Amnion-Derived Mesenchymal Stromal Cells ShowAngiogenic Properties but Resist Differentiation

into Mature Endothelial Cells

Julia Konig,1 Berthold Huppertz,1 Gernot Desoye,2 Ornella Parolini,3 Julia D. Frohlich,1,2

Gregor Weiss,1 Gottfried Dohr,1 Peter Sedlmayr,1 and Ingrid Lang1

Mesenchymal stromal cells derived from the human amnion (hAMSC) currently play an important role in stemcell research, as they are multipotent cells that can be isolated using noninvasive methods and are immuno-logically tolerated in vivo. The objective of this study was to evaluate their endothelial differentiation potentialwith regard to a possible therapeutic use in vascular diseases. hAMSC were isolated from human term placentasand cultured in Dulbecco’s modified Eagle’s medium (DMEM) (non-induced hAMSC) or endothelial growthmedium (EGM-2) (induced hAMSC). Induced hAMSC changed their fibroblast-like toward an endothelial-likemorphology, and were able to take up acetylated low-density lipoprotein and form endothelial-like networks inthe Matrigel assay. However, they did not express the mature endothelial cell markers von Willebrand factorand vascular endothelial-cadherin. Gene expression analysis revealed that induced hAMSC significantlydownregulated pro-angiogenic genes such as tenascin C, Tie-2, vascular endothelial growth factor A (VEGF-A),CD146, and fibroblast growth factor 2 (FGF-2), whereas they significantly upregulated anti-angiogenic genessuch as serpinF1, sprouty1, and angioarrestin. Analysis of protein expression confirmed the downregulation ofFGF-2 and Tie-2 (27% – 8% and 13% – 1% of non-induced cells, respectively) and upregulation of the anti-angiogenic protein endostatin (226% – 4%). Conditioned media collected from hAMSC enhanced viability ofendothelial cells and had a stabilizing effect on endothelial network formation as shown by lactate dehydro-genase and Matrigel assay, respectively. In summary, endothelial induced hAMSC acquired some angiogenicproperties but resisted undergoing a complete differentiation into mature endothelial cells by upregulation ofanti-angiogenic factors. Nevertheless, they had a survival-enhancing effect on endothelial cells that might beuseful in a variety of cell therapy or tissue-engineering approaches.

Introduction

Mesenchymal stromal cells (MSC) are promisingcandidates for tissue engineering and cell-based thera-

pies because of their low immunogenicity and their differen-tiation potential into multiple cell types [1]. A growingnumber of clinical trials shows that ex vivo expanded MSCcan be safely administered [2,3]. Most commonly, MSC areisolated from bone marrow or adipose tissue by invasiveprocedures. Depending on the age and disease stage of thedonor, proliferation and differentiation capacities of the ob-tained cells may be impaired [4,5]. Therefore, fetal or post-natal tissues are an attractive alternative source, and indeed,MSC have been isolated from umbilical cord and cord blood,placenta, and fetal membranes (amnion, chorion laeve) [6–11].

The amnion is an especially promising source of cells fortherapeutic use, as its feasibility in clinical applications hasalready been confirmed. Its first documented clinical usegoes back to 1910, where it was applied as a surgical materialin skin transplantation [12]. Since then, it has been used in avariety of clinical settings such as treatment of chemicalburns, skin ulcers, and ophthalmology. Its beneficial effectsare awarded to its anti-inflammatory, immunomodulatory,and scar formation-reducing properties, among others [13].

MSC isolated from the human amnionic membrane(hAMSC) have phenotypic and functional similarities tobone marrow-derived MSC. They have been successfullydifferentiated toward cells of the classical mesodermal (os-teogenic, adipogenic, and chondrogenic), ectodermal (neu-rogenic), and endodermal (hepatogenic, pancreatic) lineages

1Institute of Cell Biology, Histology, and Embryology and 2Clinic of Obstetrics and Gynecology, Medical University of Graz, Graz, Austria.3Centro di Ricerca E. Menni, Fondazione Poliambulanza, Istituto Ospedaliero, Brescia, Italy.

STEM CELLS AND DEVELOPMENT

Volume 21, Number 8, 2012

� Mary Ann Liebert, Inc.

DOI: 10.1089/scd.2011.0223

1309

[10]. Similar to bone marrow derived MSC, they activelysuppress T-lymphocyte proliferation [14–16] and block dif-ferentiation and maturation of monocytes into dendritic cellsin vitro [17]. After xenogeneic transplantation into neonatalswine and rats, hAMSC engraft without immunosuppression[18]. Very recently, it was shown that amnionic membraneapplication could reduce liver fibrosis in a bile duct ligationrat model [19] and improve cardiac function of ischemic rathearts [20]. In addition, isolated allo- and xenogenic hAMSCcould reduce bleomycin-induced lung fibrosis in a mousemodel [21].

These studies suggest that hAMSC hold great promise fora potential use in cell therapy and tissue engineering. One ofthe major challenges in this field lies in the vascularization ofengineered tissues and in finding a suitable cell populationfor the endothelialization of vascular grafts. MSC might bean interesting choice; however, their endothelial differentia-tion potential is still controversial. Although some groupshave reported the differentiation of MSC originating frombone marrow or adipose tissue into endothelial-like cells[22,23], opposite results showed that MSC from bone mar-row could not be differentiated toward the endothelial line-age [24,25]. Since the bone marrow and adipose tissue arehighly vascularized, it is difficult to obtain cultures free ofprimary endothelial cells. It might be possible that the iso-lated MSC populations already contained some endothelialor endothelial progenitor cells before endothelial inductionin vitro. Thus, instead of inducing endothelial differentiationof MSC, the chosen angiogenic culture conditions might haveled to a selective proliferation of already existing endothelialcells. Here, the amnion shows an additional advantage:Being avascular, it allows an easy isolation of MSC culturesfree of endothelial cells.

Based on the current controversial reports, in this study,we tested the endothelial differentiation potential of hAMSCafter excluding the presence of endothelial cells. In additionto applying phenotypic characterization and functionalstudies, we evaluated the effect of angiogenic culture con-ditions on gene and protein expression of hAMSC usingmicroarray analysis and angiogenic protein arrays. Further,we examined the paracrine effects of hAMSC on the viabilityand network formation of endothelial cells.

Materials and Methods

Isolation and culture of hAMSC

Human term placentas of normal pregnancies (range 38 to42 weeks) were obtained after spontaneous delivery or ce-sarean section with informed consent. Approval of theEthical Committee of the Medical University of Graz wasgranted (No. 21-079 ex 09/10).

Isolation of hAMSC was performed according to the pro-tocol of Soncini et al. [26]. The amnion was manually sepa-rated from the chorion and washed with sterile 0.9% saline(Fresenius Kabi, Bad Homburg, Germany) supplementedwith 150 IU/mL penicillin, 150mg/mL streptomycin (bothfrom PAA Laboratories, Pasching, Austria), and 0.4mg/mLamphotericin B (Gibco, Invitrogen, Paisley, UK). The amnionwas cut into small pieces and incubated in 2.5 U/mL dispase(BD Biosciences, Bedford, MA) at 37�C for 9 min. Subse-quently, the amnion was transferred to Dulbecco’s modifiedEagle’s medium (DMEM) low glucose (Gibco, Invitrogen)

supplemented with 15% fetal bovine serum (FBS) Gold (Gib-co, Invitrogen), 100 IU/mL penicillin, and 100mg/mL strep-tomycin for 10 min. Next, the amnion was incubated with1.0 mg/mL Collagenase A and 0.01 mg/mL DNase (bothfrom Roche, Penzberg, Germany) for 2 h. After centrifugationfor 3 min at 150 g, the supernatant was poured over a 100mmcell strainer (BD Biosciences) and centrifuged for 10 min at 300g. The cell pellet was washed in phosphate-buffered saline(PBS; Gibco, Invitrogen) and resuspended in either DMEMsupplemented with 15% FBS, 100 IU/mL penicillin, and100mg/mL streptomycin (non-induced hAMSC) or endothe-lial growth medium-2 (EGM-2; Lonza, Walkersville, MD) forendothelial induction (induced hAMSC). EGM-2 contains 2%FBS, epidermal growth factor (EGF), hydrocortisone, vascularendothelial growth factor (VEGF), fibroblast-like growth fac-tor-2 (FGF-2), insulin-like growth factor 1 (IGF-1), ascorbicacid, and heparin. Aliquots of the cell suspensions were spundown on slides immediately after isolation (cytospins). Re-maining cells were grown on culture flasks coated with 1%gelatin (Sigma-Aldrich, St. Louis, MO) and harvested withaccutase (PAA Laboratories). Medium was changed every 2–3days.

Isolation and culture of placental endothelial cells

Endothelial cells from normal term human placentas wereisolated as published earlier [27]. Briefly, after removal of theamnion, arterial chorionic blood vessels at the apical surfaceof the chorionic plate were resected. Vessels were washedwith Hank’s balanced salt solution (HBSS, Gibco) to removeresidual blood. Endothelial cells were isolated by perfusionof vessels with HBSS containing 0.1 U/mL collagenase,0.8 U/mL dispase (both from Roche), supplemented with300 IU/mL penicillin, and 300 mg/mL streptomycin, pre-warmed to 37�C. The perfusion time was limited to 7 min toavoid contamination with non-endothelial cells. The cellsuspension was centrifuged (200 g for 5 min), the pellet wasresuspended with EGM-MV medium (Lonza), and the cellswere plated on culture plates pre-coated with 1% gelatin.The endothelial identity was confirmed by positive stainingfor the classical endothelial marker von Willebrand factor(vWF; immunoglobulin fraction, rabbit anti-human, 0.7 mg/mL; Dako, Glostrup, Denmark) and absence of markersagainst fibroblasts (CD90, clone ASO2, 0.1 mg/mL, mouseIgG1; Dianova, Hamburg, Germany), and smooth musclecells (smooth muscle actin, clone 1A4, 0.2 mg/mL, mouseIgG2a and desmin, clone D33, 0.4 mg/mL, mouse IgG1, bothfrom Dako).

Flow cytometry analysis

Flow cytometry analyses were carried out at the Flow Cy-tometry Core Facility at the Center for Medical Research(ZMF) and at the Institute of Cell Biology, Histology, andEmbryology of the Medical University of Graz. The surfacemarker expression of hAMSC was analyzed with a FACSLS-RII� instrument equipped with 355 and 405 nm UV lazers, a488 nm argon ion lazer, and a 635 nm red diode lazer (BectonDickinson, Franklin Lakes, NJ). hAMSC were washed andlabeled for 30 min at 4�C at concentrations according to indi-vidual titration with monoclonal antibodies against CD105(PE-labeled, clone 2H6F11, 1:33; Caltag Laboratories, Burlin-game, CA), CD14 (FITC-labeled, clone MOP9, 1:100), CD34

1310 KONIG ET AL.

(APC-labeled, clone 8G12, 1:100), CD44 (PE-labeled, clone 515,1:33), CD45 (PE-Cy7-labeled, clone HI30, 1:100), CD73 (PE-labeled, clone AD2, 1:20), CD90 (APC-labeled, clone 5E10,1:100), and HLA-DR (PerCP-labeled, clone 243 (G46-6), 1:100,all from BD Biosciences). Appropriate isotype-matched anti-bodies were used as negative controls (BD). Data from 10,000viable cells were acquired. List mode files were analyzed withFCS Express Software (BD).

Immunohistochemistry/immunocytochemistry

Cryosections (5mm) of human term placental samples weremounted on microslides (Assistent, Karl Hecht AG, Sond-heim, Germany). hAMSC were grown on gelatin-coated glasschamber slides (Lab-Tek II, Nalgene Nunc International, Na-perville, IL) and washed with PBS before being harvested.Cryosections and cells on cytospins and chamber slides wereair-dried for at least 4 h and stored frozen. Before im-munostaining, tissue and cells were fixed in acetone for 4 min.Slides were immunolabeled using the UltraVision LP Detec-tion System (Thermo Scientific, Fremont, CA) according to themanufacturer’s instructions. The following antibodies werediluted in antibody diluent (Dako) and applied for 30 min atroom temperature: vWF (immunoglobulin fraction, rabbitanti-human, 0.7mg/mL, Dako), vascular endothelial-cadherin(VE-cadherin; clone F-8, 0.33mg/mL, mouse IgG1; Santa CruzBiotechnology, Santa Cruz, CA), and VEGF receptor-2(VEGFR-2; clone FLT-19, 10mg/mL, mouse IgG1; Sigma-Aldrich). IgG controls and normal rabbit immunoglobulinfraction control (Dako) for vWF were used in the same con-centrations as the respective antibodies. After 3 washing stepsin PBS, slides were incubated with primary antibody enhancerfor 10 min, followed by horseradish peroxidase-polymer for15 min. The slides were washed again thrice in PBS, and im-munolabeling was visualized by a 5 min exposure to 3-amino-9-ethylcarbacole (all from UltraVision kit, Thermo Scientific).The slides were counterstained with Mayer’s hematoxylin(Merck, Darmstadt, Germany), washed in distilled water, andmounted with Kaiser’s glycerol gelatin (Merck).

Proliferation of hAMSC

hAMSC were either cultured under standard conditions inDMEM supplemented with 15% FBS on gelatin coated platesor under different endothelial induction conditions, consist-ing of culture in EGM-2 in the absence or presence of 50 ng/mL VEGF (VEGF165; ReliaTech, Wolfenbuettel, Germany), ongelatin, or fibronectin (1mg/cm2; R&D Systems, Minneapolis,MN) coated plates. Cells were seeded in triplicate with adensity of 104 cells/well in 6-well plates and harvested after5 days. Cells were counted by automatic cell counting(Casy� Model TT; Scharfe Systems, Reutlingen, Germany),reseeded in a density of 104 cells/well, and again harvestedafter 4 d. Cumulative population doublings were calculatedaccording to the formula (lnN - lnN0)/ln2, with N = amountof harvested cells and N0 = amount of seeded cells. For thedetermination of the cumulative population doublings overseveral passages, cells were harvested at 90% confluence.

DiI-Ac-LDL-uptake assay

hAMSC were seeded on gelatin-coated glass chamberslides and cultured in DMEM with 15% FBS (non-induced

hAMSC) or in EGM-2 (induced hAMSC) for at least 7 days.Subsequently, cells were incubated with 10mg/mL acetylatedlow-density lipoprotein labeled with 1,1¢-dioctadecyl-3,3,3¢,3¢-tetramethylindo-carbocyanine perchlorate (DiI-Ac-LDL; BTIBiomedical Technologies, Stoughton, MA) for 4 h accordingto the manufacturer’s instructions. Cells were fixed with 4%paraformaldehyde, stained with DAPI (1:2000; Invitrogen,Eugene, OR), and observed with a Leica DM600B fluorescentmicroscope (Leica, Wetzlar, Germany) connected to anOlympus DP72 digital camera (Olympus, Tokyo, Japan).Placental endothelial cells (PlEC) served as positive controls.

Matrigel assay

hAMSC were cultured under standard (non-induced) orangiogenic (induced) conditions for at least 10 days. Thencells were seeded in a 96-well plate precoated with 40mLMatrigel (BD Biosciences) according to the manufacturer’sinstructions at a density of 104 cells/well in 100mL EGM-2.Cells were observed using a Cell-IQ Analyzer 2004-01 (Chip-Man Technologies, Tampere, Finland). PlEC served as posi-tive controls. Videos were prepared with Cell-IQ AnalyzerPro-Write v.AN 2.0.1 (Chip-Man Technologies).

Design of microarray experimentsand RNA isolation

hAMSC were isolated from 3 different placentas, and ali-quots were cultured either in DMEM + 15% FCS (non-inducedhAMSC) or EGM-2 (induced hAMSC) on gelatin-coatedculture flasks until cells reached 90% confluence (about 5–8days). Then, the cells were reseeded and cultured in the re-spective media for 14 days before RNA isolation. Total RNAwas isolated using RNeasy Mini Kit (Qiagen, Hilden, Ger-many). The integrity of each RNA sample was determinedusing an Agilent 2100 Bioanalyzer (Agilent, Foster City, CA),and only RNA samples with integrity values of 9.5–10 wereused for hybridization.

Hybridization and data analysis of microarrays

Total RNA was labeled using the Affymetrix GeneChip�

Whole Transcript Sense Target Labeling Assay and hybrid-ized to GeneChip Human 1.0 ST arrays as described by themanufacturer (Affymetrix, Santa Clara, CA). Hybridizationswere carried out at the Molecular Biology Core Facility at theCenter of Medical Research at the Medical University ofGraz. Briefly, 100 ng of total RNA were reverse transcribed tocDNA using random hexamers tagged with a T7 promotersequence. Double-stranded cDNA was subsequently used asa template in an in vitro transcription reaction followed bycDNA synthesis, fragmentation, and labeling through a ter-minal deoxynucleotidyl transferase. The hybridizationcocktail was incubated overnight at 45�C while rotating in ahybridization oven. After 16 h of hybridization, arrays werewashed and stained in an Affymetrix GeneChip fluidicsstation 450, according to the Affymetrix-recommended pro-tocol. Arrays were scanned on an Affymetrix GeneChipscanner. CEL files were imported into Partek Genomic Suitev6.4 software (Partek, Inc., St Louis, MO) and robust multi-chip average normalized (including background correction,quantile normalization across all arrays, median polished

ANGIOGENIC PROPERTIES OF MESENCHYMAL STROMAL CELLS 1311

summarization based on log transformed expression values).For statistical analysis, a paired-sample t-test was performedbetween the treatment groups. Differentially expressedgenes were selected by P < 0.005 (heat map, Tables 1 and 2) orP < 0.05 (Venn diagram) and a fold change (FC) ‡ 2. Ex-pression of genes above the background was determined bya signal intensity level > 5 after background correction.

The according data has been deposited in NCBI’s GeneExpression Omnibus (GEO) [28] and is accessible throughGEO Series accession number GSE28385 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE28385).

Protein isolation and angiogenic proteinarray analysis

hAMSC were isolated from 5 different placentas and ei-ther cultured in DMEM with 15% FBS (non-induced hAMSC)or in EGM-2 (induced hAMSC) on gelatin-coated cultureflasks until cells reached 90% confluence (about 5–8 days).Then, the cells were reseeded and cultured in the respectivemedia for 14 days. Before protein isolation, cells were wa-shed twice in PBS and lysed with RIPA-buffer (Sigma-Aldrich) containing 4% complete protease inhibitor cocktail(Roche, Mannheim, Germany) for 5 min. Lysates were clari-fied by centrifugation at 300 g for 10 min at 4�C. Supernatantwas used immediately or stored at - 80�C for further anal-ysis. Total protein concentration was determined by Lowryprotein assay. Protein of 5 hAMSC isolations was pooled,and a total of 250 mg was applied to the Human Angiogen-esis Antibody Array C1000 (RayBiotech, Norcross, GA). Thisarray contains 43 different angiogenic proteins spotted induplicates onto 2 membranes. Membranes were processedaccording to the manufacturer’s instructions. Chemilumi-nescent imaging was performed using the FluorChemQsystem, signal densities were analyzed with AlphaView

software version 2.0.1.1 (both from AlphaInnotech, CellBiosciences, Santa Clara, CA), and ratios of the respectiveprotein and internal standard densities were determined.Protein expressions by induced hAMSC are presented aspercentages of the expression of non-induced cells (setto 100%).

Preparation and analysis of hAMSC-conditionedmedia

To prepare hAMSC-conditioned medium (CdM), conflu-ent non-induced and induced hAMSC were washed withPBS and then incubated with EGM-2 for 48 h. Control me-dium (EGM-2) was prepared in parallel in culture flaskswithout cells. On harvest, hAMSC-CdM and control mediumwere centrifuged at 300 g for 10 min and then stored at -80�C.

For determining the effect of hAMSC-CdM on endothelialcell viability, lactate dehydrogenase (LDH) activity was an-alyzed in supernatants of PlEC that were cultured in 96-wellplates for 96 h either in CdM or in the respective controlmedium. LDH activity was measured using an LDH Cyto-toxicity Detection Kit (Takara Bio, Inc., Shiga, Japan) accord-ing to the manufacturer’s instructions. Experiments wereperformed with non-induced and induced hAMSC from 2and 4 different isolations, respectively, using triplicates perexperiment. Statistical analysis was performed using pairedStudent’s t-test. Data were expressed as mean – standarddeviation. P values of < 0.001 were considered statisticallysignificant.

To test the effect of hAMSC-CdM on network formationof endothelial cells, PlEC were resuspended in hAMSC-CdMor in control medium EGM-2 and cultured on GrowthFactor Reduced Matrigel (BD Biosciences) according to themanufacturer’s instructions. Cells were observed using a

FIG. 1. Expression of endo-thelial markers. hAMSC donot express vWF or VE-cad-herin in situ (tissue sections)and in vitro (cells in culture).However, a subpopulationshows expression of VEGFR-2in situ (see arrows) and in vitro(see inset for higher magnifi-cation). PlEC served as posi-tive controls. Scale bar: 50mm.hAMSC, human amnion-de-rived mesenchymal stromalcells; vWF, von Willebrandfactor; VE-cadherin, vascularendothelial-cadherin; VEGFR-2, vascular endothelial growthfactor receptor; PlEC, placen-tal endothelial cells. Colorimages available online atwww.liebertonline.com/scd

1312 KONIG ET AL.

Cell-IQ Analyzer 2004-01, and videos were prepared withCell-IQ Analyzer Pro-Write v.AN 2.0.1 (both from Chip-ManTechnologies).

Results

Immunophenotypic characterization of hAMSC

Flow cytometry revealed the expression of common MSCsurface markers. hAMSC were positive for CD90, CD73,CD105, and CD44 while being negative for CD14, CD45,HLA-DR, and CD34 (Supplementary Fig. S1; SupplementaryData are available online at www.liebertonline.com/scd).

Expression of endothelial markers by hAMSC in situand in vitro

hAMSC did not express the mature endothelial cellmarkers vWF and VE-cadherin in situ (tissue sections), di-rectly after isolation (cytospins), or in vitro (cells in culture),which confirms the amnion to be an avascular tissue. How-ever, a subpopulation of hAMSC expressed the endothelialprecursor cell marker VEGFR-2. PlEC served as positivecontrols (Fig. 1).

Proliferation of hAMSC upon endothelial induction

We tested the effect of different endothelial culture con-ditions on the proliferation of hAMSC to evaluate optimalculture conditions. Non-induced hAMSC (cultured inDMEM + 15% FBS on gelatin) showed cumulative populationdoublings of 2.5 – 0.1 after 5 days and 2.9 – 0.8 after 9 days.Endothelial culture conditions highly increased the prolifer-ation potential of hAMSC. The mean cumulative populationdoublings under endothelial conditions were 7.3 – 0.2 and13.1 – 0.3 after 5 and 9 days, respectively. There were nodistinct effects among the different endothelial conditions

consisting of culture on fibronectin or gelatin and in thepresence or absence of 50 ng/mL VEGF in EGM-2 (Fig. 2).For further studies on endothelial induction, the use ofgelatin-coating and EGM-2 medium without additionalVEGF was chosen. Supplementary Fig. S2 shows that underthese conditions the cumulative population doublings ofhAMSC were clearly higher over several passages comparedwith non-induced hAMSC.

Change of morphology and uptake of DiI-Ac-LDLupon endothelial induction

Non-induced hAMSC showed a fibroblast-like morphology(Fig. 3A). After culturing the cells in EGM-2 for a minimum of5 days, they changed their morphology to a cobblestone-likephenotype (Fig. 3B), similar to PlEC (Fig. 3C).

The uptake of DiI-Ac-LDL is specific for endothelial cellsand macrophages and occurs via a scavenger receptor. Non-induced hAMSC did not take up DiI-Ac-LDL (Fig. 3D). Afterinduction with EGM-2, hAMSC internalized DiI-Ac-LDLwith varying intensity (Fig. 3E). PlEC served as positivecontrols (Fig. 3F).

Expression of mature endothelial markers uponendothelial induction

Endothelial culture conditions did not result in the ap-pearance of vWF or VE-cadherin positive cells, even whenEGM-2 was supplemented with concentrations of VEGF ofup to 100 ng/mL and cells were grown on fibronectin. Fur-ther, neither extension of the induction period to 3 weeks norculturing the cells under low oxygen concentrations (2%)could induce expression of vWF or VE-cadherin genes orproteins (data not shown).

Gene expression changes upon endothelial induction

To identify the effect of endothelial culture on gene ex-pression changes of hAMSC, a microarray was performed.Of the 28,870 transcripts analyzed, 16,761 genes were ex-pressed (signal level > 5 after background correction). Afterfiltering with P < 0.005, 200 genes were found to be differ-entially regulated with a FC > 2, of which 92 were upregu-lated and 108 downregulated on endothelial induction (Fig.4A). Genes were ranked by FC and screened for angiogenicfunctions. The first 25 genes that were up- and down-regulated under angiogenic conditions are shown in Tables 1and 2. Selected genes of interest are displayed in a Venndiagram (Fig. 4B). The according data has been deposited inNCBI’s GEO and is accessible through GEO Series accessionnumber GSE28385 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE28385).

Interestingly, induced hAMSC significantly down-regulated typical pro-angiogenic genes such as tenascin C(FC - 27.9), Tie-2 ( - 16.8), VEGF-A ( - 5.7), CD146 ( - 3.8), andFGF-2 ( - 2.2, P < 0.05), whereas they upregulated genes withanti-angiogenic functions such as serpin peptidase inhibitorF1 (serpin F1, FC 31.4, P < 0.01), the FGF-2 signaling antag-onist sprouty1 (FC 10.6), and angioarrestin [angiopoietin(ANGPT)-like 1, 8.2]. The only factor with a possible pro-angiogenic function that we found to be upregulated wasplatelet-derived growth factor-D (PDGF-D, 26.6).

FIG. 2. Cumulative population doublings of hAMSC. Cellswere either cultured in DMEM + 15% FBS on gelatin or underdifferent endothelial conditions (EGM-2 – 50 ng/mL VEGFon gelatin or fibronectin coating). EGM-2, endothelial growthmedium-2; DMEM, Dulbecco’s modified Eagle’s medium;FBS, fetal bovine serum.

ANGIOGENIC PROPERTIES OF MESENCHYMAL STROMAL CELLS 1313

The genes of the MSC markers CD90, CD73, and CD105were expressed equally by non-induced and inducedhAMSC (FC > -2 and < 2, signal level > 5). In addition, bothexpressed VEGF-B and -C, placental growth factor (PIGF),VEGFR-1 and 2, angiopoietin-1 (see Fig. 4B), neuropilin 1and 2, tissue inhibitors of metalloproteinases 1 and 2 (TIMP1& 2), and the pericyte markers NG2 and PDGF receptor-b(PDGFR-b).

Expression of angiogenic proteins by hAMSCupon endothelial induction

hAMSC were analyzed for the presence of angiogenic pro-teins using the Human Angiogenesis Antibody Array C1000(Fig. 5). Non-induced cells expressed high levels of the pro-angiogenic factor FGF-2. After endothelial induction, its ex-pression decreased to 27% – 8% compared with non-inducedcells. Additional angiogenic factors that were downregulatedinclude interleukin-8 (IL-8) (21% – 5%), matrix metallopro-tease-1 (MMP-1) (31% – 1%), Tie-2 (13% – 1%), and urokinase-type plasminogen activator receptor (UPAR) (23% – 1%). Onthe contrary, the expression of the anti-angiogenic protein

endostatin increased to 226% – 4% as a result of angiogenicinduction. In addition, epidermal growth factor (EGF) wasupregulated to 236% – 99%.

Angiogenesis assay–network formation

The Matrigel assay is a commonly used method to eval-uate network formation by endothelial cells and was appliedto investigate whether induced hAMSC were also able toform networks.

FIG. 3. Morphology andDiI-Ac-LDL uptake by non-induced and induced hAMSCcompared with PlEC. Non-induced hAMSC show a fi-broblast-like morphology (A).After endothelial inductionfor a minimum of 5 days, theychange their morphology to acobblestone-like phenotype(B), similar to PlEC (C). Non-induced hAMSC do not takeup DiI-Ac-LDL (D). InducedhAMSC show uptake of DiI-Ac-LDL with a varying in-tensity (E). PlEC served as positive control (F). Scale bar: 100mm. DiI-Ac-LDL, 1,1¢-dioctadecyl-3,3,3¢,3¢ tetramethylindocarbo-cyanine perchlorate labeled acetylated low density lipoprotein. Color images available online at www.liebertonline.com/scd

FIG. 4. Gene expression changes on endothelial induction.(A) Heat map showing the differential gene expression ofinduced versus non-induced hAMSC (FC > 2, P < 0.005). Redcolor stands for high signal intensity and blue for low signalintensity. Cell preparations were obtained from 3 placentas(1, 2, 3). (B) Venn diagram showing the up- and down-regulation of angiogenesis-associated genes after endothelialinduction. Genes listed in the overlapping area (equally ex-pressed) are expressed with a FC > - 2 and < 2 and a signallevel > 5 after background correction. Up- and downregulatedgenes are expressed with a FC > 2 (upregulated) or < - 2(downregulated). P < 0.05. The according data has been depos-ited in NCBI’s Gene Expression Omnibus and is have accessiblethrough GEO Series accession number GSE28385 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE28385). FC, foldchange. Color images available online at www.liebertonline.com/scd

‰

1314 KONIG ET AL.

Table 1. Upregulated Genes Under Angiogenic Conditions

Gene Symbol FC P value Known angiogenic function

Dermatopontin (TRAMP) DPT 78.1 0.0010 —H19, imprinted maternally expressed transcript (nonprotein) H19 62.4 0.0036 —Prolactin receptor PRLR 53.1 0.0033 —DDelta-like 1 homolog (Drosophila) DLK1 50.2 0.0048 —FK506 binding protein 5 FKBP5 36.1 0.0011 —Absent in melanoma 1 AIM1 30.8 0.0010 —Monoamine oxidase A MAOA 29.9 0.0004 —Adenomatosis polyposis coli downregulated 1 APCDD1 27.3 0.0044 —Platelet derived growth factor D PDGFD 26.6 0.0047 Pro-angiogenic [53]Signal peptide, CUB domain, EGF-like 2 SCUBE2 17.7 0.0035 —Interleukin 1 receptor, type I IL1R1 16.4 0.0025 —LIM domain only 3 (rhombotin-like 2) LMO3 16.1 0.0009 —Interleukin-1 receptor-associated kinase 3 IRAK3 15.8 0.0046 —Serum amyloid A1 SAA1 14.4 0.0028 —Laminin, gamma 3 LAMC3 11.1 0.0025 —Family with sequence similarity 180, member A FAM180A 10.6 0.0027 —Sprouty 1 SPRY1 10.6 0.0019 Anti-angiogenic [45]Interferon-induced protein 44-like IF144L 10.3 0.0017 —Zinc finger and BTB domain containing 16 ZBZB16 10.1 0.0031 —Nidogen 1 nidogen1 10.0 0.0003 —BTB (POZ) domain containing 3 BTBD3 9.2 0.0008 —Phosphoinositide-3-kinase, regulatory subunit 1 (alpha) PIK3R1 8.9 0.0016 —Prostaglandin E receptor 2 (subtype EP2) PTGER2 8.7 0.0039 —Interleukin 1 receptor, type II IL1R2 8.6 0.0012 —Angioarrestin/Angiopoietin-like 1 ANGPTL1 8.2 0.0040 Anti-angiogenic [43]

FC, fold change.

Table 2. Downregulated Genes Under Angiogenic Conditions

Gene Symbol FC P valueKnown angiogenic

function

Carboxypeptidase A4 CPA4 - 34.3 0.0029 —Tenascin C TNC - 27.9 0.0029 Pro-angiogenic [37]Claudin 11 CLDN11 - 25.2 0.0028 —Microfibrillar associated protein 5 MFAP5 - 25.1 0.0044 Pro-angiogenic [54]Kallmann syndrome 1 sequence KAL1 - 19.9 0.0021 —KIAA1199 KIAA1199 - 18.2 0.0043 —Tie-2/TEK tyrosine kinase, endothelial TEK - 16.8 0.0048 Pro-angiogenic [39]Adenylate cyclase-associated protein, 2 (yeast) CAP2 - 13.3 0.0046 —Retinol dehydrogenase 10 (all-trans) RDH10 - 12.9 0.0039 —Bone marrow stromal cell antigen 1 BST1 - 12.7 0.0001 —V-kit Hardy-Zuckerman 4 feline sarcoma viral

oncogene homologKIT - 11.6 0.0006 —

Secretogranin II (chromogranin C) SCG2 - 11.6 0.0020 Its derivatesecretoneurin:pro-angiogenic [55]

Heparin-binding EGF-like growth factor HBEGF - 10.6 0.0011 Pro-angiogenic [56]Lysophosphatidylcholine acyltransferase 2 LPCAT2 - 8.8 0.0022 —ATPase type 13A3 ATP13A3 - 7.3 0.0001 —NCK-associated protein 5 NCKAP5 - 7.0 0.0040 —Cysteine rich transmembrane BMP regulator 1

(chordin-like)CRIM1 - 6.9 0.0032 —

Sorting nexin 25 SNX25 - 6.9 0.0001 —Integrin, alpha 2 (CD49B, alpha 2 subunit of VLA-2

receptor)ITGA2 - 6.8 0.0026 —

Glutamate receptor, ionotropic, kainate 2 GRIK2 - 6.5 0.0018 —LOC652811//similar to adlican LOC652811 - 5.8 0.0040 —Keratin 19 KRT19 - 5.7 0.0024 —Vascular endothelial growth factor A VEGFA - 5.7 0.0043 Pro-angiogenic [40]CD9 molecule CD9 - 5.2 0.0000 maybe pro-angiogenic

[57]Prostaglandin F2 receptor negative regulator PTGFRN - 5.2 0.0045 —

FC, fold change.

1315

After culture under standard (non-induced) or endothelial(induced) conditions for at least 10 days, hAMSC were see-ded on Matrigel in EGM-2. Non-induced hAMSC formednetwork-like structures within 6 h; however, these networksdisintegrated fast (Fig. 6A-D, see also Supplementary VideoS1). Induced hAMSC (Fig. 6E–H, Supplementary Video S2)formed networks similar to PlEC (Fig. 6I–L, SupplementaryVideo S3). While the endothelial networks remained static,networks of induced hAMSC became more wide-spreadover time. In addition, branches formed by PlEC showedsigns of degeneration (dark cell clusters in Fig. 6K, L) alreadyafter 24 h, whereas networks formed by induced hAMSCremained viable for about 48 h. However, cells within thesenetworks remained negative for vWF (Fig. 6M, N). PlECserved as positive control (Fig. 6O, P).

Effect of hAMSC-CdM on endothelial cells

LDH is an enzyme that is released into the culture me-dium by damaged cells [29]. When PlEC were cultured in thepresence of CdM collected from induced hAMSC, the ac-tivity of LDH in the culture supernatants was significantlyreduced to 48.7% – 14.9% of control, thus, endothelial cellviability was clearly enhanced (Fig. 7A). The addition of in-duced hAMSC-CdM to PlEC supported the formation ofnetwork-like structures in the Matrigel assay. While net-works formed by endothelial cells in control medium haddisintegrated after 72 h (Fig. 7B, see also Supplementary Vi-deo S4), networks formed by cells cultured with inducedhAMSC-CdM were still stable (Fig. 7C, see also Supple-mentary Video S5). Under the same conditions, CdMcollected from non-induced hAMSC neither enhanced en-dothelial cell viability (Fig. 7D) nor supported endothelialnetwork formation (Fig. 7E, F).

Discussion

In this study, we investigated the endothelial differentia-tion potential of MSC isolated from the amnionic membraneof human term placentas. We could show that the amnionicmesenchyme is free of mature endothelial cells as docu-mented by the absent staining with the classical endothelialmarkers vWF and VE-cadherin. Interestingly, a subpopula-tion of hAMSC is positive for VEGFR-2. Together with itsligand VEGF, VEGFR-2 plays an important role during earlyplacental and embryonic vascular development. Its expres-sion has been documented in vasculogenic and angiogenicprecursor cells found in placental villi [30]. Mice deficient inVEGFR-2 (VEGFR-2-/-) died in utero as a result of an earlydefect in the development of hematopoietic and endothelialcells [31]. Therefore, the presence of VEFGR-2 in theamnionic mesenchyme suggests an endothelial progenitorpotential of hAMSC.

Careful immunocytochemical analysis for the expressionof vWF and VE-cadherin on freshly isolated cells was per-formed to exclude a contamination with endothelial cellsbefore endothelial induction. All isolations in this study werefree of endothelial cells.

On endothelial induction with EGM-2 containing 2% FCS,EGF, hydrocortisone, VEGF, FGF-2, and IGF, hAMSCshowed some endothelial-like characteristics: They changedtheir fibroblast-like to a more endothelial cell-like morphol-ogy and acquired the ability to take up Ac-LDL. In addition,only induced hAMSC were able to form long-lasting net-works similar to endothelial cells in the Matrigel assay. Non-induced hAMSC cultured in DMEM with 15% FBS for atleast ten days initially also formed network-like structures inMatrigel. This could be explained by the fact that before theapplication in the Matrigel assay, these cells were resuspendedin EGM-2. Therefore, the network formation may be due to ashort-term stimulatory effect of abundant growth factorspresent in both the Matrigel and in EGM-2. However, thesenetworks were very unstable and disintegrated within 6 h.

Even though hAMSC were responsive to the angiogenicfactors present in EGM-2, they did not differentiate intomature endothelial cells. Induced hAMSC still expressed thecommon MSC markers CD90, CD73, and CD105. None of theangiogenic conditions used by previous studies [22,23] or

FIG. 5. Expression of angiogenic proteins. (A) HumanAngiogenesis Antibody Array. Protein of non-induced andinduced hAMSC from 5 different placentas was pooled andincubated with 2 membranes, onto which 43 angiogenicproteins are blotted in duplicates. The rectangles highlightproteins of interest. (B) Arrays were analyzed by densito-metry and normalized to the internal positive control. Proteinexpressions by induced hAMSC are shown as percentages ofprotein expression by non-induced cells (set to 100%). Colorimages available online at www.liebertonline.com/scd

1316 KONIG ET AL.

addition of high concentrations of VEGF (100 ng/mL) and theuse of fibronectin, an extracellular matrix protein known topromote endothelial differentiation [32], led to an expressionof vWF or VE-cadherin (genes or proteins). Even cells form-ing the network-like structures in the Matrigel assay re-mained negative for vWF. Also, endothelial induction under

2% oxygen did not promote endothelial differentiation, al-though low oxygen conditions are known to induce angio-genesis [33] and have a pro-angiogenic effect on MSC [34].

So far, only one other study has investigated the endothelialdifferentiation potential of hAMSC [35]. This study reportedthat culture in DMEM supplemented with 2% FBS and 50 ng/mL

FIG. 6. Network formationon Matrigel. hAMSC werecultured under standard(non-induced, A–D) or endo-thelial (induced, E–H) condi-tions for at least 10 days.Then, they were seeded onMatrigel in EGM-2. Pictureswere taken at different timepoints (6, 12, 24, 48 h). PlECserved as positive control, (I–L). For respective video files,see the Supplementary Data.Networks of induced hAMSCcultured on Matrigel for 24 hdo not express vWF (M)fluorescence, (N) phase con-trast. PlEC served as positivecontrol (O) fluorescence, (P)phase contrast. Scale bar:100 mm. Color images avail-able online at www.liebertonline.com/scd

FIG. 7. Effect of hAMSC-CdMon EC. Induced hAMSC-CdMsignificantly reduces LDH activ-ity in PlEC supernatants, shownas percentage of control after96 h of culture (A). n = 4.***P < 0.001. After 72 h on Ma-trigel, networks formed by PlECin control medium have alreadydisintegrated (B), whereas net-works formed by cells culturedwith CdM from inducedhAMSC are still stable (C). Un-der the same conditions,hAMSC-CdM collected fromnon-induced hAMSC neitherenhanced EC viability (D) norsupported endothelial networkformation (E, F). n = 2. n.s.: notsignificant. For respective videofiles see the SupplementaryData. CdM, conditioned medi-um; EC, endothelial cells.

ANGIOGENIC PROPERTIES OF MESENCHYMAL STROMAL CELLS 1317

VEGF led to a slightly increased expression of VEGFR-2 aswell as appearance of vWF-positive cells. The discrepanciesbetween the results of this previous study and our data mightbe due to differences in cell isolation protocols. In contrast toour findings, in this former study, non-induced hAMSC alsoformed stable networks in Matrigel. Thus, maybe a subset ofendothelial-like cells was present within the primary isolatedcell population, possibly caused by an incomplete separationof the amnion from the underlying vascularized chorion. Thishypothesis is supported by our Matrigel data, which clearlyshowed that non-induced hAMSC cultures devoid of endo-thelial cells did not generate stable networks.

We assume that hAMSC resist a differentiation into ma-ture endothelial cells which would be in accordance with thefact that amnionic membrane is one of the few avasculartissues that maintains its avascularity even though it is nextto a highly vascularized tissue (the chorion) [36].

Our microarray analysis did not reveal upregulation ofendothelial-specific genes in angiogenic-induced hAMSC.Under standard conditions, hAMSC express a variety ofpro-angiogenic genes. Tenascin C, a large extracellular gly-coprotein, has been shown to promote endothelial cellelongation and sprouting in bovine aortic endothelial cellsand human umbilical vein endothelial cells [37,38]. Tie-2 (orTEK) is a tyrosine-kinase transmembrane receptor that ispredominately expressed by endothelial cells. Its ligands arethe ANGPT family members ANGPT 1, 2, and 4, which aresecreted proteins with different functions in angiogenesis[39]. Both VEGF-A and FGF-2 are very potent inducers ofangiogenesis [40,41]. However, these pro-angiogenic geneswere downregulated in endothelial culture conditions, thatis, in the presence of abundant angiogenic growth factors.Instead, genes with anti-angiogenic functions were upregu-lated: Serpin F1, angioarrestin, and sprouty1. Serpin F1, orpigment epithelium-derived factor, is one of the most effec-tive natural angiogenesis inhibitors [42]. Angioarrestin, alsoknown as ANGPT-like 1, and Sprouty1 have been shown toinhibit different angiogenic processes [43–45].

The gene expression results were confirmed on the proteinlevel using an angiogenic protein antibody array. Again,FGF-2 and Tie-2 were downregulated on endothelial induc-tion, whereas the anti-angiogenic protein endostatin wasclearly upregulated compared with non-induced controls.The expressions of VEGF-A and angioarrestin in this arraywere too low compared with the internal positive control toallow reliable quantification.

It seems as if hAMSC use this upregulation of anti-angiogenic and concomitant downregulation of pro-angiogenicgenes and proteins as an autoregulatory mechanism to pro-tect themselves against a differentiation into mature endo-thelial cells. However, they do not adopt anti-angiogenicproperties toward endothelial cells. On the contrary, CdMcollected from induced hAMSC even had a positive effect onendothelial cells as shown by enhanced viability and stabi-lized network formation.

These results are consistent with studies showing thatMSC from bone marrow promote angiogenesis and supportblood vessel formation [46–51]. Au et al. could demonstratethat MSC stabilized engineered blood vessels and kept themfunctional for 130 days in vivo. The authors showed thatMSC did not differentiate into endothelial cells but insteadacted as pericyte-like cells [52].

Therefore, we assume that even though hAMSC are notideal as a substitute for endothelial cells, they might bevaluable in a variety of cell-therapeutic or tissue-engineeringapproaches where they can promote the survival of endo-thelial cells, the stabilization of pre-existing vessels, and therevascularization of ischemic tissues. A concrete examplewould be the endothelialization of vascular grafts. Here,PlEC could be used for the seeding of the luminal surface. Aco-application of hAMSC as supporting stromal or muralcells might enhance the viability of the endothelial cells and,therefore, the patency of the vascular graft.

From our results, we conclude that an endothelial induc-tion of hAMSC is advantageous for an application in thera-peutic treatments, as it promotes a survival-enhancing effecton endothelial cells. Further, careful analyses of primary cellisolations are mandatory to exclude misleading results dueto a contamination with endothelial cells.

In future studies, it needs to be determined how a co-application of PlEC and hAMSC could be used in vasculartherapies. Banking of both cell types may provide a conve-nient source for autologous therapy and for matching re-cipients with histocompatible donors.

Acknowledgments

The authors thank the research nurses Bettina Amtmann,Sandra Eppich, and Petra Wagner of the Clinic of Obstetricsand Gynecology for placenta collection, and Kerstin Hingerl,Rudolf Schmied, and Monika Siwetz from the Institute ofCell Biology, Histology, and Embryology, Medical Uni-versity of Graz, Austria, for their valuable technical assis-tance and expertise.

This work was supported by the Franz-Lanyar-Foundation(Projects # 339 and # 349) to I.L. J.D.F. and J.K. were fundedby the Medical University of Graz within the Ph.D. programMolecular Medicine.

Disclosure Statement

No competing financial interests exist.

References

1. Pittenger MF, AM Mackay, SC Beck, RK Jaiswal, R Douglas,JD Mosca, MA Moorman, DW Simonetti, S Craig and DRMarshak. (1999). Multilineage potential of adult humanmesenchymal stem cells. Science 284:143–147.

2. Horwitz EM, PL Gordon, WKK Koo, JC Marx, MD Neel, RYMcNall, L Muul and T Hofmann. (2002). Isolated allogeneicbone marrow-derived mesenchymal cells engraft and stim-ulate growth in children with osteogenesis imperfecta: Im-plications for cell therapy of bone. Proc Natl Acad of SciUSA 99:8932–8937.

3. Le Blanc K, F Frassoni, L Ball, F Locatelli, H Roelofs, I Lewis,E Lanino, B Sundberg, ME Bernardo, et al. (2008). Me-senchymal stem cells for treatment of steroid-resistant, se-vere, acute graft-versus-host disease: a phase II study.Lancet 371:1579–1586.

4. Stenderup K, J Justesen, C Clausen and M Kassem. (2003).Aging is associated with decreased maximal life span andaccelerated senescence of bone marrow stromal cells. Bone33:919–926.

5. Stolzing A and A Scutt. (2006). Age-related impairment ofmesenchymal progenitor cell function. Aging Cell 5:213–224.

1318 KONIG ET AL.

6. Hsieh JY, YS Fu, SJ Chang, YH Tsuang and HW Wang.(2010). Functional module analysis reveals differential oste-ogenic and stemness potentials in human mesenchymalstem cells from bone marrow and Wharton’s jelly of um-bilical cord. Stem Cells Dev 19:1895–1910.

7. Kita K, GG Gauglitz, TT Phan, DN Herndon and MGJeschke. (2010). Isolation and characterization of mesenchy-mal stem cells from the sub-amniotic human umbilical cordlining membrane. Stem Cells Dev 19:491–502.

8. Erices A, P Conget and JJ Minguell. (2000). Mesenchymalprogenitor cells in human umbilical cord blood. Br J Hae-matol 109:235–242.

9. Barlow S, G Brooke, K Chatterjee, G Price, R Pelekanos, TRossetti, M Doody, D Venter, S Pain, K Gilshenan and KAtkinson. (2008). Comparison of human placenta- and bonemarrow-derived multipotent mesenchymal stem cells. StemCells Dev 17:1095–1107.

10. Parolini O, F Alviano, GP Bagnara, G Bilic, HJ Buhring, MEvangelista, S Hennerbichler, B Liu, M Magatti, et al. (2008).Concise review: isolation and characterization of cells fromhuman term placenta: outcome of the first internationalWorkshop on Placenta Derived Stem Cells. Stem Cells26:300–311.

11. Parolini O, F Alviano, I Bergwerf, D Boraschi, C De Bari, PDe Waele, M Dominici, M Evangelista, W Falk, et al. (2010).Toward cell therapy using placenta-derived cells: diseasemechanisms, cell biology, preclinical studies, and regulatoryaspects at the round table. Stem Cells Dev 19:143–154.

12. Davis JW. (1910). Skin transplantation with a review of 550cases at the Johns Hopkins Hospital. Johns Hopkins Med J.15:307–396.

13. Dua HS, JA Gomes, AJ King and VS Maharajan. (2004). Theamniotic membrane in ophthalmology. Surv Ophthalmol49:51–77.

14. Magatti M, S De Munari, E Vertua, L Gibelli, GS Wenglerand O Parolini. (2008). Human amnion mesenchyme harborscells with allogeneic T-cell suppression and stimulationcapabilities. Stem Cells 26:182–192.

15. Wolbank S, A Peterbauer, M Fahrner, S Hennerbichler, Mvan Griensven, G Stadler, H Redl and C Gabriel. (2007).Dose-dependent immunomodulatory effect of human stemcells from amniotic membrane: a comparison with humanmesenchymal stem cells from adipose tissue. Tissue Eng13:1173–1183.

16. Chang YJ, SM Hwang, CP Tseng, FC Cheng, SH Huang, LFHsu, LW Hsu and MS Tsai. (2010). Isolation of mesenchy-mal stem cells with neurogenic potential from the meso-derm of the amniotic membrane. Cells Tissues Organs 192:93–105.

17. Magatti M, S De Munari, E Vertua, C Nassauto, A Albertini,GS Wengler and O Parolini. (2009). Amniotic mesenchymaltissue cells inhibit dendritic cell differentiation of peripheralblood and amnion resident monocytes. Cell Transplant18:899–914.

18. Bailo M, M Soncini, E Vertua, PB Signoroni, S Sanzone, GLombardi, D Arienti, F Calamani, D Zatti, et al. (2004). En-graftment potential of human amnion and chorion cellsderived from term placenta. Transplantation 78:1439–1448.

19. Sant’anna LB, A Cargnoni, L Ressel, G Vanosi and O Par-olini. (2010). Amniotic Membrane application reduces liverfibrosis in a bile duct ligation rat model. Cell Transplant20:441–453.

20. Cargnoni A, M Di Marcello, M Campagnol, C Nassuato, AAlbertini and O Parolini. (2009). Amniotic membrane

patching promotes ischemic rat heart repair. Cell Transplant18:1147–1159.

21. Cargnoni A, L Gibelli, A Tosini, PB Signoroni, C Nassuato,D Arienti, G Lombardi, A Albertini, GS Wengler and OParolini. (2009). Transplantation of allogeneic and xenoge-neic placenta-derived cells reduces bleomycin-induced lungfibrosis. Cell Transplant 18:405–422.

22. Oswald J, S Boxberger, B Jorgensen, S Feldmann, G Ehnin-ger, M Bornhauser and C Werner. (2004). Mesenchymalstem cells can be differentiated into endothelial cells in vitro.Stem Cells 22:377–384.

23. Liu JW, S Dunoyer-Geindre, V Serre-Beinier, G Mai, J-Lambert, RJ Fish, G Pernod, L Buehler, H Bounameaux andEKO Kruithof. (2007). Characterization of endothelial-likecells derived from human mesenchymal stem cells. J ThrombHaemost 5:826–834.

24. Zhang SJ, H Zhang, M Hou, Z Zheng, J Zhou, W Su, Y Weiand S Hu. (2007). Is it possible to obtain "true endothelialprogenitor cells" by in vitro culture of bone marrow mono-nuclear cells? Stem Cells Dev 16:683–690.

25. Fan W, R Crawford and Y Xiao. (2011). The ratio of VEGF/PEDF expression in bone marrow mesenchymal stem cellsregulates neovascularization. Differentiation 81:181–191.

26. Soncini M, E Vertua, L Gibelli, F Zorzi, M Denegri, A Al-bertini, GS Wengler and O Parolini. (2007). Isolation andcharacterization of mesenchymal cells from human fetalmembranes. J Tissue Eng Regen Med 1:296–305.

27. Lang I, A Schweizer, U Hiden, N Ghaffari-Tabrizi, G Ha-gendorfer, M Bilban, MA Pabst, ET Korgun, G Dohr and GDesoye. (2008). Human fetal placental endothelial cells havea mature arterial and a juvenile venous phenotype withadipogenic and osteogenic differentiation potential. Differ-entiation 76:1031–1043.

28. Edgar R, M Domrachev and AE Lash. (2002). Gene Expres-sion Omnibus: NCBI gene expression and hybridizationarray data repository. Nucleic Acids Res 30:207–210.

29. Decker T and M Lohmann-Matthes. (1988). A quick andsimple method for the quantitation of lactate dehydrogenaserelease in measurements of cellular cytotoxicity and tumornecrosis factor (TNF) activity. J Immunol Methods 115:61–69.

30. Demir R, UA Kayisli, Y Seval, C Celik-Ozenci, ET Korgun,AY Demir-Weusten and B Huppertz. (2004). Sequential ex-pression of VEGF and its receptors in human placental villiduring very early pregnancy: differences between placentalvasculogenesis and angiogenesis. Placenta 25:560–572.

31. Shalaby F, J Rossant, TP Yamaguchi, M Gertsenstein, XFWu, ML Breitman and AC Schuh. (1995). Failure of blood-island formation and vasculogenesis in Flk-1-deficient mice.Nature 376:62–66.

32. Wijelath ES, S Rahman, J Murray, Y Patel, G Savidge and MSobel. (2004). Fibronectin promotes VEGF-induced CD34cell differentiation into endothelial cells. J Vasc Surg 39:655–660.

33. Pugh CW and PJ Ratcliffe. (2003). Regulation of angiogen-esis by hypoxia: role of the HIF system. Nat Med 9:677–684.

34. Hu X, SP Yu, JL Fraser, Z Lu, ME Ogle, JA Wang and L Wei.(2008). Transplantation of hypoxia-preconditioned mesen-chymal stem cells improves infarcted heart function via en-hanced survival of implanted cells and angiogenesis. JThorac Cardiovasc Surg 135:799–808.

35. Alviano F, V Fossati, C Marchionni, M Arpinati, L Bonsi, MFranchina, G Lanzoni, S Cantoni, C Cavallini, et al. (2007).Term Amniotic membrane is a high throughput source formultipotent Mesenchymal Stem Cells with the ability to

ANGIOGENIC PROPERTIES OF MESENCHYMAL STROMAL CELLS 1319

differentiate into endothelial cells in vitro. BMC Dev Biol7:11.

36. Hao Y, DH Ma, DG Hwang, WS Kim and F Zhang. (2000).Identification of antiangiogenic and antiinflammatory pro-teins in human amniotic membrane. Cornea 19:348–352.

37. Schenk S, R Chiquet-Ehrismann and EJ Battegay. (1999). Thefibrinogen globe of tenascin-c promotes basic fibroblastgrowth factor-induced endothelial cell elongation. Mol BiolCell 10:2933–2943.

38. Martina E, M Degen, C Ruegg, A Merlo, MM Lino, RChiquet-Ehrismann and F Brellier. (2010). Tenascin-W is aspecific marker of glioma-associated blood vessels andstimulates angiogenesis in vitro. FASEB J 24:778–787.

39. Huang H, A Bhat, G Woodnutt and R Lappe. (2010). Tar-geting the ANGPT-TIE2 pathway in malignancy. Nat RevCancer 10:575–585.

40. Ferrara N, HP Gerber and J LeCouter. (2003). The biology ofVEGF and its receptors. Nat Med 9:669–676.

41. Szebenyi G and JF Fallon. (1998). Fibroblast Growth Factorsas Multifunctional Signaling Factors. Int Rev Cytol. 185:45–106.

42. Dawson DW, OV Volpert, P Gillis, SE Crawford, H Xu, WBenedict and NP Bouck. (1999). Pigment epithelium-derivedfactor: a potent inhibitor of angiogenesis. Science 285:245–248.

43. O’Reilly MS, L Holmgren, Y Shing, C Chen, RA Rosenthal,M Moses, WS Lane, Y Cao, EH Sage and J Folkman. (1994).Angiostatin: a novel angiogenesis inhibitor that mediates thesuppression of metastases by a Lewis lung carcinoma. Cell79:315–328.

44. Impagnatiello MA, S Weitzer, G Gannon, A Compagni, MCotten and G Christofori. (2001). Mammalian sprouty-1 and- 2 are membrane-anchored phosphoprotein inhibitors ofgrowth factor signaling in endothelial cells. J Cell Biol152:1087–1098.

45. Lee S, TM Bui Nguyen, D Kovalenko, N Adhikari, S Grindle,SP Polster, R Friesel, S Ramakrishnan and JL Hall. (2010).Sprouty1 inhibits angiogenesis in association with up-regulation of p21 and p27. Mol Cell Biochem 338:255–261.

46. Chen J, ZG Zhang, Y Li, L Wang, YX Xu, SC Gautam, M Lu,Z Zhu and M Chopp. (2003). Intravenous administration ofhuman bone marrow stromal cells induces angiogenesis inthe ischemic boundary zone after stroke in rats. Circ Res92:692–699.

47. Kinnaird T, E Stabile, MS Burnett, CW Lee, S Barr, S Fuchsand SE Epstein. (2004). Marrow-derived stromal cells ex-press genes encoding a broad spectrum of arteriogenic cy-tokines and promote in vitro and in vivo arteriogenesisthrough paracrine mechanisms. Circ Res 94:678–685.

48. Gruber R, B Kandler, P Holzmann, M Vogele-Kadletz, ULosert, MB Fischer and G Watzek. (2005). Bone marrowstromal cells can provide a local environment that favorsmigration and formation of tubular structures of endothelialcells. Tissue Eng 11:896–903.

49. Hung SC, RR Pochampally, SC Chen, SC Hsu and DJProckop. (2007). Angiogenic effects of human multipotentstromal cell conditioned medium activate the PI3K-Aktpathway in hypoxic endothelial cells to inhibit apoptosis,increase survival, and stimulate angiogenesis. Stem Cells25:2363–2370.

50. Wu Y, L Chen, PG Scott and EE Tredget. (2007). Mesench-ymal stem cells enhance wound healing through differenti-ation and angiogenesis. Stem Cells 25:2648–2659.

51. Verseijden F, SJ Posthumus-van Sluijs, P Pavljasevic, SOHofer, GJ van Osch and E Farrell. (2010). Adult human bonemarrow- and adipose tissue-derived stromal cells supportthe formation of prevascular-like structures from endothelialcells in vitro. Tissue Eng Part A 16:101–114.

52. Au P, J Tam, D Fukumura and RK Jain. (2008). Bonemarrow-derived mesenchymal stem cells facilitate engi-neering of long-lasting functional vasculature. Blood 111:4551–4558.

53. Li H, L Fredriksson, X Li and U Eriksson. (2003). PDGF-D isa potent transforming and angiogenic growth factor. Onco-gene 22:1501–1510.

54. Albig AR, DJ Becenti, TG Roy and WP Schiemann. (2008).Microfibril-associate glycoprotein-2 (MAGP-2) promotesangiogenic cell sprouting by blocking notch signaling inendothelial cells. Microvasc Res 76:7–14.

55. Kirchmair R, R Gander, M Egger, A Hanley, M Silver, ARitsch, T Murayama, N Kaneider, W Sturm, et al. (2004). Theneuropeptide secretoneurin acts as a direct angiogenic cy-tokine in vitro and in vivo. Circulation 109:777–783.

56. Abramovitch R, M Neeman, R Reich, I Stein, E Keshet, JAbraham, A Solomon and M Marikovsky. (1998). Inter-cellular communication between vascular smooth muscleand endothelial cells mediated by heparin-binding epider-mal growth factor-like growth factor and vascular endo-thelial growth factor. FEBS Lett 425:441–447.

57. Klein-Soyer C, DO Azorsa, J Cazenave and F Lanza. (2000).CD9 participates in endothelial cell migration during in vitrowound repair. Arterioscler Thromb Vasc Biol 20:360–369.

Address correspondence to:Julia Konig

Institute of Cell Biology, Histology and EmbryologyMedical University of Graz

Harrachgasse 21/7Graz 8010

Austria

E-mail: ingrid.lang@medunigraz.at

Received for publication May 5, 2011Accepted after revision July 15, 2011

Prepublished on Liebert Instant Online July 15, 2011

1320 KONIG ET AL.