Alimentary ‘green’ proteins as electrospun scaffolds for skin regenerative engineering

Alimentary ‘green’ proteins as electrospunscaffolds for skin regenerative engineeringLeko Lin1, Anat Perets1, Yah-el Har-el1, Devika Varma1, Mengyan Li1, Philip Lazarovici2,Dara L. Woerdeman3 and Peter I. Lelkes1*1School of Biomedical Engineering, Science and Health Systems, Drexel University, Philadelphia, PA, USA2School of Pharmacy, Hebrew University of Jerusalem, Israel3R&D Green Materials LLC, Philadelphia, PA, USA

Abstract

As a potential alternative to currently available skin substitutes and wound dressings, we explored the use ofbioactive scaffolds made of plant-derived proteins. We hypothesized that ‘green’ materials, derived fromrenewable and biodegradable natural sources, may confer bioactive properties to enhance wound healingand tissue regeneration. We optimized and characterized fibrous scaffolds electrospun from soy protein isolate(SPI) with addition of 0.05% poly(ethylene oxide) (PEO) dissolved in 1,1,1,3,3,3-hexafluoro-2-propanol, andfrom corn zein dissolved in glacial acetic acid. Fibrous mats electrospun from either of these plant proteinsremained intact without further cross-linking, possessing a skin-like pliability. Soy-derived scaffolds supportedthe adhesion and proliferation of cultured primary human dermal fibroblasts. Using targeted PCR arrays andqPCR validation, we found similar gene expression profiles of fibroblasts cultured for 2 and 24h on SPI sub-strates and on collagen type I at both time points. On both substrates there was a pronounced time-dependentupregulation of several genes related to ECM deposition remodelling, includingMMP-10,MMP-1, collagen VII,integrin-a2 and laminin-b3, indicating that both plant- and animal-derived materials induce similar responsesfrom the cells after initial adhesion, degrading substrate proteins and depositing extracellular matrix in a‘normal’ remodelling process. These results suggest that ‘green’ proteins, such as soy and zein, are promisingas a platform for organotypic skin equivalent culture, as well as implantable scaffolds for skin regeneration.Future studies will determine specific mechanisms of their interaction with skin cells and their efficacy inwound-healing applications. Copyright © 2012 John Wiley & Sons, Ltd.

Received 17 January 2011; Revised 20 January 2012; Accepted 24 January 2012

Keywords soy protein; zein; electrospinning; fibroblast; scaffold; wound healing; in vitro; gene expression

1. Introduction

Non-healing skin wounds afflict millions of patients world-wide and translate into global wound care expendituresamounting to $13–15 billion annually (Fonder et al.,2008). Many currently available skin substitutes useacellular collagen scaffolds, alone or in conjunction withfibroblasts and/or keratinocytes (Powell et al., 2008). Othernatural proteins, such as elastin (Rnjak et al., 2009, 2011;Rnjak-Kovacina et al., 2011) and silk fibroin (Min et al.,2004; Jeong et al., 2009), as well as polysaccharides suchas chitosan (Chen et al., 2010b) and glycosaminoglycans

such as esterified hyaluronic acid (Figallo et al., 2007;Scuderi et al., 2009) have also been investigated for skintissue engineering. However, allogeneic or xenogeneicscaffolds made from animal sources present risks of hostrejection and disease transfer, and pose cost and availabilitylimitations. Synthetic alternatives may lack the biologicalcues of natural matrix components, or do not match theirmechanical properties and are often not biodegradable(Metcalfe and Ferguson, 2007). A lesser-addressed issue isthat of the patients’ religious and ethical restrictions.Certain beliefs reject the use of human (fetus- or cadaver-derived) products, while others object to porcine- orbovine-derived material (Enoch et al., 2005). Given thesechallenges, plant proteins may offer advantages over othermaterials as a base material for tissue-engineering scaffolds.

Plant components, which are derived from renewablesources, are increasingly being considered for biomedical

* Correspondence to: P. I. Lelkes, Drexel University, BIOMED,Bossone 623, 3141 Chestnut Street, Philadelphia, PA 19104,USA. E-mail: [email protected]

Copyright © 2012 John Wiley & Sons, Ltd.

JOURNAL OF TISSUE ENGINEERING AND REGENERATIVE MEDICINE RESEARCH ARTICLEJ Tissue Eng Regen Med (2012)Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/term.1493

applications in various forms, including as scaffold materi-als for tissue engineering. For example, starch has beenused in conjunction with poly(e-caprolactone) (PCL) toform porous fibre-mesh scaffolds for the culture of ratbone marrow stromal cells (Tuzlakoglu et al., 2005) andendothelial cells (Santos et al., 2007, 2008), and veryrecently these starch–PCL scaffolds have been shown tosupport vascularization by outgrowth endothelial cells inco-culture with primary human osteoblasts (Ghanaatiet al., 2011). Wheat gluten was most likely the first plantprotein to have been electrospun (Woerdeman et al.,2005), paving the way for fabrication of nanofibrous orsubmicron fibrous scaffolds from other plant-derivedproteins. For example, corn zein has been shown to be asuitable biomaterial in the form of films (Sun et al.,2005) and porous scaffolds (Gong et al., 2006; Wanget al., 2007; Tu et al., 2009; Xu et al., 2010), and has beenelectrospun by different groups (Yao et al., 2007; Sellinget al., 2007) using a variety of solvents.

The suitability of soy protein for biomedical applica-tions has been well demonstrated (Vaz et al., 2003a,2003b, 2003c, 2004; Silva et al., 2003, 2005; Curt et al.,2009; Luo et al., 2010). For example, Vaz et al. (2003a)compared cast films made from soy protein with thosemade from gelatin, casein and sodium caseinate, andfound that soy films were most resistant to hydrolysis,whether or not they were crosslinked; this was attributedto the globular structure of soy protein, unlike thecoiled or helical structure of other proteins. Reddy andYang (2009) developed soy protein fibres using wetspinning without additives or crosslinking. These fibreswith diameters of 50–150 mm supported the growth ofmouse fibroblasts for 7 days. Recently, we and othershave developed the methodology to electrospin soyproteins (Lelkes et al., 2008; Vega-Lugo and Lim, 2008;Fung et al., 2010).

Electrospinning is an attractive technique for fabricat-ing fibrous scaffolds from natural and synthetic polymers,alone or in combination (Bhardwaj and Kundu, 2010).This platform technology enables a good degree of controlover the fibre size and porosity of the ensuing fibrousmeshes, which depends on the fibre diameter as well asthe thickness (Rnjak-Kovacina et al., 2011; Solimanet al., 2011). Electrospun scaffolds support the spreadingand in some cases also the infiltration of cells, such assmooth muscle cells, inside an electrospun vascular graft(Han et al., 2011). Our laboratory recently establishedcarefully optimized electrospinning parameters fornatural proteins, such as collagen and elastin, and blendsof natural and synthetic polymers (Li et al., 2005; Lelkeset al., 2008; Han et al., 2011). Submicron-scale fibresare desirable for tissue-engineering applications, as theymaximize the surface area:volume ratio of the fibresand thus provide more topology for cellular support fora more organized, continuous epithelium (Powell andBoyce, 2007).

In this study we focused on electrospinning two plant-derived natural polymers. Soy protein was selected forits potential in tissue-engineering applications (Vaz et al.,

2003a, 2003b, 2003c, 2004; Silva et al., 2003, 2005;Curt et al., 2009; Luo et al., 2010). Soy protein isolate(SPI), commercially purified to a purity of> 90%,consists of two main fractions of protein, namely a 7 Sfraction, which is pure glycinin, and an 11 S fraction,comprising mostly b-conglycinin as well as small quanti-ties of g-conglycinin, lipoxygenases, a-amylases andhaemagglutinins (Nielsen, 1985). SPI is extremelyversatile in that it can be formulated into films, powders,coatings, solids, gels or fibres (Zhang et al., 2003; Jianget al., 2007a; Mauri and Añon, 2006; Maltais et al., 2007;Tian et al., 2010). From a biological standpoint, soy proteinoffers certain advantages as a possible biomaterial, due tothe bioactivity of individual ‘cryptic’ peptides (Maruyamaet al., 2003) and isoflavones that may be metabolized bythe body upon resorption of the scaffold. Among thepeptides, lunasin and the Bowman–Birk inhibitor (BBI)have been shown to suppress carcinogenesis in vitro aswell as in animal models in vivo (Armstrong et al., 2003;de Lumen, 2005).

Zein, the major storage protein of corn, comprises mainlya- and b-fractions of 14–24kDa (Shukla and Cheryan,2001). As a film, it has been shown to promote the adhesionof HL-7702 human liver cells to a greater extent than acollagen coating, and its degradation products wereshown to stimulate the proliferation of HL-7702 cellswhen added to cell culture medium (Sun et al., 2005).Zein hydrolysis also yields bioactive peptides, such asangiotensin-converting enzyme (ACE)-inhibiting peptide(Miyoshi et al., 1991), which incidentally can also bederived from soy protein (Mallikarjun Gouda et al.,2006). ACE upregulation has been implicated in fibroticremodelling in pathological scars, as it appears to havesignificantly higher activity in fibrotic scars than innormal and wounded skin tissue (Morihara et al., 2006).

We hypothesize that wound dressings made of plantproteins might contribute to a more natural healingprocess and be a suitable alternative to currently avail-able treatments for wounds. As a first step towardstesting our hypothesis, we describe here, for the firsttime, the optimization of the conditions for consistentlyelectrospinning bead-free submicron-size fibres from SPIwith a minimal addition of high molecular weight syn-thetic polymer, poly(ethylene oxide) (PEO), to increasechain entanglements, and from zein without any addi-tives. Furthermore, we tested the mechanical propertiesof soy-protein based mats and demonstrated that suchscaffolds support the adhesion and growth of humanfibroblasts. To gain insight into the mechanisms ofcellular adhesion to foreign material substrates, weutilized focused gene arrays to compare differentialexpression of some integrins and matrix metalloproteinases(MMPs) in primary human dermal fibroblasts cultured onsurfaces coated with, respectively, soy protein and collagentype I, the latter being a ubiquitous native extracellularmatrix protein. Our results indicate that the cells expresssimilar sets of extracellular matrix proteins, integrins andmetalloproteinases to attach to and to remodel thesedissimilar substrates.

L. Lin et al.

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

2. Materials and methods

2.1. Production of the scaffolds

2.1.1. Preparation of protein solutions

Soy protein isolate (SPI; obtained from Cargill Health andFood Technologies, Minneapolis, MN, USA) was blendedwith poly(ethylene oxide) (PEO; Sigma, St. Louis, MO,USA) by first dissolving 0.5% w/v PEO in 1,1,1,3,3,3-hexafluoro-2-propanol (HFP; Sigma) and adding appro-priate volumes from this stock solution to 5%, 6%, 7%and 8% w/v SPI in HFP, respectively. Blend solutions wereleft to stir at room temperature (RT) for at least 48h andfor up to ca. 2weeks before electrospinning to ensurecomplete and homogeneous dissolution.

Corn zein (obtained from Freeman Industries LLC,Tuckahoe, NY, USA) was dissolved at 35%, 40% and 45%w/v in glacial acetic acid (Fisher Scientific, Pittsburgh, PA,USA) and left to stir for at least 3 h and up to ca. 1weekbefore electrospinning.

2.1.2. Electrospinning of protein solutions

Fibres were electrospun in a horizontal set-up, as previouslydescribed (Li et al., 2005). In brief, a digital precisionsyringe pump (KD Scientific Single Syringe Infusion Pump,Fisher Scientific) was used to eject solution from a dispos-able 3ml syringe (BD Biosciences, Franklin Lakes, NJ,USA) through an 18-gauge needle. In preliminary paramet-ric studies we identified optimal settings that would yieldconsistent, bead-free fibres. For further experiments, SPI/PEO blend solutions were electrospun at the optimizeddelivery rate of 0.8ml/h, air gap distance of 15 cm andaccelerating voltage of 12 kV. For zein solutions, theoptimized delivery rate was 0.5ml/h, air gap distance15 cm and accelerating voltage 20kV for 35% and 40%solutions and 25kV for 45% solutions. As a syntheticmaterial for comparison, 20% w/v 90:10 poly(lactic-co-glycolic acid) (PLGA) was electrospun at 0.5ml/h,15 kV and 15 cm. As a natural, animal-derived material forcomparison, 8% w/v gelatin (Sigma) was electrospun at0.8ml/h, 12 kV and 20 cm. To capture the electrospunfibres, circular glass cover slips (Fisher), 15mm in diameter,were attached to a rectangular aluminium collector. Formicroscopic analysis of individual fibres, 0.2ml of solutionwas electrospun; as cell culture substrates, 1ml waselectrospun onto the cover slips, yielding scaffolds of ca.20mm thickness. For mechanical testing, fibrous mats(~0.1–0.2mm thickness) were electrospun and collecteddirectly on the aluminium collector.

2.2. Materials characterization

2.2.1. Measurement of fibre diameters

Glass cover slips coatedwith electrospunfibresweremountedonto metal stubs with carbon tape and sputter-coated for

30 s with platinum and palladium (Pt/Pd) prior to visuali-zation in an environmental scanning electron microscope(ESEM; XL-30 SEM-FEG, FEI, Hillsboro, OR, USA) at anaccelerating voltage of 10 kV and spot size 3. For eachsample, at least 10 measurements/image were taken from10 images of random areas. Fibre diameters were measuredusing University of Texas Health Science Center at SanAntonio (UTHSCSA) ImageTool 3.0 software (for eachsample, 100 individual fibres were evaluated).

Dry samples were mounted as spun. To investigatethe degree of swelling with hydration, samples ofSPI/PEO (soy) and zein scaffolds were immersed in1� phosphate-buffered saline (PBS) solution withoutcalcium and magnesium (Mediatech Inc., Herndon, VA,USA) for 3 h and then dehydrated before mounting forSEM. Soy scaffolds were dehydrated first in 100% ethanol(EtOH) before critical point drying (CPD7501, SPI Supplies,West Chester, PA, USA) to preserve structural morphology.Since zein fibres dissolved in EtOH and were damaged byacetone, an alternative dehydration process of oven-dryingat 37 �C overnight was used instead.

2.2.2. Mechanical testing

Mats were electrospun and cut into rectangular pieces(approximately 5 � 25 � 0.1mm). Tensile tests wereperformed in triplicate for at least three independentsamples, using an Instron 5564, in dry and hydrated states.For testing hydrated samples, the mats were immersed inDulbecco’s modified Eagle’s medium (DMEM; Mediatech)supplemented with 10% fetal bovine serum (FBS; AtlantaBiologicals, Lawrenceville, GA, USA). We used DMEM+10% FBS rather than just PBS as ‘simulated body fluid’,because the presence of proteases in serum more closelysimulates the physiological environment of body fluids thandoes PBS alone. For testing the effect of hydration on tensileproperties over time, scaffolds electrospun from 7% SPI/0.05% PEO were immersed in DMEM+10% FBS, 1� PBSor distilled deionized water for 3 h (0days), 7 days and14days at 37 �C and 5% CO2. Samples were tested tobreakage at a gauge length of 15mm on a 10N load cell.Crosshead speed was 1mm/min for dry samples and10mm/min for wet samples, as the higher speed wasmore appropriate to account for the higher ductility ofwet samples.

Viscoelastic properties of the scaffolds were measuredusing a Stresstech instrument (ATS Rheosystems, Borden-town, NJ, USA). Soy scaffolds were electrospun onto a5 � 5 cm target, as described above. For each scaffold,a dynamic oscillatory strain sweep was conducted todetermine the linear viscoelastic region, followed by afrequency sweep. After the frequency sweep, the samplewas hydrated with distilled water and the same measure-ments were repeated.

2.2.3. In vitro degradation

Soy scaffolds were electrospun and cut into rectangularpieces (ca. 10 � 25 � 0.1mm). Each piece was placed in

Electrospun soy scaffold for regenerative engineering

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

a 60mm diameter Petri dish and its dry weight was takenas a baseline. Pieces were then sterilized by UV exposureand divided into four groups, with at least threespecimens/group/time-point, to be hydrated in 3mldistilled deionized water, DMEM, DMEM+10% FBS, andDMEM/equine serum (ES) in a 1:1 ratio, respectively.Samples were kept in a humidified incubator at 37 �C and5% CO2; the hydration media were changed daily. Attime-points of 1, 4, 7, 14, 21, 28, 35 and 42days, the mediawere removed and the specimens washed twice with waterto remove residual salts and placed in a vacuum oven at50 �C overnight for complete dehydration. The specimenswere weighed in their dishes and the percentage mass losswas calculated as:

%mass loss ¼ 100� original mass� final massð Þ=original mass

2.3. In vitro cytocompatibility

2.3.1. Cell culture

Primary human dermal fibroblasts (HDFs) from neonatalforeskin (Clonetics, San Diego, CA, USA) were cultured inT-75 culture flasks in DMEM with 4.5 g/l glucose supple-mented with 10% FBS and 50 IU/ml penicillin, 50mg/mlstreptomycin (complete medium, from Mediatech) andused at passages 18–20. Flasks were kept in humidifiedincubators at 37 �C and 5% CO2; the culture medium waschanged twice weekly. At confluence, cells were detachedfrom flasks by incubating in 0.25% trypsin/2mM EDTA(Mediatech) for 5min and centrifuged for 5min in com-plete medium prior to counting and seeding onto scaffolds.

2.3.2. Seeding of fibroblasts onto scaffolds

Scaffolds were sterilized by immersion in a 1:10 dilutionof antibiotic–antimycotic solution (ABAM; Mediatech) in1� PBS for 48 h, followed by immersion in a 1:50 dilutionof ABAM for 24 h. The scaffolds were then washed threetimes with 1� PBS without calcium and magnesium toremove traces of ABAM before transferring to 24-wellplates, using sterile forceps inside a laminar flow hood,and securing with Viton O-rings (Cole-Parmer, VernonHills, IL, USA; Li et al., 2006; Han et al., 2011).

Fibroblasts were seeded at a density of 30 000 cells/wellin 200ml volumes onto scaffolds in triplicate in a 24-wellplate and cultured for up to 8days. For each type ofscaffold, one scaffold was kept with no cells as a blank,i.e. to compensate for any effect of scaffold material onthe AlamarBlue readings (see below). Soy protein scaffoldswere prepared from 5%, 6%, 7% and 8% SPI, each with0.05% PEO; zein scaffolds were prepared at 35%, 40%and 45%. Cells were also seeded in triplicate onto scaffoldselectrospun from 20% poly(lactic-co-glycolic acid) (PLGA)and 8% gelatin as synthetic and natural control materials,respectively, and also onto glass cover slips and directly intothe well, as standard controls.

2.3.3. AlamarBlue assay

Cell proliferation was determined using the alamarBlueW

assay, as previously described (Nikolaychik et al., 1996;Li et al., 2005, 2006; Guo et al., 2007; Han et al., 2011).After allowing the cells to attach to the scaffolds for 1 h,complete medium with 10% v/v AlamarBlue (BD Biosci-ence) was added to each well to a total volume of 0.5ml.The plates were incubated at 37 �C, 5% CO2 for 3 h before200ml supernatant from each well was transferred induplicate to a 96-well plate. AlamarBlue fluorescence wasread on a Cytofluor fluorescence microplate reader (excita-tion, l=530nm; emission, l=580nm) and calibratedagainst standard curves, using known amounts of cellsgrown at a range of densities (1000–500 000 cells/well)on tissue culture polystyrene (TCPS) inside a 24-well plate.In preliminary studies, we determined the optimal amountof cells for proliferation studies to be 30 000 cells/well.

2.3.4.Visualization of cell morphology on scaffolds

The morphology of cells on the scaffolds was assessedby fluorescence microscopy and SEM, as previouslydescribed (Li et al., 2005, 2006; Han et al., 2011). Forfluorescence microscopy, the specimens were fixed in3.7% paraformaldehyde (Fisher) for 10min, washedthree times with 1� PBS, and incubated with 4mg/mlHoechst 33258 (BBZ, bisbenzimide; Sigma) and 2mg/mlTRITC-conjugated rhodamine phalloidin (phalloidintetramethyl-rhodamine B isothiocyanate; Sigma) in 1�PBS with 0.2% Triton-X100 (Sigma) for 15min forstaining nuclei and F-actin cytoskeleton, respectively.Samples were visualized in a Leica DMRX upright micro-scope equipped with the appropriate fluorescence filters.Digital images were acquired with a Leica DC 300FXcamera. For SEM visualization, the samples were fixed with2.5% glutaraldehyde in PBS for 1 h at 4 �C, washed threetimes with 1� PBS, and dehydrated in an ethanols gradient15%, 30%, 50%, 75%, 85%, 90%, 95% and 100%, aspreviously described (Li et al., 2005, 2006). Dehydratedsamples were critical point-dried and mounted onto stubsfor SEM visualization.

2.4. Polymerase chain reaction (PCR) arrayanalysis and validation

2.4.1. Substrates

In order to investigate the role of soy protein on celladhesion, and not to confound the results with potentialcontributions of three-dimensional (3D) fibre topology,solutions of SPI/PEO were prepared as for electrospinningand then diluted in HFP to a final concentration of 0.1%w/v SPI with 0.001% w/v PEO. Glass cover slips, 15mmin diameter, were coated with this diluted solution bydrop-wise pipetting and allowing the solvent to evaporatein a fume hood overnight. These substrates were thensterilized by UV exposure for 30min, transferred into a

L. Lin et al.

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

24-well plate and immersed in 500ml 1� PBS to equilibratepH. For collagen substrates, 200ml of a 10mg/ml solution ofcollagen type I from rat tail (BD Biosciences) was pipetteddrop-wise into each well of the same 24-well plate contain-ing the sterile glass cover slips and allowed to bind for30min in a humidified incubator at 37 �C and 5% CO2. Allwells were aspirated and washed twice with 1� PBS priorto cell seeding. Cells were allowed to attach for 2 h or 24hon each substrate before RNA isolation.

2.4.2. RNA isolation

Primary human dermal fibroblasts were seeded onto coatedsubstrates in triplicate wells at a density of approximately150 000 cells/well in serum-free DMEM. At 2h and 24hpost-seeding, respectively, the medium was removed andthe cells were washed and lysed for RNA isolation, usingthe RNeasy Mini Kit (Qiagen, Valencia, CA, USA) accordingto the manufacturer’s instructions. RNA concentration andpurity were quantified using a NanoDrop ND-1000 spectro-photometer (Thermo-Scientific, West Palm Beach, FL,USA), where a ratio of absorbance at 260nm and 230nm(A260/A230)> 1.7 was taken as an indication of qualityRNA. Agarose gel electrophoresis and ethidium bromidestaining (not shown) displayed two bands of 28S and 18Sribosomal RNA (rRNA), further emphasizing RNA integrity.

2.4.3. Focused PCR arrays for human extracellularmatrix (ECM) and adhesion molecules

High-quality RNA (1mg) obtained as above was used tosynthesize complementary DNA (cDNA), using the RT² NanoPreAMP cDNA Synthesis Kit (SABiosciences, Frederick, MD,USA); DNA was diluted, added to a master mix accordingto the manufacturer’s instructions and loaded into eachwell of a 96-well PCR array plate, targeting human ECMand cellular adhesion molecules (PAHS-013, SABios-ciences). PCR was performed on a Stratagene MX3000P(Agilent Technologies, La Jolla, CA, USA). The data wereanalysed using RT2 Profiler PCR Array Data Analysissoftware (SABiosciences) and MultiExperiment Viewersoftware (MeV v4.5, Dana-Farber Cancer Institute, Boston,MA, USA; Saeed et al., 2006).

2.4.4. Validation of PCR array data

To validate the microarray data, we assessed differentialexpression of select target mRNA species, using quantitativereal-time PCR. Fibroblasts were cultured, as above, on glasscover slips coated with SPI, PLGA and collagen type Isolutions and on uncoated glass cover slips for 2 h and24h. At each time point, the cells were lysed with TRIzol(Invitrogen, Carlsbad, CA, USA) and RNA was isolated bythe phenol–chloroform method. cDNA was synthesizedusing the High Capacity cDNA Reverse Transcription Kit(Applied Biosystems) and PCR was performed on a Strata-gene MX3000P (Agilent), using TaqMan Fast UniversalPCR Master Mix (Applied Biosystems) and inventoriedTaqMan primers (Applied Biosystems) with FAM reporter

dye for the following human genes: matrix metallopepti-dase 10 (MMP-10, Hs00233987_m1, NCBI referencesequence NM_002425.1); MMP-1 (Hs00899660_g1,NM_001145938.1); integrin-a1 (ITGA2, Hs00158127_m1,NM_002203.3); collagen type VII-a1 (COL7A1, Hs01574741_g1,NM_000094.3); and laminin-b3 (LAMB3, (Hs00165078_m1,NM_001017402.1). These genes were chosen because theywere identified as highly upregulated using the PCR arrays.Gene expression was normalized to that of peptidylprolylisomerase A (PPIA, Hs99999904_m1, NM_021130.3) asthe invariant housekeeping gene.

2.5. Statistical analysis

All experiments were repeated at least three times intriplicate. Data are expressed as mean� standard deviation(SD), where applicable. One-way analysis of variance(ANOVA) was used to determine statistical significance(p< 0.05). The Holm–Sidak method for post hoc testingwas employed, using SigmaStat software (Systat Software,Chicago, IL, USA).

3. Results and discussion

3.1. Fabrication of scaffolds

3.1.1. Soy protein isolate

Soy protein isolate (SPI) containing two major proteinfractions, of molecular weights ca. 180 and 350kDa (Huaet al., 2005), could be dissolved in weak (0.1 N) and strong(neat) alkali, including ammonium hydroxide and sodiumhydroxide (NaOH), as well as HFP, but not in acetic acidor other common organic solvents, such as dimethylforma-mide (DMF) and tetrahydrofuran (THF). Pure SPI solutionscould not be electrospun into fibres at any concentration orset of parameters tested, as only electrospraying occurred atlower concentrations and complete occlusion of the syringeneedle occurred at higher concentrations. The addition oflow concentrations of PEO, down to as little as 0.05% w/vfor SPI dissolved in HFP or 0.5% w/v for SPI dissolved in0.1 N NaOH, allowed the formation of fibres. We believethat the beneficial effect of such low amounts of PEO canbe attributed to increased chain entanglements in thesolution (Shenoy et al., 2005), due to the presence of highmolecular weight PEO (MW=106 g/mol). This finding isin line with the recent report by Vega-Lugo and Lim(2008), who used a combined thermal and alkaline treat-ment to dissolve SPI in 1% NaOH and also found that SPIdid not yield electrospun fibres when dissolved alone,but required the addition of ~0.8% w/w PEO to formconsistent fibres. They obtained fibres of ~200–300 nmin diameter from solutions of 10–15% SPI with 0.8%PEO and 0.5–1.0% Triton X-100 as a surfactant. Recently,Fung et al. (2010) also electrospun the supernatant of thesoluble fraction of soy fibres extracted from soy residue,dissolved in dilute NaOH, with the addition of 5% w/v

Electrospun soy scaffold for regenerative engineering

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

PEO, and obtained tubular fibres with diameters< 500nm.By contrast to the results reported by Vega-Lugo and Lim,our preliminary studies (Lelkes et al., 2008) indicated thatSPI/PEO solutions in NaOH could be electrospun into fibresof similar size even without heating the solution, but thatthe resulting fibres were prone to hydrolysis. Thus, oursubsequent studies were carried out using HFP as a solvent,yielding several benefits. The amount of PEO needed toform fibres is more than one order of magnitude lower thanthat reported by Vega-Lugo and Lim (2008). Significantly,soy fibrous mats electrospun from HFP solutions werestable in aqueous media without the need for furthercrosslinking, exhibited better ductility, and were signifi-cantly more durable than those electrospun from aqueousNaOH solutions. We believe this reflects the enhancedsolvent power of HFP over aqueous NaOH.

In preliminary studies, we ascertained that the qualityof the resultant electrospun fibres was not reduced whensolutions were left to stir for> 2weeks (data not shown).We tested a range of SPI concentrations from 5% to 8%:below 5% the solution yielded only inconsistent fibres,

and above 8% the viscous solution obstructed the needletip during electrospinning. Figure 1 shows the distinctmorphologies of soy fibres at different SPI concentrationsand different relative contents of PEO. Within the range ofSPI concentrations (5–8%) tested, adding a minimumof 0.05% PEO resulted in consistent fibre formation at bothlow and high concentrations of SPI (Table 1). At allconcentrations tested, soy fibres possessed a flattened,ribbon-like morphology (Figure 1d, inset). Interestingly,this flattened morphology is shared with other electrospunglobular proteins, such as haemoglobin and myoglobin(Barnes et al., 2006).

It appears that while a higher SPI concentration corre-lates to a larger fibre size, as also reported for other naturalproteins (Li et al., 2005; Bhardwaj and Kundu, 2010), thePEO content determines the consistency of fibre formation.Figure 2 shows increasing fibre diameter with SPI concen-trations, with PEO concentration kept constant at either0.025% or 0.05%. Surprisingly, a lower content of PEOdid not correlate with smaller fibre diameters (Figure 2).This may be because the increased chain entanglements in

Figure 1. Representative images showing consistent SPI/PEO fibre formation as a function of both SPI and PEO concentrations: (a)5% SPI, 0.025% PEO; (b) 5% SPI, 0.05% PEO; (c) 8% SPI, 0.025% PEO (inset shows fine strands overlapping fibres); (d) 8% SPI,0.05% PEO (inset shows ribbon-like flat morphology). Scale bar=5mm (insets, 1mm)

Table 1. Characteristics of fibres electrospun from different ratios of SPI and PEO

SPI (%) PEO (%) Characteristics of electrospun fibres

5 0.025 Heterogeneous mixture of fibres and electrosprayed debris,which might represent protein aggregates in the solution

8 0.025 Fibres with some very fine strands and aggregation at some inter-fibre junctions5 0.05 Consistent fibre formation8 0.05 Consistent fibre formation

L. Lin et al.

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

solution with the higher PEO content (Shenoy et al., 2005)leads to decreased protein aggregation, allowing for extru-sion into finer fibres. Across the range of concentrations ofSPI tested, all SPI/PEO blend solutions yielded submicronto micron-sized fibres, with approximately 702� 201nmwidth (heterogeneity 29%) at 5% SPI, 0.05% PEO and1251�306nm width (heterogeneity 24%) at 8% SPI,0.05% PEO (Figure 2). These diameters are significantlylarger than the ~200–300nm obtained by Vega-Lugo andLim (2008), which may be explained by the difference insolvents used. HFP is a more volatile solvent, with muchlower surface tension than NaOH, and this leads to rapidevaporation when the polymer leaves the needle tip,collapsing to form fibres with flat, ribbon-like morphology(Koombhongse et al., 2001).

3.1.2. Zein

In contrast to SPI, smooth, consistent tubular fibres wereelectrospun from zein solutions in glacial acetic acid, withno additives, at every concentration tested from 35% to45% (Figure 3). At concentrations< 35%, the solutionyielded beaded fibres; at 45%, the solution was very

viscous and tended to obstruct the syringe needle. Theability of zein, but not soy, to form beadless electrospunfibres without additives such as PEO may be attributableto the higher concentration of solids in solution (inthe case of zein), or to differences in molecular weightdistribution within the two proteins: soy protein has arather broad molecular weight distribution (which spansseveral decades), while zein has only two narrow peaks(Lelkes et al., 2008).

As shown in Figure 4, zein fibre diameters increaseapproximately linearly with increasing concentrationsacross the range we tested (35–45%). Resulting fibresare considerably smaller than soy fibres and close to the‘nanoscale’ range, from 153�18nm (heterogeneity 12%)at 35% to 344� 27nm (heterogeneity 8%) at 45%. Worthnoting is that the heterogeneity of zein fibres is muchsmaller than that of soy fibres. Again, this is likely due tothe uniformity of the solution with only one polymer, andthe lower polydispersity of zein (Lelkes et al., 2008). Aszein is a low molecular weight protein of 14–24 kDa(Shukla and Cheryan, 2001), a higher minimum concen-tration of solids in solution is required to reach sufficientviscosity to electrospin.

3.2. Mechanical properties

Soy and zein scaffolds were tensile-tested to break. Nosignificant difference was observed between mechanicalproperties of dry soy scaffolds with variation in the proteinconcentrations investigated (Figure 5). Hydrated sampleshad ultimate tensile strengths roughly one order of magni-tude lower than dry samples, with a strain at break oneorder of magnitude higher than dry samples. We notethat the strain at break values for hydrated soy scaffolds(~83%) are in the same order of magnitude as thosereported for human skin, which is approximately 100%(Diridollou et al., 2000). The ultimate tensile strength(UTS) of hydrated soy scaffolds (~0.1MPa) is about oneorder of magnitude less than that reported for skin, whichis approximately 7.7MPa (Adekogbe and Ghanem, 2005).

Using a parallel plate rheometer, we determined theviscoelastic properties of the corn and soy solutions aswell as the viscoelasticity of scaffolds made from 7%SPI/0.05% PEO. As expected, the corn solutions showed

Figure 2. Variation of fibre diameter with relative SPI and PEOconcentrations. While increasing the concentration of SPIincreased fibre diameter, increasing the concentration of PEO from0.025% to 0.05% resulted in decreased diameters for 6%, 7% and8% SPI. *Statistical significance between samples with 0.025%and 0.05% PEO (p<0.05)

Figure 3. Electrospun fibres from: (a) 35% zein; (b) 40% zein; (c) 45% zein. Scale bar=5mm; insets highlight consistent tubularmorphology; scale bar=1mm

Electrospun soy scaffold for regenerative engineering

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

increasing viscosity as the concentration increased, withvalues of 0.54Pa s, 0.80Pa s, and 0.98Pa s at 35%, 40%,and 45% zein, respectively. However, we could not properlyassess the viscoelastic properties of the soy solutions. Theorganic solvent used to electrospin soy, HFP, evaporatesquickly when exposed to air. This made it impossible tocollect reliable rheological data using a standard rheometer.

The soy scaffolds exhibited almost complete elasticbehaviour, both when dry and wet, with a small phase

lag (6 and 9.5 degrees, respectively, at 1Hz). However,the complex modulus, G, of the dry scaffold (74.7 kPa)was approximately 50-fold greater than that of thehydrated scaffolds (1.5 kPa), indicating that whenhydrated, the scaffolds remain elastic, but become muchsofter and pliable.

The UTS of zein and soy scaffolds were similar to eachother in both dry and hydrated conditions (Figure 5). TheUTS of hydrated scaffolds is roughly one order of magnitudelower than that of dry scaffolds, while the strain to break ofwet scaffolds is roughly an order of magnitude higher thanthat of dry scaffolds, for both soy and zein. The strain tobreak of hydrated zein scaffolds (317–443%) was morethan four-fold higher than that of hydrated soy scaffolds(73–85%), although the UTS of hydrated soy and zeinscaffolds were similar (0.06–0.14MPa), with the exceptionof 35% zein (0.26MPa). This difference may reflect thenumber of internal defects in each of the fibres, as theuniformity of the microstructure decreases from skin to core(Lv et al., 2009). As discussed above in Section 3.1, themolecular weight distribution of corn zein is much lessdisperse than that of soy protein; therefore, corn zeinfibres are expected to contain fewer internal defects thansoy fibres.

The change in mechanical properties of zein scaffoldswith hydrationmay also be explained by a partial ‘refolding’of the zein induced by contact with water. Water moleculesmay break the intermolecular hydrogen bonds, stabilizingthe antiparallel b-sheet component of zein while increas-ing the a-helix component (Mizutani et al., 2003). Hence,the zein fibres may have rearranged into a more coiledstructure, which would allow for a greater extent ofstretching under tensile forces.

Based on their relative ease of electrospinning andreasonably maximizing the amount of protein spun persolution volume, scaffolds composed of 7% SPI/0.05%PEO and 40% zein, respectively, were chosen for furtherexperiments.

3.3. Stability of scaffolds in aqueous media

3.3.1. Variation of fibre diameters with hydration

Fibres spun from 7% SPI/0.05% PEO and 40% zein respec-tively were hydrated in 1� PBS. Figure 6 shows themorphology of soy (Figure 6a) and zein (Figure 6b) fibresafter hydration for 2 h in PBS. Both fibre types remainintact. As seen in Table 2, soy fibres swelled to almost dou-ble their dry diameters after 2 h of hydration. By contrast,zein fibres swelled to a lesser degree, possibly due tothe original higher density of material in the fibres, de-creasing their swelling capacity. Several explanations arepossible for the difference between the relative amountsof swelling of soy and zein fibres. Besides soy beingintrinsically more hydrophilic than zein (Chen et al.,2010a), soy scaffolds contain another hydrophilic compo-nent, PEO, which would facilitate hydration, as opposedto hydrophobic zein alone.

Figure 4. Variation of fibre diameter with zein concentration.Data are presented as mean� standard deviation (SD; n=100for each data point). Differences in the diameters are statisticallysignificant between each concentration represented (p<0.05)

Figure 5. (a)Ultimate tensile strength (UTS). (b) Strain to break ofvarious formulations of soy and zein scaffolds in dry and hydratedconditions. Hydrated scaffolds have a UTS of roughly one order ofmagnitude lower than dry scaffolds and a strain to break of roughlyan order of magnitude higher than dry scaffolds

L. Lin et al.

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

3.3.2. Variation of mechanical properties overtime with hydration

In order to assess the effect of prolonged hydrolysis (biode-gradability) on the mechanical properties, 7% SPI/0.05%PEO scaffolds were hydrated for 3 h (0days), 7 days and14days in DMEM+10% FBS, PBS and H2O to investigatethe effects of hydrolysis with and without osmoticpressure stabilization (salts in PBS) and possible proteolyticdegradation due to proteases present in serum. In DMEM+10% FBS, both UTS and strain at break were statisticallysignificantly lower (p< 0.05) at 7 and 14days comparedto 3h (Figure 7). In PBS, UTS was statistically significantlyhigher at 7 and 14days compared to 3h. In water, UTS wasextremely statistically significantly higher (p< 0.01) at 7and 14days compared to 3h (Figure 7a). Strain at breakwas statistically significantly lower at 7 days compared to3h in PBS and extremely statistically significantly higherat 7 and 14days compared to 3h in water (Figure 7b).These differences may be explained by PBS and DMEMboth being osmotically balanced, while water not onlycauses osmotic pressure for the proteins in the scaffoldbut is also a poor solvent for SPI, which may cause aggrega-tion of the hydrophobic units over time (Kuipers andGruppen, 2008). This would result in both higher strengthand ductility of the scaffolds as the aggregates arebeing stretched.

3.3.3. In vitro degradation of soy scaffolds

The soy scaffolds lost ~10% of their original mass in the50:50 mix of DMEM and equine serum (ES), ~25% in

DMEM with or without FBS, and ~40% in distilleddeionizedwater within the first day of hydration (Figure 8).After that they remained stable for 42days. Scaffolds mayhave degraded more in water due to hydrolysis coupledwith osmotic imbalance, whereas the effect of hydrolysisin DMEM was reduced by the osmotic pressure balancingpresence of salts. The presence of serum did not increasescaffold degradation in the case of both 10% and 50%serum in the buffer. This effect might be surprising, sinceexposure to purified proteolytic enzymes, such as trypsinand collagenase, resulted in the effective degradation ofsoy protein (data not shown). However, the inhibitory effectof serum proteins on serine proteases present in serum isa known phenomenon in the in vitro digestion of ECMproteins in mammalian cells (e.g. Wang and Zheng, 2009).

3.4. In vitro cytocompatibility

3.4.1. Soy and zein scaffolds support fibroblastadhesion, spreading and proliferation

At confluence, fluorescent staining for F-actin revealedsimilar cytotypic elongated ‘fibroblast-like’ morphologieson all scaffolds (Figure 9). No differences were seen inthe morphology of cells cultured on scaffolds electrospunfrom different concentrations of soy (Figure 9a, b) andzein (Figure 9c), or between the two ‘green’ materials.Inspection of the cultures by SEM showed that the cellsfully spread over the scaffolds, expressed lamellipodiaand secreted their native ECM (Figure 10).

3.4.2. Soy and zein scaffolds support cellulargrowth and proliferation

Cells on all substrates followed a similar trend of increasingAlamarBlue fluorescence, indicative of proliferation overthe 8days in culture. We noted that the cells on the electro-spun scaffolds demonstrated a trend towards highernormalized fluorescence values than the ‘gold standard’,TCPS (Figure 11). This might indicate that tensile/elasticfibre properties in concert with the 3D structure and/orporosity of the scaffolds may favour the formation ofmultilayered cell assemblies. We recently reported similar

Figure 6. (a) 7% SPI/0.05% PEO. (b) 40% zein fibres after hydration for 2h in PBS; scale bar= (a) 10mm; (b) 5mm. Both fibre typesremain intact with slight coiling compared to dry fibres after hydration

Table 2. Variation of fibre diameters with hydration for 2h inPBS

Fibre diameter (nm)

As spun After hydration

Soy 852�373 1473�328Zein 236�21 297�26

Soy fibres swelled to a greater degree than zein fibres. Bothincreases in fibre diameter were statistically significant (p<0.05).

Electrospun soy scaffold for regenerative engineering

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

observations for electrospun scaffolds made of a tertiaryblend of natural and synthetic polymers (Han et al., 2011).

We conclude that ‘green’ proteins are at least as wellsuited as the more established mammalian substrates toserve as scaffolding biomaterials for human dermalfibroblast and other cell culture. Taken together, our datasupport the further development of these materials fortissue-engineering applications, such as wound healing.

3.5. Differential gene expression on SPI andcollagen type I substrates

To address the question of how mammalian cells adhereto and interact with ‘green’ plant protein substrates, weseeded fibroblasts onto cover slips that had been coatedwith either SPI or type I collagen solutions (see Materialsand methods). The materials were studied as coatings oncover slips, instead of as electrospun fibres, to eliminatethe effects that topology and the different fibre shapesand diameters might have on the initial cell–substrateinteractions. We evaluated 2 and 24 h after seeding earlyand sustained differential expression of mRNA speciesfor a number of integrins, ECM proteins and matrix

Figure 8. Variation of percentage mass loss of soy scaffolds overtime in distilled deionized water, DMEM, DMEM+10% FBS and50:50 DMEM:ES. No statistically significant differences were ob-served between 42days and 1day for any of the conditions. Massloss was significantly (*p<0.05) lower in water than 50:50DMEM:ES at day 1 and significantly lower in water than DMEM+10% FBS and water at day 42

Figure 7. Variation of (a) ultimate tensile strength and (b) strain at break of 7%SPI/0.05% PEO scaffolds with hydration in DMEM+10%FBS, PBS andwater for 0, 7 and 14days. *Statistical significance (p<0.05); **extreme statistical significance (p<0.01) compared to day 0

L. Lin et al.

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

metalloproteinases, using a targeted PCR array (SABiosciences). Overall, we found minimal qualitativedifferences between the various substrates, i.e. in no casedid we find any gene that was upregulated on SPI to bedownregulated on collagen, or vice versa; the differenceswere rather quantitative in nature. As expected, there wassignificant upregulation of a number of ECM-related genes(Table 3) on SPI and collagen type I after 24h as comparedto the same substrates at 2 h. For both these substrates,upregulation of matrix metalloproteinase 10 (MMP-10)was the highest, followed by MMP-1, integrin a2, collagentype VII-a1 and laminin-b3.

To validate the PCR array data we also performed quan-titative real-time PCR for select genes and compared theirexpression in fibroblasts seeded at similar densities on thesame substrates (SPI, collagen) for 2 and 24h. As seen inTable 3, qPCR analysis yielded qualitatively similar patterns

as the gene arrays, albeit somewhat lower absolute valuesfor the expression and upregulation of ECM-related genesafter 24h, with, respectively, MMP-10 again showingthe highest levels (~500–1000-fold), collagen VII-a1 andlaminin-b3 showing the lowest levels (~5–8-fold) and littledifference between cells plated on the various substrates.

The combined upregulation of MMP-1, MMP-10 andMMP-7 has previously been associated with intestinalwound healing, as expressed by migrating enterocytes inischaemic colitis-associated ulcers (Salmela et al., 2004).Dermal fibroblasts also upregulate MMP-10 as well asMMP-1, -3, -8 and MT5-MMP (membrane-type MMP-24)in response to stratifin, a protein molecule secreted bydifferentiated keratinocytes, that may thus play a role inpreventing hyperproliferative disorders associated withoverproduction of dermal ECM in wound healing (Medinaet al., 2007). In the absence of stimuli from other celltypes, cell–cell contact between human dermal fibroblastsin vitro activates an inflammatory response that leads tonon-apoptotic cell death, correlating with a high upregu-lation of MMP-10 (106-fold, as compared to non-contactactivated fibroblasts; Sirén et al., 2006). In our case, asimilar mechanism may have been responsible for thehigh upregulation of MMP-10 on different kinds ofsubstrates, since this was also observed when cells werecultured on PLGA and uncoated glass for 2 h and 24 h(data not shown).

MMP-1 is commonly associated with skin wound healingas well as cancer, and is often expressed by normal fibro-blasts in vitro (Pardo and Selman, 2005). In vitro, cultureof dermal fibroblasts with medium conditioned by bonemarrow-derived fibrocytes ‘reprogrammed’ by the depriva-tion of transforming growth factor-b (TGFb) significantlyincreased MMP-1 expression and reduced type I collagenexpression, showing an anti-fibrotic response that may

Figure 9. Human dermal fibroblasts grew to confluence on day 8 of culture on various scaffolds (nuclei, blue; F-actin cytoskeleton,red): (a) 5% SPI, 0.05% PEO; (b) 8% SPI, 0.05% PEO; (c) 40% zein; (d) gelatin; (e) PLGA; (f) glass. Original magnifications, �200

Figure 10. Scanning electron micrograph of primary humandermal fibroblasts cultured on electrospun soy scaffolds on day 8.Fibres are intact and fibroblasts have secreted native extracellularmatrix (arrows) on the scaffold. Scale bar=5mm

Electrospun soy scaffold for regenerative engineering

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

increase the tissue-remodelling capacity of these fibroblasts(Medina and Ghahary, 2010).

From our data it appears that the response of fibroblastsdoes not differ between distinct substrates upon initialadhesion at the early time point at 2 h, when interactionsbetween cells and substrates are largely dependent onelectrostatic parameters (van der Valk et al., 1983), or afterequilibrating with the environment at the later time point of24h. The high upregulation of MMPs suggests activebreakdown of substrate protein, followed by the expressionof native ECM components collagen type VII and lamininb3, as well as natural receptors for ECM, especially thecollagen receptor integrin a2. Integrin a2b1, a predominantcollagen receptor, is also more highly expressed in ageddermis compared to young dermis, and its activation maybe the cause of the increased expression of MMP-1 (Fisheret al., 2009). Integrin-a subunits are more specialized than-b subunits, as one -b subunit can combine with one orseveral different -a subunits (Noszczyk et al., 2002). Itshould be noted that integrin a2b1 is often associated withMMP-1 expression, and has a contrasting role to integrina1b1 in the activation of human hepatic stellate cells from

quiescence to myofibroblast-like cells (Znoyko et al.,2006). Integrin a2b1 upregulates, while integrin a1b1downregulates, MMP-1 mRNA expression and MMP-1promoter activity, and the stimulation of a2b1 or theinhibition of a1 both promoted the development of well-developed multicellular networks (Znoyko et al., 2006).In the context of early dermal wound healing, fibroblaststransiently express mainly a2 and av subunits associatedwith upregulation of the b1 subunit (Noszczyk et al., 2002).

The moderate upregulation of collagen type VIIsuggests the constructive side of the remodelling process.Collagen type VII is a necessary component of the anchor-ing fibrils at the dermal–epidermal junction; in disorderswhere it is absent, such as dystrophic epidermolysisbullosa (DEB), it has been shown in mouse models to berestored by the delivery of allogenic fibroblasts (Woodleyet al., 2007; Wong et al., 2008). In human subjects withrecessive DEB, allogeneic fibroblasts injected intrader-mally could promote sustained increase in collagen VIIat the dermal–epidermal junction, although apparentlydue to increased expression in the patients’ own COL7A1gene (Wong et al., 2008).

Table 3. Time-dependent upregulation of select genes in primary human dermal fibroblasts cultured on SPI and collagen type Isubstrates

Gene symbol Description

PCR arrays PCR validation

Fold regulation (p) Fold regulation (p)

SPI Col I SPI Col I

MMP10 Matrix metallopeptidase 10(stromelysin 2)

1152.7255 (0.0232) 471.4083 (0.0525) 554.4832 (0.0131) 996.1905 (0.011)

MMP1 Matrix metallopeptidase 1(interstitial collagenase)

53.7242 (0.0068) 36.7796 (0.039) 236.6591 (0.0078) 196.2660 (0.0533)

ITGA2 Integrin, a2 (CD49B, a2subunit of VLA-2 receptor)

16.8052 (0.001) 12.5606 (0.0245) 30.0821 (0.0046) 81.1177 (0.0071)

COL7A1 Collagen, type VII, a1 6.6231 (0.0008) 6.0106 (0.0085) 7.0902 (0.0222) 8.2344 (0.0897)LAMB3 Laminin, b3 6.4718 (0.0002) 7.0494 (0.0078) 6.5375 (0.1086) 5.6016 (0.055)

Data are presented as the ratio of fold regulation at 24h vs 2h of culture on the two substrates. Although in some cases the absolute valuesbetween the PCR arrays and RT–PCR data differ, the trends are quite similar when comparing the expression of genes for each substrate.

Figure 11. Increase in AlamarBlue™ fluorescence with time normalized to fluorescence readings on day 0 (n=3). Statistically significant(*p<0.05) increase compared to TCPS for each time point

L. Lin et al.

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

Taken together, the gene array and the qPCR data showgood quantitative and qualitative agreement, suggestingthat fibroblasts seeded onto SPI will deposit and remodeltheir own ECM substrates in a manner similar to otherestablished substrates, such as collagen and PLGA. Thesedata from our in vitro studies suggest that SPI is a usefulbiomaterial that will induce a cellular tissue remodellingresponse similar to that of native ECM proteins.

4. Conclusions

In this proof-of-concept study, we demonstrated theusefulness of a novel material category for tissue-engineering scaffolding, namely electrospun alimentaryplant proteins. We have shown that SPI, with the additionof trace amounts of PEO, can be electrospun into astable, hydrolysis-resistant fibrous scaffold that supportscellular attachment, growth and proliferation. Zein canbe electrospun into nanofibres that remain stable in anaqueous environment without the need for syntheticadditives or crosslinking. Both these materials yieldcytocompatible scaffolds, which support the adhesionand proliferation of human dermal fibroblasts. To addressthe interesting mechanistic questions of how mammaliancells adhere to and interact with plant proteins, weidentified select genes that are similarly upregulated inthese fibroblasts when cultured on SPI or collagen I

substrates, and observed a comparable initial remodellingresponse on both substrates.

Based on the interactions between our electrospun plant-protein derived scaffolds and human dermal fibroblasts, aswell as the thin structure and skin-like pliability of thesescaffolds, we conclude that electrospun ‘green’ proteinscaffolds provide a potential platform for skin regeneration.These scaffolds may be suitable to be applied directly on awound as a dressing, or used as a base for an in vitro skinequivalent. Further studies on the direct effect of soyscaffolds or their bioactive degradation products on woundhealing in vivo will help to clarify the benefits of this novelclass of biomaterials.

Acknowledgements

This study was supported by a translational grant from theWallaceH. Coulter Foundation (PIL), a National Science FoundationIntegrative Graduate Education and Research Traineeship (NSF-IGERT, Grant nos DGE-0221664 and DGE-0654313, to L.L.), aSeed Grant from the Nanotechnology Institute of South-EasternPennsylvania (to P.I.L. and D.W.) and a Collaborative Grant fromthe Louis and Bessie Stein Family Foundation (to P.I.L. and P.L.).We thank Professor Karen Winey, University of Pennsylvania, foruse of the Instron and Dr P. Uttayarat, Dr M. Moniruzzaman andDr A. Kota for technical assistance. We thank Professor JaneClifford for use of the Nanodrop and image station, Professor MarkStearns for use of the thermal cycler and Professor Gregg Johannesfor use of the Stratagene MX3000P.

References

Adekogbe I, Ghanem A. 2005; Fabricationand characterization of DTBP-crosslinkedchitosan scaffolds for skin tissue engineer-ing. Biomaterials 26: 7241–7250.

Armstrong WB, Wan XS, Kennedy AR, et al.2003; Development of the Bowman–Birkinhibitor for oral cancer chemopreventionand analysis of Neu immunohistochemicalstaining intensity with Bowman–Birk inhibi-tor concentrate treatment. Laryngoscope113: 1687–1702.

Barnes CP, Smith MJ, Bowlin GL, et al. 2006;Feasibility of electrospinning the globularproteins hemogloblin and myoglobin.J Eng Fibers Fabrics 1: 16–28.

Bhardwaj N, Kundu SC. 2010; Electrospin-ning: a fascinating fiber fabrication tech-nique. Biotechnol Adv 28: 325–347.

Chen L, Hébrard G, Beyssac E, et al. 2010a; Invitro study of the release properties of soy–zein protein microspheres with a dynamicartificial digestive system. J Agric FoodChem 58: 9861–9867.

Chen H, Huang J, Yu J, et al. 2010b; Electro-spun chitosan–graft–poly(e-caprolactone)/poly(e-caprolactone) cationic nanofibrousmats as potential scaffolds for skintissue engineering. Int J Biol Macromol48: 13–19.

Curt S, Subirade M, Rouabhia M. 2009;Production and in vitro evaluation of soyprotein-based biofilms as a support forhuman keratinocyte and fibroblast culture.Tissue Eng Part A 15: 1223–1232.

de Lumen BO. 2005; Lunasin: a cancer-preventive soy peptide. Nutr Rev 63: 16–21.

Diridollou S, Patat F, Gens F, et al. 2000; Invivo model of the mechanical propertiesof the human skin under suction. Skin ResTech 6: 214–221.

Enoch S, Shaaban H, Dunn KW. 2005;Informed consent should be obtained frompatients to use products (skin substitutes)and dressings containing biological mate-rial. J Med Ethics 31: 2–6.

Fisher GJ, QuanT, Purohit T, et al. 2009; Colla-gen fragmentation promotes oxidative stressand elevates matrix metalloproteinase-1 infibroblasts in aged human skin. Am J Pathol174: 101–114.

Fonder MA, Lazarus GS, Cowan DA, et al.2008; Treating the chronic wound: apractical approach to the care of nonhealingwounds and wound care dressings. J AmAcad Dermatol 58: 185–206.

Fung WY, Yuen KH, Liong MT. 2010; Charac-terization of fibrous residues from agro-wastes and the production of nanofibers.J Agric Food Chem 58: 8077–8084.

Ghanaati S, Fuchs S, Webber MJ, et al. 2011;Rapid vascularization of starch–poly(capro-lactone) in vivo by outgrowth endothelialcells in co-culture with primary osteoblasts.J Tissue Eng Regen Med 5: e136–e143.

Gong S, Wang H, Sun Q, et al. 2006;Mechanical properties and in vitro biocom-patibility of porous zein scaffolds. Bioma-terials 27: 3793–3799.

Guo Y, Li M, Mylonakis A, et al. 2007;Electroactive oligoaniline-containing self-assembled monolayers for tissue engi-neering applications. Biomacromolecules8: 3025–3034.

Han J, Lazarovici P, Pomerantz C, et al. 2011;Co-electrospun blends of PLGA, gelatin,and elastin as potential nonthrombogenicscaffolds for vascular tissue engineering.Biomacromolecules 12: 399–408; DOI:10.1021/bm101149r.

Hua Y, Yuang Y, Qiu A, et al. 2005; Propertiesof soy protein isolate prepared from aque-ous alcohol washed soy flakes. Food ResInt 38: 273–279.

Figallo E, Flaibani M, Zavan B, et al. 2007;Micropatterned biopolymer 3D scaffoldfor static and dynamic culture of humanfibroblasts. Biotechnol Prog 23: 210–216.

Jeong L, Yeo IS, Kim HN, et al. 2009; Plasma-treated silk fibroin nanofibers for skinregeneration. Int J Biol Macromol 44:222–228.

Jiang Y, Tang C-H, Wen Q-B, et al. 2007a;Effect of processing parameters on theproperties of transglutaminase-treatedsoy protein isolate films. Innovat Food SciEmerg Tech 8: 218–225.

Jiang H, Zhao P, Zhu K. 2007b; Fabricationand characterization of zein-based nanofi-brous scaffolds by an electrospinningmethod. Macromol Biosci 7: 517–525.

Koombhongse S, Liu W, Reneker DH. 2001;Flat polymer ribbons and other shapes by

Electrospun soy scaffold for regenerative engineering

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

electrospinning. J Polymer Sci B PolymerPhys 39: 2598–2606.

Kuipers BJ, Gruppen H. 2008; Identificationof strong aggregating regions in soyglycinin upon enzymatic hydrolysis. J AgricFood Chem 56: 3818–3827.

Lelkes PI, Li M, Perets A, et al. 2008; Electro-spinning of natural proteins for tissueengineering scaffolding. In Handbook ofNatural-based Polymers for BiomedicalApplications, Reis RL (ed.). Woodhead:Cambridge, UK; 446–482.

Li M, Mondrinos MJ, Gandhi MR, et al. 2005;Electrospun protein fibers as matricesfor tissue engineering. Biomaterials 26:5999–6008.

Li M, Guo Y, Wei Y, et al. 2006; Electrospin-ning polyaniline-contained gelatin nanofi-bers for tissue engineering applications.Biomaterials 27: 2705–2715.

Luo LH, Zhang YF, Wang XM, et al. 2010;Preparation, characterization, and in vitroand in vivo evaluation of cellulose/soyprotein isolate composite sponges. J Bio-mater Appl 24: 503–526.

Lv M-Y, Ge H-Y, Chen J. 2009; Study on thechemical structure and skin-core structureof polyacrylonitrile-based fibers duringstabilization. J Polym Res 16: 513–517.

Mallikarjun Gouda KG, Gowda LR, RaoAG, et al. 2006; Angiotensin I-convertingenzyme inhibitory peptide derived fromglycinin, the 11 S globulin of soybean(Glycine max). J Agric Food Chem 54:4568–4573.

Maltais A, Remondetto GE, Subirade M. 2007;Mechanisms involved in the formation andstructure of soya protein cold-set gels: amolecular and supramolecular investiga-tion. Food Hydrocolloids 22: 550–559.

Maruyama N, Maruyama Y, Tsuruki T, et al.2003; Creation of soybean b-conglycinin-b with strong phagocytosis-stimulating ac-tivity. Biochim Biophys Acta 1648: 99–104.

Mauri AN, Añon MC. 2006; Effect of solutionpH on solubility and some structural prop-erties of soybean protein isolate films. J SciFood Agric 86: 1064–1072.

Medina A, Ghaffari A, Kilani RT, et al.2007; The role of stratifin in fibroblast–keratinocyte interaction. Mol Cell Biochem305: 255–264.

Medina A, Ghahary A. 2010; Fibrocytes canbe reprogrammed to promote tissue remo-deling capacity of dermal fibroblasts. MolCell Biochem 344: 11–21.

Metcalfe AD, Ferguson MW. 2007; Tissueengineering of replacement skin: the cross-roads of biomaterials, wound healing,embryonic development, stem cells andregeneration. J R Soc Interface 4: 413–437.

Min B-M, Lee G, Kim SH, et al. 2004; Elec-trospinning of silk fibroin nanofibers andits effects on the adhesion and spreadingof normal human keratinocytes andfibroblasts in vitro. Biomaterials 25:1289–1297.

Miyoshi S, Ishikawa H, Kaneko T, et al. 1991;Structures and activity of angiotensin-converting enzyme inhibitors in an a-zeinhydrolysate. Agric Biol Chem 55: 1313–1318.

Mizutani Y, Matsumura Y, Imamura K, et al.2003; Effects of water activity and lipid ad-dition on secondary structure of zein inpowder systems. J Agric Food Chem 51:229–235.

Morihara K, Takai S, Takenaka H, et al. 2006;Cutaneous tissue angiotensin-converting

enzyme may participate in pathologic scarformation in human skin. J Am AcadDermatol 54: 251–257.

Nielsen NC. 1985; Structure of soy proteins.In New Protein Foods, vol 5, Altschul AA,Wilcke HL (eds). Academic Press: NewYork; 27–64.

Nikolaychik VV, Samet MM, Lelkes PI. 1996; Anewmethod for continual quantitation of vi-able cells on endothelialized polyurethanes.J Biomater Sci Polym Ed 7: 881–891.

Noszczyk BH, Klein E, Holtkoetter O, et al.2002; Integrin expression in the dermisduring scar formation in humans. ExpDermatol 11: 311–318.

Pardo A, Selman M. 2005; MMP-1: the elderof the family. Int J Biochem Cell Biol 37:283–288.

Powell HM, Boyce ST. 2007; Fiber density ofelectrospun gelatin scaffolds regulatesmorphogenesis of dermal–epidermal skinsubstitutes. J Biomed Mater Res A 84:1078–1086.

Powell HM, Supp DM, Boyce ST. 2008; Influ-ence of electrospun collagen on woundcontraction of engineered skin substitutes.Biomaterials 29: 834–843.

Reddy N, Yang Y. 2009; Soyprotein fiberswith high strength and water stability forpotential medical applications. BiotechnolProg 25: 1796–1802.

Rnjak J, Li Z, Maitz PK, et al. 2009; Primaryhuman dermal fibroblast interactions withopen weave three-dimensional scaffoldsprepared from synthetic human elastin.Biomaterials 30: 6469–6477.

Rnjak J, Wise SG, Mithieux SM, et al. 2011;Severe burn injuries and the role of elastinin the design of dermal substitutes. TissueEng Part B Rev 17: 81–91.

Rnjak-Kovacina J, Wise SG, Li Z, et al. 2011;Tailoring the porosity and pore size of elec-trospun synthetic human elastin scaffoldsfor dermal tissue engineering. Biomaterials32: 6729–6736.

Saeed AI, Bhagabati NK, Braisted JC, et al.2006; TM4 microarray software suite.Methods Enzymol 411: 134–193.

SalmelaMT, Pender SL, Karjalainen-LindsbergML, et al. 2004; Collagenase-1 (MMP-1),matrilysin-1 (MMP-7), and stromelysin-2(MMP-10) are expressed by migratingenterocytes during intestinal woundhealing. Scand J Gastroenterol 39:1095–1104.

Santos MI, Fuchs S, Gomes ME, et al. 2007;Response of micro- and macrovascular en-dothelial cells to starch-based fiber meshesfor bone tissue engineering. Biomaterials28: 240–248.

Santos MI, Tuzlakoglu K, Fuchs S, et al. 2008;Endothelial cell colonization and angio-genic potential of combined nano- andmicro-fibrous scaffolds for bone tissueengineering. Biomaterials 29: 4306–4313.

Scuderi N, Anniboletti T, Carlesimo B, et al.2009; Clinical application of autologousthree-cellular cultured skin substitutesbased on esterified hyaluronic acid scaffold:our experience. In Vivo 23: 991–1003.

Selling GW, Biswas A, Patel A, et al. 2007;Impact of solvent on electrospinning ofzein and analysis of resulting fibers.Macromol Chem Phys 208: 1002–1010.

Shenoy SL, Bates WD, Frisch HL, et al. 2005;Role of chain entanglements on fiberformation during electrospinning of poly-mer solutions: good solvent, non-specific

polymer–polymer interaction limit. Poly-mer 46: 3372–3384.

Shukla R, Cheryan M. 2001; Zein: the indus-trial protein from corn. Ind Crop Prod13: 171–192.

Silva GA, Vaz CM, Coutinho OP, et al. 2003;In vitro degradation and cytocompatibilityevaluation of novel soy and sodium casein-ate-based membrane biomaterials. J MaterSci Mater Med 14: 1055–1066.

Silva SS, Santos MI, Coutinho OP, et al. 2005;Physical properties and biocompatibility ofchitosan/soy blended membranes. J MaterSci Mater Med 16: 575–579.

Sirén V, Salmenperä P, Kankuri E, et al. 2006;Cell–cell contact activation of fibroblastsincreases the expression of matrix metallo-proteinases. Ann Med 38: 212–220.

Soliman S, Sant S, Nichol JW, et al. 2011;Controlling the porosity of fibrous scaf-folds by modulating the fiber diameterand packing density. J Biomed Mater ResA 96: 566–574.

Sun QS, Dong J, Lin ZX, et al. 2005;Comparison of cytocompatibility of zein filmwith other biomaterials and its degradabil-ity in vitro. Biopolymers 78: 268–274.

Tian K, Shao Z, Chen X. 2010; Natural elec-troactive hydrogel from soy protein isola-tion. Biomacromolecules 11: 3638–3643.

Tu J, Wang H, Li H, et al. 2009; The in vivobone formation by mesenchymal stemcells in zein scaffolds. Biomaterials 30:4369–4376.

Tuzlakoglu K, Bolgen N, Salgado AJ, et al.2005; Nano- and microfiber combinedscaffolds: a new architecture for bonetissue engineering. J Mater Sci MaterMed, 16: 1099–1104.

van der Valk P, van Pelt AW, Busscher HJ,et al. 1983; Interaction of fibroblasts andpolymer surfaces: relationship betweensurface free energy and fibroblast spread-ing. J Biomed Mater Res 17: 807–817.

Vaz CM, De Graaf LA, Reis RL, et al. 2003a;Effect of crosslinking, thermal treatmentand UV irradiation on the mechanicalproperties and in vitro degradation behav-ior of several natural proteins aimed to beused in the biomedical field. J Mater SciMater Med 14: 789–796.

Vaz CM, De Graaf LA, Reis RL, et al. 2003b;In vitro degradation behaviour of biode-gradable soy plastics: effects of crosslink-ing with glyoxal and thermal treatment.Polymer Degrad Stabil 81: 65–74.

Vaz CM, van Doeveren PF, Reis RL, et al.2003c; Soy matrix drug delivery systemsobtained by melt-processing techniques.Biomacromolecules 4: 1520–1529.

Vaz CM, van Doeveren PF, Dias GR, et al. 2004;Controlled delivery achieved with bilayermatrix devices produced by co-injectionmoulding. Macromol Biosci 4: 795–801.

Vega-Lugo A, Lim L. 2008; Electrospinning ofsoy protein isolate nanofibers. J BiobasedMater Bioenerg, 2: 223–230.

Wang HJ, Gong SJ, Lin ZX, et al. 2007; In vivobiocompatibility and mechanical propertiesof porous zein scaffolds. Biomaterials 28:3952–3964.

Wang Q, Zheng YP. 2009; Ultrasound biomi-croscopy imaging for monitoring progres-sive trypsin digestion and inhibition inarticular cartilage. Ultrasound Med Biol35: 1535–1545.

Woerdeman DL, Ye P, Shenoy S, et al. 2005;Electrospun fibers from wheat protein:

L. Lin et al.

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term

investigation of the interplay betweenmolecular structure and the fluid dynam-ics of the electrospinning process. Bioma-cromolecules 6: 707–712.

Wong T, Gammon L, Liu L, et al. 2008;Potential of fibroblast cell therapy forrecessive dystrophic epidermolysis bul-losa. J Invest Dermatol 128: 2179–2189.

Woodley DT, Remington J, Huang Y, et al.2007; Intravenously injected human

fibroblasts home to skin wounds, delivertype VII collagen, and promote woundhealing. Mol Ther 15: 628–635.

Xu YZ, Wu JJ, Chen YP, et al. 2010; The useof zein and Shuanghuangbu for periodon-tal tissue engineering. Int J Oral Sci 2:142–148.

Yao C, Li X, Song T. 2007; Electrospinningand crosslinking of zein nanofiber mats.J Appl Polymer Sci 103: 380–385.

Zhang X, Min BG, Kumar S. 2003; Solu-tion spinning and characterization ofpoly(vinyl alcohol)/soybean protein blendfibers. J Appl Polymer Sci 90: 716–721.

Znoyko I, Trojanowska M, Reuben A.2006; Collagen binding a2b1 and a1b1integrins play contrasting roles in regu-lation of Ets-1 expression in humanliver myofibroblasts. Mol Cell Biochem282: 89–99.

Electrospun soy scaffold for regenerative engineering

Copyright © 2012 John Wiley & Sons, Ltd. J Tissue Eng Regen Med (2012)DOI: 10.1002/term