K13), the major differentiation-associated marker ofnon-keratinized oral mucosa, was very strongly expressed in theOME as in the native oral mucosa (Fig. 4a and b). It was expressed inthe suprabasal layers of our model but not in the basal layer, asreported in literature for native non-keratinized oral mucosa [37].

Fig. 4. Immunofluorescence analysis of the oral mucosa equivalent in comparison to the native oral mucosa. Immunolabelling of keratin 13 (K13) (a and b), the major differentiationmarker of non-keratinized oral mucosa epithelium; the basement membrane protein laminin 332 (c and d); and the nuclear antigen Ki67 (arrows in e and f), marker of proliferatingcells, in the oral mucosal equivalent based on the ELR-collagen nanofibrous scaffold (a,c,e), and in the native oral mucosa (b,d,f). Immunolabelling is shown in green (fluorescence,for K13 and laminin 332) or in brown/black (DAB, for Ki67). Cell nuclei are shown in red. Bars ¼ 50 mm (For interpretation of the references to colour in this figure legend, the readeris referred to the web version of this article.)

B. Kinikoglu et al. / Biomaterials 32 (2011) 5756e57645760

The basement membrane protein laminin 332 (formerly laminin 5),a notable cell adhesion molecule which strongly promotes cellularadhesion and migration much more efficiently than other ECMproteins [38], was strongly expressed all along the interfacebetween lamina propria and epithelium in the reconstructed oralmucosa, as well as in native oral mucosa (Fig. 4c and d). Ki67antigen, an overall marker of proliferative activity, was detected inthe basal cells of the OME throughout the epithelium (Fig. 4e). Thenumber of proliferating cells labeled using anti-Ki67 was as high inthe OME as in native oral mucosa (Fig. 4e and f). As a result, theepithelium of the OME was as thick as that of the native oralmucosa (Fig. 4a and b). In the control oral mucosal equivalent basedon the collagen electrospun scaffold without ELR, a few Ki67positive cells were observed, and its epithelium was thin (Fig. 5).The average number of Ki67 positive cells/area in the ELR-collagenscaffold and the control collagen scaffold were 48.3 � 3.4 and

5.5� 0.3, respectively. This difference between the twomodels wasstatistically significant (ManneWhitney test, p < 0.001).

Keratinized epithelium has the advantage over non-keratinizedepithelium of providing a better barrier against infection due to thepresence of layers such as stratum corneum [39]. Nonkeratinizedepithelium on the other hand is thicker than the keratinized one innative tissues, probably to compensate for the absence of a stratumcorneum in providing the barrier function. Therefore, the oralmucosa reconstructed in the present study not only has theadvantage of high self-renewing potential due to the high pop-ulation of proliferative basal cells, but also has the advantage ofhaving consequently a thick epithelium which would providea better barrier function.

Ki67 antigen is expressed within the nuclei of cells that are ina proliferating stage (G1, S, G2, andM phases) and is also observed inthe basal layer cells of oral mucosa along the basement membrane

Fig. 5. Influence of the ELR on the thickness of the reconstructed epithelium and on the expression of keratin 13. a) Oral mucosal equivalent based on the ELR-collagen electrospunscaffold (10% protein concentration) had a thick epithelium containing a high number of proliferative basal cells Ki67 positive (b), c) in the control collagen electrospun scaffold (10%protein concentration) the epithelium was thin, containing a few cell layers, and a few proliferative basal cells Ki67 positive (d). Keratin 13 was expressed strongly in both.Bars ¼ 50 mm.

B. Kinikoglu et al. / Biomaterials 32 (2011) 5756e5764 5761

[11]. Therefore, besides being considered as an overall marker ofproliferative activity (identified throughout the cell cycle) [40], it isalso frequently used to assess cell epithelial turnover in the oralmucosa [41]. In normal oral mucosa, Ki67 antibody stains less than20% of the epithelial cells and Ki67 positive cells are limited to thebasal layer. However, inpremalignant ormalignantmucosaabnormalKi67positivity is observedalso at suprabasal levels [41]. In ourmodel,theKi67positive epithelial cellswere located strictly at the basal levelwith the same spatial distributionas innormal oralmucosa, andwereuniformly distributed throughout the oral mucosal equivalent. Itshouldbenoted thateven if the basal cells of the reconstructedmodelwere proliferative and undifferentiated, the suprabasal cells werewell-differentiated as in normal oral mucosa, expressing stronglykeratin 13, the major differentiation marker of non-keratinized oralmucosa.

There might be two reasons/potential mechanisms for thisphenomenon: ELR containing RGD might have increased theinitial adhesion of highly proliferating cells such as epithelialprogenitor/stem cells and might have provided a better niche fortheir long-term culture, or it might also have increased theproliferation of the cells in the long run considering that RGD isalso the active motif of the epidermal growth factor (EGF), whichis a known mitogen [19]. Indeed, others have found that ELRcontaining RGD either caused an increase in initial cell adhesion(bone marrow stem cells [4], SaOS-2 [20]) or in both epithelial celladhesion and proliferation [5]. The mechanism by which the ELRshows its effect is worth investigating for both tissue engineeringand regenerative medicine applications and for fundamentalresearch on epithelial disorders such as carcinogenesis. In ourstudy the benefit of using this ELR was obtaining an oral mucosaequivalent that mimicked the native tissue more closely as shownby its thicker epithelium and higher number of proliferating basalepithelial cells, which would ensure the renewal of its epitheliumin vivo.

3.5. Transmission electron microscopy

Transmission electron microscopic analysis of the reconstructedoral mucosa based on the ELR-collagen nanofibrous scaffoldshowed the ultrastructural organization of the epithelium (Fig. 6),the lamina propria equivalent (Fig. 7) and the basement membrane(Fig. 7). In the epithelium, epithelial cells progressively flattenedwhile progressing towards the culture surface and their nuclei werescarcely observed in the upper layers (Fig. 6a). In the superficiallayers the cell periphery showed an intricate pattern of villousplasma membranes filling the intercellular spaces (Fig. 6b). Basalcells were cubic and smaller, compared to the suprabasal andsuperficial ones (Fig. 6c). Numerous intercellular junctions such asdesmosomes were detected between adjacent epithelial cells in allepithelial layers. Their number was lower in basal layers, and theywere also smaller in size in these layers (Fig. 6d). When observed ata higher magnification, their 2 desmosomal plaques became visible(Fig. 6e). In the suprabasal layers, the number of desmosomes washigher and they were better structured and bigger in size (Fig. 6f).Keratin filaments were detected around desmosomes at the highermagnification (Fig. 6g).

A continuous and ultrastructurally well-organized basementmembrane composed of lamina lucida and lamina densa hasformed on the lamina propria equivalent all along the interfacebetween the epithelium and the lamina propria (Fig. 7a and b).Hemidesmosomes were also observed around the basementmembrane (Fig. 7b). These are attachment complexes linked tointermediate filaments that connect epithelial cells to the extra-cellular matrix. They provide tissue integrity and resistance tomechanical forces [42]. Anchoring filaments were observed to linkthese hemidesmosomes to the basal lamina (Fig. 7b). In the laminapropria equivalent of the reconstructed oral mucosa, large amountsof newly synthesized collagen were detected around functionalfibroblasts by transmission electron microscopy (Fig. 7c). At higher

Fig. 6. Ultrastructural analysis of the epithelium of the oral mucosal equivalent based on the ELR-collagen nanofibrous scaffold by transmission electron microscopy (TEM). a) In thebasal and suprabasal layers of the epithelium, cell nuclei (N) were easily detectable; whereas, as the cells differentiated towards the culture surface (arrow) they became flat,elongated and less voluminous, b) superficial layers of the epithelium presented a number of microvilli, c) basal cells of the epithelium were much more voluminous than thesuperficial ones, d) desmosomes (D) were detected between adjacent epithelial cells in all epithelial layers, e) desmosomes observed at higher magnification: the 2 desmosomalplaques became visible, f) in the suprabasal layers, the number of desmosomes was higher and they were better structured and bigger in size, g) keratin (K) filaments detectedaround desmosomes at higher magnification.

B. Kinikoglu et al. / Biomaterials 32 (2011) 5756e57645762

Fig. 7. Ultrastructural analysis of the basement membrane and the lamina propria equivalent of the tissue engineered oral mucosa based on the ELR-collagen nanofibrous scaffoldby transmission electron microscopy (TEM). a) The basement membrane (BM) anchoring firmly the epithelium composed of epithelial cells and keratin (K) filaments to theunderlying lamina propria equivalent which contains newly synthesized collagen fibers (C), b) basement membrane composed of lamina densa (LD) and lamina lucida (LL) at highermagnification. Anchoring fibrils (AF) and a number of hemidesmosomes (HD) were detected, c) in the lamina propria equivalent, an active fibroblast (F) is synthesizing collagen (C),d) transversal and longitudinal sections of newly-synthesized collagen fibers with their characteristic striations visible at higher magnification in the lamina propria equivalent.

B. Kinikoglu et al. / Biomaterials 32 (2011) 5756e5764 5763

magnification, transverse or longitudinal sections with typicalstriations of the collagen fibrils became visible (Fig. 7d).

4. Conclusions

In the present study, using easily accessible humanadult cells, andincorporation of an ELR into a collagen scaffold, we demonstrated forthe first time the possibility to obtain a 3D oral mucosa equivalentwith a self-renewing potential as high as native oral mucosa, with allits basal epithelial cells in proliferative stage. This suggests that evenafter 6 weeks of in vitro culture, the equivalent would still be able toself-renewwhen transplanted in vivo. The reconstructed oralmucosawas shown to closely mimic the native one in other aspects as well,such as its thick epithelium, the expression of markers characteristicof oral epithelial differentiation (keratin 13), the presence of anultrastructurally well-organized basement membrane expressinglaminin 332, and a lamina propria equivalentwithnewly synthesizedcollagen fibers. The scaffold proposed heremight also be used for thereconstruction of soft tissues other than oralmucosa. As futurework,it would be interesting to investigate the dose-effect relationship byusing different ELR-collagen ratios. Hence, the 3D oral mucosaequivalent developed might also be used as a model to study

epidermal proliferation and differentiation, and in studies of epithe-lial disorders including carcinogenesis.

Acknowledgments

This work was supported by METU, Hospices Civils de Lyon,MICINN (project MAT 2007-66275-C02-01), JCyL (projectsVA016B08 and VA030A08) and CIBER-BBN (project CB06-01-0003).B. Kinikoglu was supported by the Scientific and TechnologicalResearch Council of Turkey (TUBITAK) and the French Government.

References

[1] Rodríguez-Cabello JC, Martín L, Alonso M, Arias FJ, Testera AM. “Recombi-namers” as advancedmaterials for the post-oil age. Polymer 2009;50:5159e69.

[2] Chilkoti A, Christensen T, MacKay JA. Stimulus responsive elastin biopolymers:applications in medicine and biotechnology. Curr Opin Chem Biol 2006;10:652e7.

[3] Rodríguez-Cabello JC, Prieto S, Reguera J, Arias FJ, Ribeiro A. Biofunctionaldesign of elastin-like polymers for advanced applications in nano-biotechnology. J Biomater Sci Polym Ed 2007;18:269e86.

[4] Ozturk N, Girotti A, Kose GT, Rodríguez-Cabello JC, Hasirci V. Dynamic cellculturing and its application to micropatterned, elastin-like protein-modifiedpoly(N-isopropylacrylamide) scaffolds. Biomaterials 2009;30:5417e26.

B. Kinikoglu et al. / Biomaterials 32 (2011) 5756e57645764

[5] Martínez-Osorio H, Juárez-Campo M, Diebold Y, Girotti A, Alonso M, Arias FJ,et al. Genetically engineered elastin-like polymer as a substratum to culturecells from the ocular surface. Curr Eye Res 2009;34:48e56.

[6] Betre H, Ong SR, Guilak F, Chilkoti A, Fermor B, Setton LA. Chondrocyticdifferentiation of human adipose-derived adult stem cells in elastin-likepolypeptide. Biomaterials 2006;27:91e9.

[7] MacNeil S. Progress and opportunities for tissue-engineered skin. Nature2007;445:874e80.

[8] Obokata H, Yamato M, Yang J, Nishida K, Tsuneda S, Okano T. Subcutaneoustransplantation of autologous oral mucosal epithelial cell sheets fabricated ontemperature-responsive culture dishes. J Biomed Mater Res A 2008;86A:1088e96.

[9] Tobita T, Izumi K, Feinberg SE. Development of an in vitro model for radiation-induced effects on oral keratinocytes. Int J Oral Maxillofac Surg 2010;39:364e70.

[10] Andreadis ST, Hamoen KE, Yarmush ML, Morgan JR. Keratinocyte growthfactor induces hyperproliferation and delays differentiation in a skin equiv-alent model system. FASEB J 2001;15:898e906.

[11] Tomakidi P, Breitkreutz D, Fusenig NE, Zöller J, Kohl A, Komposch G. Estab-lishment of oral mucosa phenotype in vitro in correlation to epithelialanchorage. Cell Tissue Res 1998;292:355e66.

[12] Izumi K, Takacs G, Terashi H, Feinberg SE. Ex vivo development of a compositehuman oral mucosal equivalent. J Oral Maxillofac Surg 1999;57:571e7.discussion 577e578.

[13] Liu J, Lamme EN, Steegers-Theunissen RP, Krapels IP, Bian Z, Marres H, et al.Cleft palate cells can regenerate a palatal mucosa in vitro. J Dent Res 2008;87:788e92.

[14] Collombel C, Damour O, Gagnieu C, Marichy J, Poinsignon F. Biomaterials witha base of collagen, chitosan and glycosaminoglycans; process for preparingthem and their application in human medicine. French Patent No. 8708252,1987. European Patent No. 884101948, 1988. US Patent No. PCT/FR/8800303,1989.

[15] Kinikoglu B, Auxenfans C, Pierrillas P, Justin V, Breton P, Burillon C, et al.Reconstruction of a full-thickness collagen-based human oral mucosalequivalent. Biomaterials 2009;30:6418e25.

[16] Peña I, Junquera LM, Meana A, García E, García V, De Vicente JC. In vitroengineering of complete autologous oral mucosa equivalents: characteriza-tion of a novel scaffold. J Periodontal Res 2010;45:375e80.

[17] Xiong X, Zhao Y, Zhang W, Xie W, He S. In vitro engineering of a palatalmucosa equivalent with acellular porcine dermal matrix. J Biomed Mater ResA 2008;86A:544e51.

[18] Xiong X, Jia J, He S, Zhao Y. Cryopreserved lip mucosa tissue derived kerati-nocytes can fabricate tissue engineered palatal mucosa equivalent. J BiomedMater Res B Appl Biomater 2010;94:165e70.

[19] Berk BC, Brock TA, Webb RC, Taubman MB, Atkinson WJ, Gimbrone Jr MA,et al. Epidermal growth factor, a vascular smooth muscle mitogen, induces rataortic contraction. J Clin Invest 1985;75:1083e6.

[20] Costa RR, Custódio CA, Testera AM, Arias FJ, Rodríguez-Cabello JC, Alves NM,et al. Stimuli-responsive thin coatings using elastin-like polymers forbiomedical applications. Adv Funct Mater 2009;19:3210e8.

[21] Zorlutuna P, Elsheikh A, Hasirci V. Nanopatterning of collagen scaffoldsimprove the mechanical properties of tissue engineered vascular grafts. Bio-macromolecules 2009;10:814e21.

[22] Kim TG, Chung HJ, Park TG. Macroporous and nanofibrous hyaluronic acid/collagen hybrid scaffold fabricated by concurrent electrospinning and depo-sition/leaching of salt particles. Acta Biomaterialia 2008;4:1611e9.

[23] Ekaputra AK, Prestwich GD, Cool SM, Hutmacher DW. Combining electrospunscaffolds with electrosprayed hydrogels leads to three-dimensional cellula-rization of hybrid constructs. Biomacromolecules 2008;9:2097e103.

[24] Jin HJ, Chen JS, Karageorgiou V, Altman GH, Kaplan DL. Human bone marrowstromal cell responses on electrospun silk fibroin mats. Biomaterials 2004;25:1039e47.

[25] Min BM, Lee G, Kim SH, Nam YS, Lee TS, Park WH. Electrospinning of silkfibroin nanofibers and its effect on the adhesion and spreading of normalhuman keratinocytes and fibroblasts in vitro. Biomaterials 2004;25:1289e97.

[26] Mo XM, Xu CY, Kotaki M, Ramakrishna S. Electrospun P(LLA-CL) nanofiber:a biomimetic extracellular matrix for smooth muscle cell and endothelial cellproliferation. Biomaterials 2004;25:1883e90.

[27] HeW, Yong T, Teo WE, Ma Z, Ramakrishna S. Fabrication and endothelializationof collagen-blended biodegradable polymer nanofibers: potential vascular graftfor blood vessel tissue engineering. Tissue Eng 2005;11:1574e88.

[28] Shin HJ, Lee CH, Cho IH, Kim YJ, Lee YJ, Kim IA, et al. Electrospun PLGAnanofiber scaffolds for articular cartilage reconstruction: mechanical stability,degradation and cellular responses under mechanical stimulation in vitro.J Biomater Sci Polym Ed 2006;17:103e19.

[29] Venugopal JR, Zhang YZ, Ramakrishna S. In vitro culture of human dermalfibroblasts on electrospun polycaprolactone collagen nanofibrous membrane.Artif Organs 2006;30:440e6.

[30] Nam J, Huang Y, Agarwal S, Lannutti J. Improved cellular infiltration in elec-trospun fiber via engineered porosity. Tissue Eng 2007;13:2249e57.

[31] Lee YH, Lee JH, An IG, Kim C, Lee DS, Lee YK, et al. Electrospun dual-porositystructure and biodegradation morphology of Montmorillonite reinforced PLLAnanocomposite scaffolds. Biomaterials 2005;26:3165e72.

[32] Eichhorn SJ, Sampson WW. Statistical geometry of pores and statistics ofporous nanofibrous assemblies. J R Soc Interface 2005;2:309e18.

[33] Kinikoglu B, Rodríguez-Cabello JC, Damour O, Hasirci V. A smart bilayerscaffold of elastin-like recombinamer and collagen for soft tissue engineering.J Mater Sci Mater Med. doi:10.1007/s10856-011-4315-6.

[34] Dong B, Arnoult O, Smith ME, Wnek GE. Electrospinning of collagen nanofiberscaffolds from benign solvents. Macromol Rapid Commun 2009;30:539e42.

[35] Telemeco TA, Ayres C, Bowlin GL, Wnek GE, Boland ED, Cohen N, et al.Regulation of cellular infiltration into tissue engineering scaffolds composedof submicron diameter fibrils produced by electrospinning. Acta Biomater2005;1:377e85.

[36] Sell SA, McClure MJ, Garg K, Wolfe PS, Bowlin GL. Electrospinning of collagen/biopolymers for regenerative medicine and cardiovascular tissue engineering.Adv Drug Deliv Rev 2009;61:1007e19.

[37] Moharamzadeh K, Brook IM, Van Noort R, Scutt AM, Thornhill MH. Tissue-engineered oral mucosa: a review of the scientific literature. J Dent Res 2007;86:115e24.

[38] Miyazaki K. Laminin-5 (Iaminin-332): unique biological activity and role intumor growth and invasion. Cancer Sci 2006;97:91e8.

[39] Nouri K. Skin Cancer. The McGraw-Hill Companies, Inc; 2007. 2.[40] Hamadah O, Goodson ML, Thomson PJ. Clinicopathological behaviour of

multiple oral dysplastic lesions compared with that of single lesions. Br J OralMaxillofac Surg 2010;48:503e6.

[41] Montebugnoli L, Badiali G, Marchetti C, Cervellati F, Farnedi A. FoschiniMP.Prognostic value of Ki67 from clinically and histologically ‘normal’ distantmucosa in patients surgically treated for oral squamous cell carcinoma:a prospective study. Int J Oral Maxillofac Surg 2009;38:1165e72.

[42] Zhang H, Labouesse M. The making of hemidesmosome structures in vivo. DevDyn 2010;239:1465e76.