Available online at www.sciencedirect.com

Acta Biomaterialia 5 (2009) 2983–2994

www.elsevier.com/locate/actabiomat

Enhanced angiogenesis of porous collagen scaffolds by incorporationof TMC/DNA complexes encoding vascular endothelial growth factor

Zhengwei Mao a,1, Haifei Shi b,1, Rui Guo a, Lie Ma a,*, Changyou Gao a,*,Chunmao Han b, Jiacong Shen a

a Key Laboratory of Macromolecular Synthesis and Functionalization, Ministry of Education, Department of Polymer Science and Engineering,

Zhejiang University, Hangzhou 310027, Chinab Faculty of Burn, Second Affiliated Hospital of Zhejiang University, Hangzhou 310027, China

Received 4 August 2008; received in revised form 26 March 2009; accepted 3 April 2009Available online 9 April 2009

Abstract

Angiogenesis of an implanted construct is one of the most important issues in tissue engineering and regenerative medicine, and canoften take as long as several weeks. The vascular endothelial growth factor (VEGF) shows a positive effect on enhancing angiogenesisin vivo. But the incorporation of growth factors has many limitations, since they typically have half-lives only on the order of minutes.Therefore, in this work the DNA encoding VEGF was applied to enhance the angiogenesis of a collagen scaffold. A cationic gene deliveryvector, N,N,N-trimethyl chitosan chloride (TMC), was used to form complexes with the plasmid DNA encoding VEGF. The complexeswere then incorporated into the collagen scaffold, the loading being mediated by the feeding concentration and release in a sustainedmanner. In vitro cell culture demonstrated a significant improvement in the VEGF expression level from the TMC/DNA complexes con-taining scaffolds, in particular with a large amount of DNA. The scaffolds containing the TMC/DNA complexes were subcutaneouslyimplanted into Sprague–Dawley mice to study their angiogenesis via macroscopic observation, hematoxylin–eosin staining and immu-nohistochemical staining. The results demonstrated that the incorporation of TMC/DNA complexes could effectively enhance the in vivoVEGF expression and thereby the angiogenesis of implanted scaffolds.� 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Angiogenesis; Scaffold; N,N,N-Trimethyl chitosan chloride; Vascular endothelial growth factor; Gene delivery

1. Introduction

One of the most important issues for engineered tissuesis the vascularization of the constructs [1,2], since a tissuethat is more than a few millimeters in size generally cannotsurvive by only the diffusion of nutrients and metabolicproducts. Therefore, the rapid formation of new blood cap-illaries is essential to supply the necessary nutrients andoxygen to and remove waste products from the cells [3,4].Unfortunately, it usually takes several weeks for a con-

1742-7061/$ - see front matter � 2009 Acta Materialia Inc. Published by Else

doi:10.1016/j.actbio.2009.04.004

* Corresponding authors. Tel./fax: +86 571 87951108.E-mail addresses: liema@zju.edu.cn (L. Ma), cygao@zju.edu.cn

(C. Gao).1 These two authors contributed equally to this work.

struct to become fully vascularized [5]. Hence, accelerationof the angiogenesis rate is urgently required and can resultin better healing of the affected tissues [6].

Several approaches have been developed to enhance thevascularization of the tissue-engineered constructs. Forexample, Gibson et al. [7] adjusted the pore size of the scaf-fold and found an optimum diameter (�100 lm) for cellu-lar adhesion and migration. Pieper et al. [4] proved thatincorporation of glycosaminoglycans can increase angio-genesis degree in vivo. However, a sufficient vasculaturestill takes more than 4 weeks to develop. The use of angio-genetic factors is a popular approach to inducing neovascu-larization in the engineered tissues. An importantstimulating factor in angiogenesis is vascular endothelialgrowth factor (VEGF), which acts on the VEGF receptors.

vier Ltd. All rights reserved.

2984 Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994

VEGF stimulates the cells to produce matrix metallopro-teinases (MMPs), which degrade the basement membraneand surrounding extracellular matrix. As a result, endothe-lial cells proliferate and migrate towards the interstitium,where they start sprouting. Subsequently, the cells prolifer-ate and migrate towards the newly formed sprouts andmature by forming a single cell layer around the sprout[1,8,9]. Other important growth factors known for theangiogenic potential include angiogenin (including angio-genin-1 and angiogenin-2), basic fibroblast growth factor(bFGF), platelet-derived growth factor (PDGF) and trans-forming growth factor (TGF). Our and other works haveshown that, in a rabbit ear chamber model, induced angio-genesis and full vascularization by local administration ofangiogenin or bFGF requires 2–3 weeks [10,11].

Although the growth factors are very effective, theirapplication for tissue regeneration must overcome numer-ous difficulties, because their half-lives are only on theorder of minutes [12–14]. As DNA is much more stablethan the growth factors, the scaffold that releases DNAencoding angiogenesis factors represents a powerful alter-native to the direct delivery of the proteins. When the scaf-folds are implanted into an animal or human body, DNAwill be released to transfect the cells in the surrounding tis-sues or the cells invading the scaffolds. The transfected cellscan subsequently act as bioreactors for the local produc-tion of tissue-inductive factors [15,16]. Shea et al. [17,18]encapsulated plasmid DNA and DNA complex separatelywithin a poly(lactide-co-glycotide) scaffold during the fab-rication process. The localized plasmid DNA released fromthe scaffolds can promote gene expression and is then ableto promote tissue formation, such as angiogenesis andbone formation. Planck et al. [19] loaded liposome/DNAand cationic polymer/DNA complexes separately withina collagen sponge by a lyophilization method. The gene-loaded scaffolds could transfect cells in vitro and in vivo.Zhang et al. [20] incorporated plasmid DNA encodingTGF-b1 into a collagen–gelatin scaffold. The gene-trans-fected chondrocytes could express TGF-b1 stably for3 weeks in vitro. Nonetheless, the use of in vivo transfec-tion of the tissue cells by an implanted scaffold to enhanceangiogenesis has still not been studied thoroughly.

In our previous study, we developed a collagen-basedporous scaffold (with a pore size of 50–100 lm and porosityover 98%) and applied it to skin regeneration. However,the full vascularization of the scaffolds embedded subcuta-neously in a rabbit ear needed 6 weeks. The slow angiogen-esis process delays the transplantation of split-thicknessskin grafts and thereby hampers the healing of damagedskin. To solve this problem, bFGF and angiogenin wereloaded into the scaffold to accelerate the angiogenesis pro-cess. Unfortunately, the high cost and short half-lives ofgrowth factors limit this trial in practice. In order to over-come the drawbacks of growth factors, a relatively stableand cheap gene is considered here as an alternative. Inour laboratory, a non-viral gene vector, N,N,N-trimethylchitosan chloride (TMC), was synthesized [21]. Its

in vitro gene delivery efficiency is rather high, and it haslow cell toxicity. However, the feasibility of using this kindof vector for in vivo applications is still unknown. Herein,one kind of TMC (Mw � 6 kDa, quaternization degree�40%), which has optimal gene transfection efficiencyand low cytotoxicity, is applied to protect the plasmidDNA encoding VEGF. Then the TMC/DNA complexesare incorporated into the collagen scaffolds, whosein vivo angiogenesis ability is studied and compared withDNA-free collagen scaffold in a mouse model.

2. Experimental

2.1. Materials

DNA (fish sperm, sodium salt, used as a model to studythe physicochemical property of TMC/DNA particles) waspurchased from AMRESCO. Plasmid encoding enhancedgreen fluorescence protein (pDNA-EGFP) was a gift fromDr. Jun Li, State Key Laboratory of Diagnosis and Treat-ment for Infectious Diseases, China. Plasmid DNA encod-ing vascular endothelial growth factor (pDNA–VEGF)was donated from Prof. Changyong Wang and Dr. HongJiang, Military Medical Academy of PLA, China. Theplasmids were amplified in Escherichia coli and purifiedby a differential precipitation method [22]. The DNA con-centration and purity were assessed via ultraviolet opticalintensity at 260 and 280 nm, respectively. The plasmidDNA was lyophilized and stored at �20 �C until the trans-fection experiments. All other reagents were of analyticalgrade and were used as received. Chitosan (degree ofdeacetylation 90%, Mw � 6 k) was purchased from Haide-bei Co. Ltd., Qingdao, China. Ethidium bromide was pro-vided by Fluka. Triple-distilled water was used throughoutthe study. Human embryonic kidney cells (HEK293) weremaintained in Dulbecco’s modified Eagle’s medium(DMEM; Gibco) supplemented with penicillin/streptomy-cin and 10% fetal bovine serum (FBS; Sijiqing Co. Ltd.,Hangzhou, China). The cells were incubated at 37 �C in ahumidified atmosphere containing 5% CO2 and used atan appropriate degree of confluence. The antibodies ofVEGF, CD31 and a-SMA, used for immunohistochemicalanalysis, were purchased from Neomarker and Sigma.

2.2. TMC synthesis

TMC was prepared as reported previously [21]. Briefly,5 g chitosan was dispersed in 1-methyl-2-pyrrolidinone con-taining 13 g of sodium iodide at 60 �C under agitation. Next,27.5 ml of NaOH (15% w/v) solution was added to maintainan alkaline environment throughout the reaction. Methyla-tion was produced through nucleophillic substitution bythe addition of 30 ml of methyl iodide and incubation for30 min. The product was precipitated by the addition ofdiethyl ether/ethanol (1:1 v/v). After being centrifuged at6000 RCF, the supernatant was discarded. The productwas dried in nitrogen atmosphere, redissolved in 0.5 M NaCl

Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994 2985

and precipitated again with diethyl ether/ethanol to substi-tute the iodide by chloride. The centrifuged precipitate wasthen thoroughly washed with diethyl ether/ethanol and driedagain. The final pellets were dissolved in water and dialyzedagainst water for 3 days, then freeze-dried.

2.3. Preparation of TMC/DNA complexes

Fish sperm DNA was used as a model for in vitro load-ing and release, and pDNA–VEGF was used for in vitroand in vivo transfection. The theoretical charge ratio (±)(molar ratio of amine group to phosphate group) isdenoted by N/P ratio. The N/P ratio used in this studywas fixed at 10 since the TMC/DNA complexes with thisratio have the highest transfection efficiency in vitro. Ourprevious results show that the transfection efficiency ofTMC/DNA complexes is nine times higher than that ofnaked pDNA, which is also higher than that of PEI butstill lower than that of lipofectamine. At this N/P ratio,the mean diameter of the TMC/DNA complex is about180 nm and the zeta potential of this complex is about+10 mV in DMEM and +3 mV in DMEM containing10% FBS [21]. TMC and DNA were dissolved in phos-phate-buffered saline (PBS, pH 7.2) to form solutions of36 and 6 mg ml�1, respectively, and filtered through a syr-inge filter (with a pore size of 220 nm) for sterilization.Then 1 ml of the TMC solution was added to 1 ml of theDNA solution and vortexed violently for 30 s, before beingincubated for 30 min at 37 �C.

2.4. Fabrication of DNA-loaded collagen scaffold

Collagen was dissolved in 0.5 M acidic acid solution toprepare a 0.5% (w/v) solution. After deaeration under vac-uum to remove any entrapped air bubbles, the collagen solu-tion was put into a homemade mold, frozen at�20 �C for 2 hand then lyophilized for 24 h to obtain a porous collagenscaffold with a diameter of 15 mm, a thickness of 2 mmand a weight of 2 mg. After incubation in 0.05 M acetic acidsolution to regain its swelled state, the collagen scaffold wasfurther treated with 0.1% (w/v) glutaraldehyde (GA) at 4 �Cfor another 24 h. The scaffolds were then rinsed thoroughlyto remove any excess GA, and lyophilized again to obtainthe GA-crosslinked collagen scaffold [23,24].

PBS (0.4 ml) containing different amounts of TMC/DNA complex was dropped onto the dried collagen scaf-fold, which was then kept at 4 �C overnight for incorpora-tion of the TMC/DNA complexes into the scaffold. TheseDNA-loaded scaffolds were carefully washed twice in PBS,then used for further experiments.

Each of the collagen scaffolds containing the TMC/DNAcomplexes was immersed into 0.5 ml of 25 mg ml�1 papain/PBS solution containing 4 mg ml�1 EDTA and digestedovernight at 60 �C. The final volume of the solution wasadjusted to 1 ml. The quantity of DNA immobilized on thecollagen scaffolds was tested by fluorometer (LS55, Perkin-Elmer, UK) with the fluorescent dye Hoechst 33258 [25].

2.5. DNA release profile in vitro

An in vitro release assay was conducted to determine therelease kinetics of DNA from the collagen scaffolds. Eachof the collagen scaffolds containing the TMC/DNA com-plexes was immersed in 1 ml of sterile PBS and gently stir-red at 37 �C. At desired time intervals, 0.2 ml of thesupernatant was collected for analysis and replaced withan equal volume of fresh PBS. The 0.2 ml supernatantwas added into 0.2 ml of 25 mg ml�1 papain/PBS solutioncontaining 4 mg ml�1 EDTA and digested overnight at60 �C. The quantity of the released DNA was tested byfluorometer with the fluorescent dye Hoechst 33258. Alldata were averaged from three parallel measurements andexpressed as mean ± standard deviation (SD).

2.6. In vitro gene transfection

The extent of transgene expression in vitro was deter-mined by a static three-dimensional (3-D) culture methodusing pDNA–VEGF and HEK293 cells. For this, 5 � 105

cells were seeded into the DNA-incorporated scaffold.After incubation for fixed time intervals, the expressionof human VEGF in the supernatant was tested by anELISA kit (DVE00, R&D systems, Minneapolis) followingthe procedures described in the user’s manual. The cellnumber was determined by DNA quantity as describedpreviously [25]. After correlating the DNA quantity withthe cell number by a standard curve, the average VEGFexpression level per 106 cells was calculated. A DNA-freescaffold and a scaffold containing 356 lg of naked plasmidDNA (naked DNA356) were used as controls.

2.7. In vivo transfection and angiogenesis

The in vivo transfectability of the DNA-impregnatedscaffolds and the acceleration of angiogenesis was accessedby a subcutaneous implantation test in Sprague–Dawleymice following the institutional guidelines. Adult mice, aged3 months and weighing 102 ± 5.8 g, were purchased fromthe Laboratory Animal Center of Zhejiang University. Thenight before the experiments took place, the backs of themice were denuded with 8% Na2S aqueous solution. Themice were anesthetized by intraperitoneal injection of pento-barbital sodium (Sigma) at a dosage of 30 mg kg�1, in accor-dance with the instructions. The DNA-containing scaffoldsand DNA-free scaffolds (12 mm in diameter, 2 mm in thick-ness) were implanted into the subcutaneous tissue of theSprague–Dawley mice. Four scaffolds were implanted intoeach mouse, with a distance of 3 cm between adjacent scaf-folds. Three time intervals were set for each group, with fourparallel scaffolds for each time interval. There was thus atotal of 12 samples for each group.

After 1–3 weeks of implantation, samples were harvestedand fixed in 4% formaldehyde/PBS solution at 4 �C, dehy-drated with a graded series of ethanol and embedded in par-affin. The samples were sectioned (5 lm), stained by

2986 Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994

hematoxylin–eosin (H&E) and visualized by light micros-copy. Immunohistochemical staining of VEGF was used tostudy the VEGF expression level in vivo after transfection.Immunohistochemical staining of alpha smooth muscleactin and CD31 was used to characterize the newly formedblood vessels [26]. Different areas were photographed afterH&E and immunohistochemical staining. Quantificationof the histological sections was conducted by capturingimages on a computer and subsequent analysis. The bloodvessel number was determined at 200� magnification, andthe density of the blood vessels was normalized to the num-ber of blood vessels per mm2.

2.8. Statistical analysis

Data are expressed as mean ± SD. Statistical analysiswas performed using the two-population Student’s t-test(Fig. 1) and the chi-square test (Figs. 3 and 10). The signif-icant level was set at p < 0.05.

3. Results and discussion

3.1. The loading and release of TMC/DNA complexes in

collagen scaffold

It is known that the amount of DNA can significantlyinfluence the transfected cell numbers, and thereby theamount of products secreted, such as cell growth factors. Aporous collagen scaffold has such a strong ability to adsorbDNA complexes, so that a DNA-loaded scaffold can be eas-ily obtained by dropping a solution of TMC/DNAcomplexes into the scaffold. Fig. 1 shows that the immobi-lized DNA amount increased monotonously with increasingfeeding TMC/DNA complex concentration. Each piece ofcollagen scaffold (2 mg) adsorbed only 20 lg of DNA witha feeding concentration of 0.1 mg ml�1; this increased to72, 170 and 322 lg when the feeding concentration was

0.1 0.3 1 30

100

200

300

*

*

*

Inco

rpor

ated

DN

A p

er s

caff

old

(µg/

2mg)

DNA concentration (mg/ml)

Fig. 1. Incorporated DNA amount of TMC/DNA complexes (N/P 10)per 2 mg collagen scaffold as a function of feeding concentration.*p < 0.01.

increased to 0.3, 1 and 3 mg ml�1, respectively. Therefore,within the investigated range, the amount of DNA adsorbedin the scaffolds can be mediated by its feeding concentration.

The crosslinked porous collagen scaffold swells andretains water inside to form a gel-like morphology [23].In this sense, ‘‘adsorption” should be the main status ofthe incorporated TMC/DNA complexes because most ofthem should adsorb onto the pore walls instead of enteringthe collagen dense walls. Of course, it cannot be excludedthat the complexes are geometrically loaded into the poresand suspended in the liquid. Therefore, release of theloaded DNA complexes can be expected, as typicallyshown in Fig. 2. Here the collagen scaffolds containing322 lg (TMC/DNA322) and 170 lg of DNA (TMC/DNA170) were tested because only these two sampleshad enough DNA for the sustained release test via the cur-rent method. Unlike the expected fast release, the releasetime lasted up to 96 h, with no apparent burst releasebehavior. The DNA was released rapidly and linearly dur-ing the first 48 h, with a total amount released of �70%.This is quite normal because of the large imbalance ofDNA concentration between the scaffold interior and thesolution at this stage. After this time, the release ratedecreased. At the end of the investigation (96 h), more than80% of the adsorbed DNA had been released. Fig. 2 alsoshows that a slight higher ratio of DNA was released fromthe TMC/DNA170-loaded scaffold. The release profile ofnaked DNA from the scaffold was also studied (data notshown), but no significant difference could be foundbetween the naked DNA and the TMC/DNA complex,indicating that the format of the DNA is not key to con-trolling the release of DNA.

3.2. In vitro gene transfection

In order to evaluate the VEGF secretion level in vitro,a 3-D static culture model was used. Fig. 3a shows the

0 20 40 60 80 100

0

20

40

60

80

Cum

ulat

ive

rel

ease

(%

)

Time (h)

TMC/DNA 170 TMC/DNA 322

Fig. 2. Cumulative released amount of DNA (TMC/DNA complexes, N/P 10) from the scaffolds as a function of incubation time. The collagenscaffolds containing 322 lg (TMC/DNA322) and 170 lg of DNA (TMC/DNA170), respectively, were tested.

Control Naked DNA356 TMC/DNA70 TMC/DNA3220

2000

4000

6000*

VE

GF

(pg

/ml p

er

10

6 cel

ls)

4d 8d

*(a)

Control Naked DNA356 TMC/DNA70 TMC/DNA3220

50

100

150

200

Cel

l Num

ber

(10

4)

4d 8d

(b)

Fig. 3. (a) VEGF secretion from HEK293 cells and (b) number of HEK293 cells in different samples after transfection at day 4 and 8. *p < 0.01.

2 For interpretation of the references to colour in this figure legend, thereader is referred to the web version of this article.

Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994 2987

VEGF expression of HEK293 cells cultured for 4 and8 days in the collagen scaffolds that were incorporatedwith the TMC/DNA complexes and naked plasmidDNA, respectively. The virgin and naked DNA-loadedcollagen scaffolds were chosen as the controls. Theimpregnation of naked pDNA–VEGF (356 lg of DNAper scaffold, named as naked DNA356) enhanced theVEGF expression level during the first 4 days, whichwas only about 50% higher than that of the DNA-freecontrol. This should be attributed to the low transfectionefficiency of the naked plasmid DNA. [21] However, theTMC/DNA complexes-incorporated scaffold increasedthe VEGF expression level by a factor of 3.5 over theDNA-free scaffold at day 4 even at a relative lower dose(70 lg of DNA per scaffold, named as TMC/DNA70,p < 0.01). With an increase in the dose of the loadedTMC/DNA complexes, the level of VEGF expressionwas improved again. For example, the secreted VEGFby TMC/DNA322 group was eight times higher than thatof the DNA-free group at day 4 (p < 0.01). From day 5 today 8, although the VEGF expression level of the TMC/DNA complexes loaded scaffolds was still higher thanthat of the DNA-free control, it was smaller than thatof the first 4 days (p < 0.01). Furthermore, there was noobvious difference between the naked DNA356 scaffoldand the DNA-free control at this period. This could beattributed to the temporal nature of expression of ourgene delivery vector. In many cases, the polymeric genedelivery vector and naked DNA could not sustain thegene expression for a long time. The cell proliferationdata post-transfection is shown in Fig. 3b. The similar cellnumbers of the gene-transfected samples and the DNA-free controls indicated the low cytotoxicity of the DNAand TMC/DNA complexes at the given dosages.

3.3. Gross views of angiogenesis of scaffolds

Fig. 4 shows the gross views of the implanted scaffolds2 weeks after implantation. The colors of the DNA-freescaffold (Fig. 4a) and the TMC/DNA-EGFP scaffold(pDNA encoding enhanced green fluorescence protein,Fig. 4c) are pale, with an absence of blood vessels on their

surfaces. On the naked DNA356 scaffold, a lot of bloodvessels can be seen surrounding the scaffold, but only afew of them are in the scaffold (Fig. 4b). By contrast, thecolor of the TMC/DNA70 scaffold (Fig. 4d) is red2, andthere are a lot of blood vessels on its surface. With a stillhigher amount of DNA (the TMC/DNA322 scaffold),more blood vessels were found on the scaffold surface(Fig. 4e). This gross observation confirms primarily thatthe large amount of naked DNA–VEGF and TMC/DNA–VEGF complexes can enhance the angiogenesis ofthe collagen scaffold.

3.4. Histological and immunohistochemical examination

To examine the angiogenesis process of the implantedscaffolds, H&E and immunohistochemical staining werecarried out at different implantation intervals. In theH&E-stained pictures (Figs. 5–7), granulation tissue wasalways found surrounding the scaffolds. Deformation ofthe scaffolds could be observed because of the stress afterimplantation and the relatively low mechanical strengthof the highly porous collagen scaffold.

After implantation for 1 week, the DNA-free scaffold(Fig. 5a) showed minor infiltration of granulocytes andmacrophages only at the margin of the scaffold, whereasthe naked DNA356 scaffold (Fig. 5b), the TMC/DNA-EGFP scaffold (Fig. 5c), the TMC/DNA70 scaffold(Fig. 5d) and the TMC/DNA322 scaffold (Fig. 5e) allshowed a moderate infiltration of macrophages, fibroblastsand some granulocytes, some of which were even in themiddle of the scaffolds. The blood vessels (circular struc-tures containing red blood cells) were found in the sur-rounding tissues (Fig. 5a and c) but not in the scaffolds.By contrast, the blood vessels were already found at themarginal areas of the naked DNA356 scaffold and theTMC/DNA70 and TMC/DNA322 scaffolds (Fig. 5b, dand e, arrowheads), although still no blood vessels wereobserved in the interior parts of these samples.

Fig. 4. Gross views of the scaffolds after implantation for 2 weeks. (a) DNA-free scaffold, (b) naked DNA356 scaffold, (c) TMC/DNA-EGFP scaffold, (d)TMC/DNA70 scaffold, and (e) TMC/DNA322 scaffold.

2988 Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994

Immunohistochemical staining of VEGF was used toexamine the expression of VEGF after in vivo transfectionin the DNA-free scaffold, the naked DNA356 scaffold andthe TMC/DNA322 scaffold (Fig. 8a–c). Since our antibodyis against human and mouse VEGF, positive VEGF stain-ing can also be observed in the DNA-free scaffold (Fig. 8a).This VEGF signal (brown color) is located mainly in thecytoplasms of cells at the marginal scaffold and close extra-cellular matrix. As only few cells infiltrated into the DNA-free scaffold during the first week (Fig. 5a), the weakVEGF signal is understandable. By contrast, stronger posi-tive staining of VEGF was observed in the DNA356 scaf-fold (Fig. 8b) and the TMC/DNA322 scaffold (Fig. 8c),in both the cytoplasm of the infiltrated cells and the extra-cellular matrix in the scaffolds. These results qualitativelyconfirm that the pDNA encoding VEGF could enhancethe secretion of VEGF in tissues after in vivo transplanta-tion of both naked pDNA and TMC/DNA-containingscaffolds. There was no obvious VEGF trace in the sur-rounding tissues at this stage. The reason for this mightbe that the cells in the normal tissues have a low concentra-tion of VEGF, and these cells are not easily transfected incomparison with the newly proliferated cells as the genetransfection is more likely to happen during the cell prolif-eration process. Immunohistochemical staining of CD31was used to confirm neo-vessel formation (Fig. 9a) as itrepresents a highly specific marker for vascular and endo-thelial cells. Immunohistochemical staining of SMA wasused as a marker for mature blood vessels in the TMC/DNA322 scaffold (Fig. 9d, indicated by the arrow). Both

assays show similar blood vessel density and location asthe H&E staining (Fig. 5e). In the SMA case, the bloodvessels had a circular structure, indicating that the newlyformed blood vessels have smooth muscle cells in theirwalls. Although the blood vessel maturation was reportedin the presence of PDGF or FGF-2 [8,11,17] here we gotmature blood vessels with a smooth muscle cell layer withonly VEGF. The reason might be that VEGF is capable ofupregulating the production of PDGF or FGF-2, whichaccelerates the maturation of blood vessels. Another possi-bility is that the inflammatory response induced by theextraneous DNA can affect the expression of some cyto-kines, which then upregulate the expression of PDGF orFGF-2.

Fig. 10 summarizes the average blood vessel density permm2 for the different kinds of scaffolds. The average num-bers of blood vessels are 1.2, 1.5, 1.6, 1.5 and 2.2 per mm2

for the DNA-free, naked DNA356, TMC/DNA-EGFP,TMC/DNA70 and TMC/DNA322 scaffolds, respectively.No significant difference was detected between thesescaffolds.

At week 2, granulocytes and macrophages disappearedfrom the DNA-free scaffold (Fig. 6a). Few fibroblasts haveinfiltrated into the interiors of the scaffolds at this stage.Only a moderate number of macrophages and a few fibro-blasts were observed in the entire naked DNA356 (Fig. 6b),TMC/DNA-EGFP (Fig. 6c), TMC/DNA70 (Fig. 6d) andTMC/DNA322 (Fig. 6e) scaffolds, although the cell densityin the scaffold centers was much lower than that in the mar-ginal parts. As expected, the gene-incorporated scaffolds

Fig. 5. Optical micrographs of H&E-stained sections of the scaffolds after implantation in Sprague–Dawley mice for 1 week. (a) DNA-free scaffold, (b)naked DNA356 scaffold, (c) TMC/DNA-EGFP scaffold, (d) TMC/DNA70 scaffold, and (e) TMC/DNA322 scaffold. The bar indicates 200 lm. M, theimplanted collagen/chitosan scaffold; T, the subcutaneous connective tissue. Arrows indicate blood vessels in the scaffolds.

Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994 2989

attracted a larger number of cells than the DNA-free scaf-fold. It is known that the macrophages attracted by theexpressed VEGF can release additional VEGF as well asMMPs to the surrounding tissues [27]. The VEGF alsostimulates proliferation and differentiation of granulocytes,endothelial cells, fibroblasts and many other cells. vanKuppevelt et al. [1] also found that the addition of VEGFand FGF2 to a scaffold attracted more cells. An increasedcellular reaction may be necessary to trigger the angiogen-esis [28].

The H&E staining also reveals a difference in the angio-genesis (Fig. 6). The DNA-free, naked DNA356 and TMC/DNA-EGFP scaffolds only showed blood vessels at theirmarginal areas (Fig. 6a–c, arrowheads), where the deepestinfiltration depth of the blood vessel was less than 0.2 mm.However, blood vessels were present in the more interiorparts of the TMC/DNA70 and TMC/DNA322 scaffolds(Fig. 6d, arrowheads). The greatest depth of the blood ves-sels was more than 0.5 mm (the total thickness of the scaf-fold was 2 mm).

Fig. 6. Optical micrographs of H&E-stained sections of the scaffolds after implantation in Sprague–Dawley mice for 2 weeks. (a) DNA-free scaffold, (b)naked DNA356 scaffold, (c) TMC/DNA-EGFP scaffold, (d) TMC/DNA70 scaffold, and (e) TMC/DNA322 scaffold. The bar indicates 200 lm. M, theimplanted scaffolds; T, the subcutaneous connective tissue. Arrows indicate blood vessels in the scaffolds.

2990 Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994

The VEGF expression in the DNA-free scaffold(Fig. 8d) was much stronger than that in the first week,conveying that during the tissue-healing process the scaf-fold itself can function as an analogue of the extracellularmatrix to regulate the VEGF expression level of cells.The areas of VEGF signal in the naked DNA356 andTMC/DNA322 scaffolds are obviously larger than that inthe DNA-free scaffold. The results demonstrate again thatthe scaffold loaded with pDNA–VEGF can successfullyenhance the expression level of VEGF in tissues. TheVEGF-stained area in the TMC/DNA322 scaffold waseven larger than that of the naked DNA356 scaffold, and

VEGF was not restricted to the scaffold but was alsolocated in the surrounding tissues. The results confirm that,because of the protection effect of TMC, the gene transfec-tion in vivo is more efficient. It is worth noting that theVEGF concentration in the TMC/DNA322 scaffold is alsohigher in the second week than during the first week. Tak-ing into account the fast in vitro release rate (Fig. 3a, 70%DNA released in 2 days) and the in vitro gene expressionresults (Fig. 4a, the gene expression is highest during theinitial 4 days), it is likely that the gene transfectionin vivo cannot be sustained for 2 weeks. Our explanationis that the release behavior of DNA should be different in

Fig. 7. Optical micrographs of H&E-stained sections of the scaffolds after implantation in Sprague–Dawley mice for 3 weeks. (a) DNA-free scaffold, (b)naked DNA356 scaffold, (c) TMC/DNA-EGFP scaffold, (d) TMC/DNA70 scaffold, and (e) TMC/DNA322 scaffold. The bar indicates 200 lm. M, theimplanted collagen/chitosan scaffold; T, the subcutaneous connective tissue. Arrows indicate blood vessels in the scaffolds.

Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994 2991

the in vitro and in vivo experiments. Due to the poorerexchange of media in vivo, the release time of DNAin vivo should be much longer than that in vitro. More-over, even when the cells were successfully transfectedand the expression of VEGF started immediately, the levelof VEGF secretion should be still low inside the scaffoldbecause of the insufficient number of cells infiltrating intothe scaffold during the first week after transplantation.Thirdly, in the in vivo experiment, the secreted VEGFmay be bonded to extracellular matrix molecules, such aschondroitin sulfate and heparin, or other receptors inside

cells, which can play a storage role and thereby extendthe lifetime of the VEGF. Consequently, the in vivo angio-genesis could still be enhanced even after 2 weeks, althoughthe expression level of VEGF was lower at this time. Theblood vessels in the TMC/DNA322 scaffold were con-firmed by CD31 staining (Fig. 9b, indicated by arrows)too. The immunoassay of SMA shows the existence ofsmooth muscle cells on the blood vessel walls (Fig. 9e, indi-cated by arrows).

At this stage, the average numbers of blood vessels permm2 for the DNA-free and TMC/DNA-EGFP scaffolds

Fig. 8. Optical micrographs of VEGF-stained sections of scaffolds after implantation in Sprague–Dawley mice for 1 week (a–c), 2 weeks (d–f) and 3 weeks(g–i). DNA-free scaffolds (a, d, and g), naked DNA356 scaffolds (b, e, and h) and TMC/DNA322 scaffolds (c, f, and i) were selected for evaluation. Thebar indicates 200 lm. The brown color represents the VEGF positive signal.

2992 Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994

were 2.1 and 2.5, respectively, which were slightly increasedcompared to week 1 (Fig. 10, p < 0.05). No significant dif-ference was found between the DNA-free scaffold and theTMC/DNA-EGFP scaffold. By contrast, the average num-bers of the blood vessels were significantly increased to 5.3,4.6 and 8.2 for the naked DNA356, TMC/DNA70 andTMC/DNA322 scaffolds, respectively, all of which are sig-nificantly larger than those of the DNA-free and the TMC/DNA-EGFP scaffolds (p < 0.01). This result is consistentwith the observation shown in Fig. 4. The higher blood ves-sel density in the TMC/DNA322 scaffold than in the othersis understandable since it has the highest amount of VEGF(Fig. 8).

At week 3, a larger number of macrophages and fibro-blasts were observed in the entire scaffolds of all the sam-ples (Fig. 7). The pores of the marginal parts of thescaffolds were filled with extracellular matrix. The bloodvessels were found in deeper locations of the scaffolds forall the samples. The blood vessels were found throughoutthe whole TMC/DNA322 scaffold (Fig. 7e), whereas stillno blood vessels were detected in the middle of the otherscaffolds. The VEGF area of all the samples at this stageis smaller than during the second week (Fig. 8g–i), implyingthat DNA starts to deactivate or be degraded at this stagein vivo. The CD31 and SMA staining results (Fig. 9c and f)show that some of the newly formed blood vessels in theTMC/DNA322 scaffold are mature vessels with an intact

endothelial cell layer and a smooth muscle cell layer in theirwalls. The average numbers of blood vessels of the DNA-free, naked DNA356, TMC/DNA-EGFP, TMC/DNA70and TMC/DNA322 scaffolds are 4.3, 5.8, 5.6, 6.7 and12.6 per mm2, respectively, and all are higher than at week2 (p < 0.05) except the naked DNA scaffold.

In view of the above results, one can conclude thatthe impregnation of naked pDNA–VEGF or TMC/DNA–VEGF complexes can obviously enhance the angi-ogenesis of the scaffolds, resulting in a higher blood ves-sel density and deeper infiltration of the blood vessels,while incorporation of TMC/DNA-EGFP has no obvi-ous effect on angiogenesis. Introducing the cationic vec-tor TMC could obviously increase the in vivo genetransfection ability, leading to higher VEGF expressionlevel and blood vessel density. The blood vessel densityand infiltration depth both increase with increasingamount of DNA incorporated. These results also matchthe results of the in vitro VEGF expression, demonstrat-ing that a higher DNA dose results in a higher VEGFexpression, possibly leading to more cell infiltration,more extracellular matrix synthesis and faster angiogene-sis in vivo. Our study indicates that the combination ofTMC/DNA–VEGF complexes with a porous scaffold isa simple and effective way to enhance the angiogenesisof implants, thereby accelerating the process of woundhealing and tissue regeneration.

Fig. 9. Optical micrographs of CD31 (a–c) and SMA-stained (d–f) sections of the TMC/DNA322 scaffolds after implanted in Sprague–Dawley mice for1 week (a and d), 2 weeks (b and e) and 3 weeks (c and f). The bar indicates 200 lm. Arrows indicate blood vessels in the scaffolds.

Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994 2993

4. Conclusion

A collagen scaffold containing plasmid DNA encodingVEGF was fabricated and showed faster angiogenesis inin vivo experiments. The plasmid DNA encoding VEGFwas first complexed with the cationic gene delivery vectorTMC. The TMC/DNA complexes formed were incorpo-rated into the porous collagen scaffolds by simple physicaladsorption. The amount of loaded DNA can be mediatedby its feeding concentration, and the impregnated DNAcan be released in a sustained manner. The in vitro trans-gene experiments showed that the TMC/DNA–VEGFcomplexes loaded in the collagen scaffolds could signifi-

cantly improve the expression level of VEGF in HEK293cells. The in vivo implantation results demonstrated thatcollagen scaffolds impregnated with the TMC/DNA–VEGF complexes could enhance the expression level ofVEGF, and thereby cell infiltration, extracellular matrixsynthesis and angiogenesis, in a dose-dependent manner.Some of the newly formed blood vessels produced by theTMC/DNA–VEGF complexes contained an intact smoothmuscle cell layer, illustrating their maturity. This researchproves that incorporation of TMC/DNA–VEGF com-plexes, which may function as an alternative to water-solu-ble growth factors, is a feasible way to obtain a bioactivescaffold with the property of enhanced angiogenesis.

7 14 21

0

5

10

15

*

Ave

rage

blo

od v

esse

l num

ber

per

mm

2

Time (day)

DNA Free Control Naked DNA356 TMC/DNA-EGFP TMC/DNA70 TMC/DNA322

*

Fig. 10. The average blood vessel density within different kinds ofscaffolds after implantation for 1–3 weeks. *p < 0.01.

2994 Z. Mao et al. / Acta Biomaterialia 5 (2009) 2983–2994

Acknowledgements

We gratefully acknowledge Dr. Jun Li, State Key Labora-tory of Diagnosis and Treatment for Infectious Diseases,China, and Prof. Changyong Wang and Dr. Hong Jiang,Military Medical Academy of PLA, China, for their kinddonation of plasmid DNA. We also thank Dr. GuopingSheng, Medical School of Zhejiang University, for his kindhelp in the amplification of plasmid DNA. This study wasfinancially supported by the Major State Basic ResearchProgram of China (2005CB623902), the National High-techResearch and Development Program (2006AA03Z442,2006AA02A140), the Science and Technology Program ofZhejiang Province (2006C13022, 2007C23014) and the Nat-ural Science Foundation of China (50873088).

References

[1] Nillesen STM, Geutjes PJ, Wismans R, Schalkwijk J, Daamen WF,van Kuppevelt TH. Increased angiogenesis and blood vessel matu-ration in acellular collagen–heparin scaffolds containing both FGF2and VEGF. Biomaterials 2007;28:1123–31.

[2] Nomi M, Atala A, De Coppi P, Soker S. Principals of neovascular-ization for tissue engineering. Mol Aspects Med 2002;23:463–83.

[3] Mooney DJ, Mikos AG. Growing new organs. Sci Am 1999;280:60–5.[4] Pieper JS, Hafmans T, van Wachem PB, van Luyn MJ, Brouwer LA,

Veerkamp JH, van Kuppevelt TH. Loading of collagen–heparinsulfate matrices with bFGF promotes angiogenesis and tissuegeneration in rats. J Biomed Mater Res 2002;62:185–94.

[5] Patel ZS, Mikos AG. Angiogenesis with biomaterial-based drug- andcell-delivery systems. J Biomater Sci Polym Ed 2004;15:701–26.

[6] Lee KY, Peters MC, Mooney DJ. Comparison of vascular endothelialgrowth factor and basic fibroblast growth factor on angiogenesis inSCID mice. J Control Release 2003;87:49–56.

[7] O’Brien FJ, Harley BA, Yannas IV, Gibson LJ. The effect of pore size oncell adhesion in collagen–GAG scaffolds. Biomaterials 2005;26:433–41.

[8] Bruick RK, McKnight SL. Building better vasculature. Genes Dev2001;15:2497–502.

[9] Hoeben A, Landuyt B, Highley MS, Wildiers H, van Oosterom AT,De Bruijn EA. Vascular endothelial growth factor and angiogenesis.Pharmacol Rev 2004;56:549–80.

[10] Komori M, Tomizawa Y, Takada K, Ozaki M. A single localapplication of recombinant human basic fibroblast growth factoraccelerates initial angiogenesis during wound healing in rabbit earchamber. Anesth Analg 2005;100:830–4.

[11] Shi HF, Han CM, Mao ZW, Ma L, Gao CY. Enhanced angiogenesisin porous collagen-chitosan scaffolds loaded with angiogenin. TissueEng A 2008;14:1775–85.

[12] Kim DC, Sugiyama Y, Satoh H, Fuwa T, Iga T, Hanano M. Kineticanalysis of in vivo receptor-dependent binding of human epidermalgrowth factor by rat tissues. J Pharm Sci 1988;77:200–7.

[13] Konturek SJ, Pawlik W, Mysh W, Gustaw P, Sendur R, Mikos E, et al.Comparison of organ uptake and disappearance half time of humanepidermal growth factor and insulin. Regul Pept 1990;30:137–48.

[14] Edelman ER, Nugent MA, Karnovsky MJ. Perivascular and intra-venous administration of basic fibroblast growth factor: vascular andsolid organ deposition. Proc Natl Acad Sci USA 1993;90:1513–7.

[15] Bonadio J, Goldstein SA, Levy RJ. Gene therapy for tissue repair andregeneration, Adv. Drug Deliv Rev 1998;33:53–69.

[16] Jang JH, Shea LD. Controllable delivery of non-viral DNA fromporous scaffolds. J Control Release 2003;86:157–68.

[17] Shea LD, Smiley E, Bonadio J, Mooney DJ. DNA delivery frompolymer matrices for tissue engineering. Nat Biotechnol 1999:551–4.

[18] Jang JH, Rives CB, Shea LD. Plasmid delivery in vivo from poroustissue-engineering scaffolds: transgene expression and cellular trans-fection. Mol Ther 2005;12:475–83.

[19] Scherer F, Schillinger U, Putz U, Stemberger A, Plank C. Nonviralvector loaded collagen sponges for sustained gene delivery in vitroand in vivo. J Gene Med 2002;4:634–43.

[20] Guo T, Zhao JN, Chang JB, Ding Z, Hong H, Chen JN, et al. Porouschitosan–gelatin scaffold containing plasmid DNA encoding trans-forming growth factor-b1 for chondrocytes proliferation. Biomateri-als 2006;27:1095–103.

[21] Mao ZW, Ma L, Jiang Y, Yan M, Gao CY, Shen JC. N,N,N-Trimethyl chitosan chloride as a gene vector: synthesis and applica-tion. Macromol Biosci 2007;7:855–63.

[22] Mao HQ, Roy K, Troung-Le VL, Janes Kevin A, Lin KevinY, YanWang J, et al. Chitosan–DNA nanoparticles as gene carriers:synthesis, characterization and transfection efficiency. J ControlRelease 2001;70:399–421.

[23] Ma L, Gao CY, Mao ZW, Zhou J, Shen JC. Enhanced biologicalstability of collagen porous scaffolds by using amino acids as novelcross-linking bridges. Biomaterials 2004;25(15):2997–3004.

[24] Shi YC, Ma L, Zhou J, Mao ZW, Gao CY. Collagen/chitosan–silicone membrane bilayer scaffold as a dermal equivalent. Polym AdvTechnol 2005;16:789–94.

[25] Hong Y, Song HQ, Gong YH, Mao ZW, Gao CY, Shen JC. Covalentlycrosslinked chitosan hydrogel: properties of in vitro degradation andchondrocyte encapsulation. Acta Biomater 2007;3:23–31.

[26] Taylor AP, Rodriguez M, Adams K, Goldenberg DM, BlumenthalRD. Altered tumor vessel maturation and proliferation in placentagrowth factor-producing tumors: potential relationship to posttherapytumor angiogenesis and recurrence. Int J Cancer 2003;105:158–64.

[27] Blau HM, Banfi A. The well-tempered vessel. Nat Med 2001;7:532–4.[28] van Amerongen MJ, Molema G, Plantinga J, Moorlag H, van Luyn

MJ. Neovascularization and vascular markers in a foreign bodyreaction to subcutaneously implanted degradable biomaterial in mice.Angiogenesis 2002;5:173–80.