Toxicology in Vitro 23 (2009) 719–727

Contents lists available at ScienceDirect

Toxicology in Vitro

journal homepage: www.elsevier .com/locate / toxinvi t

Cytotoxicity of 5-fluorouracil: Effect on endothelial differentiation via cellcycle inhibition in mouse embryonic stem cells

Gi Dae Kim a, Gyu-Seek Rhee b, Hyung-Min Chung c, Kew-Mahn Chee a,1, Gi Jin Kim c,*,1

a School of Life Sciences and Biotechnology, Korea University, Seoul 136-701, Republic of Koreab Department of Reproductive and Developmental Toxicology, National Institute of Toxicological Research, KFDA, Seoul 122-704, Republic of Koreac Graduate School of Life Science and Biotechnology, CHA Stem Cell Institute, College of Medicine, Pochon CHA University, Seoul 135-081, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history:Received 5 November 2008Accepted 22 February 2009Available online 9 March 2009

Keywords:Mouse embryonic stem cellsViabilityProliferationDifferentiation5-FluorouracilCytotoxicitycDNA microarrayCell cycle

0887-2333/$ - see front matter � 2009 Elsevier Ltd. Adoi:10.1016/j.tiv.2009.02.012

Abbreviations: Mouse ESCs, mouse embryonic stemEGM-2, endothelial growth medium-2; LIF, leukaemouse embryonic fibroblast; MTT, 3-(4,5-dimethylthizolium bromide; 5-FU, 5-fluorouracil; PECAM, platemolecule; Probucol, 4,40-(isopropylidenedithio) bis (Cyclin-dependent kinase; CDKI, Cyclin-dependent kinating cell nuclear antigen; GAPDH, glyceraldehyde-3

* Corresponding author. Tel.: +82 2 3468 3687; faxE-mail address: gjkim@cha.ac.kr (G.J. Kim).

1 These two authors contributed equally to this wor

Embryonic stem cells (ESCs) are known to characteristics for pluripotency and self-renewal, but the pre-cise mechanisms of ES-derived cells to specific toxicants have not been determined. Here, we evaluatedthe cytotoxicity of 5-fluorouracil (5-FU) and see its effect on cell viability, proliferation, and differentia-tion in mouse ESC-derived endothelial differentiation. Mouse ESCs were exposed to 5-FU (10 lM) andcombined with probucol (50 lM) for 24 h, which is an antagonist of 5-FU. Changes in gene expressionas a result of 5-FU exposure in mouse ESC-derived endothelial precursor cells (ES-EPCs) were assessedusing an oligonucleotide microarray (AB1700). The expression of Oct-4 was decreased during the differ-entiation of mouse ESCs into endothelial cells; otherwise, the expression of PECAM was increased. MouseES-EPCs were shown to have a decrease in viability (49.8%) and PECAM expression, and induce G1/Sphase (31.1%/60.6%) when compared with/without treatment of 5-FU. Expression of cell cycle-relatedproteins was increased in endothelial precursor cells exposed to 5-FU without probucol treatment. Fromtheses results suggest that 5-FU inhibit endothelial differentiation as well as inducing the G1/S phasearrest. We propose that mouse ES-EPCs might be a useful tool for screening the cytotoxicity of com-pounds in endothelial cells.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Embryotoxicity tests using animals are a traditional strategy toidentify potentially hazardous chemicals. They can also be used toconfirm the absence of toxic properties in the development ofpotentially useful new substances (Brown et al., 1995). Thesein vivo tests have several limitations, such as validation of animalmodels, enormous cost, high labour intensity, time required togenerate meaningful results, and ethics for animal experiments(Knight, 2007; Bremer and Hartung, 2004). Therefore, there is aneed for alternative methods to evaluate the potential reproduc-tive toxicity of chemical substances, by in vitro systems. To developa new alternative screening test, many scientists have tried to use

ll rights reserved.

cells; EBs, embryoid bodies;mia inhibitory factor; MEF,azol-2yl)-2,5,-diphenyl tetra-let endothelial cell adhesion2,6-di-t-butylophenol); CDK,ase inhibitor; PCNA, prolifer-phosphate dehydrogenase.: +82 2 538 4102.

k.

cell lines, primary cell cultures of dissociated cells from mice or ratembryo limb buds, midbrains for micromass tests, or whole em-bryos from rat (Steele et al., 1983).

Mouse embryonic stem cells (mouse ESCs) are able to differen-tiate into various cell types, including three germ layers as plurip-otent cells derived from the inner cell mass of blastocysts. They canalso undergo unlimited self-renewal (Evans and Kaufman, 1981;Martin, 1981; Ramalho-Santos et al., 2002). Therefore, an embry-onic stem cell test (EST), which mirrors growth and differentiation,is an in vitro test system well-suited for the evaluation of theembryotoxic potential of substances (Spielmann et al., 1997).

Endothelial, endothelial-like cells, and endothelial precursorcells (EPCs) derived from stem cells have been explored to estab-lish a toxicity screening system for endothelial-specific toxicants(Kim and von Recum, 2008). The feasibility of these screening sys-tems depends on the differentiation processes of the ESCs used;guided differentiation into target cell types and accurate investiga-tion of the mechanisms of endothelial toxicity are necessary. Re-cently, we reported that endothelial-like and endothelial cellsderived from mouse ESCs using EGM medium and optimal proto-cols are more sensitive to 5-FU toxicity than undifferentiated endo-thelial cells as well as a mouse endothelial cell line (Kim et al.,2008).

720 G.D. Kim et al. / Toxicology in Vitro 23 (2009) 719–727

5-Fluorouracil (5-FU) is one of the most widely used as an anti-cancer or anti-angiogenesis agents for advanced carcinoma andworks through G1/S cell cycle arrest and the induction of apoptoticdeath of the cancer cells (Lewin et al., 1987; Dimery and Hong,1993). Also, 5-FU induces a direct toxic effect on the endothelium(Kinhult et al., 2003). However, the exact molecular biological roleof 5-FU on cell cycle regulation in the endothelial differentiation ofmouse ESCs has not been fully explained yet.

The objectives of this study were to investigate the specific ac-tion of 5-FU on the endothelial differentiation of cells derived frommouse ESCs. Secondly, to investigate the correlation between cellcycle regulation and endothelial differentiation in mouse ESCs ex-posed by 5-FU.

Table 1Sequences of oligonucleotide primers used for RT-PCR analysis.

Gene Primer sequences Product size (bp)

PECAM F 50-GCCTGGAGAGGTTGTCAGAG-30 357R 50-GGTGCTGAGACCTGCIIII C-30

GAPDH F 50-TGTTCCTACCCCCAATGTGT-30 R 39650-TGTGAGGGAGATGCTCAGTG-30

2. Materials and methods

2.1. Cell culture conditions and endothelial cell differentiation

Mouse D3 ESCs (ATCC Cat. No. CRL-1934, Rockville, MD, USA)were co-cultured with mitomycin C-treated mouse embryonicfibroblast (MEF) cells in high glucose DMEM (Gibco-BRL, Invitro-gen, Carlsbad, CA) containing 15% fetal bovine serum (FBS; Hy-clone, Ogden, UT), 1000 U/ml LIF/ESGRO (Chemicon, Temecula,CA), and basic ES medium components [50 U/ml penicillin and50 lg/ml streptomycin (Gibco-BRL, Invitrogen, Carlsbad, CA), 1%non-essential amino acids (Gibco-BRL, Invitrogen, Carlsbad, CA)and 0.1 mM b-mercaptoethanol (Gibco-BRL, Invitrogen, Carlsbad,CA)]. The hanging drops method (20 ll per drop; 1 � 105

cells ml�1) was used to induce differentiation as describedby Heuer and theirs colleagues (Heuer et al., 1993) with minormodifications. After incubation for 3 days, embryoid bodies(EBs) were transferred to gelatine-coated wells of chamber slides(Nunc, Denmark) or 60 mm dishes to allow attachment. To pro-mote endothelial cell differentiation, 3-day-old EBs were placedin medium consisting of EBM-2, 5% FBS, and growth factor cock-tail (EGM2-MV Bullet Kit; Clonetics/BioWhittaker, Walkersville,MD).

2.2. Cytotoxicity analysis

To determine cytotoxic effects of 5-FU on mouse ESCs, theMTT assays were performed in the absence of mLIF as previouslydescribed (Spielmann et al., 1997; Scholz et al., 1999). Briefly,1000 cells were seeded into each well of a 96-well microtitreplate and grown in the presence of a concentration range of 5-FU and probucol. A negative control containing solvent dilutedin medium was also included. At day 9, the cells were exposedto 5-FU (10 lM) with/without probucol (50 lM) in a total volumeof 200 ll for 24 h. The 5-FU and probucol were dissolved in cellculture medium and ethanol, respectively. The final ethanol con-centration in the wells was 0.1%. The controls were incubatedwith equal volumes of drug solvents to avoid changes that couldbe due to solvent. About 20 ll of MTT (5 mg/ml) was added to200 ll culture medium on day 10, followed by incubation at37 �C for 4 h. After incubation, the MTT solution was carefully re-moved and 150 ll of DMSO (Sigma, St Louis, MO) was added toeach well. The plates were shaken on a plate mixer until all crys-tals had dissolved. The absorbance of the resulting coloured solu-tion was measured at 570 nm with a Genios luminometer(TECAN, Austria) at a reference wavelength of 630 nm. Cytotoxic-ity was expressed as a percentage of cells surviving, relative tountreated cultures, and the concentration required to inhibit cellgrowth by 50% (IC50) was calculated. Each experiment was per-formed using six replicates for each drug concentration and re-peated in triplicate.

2.3. RNA isolation and reverse transcription-polymerase chain reaction(RT-PCR)

Cells, including control and test groups, that had been exposedto 5-FU with/without probucol were directly harvested into tubescontaining Trizol (Gibco-BRL, Invitrogen, USA) and mRNA was ex-tracted according to the manufacturer’s protocol. The isolated totalRNAs were quantified using a spectrophotometer (SmartSpec 3000,Bio-Rad). First-strand cDNA was synthesised from 2 lg of totalRNA using an oligo (dT) primer and a SuperScript II First-StrandSynthesis System for RT-PCR (Invitrogen, Carlsbad, CA), accordingto the manufacturer’s instructions. First-strand cDNAs were ampli-fied in a final volume of 25 ll containing 0.5 U Ex Taq DNA poly-merase (TaKaRa Biotechnology, Korea) and 10 pmol of eachtarget primer (Table 1). PCR conditions were as follows: 5 min at94 �C, 30 amplification cycles (denaturation at 94 �C for 1 min,annealing at 55 �C for 1 min, extension at 72 �C for 1 min), followedby a final extension at 72 �C for 5 min. The amplified products wereseparated on 1.5% agarose gels and visualised with ethidium bro-mide staining. cDNA samples were adjusted to yield equal GAPDHamplifications.

2.4. Immunocytochemistry

At day 9, the cells were exposed to 5-FU with/without probucolfor 24 h. After the 24 h incubation, cells were fixed with freshlyprepared MeOH/DMSO (4:1) for overnight incubation at 4 �C. Cellswere blocked with blocking solution containing 1% BSA and 0.1%Tween 20 for 30 min, and then incubated with rat anti-mouse PE-CAM-1 (1:100) (MEC 13.3, Santa Cruze, Biotechnology, Inc.), or rab-bit anti-mouse PCNA (1:100) (Santa Cruze, Biotechnology, Inc) at4 �C for overnight. After washing, cells were incubated with goatanti-rat IgG-FITC (1:100) (Santa Cruz, Biotechnology, Inc) or goatanti-rabbit IgG-TRITC (1:100) (Chemicon, Temecula, CA) for pri-mary antibodies, respectively. After staining, coverslips weremounted in 30% Mowiol (Calbiochem-Novabiochem, Schwalbach,Germany). Images were obtained and analysed using a Bio-Radconfocal microscope (Radiance 2000 FCMP, Bio-Rad, USA).

2.5. FACS analysis

In order to analyse how Oct-4 expression varies in mouse ESCsthrough different stages of endothelial cell differentiation, endo-thelial differentiation induced cells for 0, 4, 7 and 10 days wereharvested using cell dissociation buffer (Sigma, St. Louis, MO). Toanalyse PCNA expression, the cells were exposed to 5-FU with/without probucol for 24 h at day 9. After 24 h incubation, controlcells and treated cells were harvested using cell dissociation buffer(Sigma, St. Louis, MO). Cells were re-suspended at 106 cells/100 llin suspension buffer and then incubated with 1 lg/100 ll of rabbitanti-mouse Oct-4 (Santa Cruz, Biotechnology, Inc.), or rabbit anti-mouse PCNA (1:100) (Santa Cruze, Biotechnology, Inc.) for 1 h at4 �C. Negative controls were incubated for 1 h at 4 �C with fluoro-chrome labelled irrelevant isotype control antibodies: 1 lg/100 llgoat anti-rabbit FITC-conjugated IgG (Chemicon) or goat anti-rab-bit IgG-TRITC (1:100) (Chemicon, Temecula, CA) for primary anti-bodies, respectively. After staining, cells were analysed without

G.D. Kim et al. / Toxicology in Vitro 23 (2009) 719–727 721

fixation using a FACS Calibur (Becton Dickinson, MA) using 5 lg/mlof propidium iodide (Sigma, St. Louis, MO) to exclude dead cells.

For cell cycle analysis in differentiation into endothelial cell ofmouse ESCs, the treated cells were trypsinised and fixed in cooled70% ethanol at 4 �C. The cells were then incubated in 0.3 ml of DNAstaining solution (100 lg/ml PI, 1 mg/ml RNase A (DNase-free) inPBS). The cells were then transferred into D-Hank’s solution. Thecell suspension was stored on ice in a dark room for a minimumof 30 min and analysed within 2 h. Data analysis was carried outusing CellQuest software (Becton Dickinson, MA).

2.6. Microarray

For cDNA microarray analysis, differentiated mouse ESCs trea-ted with/without 5-FU (10 lM) for 24 h were collected. Their totalRNA was extracted using Trizol (Gibco-BRL, Invitrogen, USA) andmRNA was extracted according to the manufacturer’s protocol.The quantity and quality of total RNA and amplified RNA were as-sessed by using a Bioanalyzer 2100 (Agilent Technologies). The Ap-plied Biosystems Mouse Genome Survey Microarray contains32,996 60-mer oligonucleotide probes, representing 32,181 indi-vidual mouse genes. Digoxigenin-UTP labelled cRNA was gener-ated and linearly amplified from 2 lg of total RNA using AppliedBiosystems Chemiluminescent RT-IVT Labelling Kit v.2.0 and man-ufacturer’s protocol (Applied Biosystems). Array hybridization (fivearrays per sample), chemiluminescence detection, image acquisi-tion and analysis were performed using Applied BiosystemsChemiluminescence Detection Kit (Applied Biosystems) and Ap-plied Biosystems 1700 Chemiluminescent Microarray Analyzer(Applied Biosystems), following the manufacturer’s protocol.Images were auto-gridded, then spot and spatially normalised.Chemiluminescent signals were quantified, corrected for back-ground, and the final images and feature data were processed usingthe Applied Biosystems 1700 Chemiluminescent Microarray Ana-lyzer software v1.1. Data and images were collected through anautomated process for each microarray using the 1700 analyzer.A total of 10 arrays were run for the two groups (five technical rep-licates for each group). A global median normalisation, whichnormalises signal intensities across all microarrays to achieve thesame median signal intensities for each array, was performed onthe Applied Biosystems data sets. For Applied Biosystems arrays,the detection threshold was set as S/N > 3 with a quality flag < 100.Correlation and coefficient of variation (CV) analyses were per-formed using Matlab� software (Mathworks, Natick, MA). Differen-tial expression analysis was done using two different statisticalmethods: (1) ANOVA analysis was performed using Avadis soft-ware. Differentially expressed genes between control and 5-FU(10 lM) treated groups were determined based on the followingcriteria: (a) p < 0.001 in ANOVA analysis; (b) average change be-tween control and 5-FU (10 lM) treated groups > two-fold; (c)detectable in more than 50% samples. (2) Significance Analysis ofMicroarray (SAM: http://www-stat.stanford.edu/~ibs/SAM), asupervised learning statistical software that performs a modifiedt-test to identify genes with significant changes in expression,and uses permutations to estimate the false discovery rate (FDR).Hierarchical clustering of log ratios was performed using the soft-ware (http://rana.lbl.gov/EissenSoftware.htm) Cluster and Tree-view; Euclidean correlation, median centring and completelinkage were applied in all clustering applications.

2.7. Western blot analysis

Control cells and treated Cells were lysed in RIPA buffer (1% NP-40, 150 mM NaCl, 0.05% DOC, 1% SDS, 50 mM Tris) containing pro-tease inhibitor for 1 h at 4 �C. The supernatant was separated bycentrifugation, and protein concentration was determined with a

Bradford protein assay kit II (Bio-rad). Proteins (25 lg/well) dena-tured with Laemmli sample buffer (Sigma) were separated by 10%SDS–polyacrylamide gel (Bio-Rad) under a constant current of50 mV. Proteins were transferred onto nitrocellulose membranes(0.45 mm, Amersham Life Sciences). The membranes were blockedwith a 5% BSA solution for 3 h, washed with PBS containing 0.2%Tween 20, then incubated with the primary antibody overnightat 4 �C. Human specific antibodies against CDK-2, CDK-4, CyclinD1, Cyclin E, p21WAF1/CIP1, p27Kip1, p53, and b-actin from Santa CruzBiotechnology (Santa Cruz Biotechnology) were used to probe theseparate membranes. The immunoreaction was continued withthe secondary goat anti-rabbit horseradish-peroxidase conjugatedantibody after washing for 2 hours at room temperature. The spe-cific protein bands were detected by enhanced chemiluminescence(Pierce), with X-Omat AR films (Kodak). As a protein loading con-trol, parallel gels were subjected to Western blot analysis using ab-actin antibody (Santa Cruz Biotechnology).

2.8. Statistical analysis

All results are expressed as a percentage of the untreated con-trol values, or as a mean ± SD of three independent experiments,each with six replicates. Statistical significance was determinedusing the Student’s t-test for paired data. A P value of < 0.05 wasregarded as significant, and IC50 values were calculated using Sig-maplot version 9.0.

3. Results

3.1. Effect of 5-FU on the endothelial differentiation of cells derivedfrom mouse ESCs

To study the role of 5-FU in endothelial differentiation, mouseESCs were differentiated into endothelial precursor cells and trea-ted with 5-FU (10 lM). Oct-4 expression and cell viability wereanalysed. The expression of Oct-4 gradually decreased, with100%, 90.6%, 81.3%, 50.5% expression on differentiation days 0, 4,7, and 10, respectively (Fig. 1A). In a previous study, we found thatthe percentage of PECAM expression increased by up to 44.5% inmouse ES-derived endothelial precursor cells, grown in EGM-2medium for 10 days. There is a strong correlation between the de-creased Oct-4 expression and increased PECAM expression duringthe endothelial differentiation of cells derived from mouse ESCs.Next, we examined the viability of endothelial precursor cellswhen they were exposed to 5-FU (10 lM), using MTT assays. Asshown in the supplemental data, cells exposed to various concen-trations of 5-FU and probucol for 24 h displayed a decrease in cellviability in a concentration dependent manner. The IC50 of 5-FUand the concentration of probucol needed to avoid affecting cellviability were 10 lM, 50 lM, respectively. The viability was signif-icantly decreased to 49.8% of the control in the endothelial differ-entiated cells exposed to 5-FU (10 lM) for 24 h. The viability wassignificantly recovered to 65.7% in cells exposed to 5-FU (10 lM)combined with probucol (50 lM) (P < 0.01) (Fig. 1B).

3.2. 5-FU inhibits cell proliferation in endothelial differentiated cellsderived from mouse ESCs

The morphology of mouse ES-derived endothelial precursorcells after treatment with the IC50 values of 5-FU (10 lM) for24 h was changed and detached, compared to the control cellswhich grew as confluent aggregates with rounded and polygonalcell morphology. However, probucol (50 lM) treatment to the 5-FU treatment groups for 24 h induced a morphology similar to thatof the control cells (Fig. 2A, top panel). In order to investigate the

Cell

viab

ility

(% o

f con

trol)

0

50

100

150

5-FU - 10 10 (uM)Probucol - - 50 (uM)

**

*

A

B

Fig. 1. The expression of Oct-4 and the cytotoxicity of 5-fluorouracil in endothelial precursor cells, derived from mouse embryonic stem cells. The expression of Oct-4gradually decreased during endothelial differentiation (A). After inducing endothelial differentiation for 9 days with EGM-2 medium, the cells were exposed to 5-fluorouracil(10 lM) with/without probucol (50 lM) for 24 h. Analysis of the cell viability of the endothelial differentiated cells exposed to 5-fluorouracil were assessed by MTT assays (B).The experiments were performed in three independent runs (n = 6). Standard error bars are shown. Significance was tested by the student t-test (*p < 0.01). Numbers indicatethe percentages of target antibody-positive cells.

722 G.D. Kim et al. / Toxicology in Vitro 23 (2009) 719–727

effects of 5-FU on the proliferation of endothelial precursor cellsderived from mouse ESCs, PCNA expression was assessed byimmunocytochemistry and FACS analysis. After inducing endothe-lial differentiation for 9 days, we observed that the endothelial pre-cursor cells exposed to 5-FU (10 lM) for 24 h had decreased PCNAexpression, whilst the probucol (50 lM) treated group had recov-ered its PCNA expression (Fig. 2A, bottom panel). These results cor-related with the PCNA expression of the endothelial precursor cellsexposed to 5-FU (47.7%), as shown by FACS analysis.

3.3. Expression of PECAM in endothelial differentiated cells is down-regulated by 5-FU

To test whether the treatment of mouse ES-derived endothelialprecursor cells with 5-FU can influence endothelial differentiation

through expression of endothelial-specific genes, the expressionlevels of PECAM were analysed by immunocytochemistry andRT-PCR. The expression of PECAM was dramatically decreased inthe 5-FU treatment group, compared to the control group. How-ever, the expression of PECAM was maintained in the probucoltreatment group (Fig. 3A and B). These findings demonstrate that,in accordance with the morphological differentiation analysis, 5-FU decreases endothelial-specific mRNA levels and so inhibits theexpression of genes involved in endothelial differentiation.

3.4. Gene profiling of mouse embryonic stem cells exposed to 5-FUusing microarray analysis

In order to determine gene expression changes in cells exposedto 5-FU (10 lM), the RNA contained in both the control cells, and

Fig. 2. Anti-proliferative affect of 5-fluorouracil in endothelial precursor cells. Morphology (A, top panel) and PCNA expression (A, bottom panel) of control and cells exposedto 5-fluorouracil (10 lM) with/without probucol (50 lM) for 24 h. Cell nuclei were stained with DAPI. The percentages of cells with PCNA expression were analysed by FACS(B). Numbers indicate the percentages of target antibody-positive cells. Original magnification: 40� for (A).

G.D. Kim et al. / Toxicology in Vitro 23 (2009) 719–727 723

the cells exposed to 5-FU treatment for 24 h at differentiation day9 was collected. As shown a Fig. 4, a total of 11,668 genes of 32,996cDNAs were selected as differentially expressed genes by the ANO-VA test (p < 0.001). A hierarchical cluster analysis yielded majorclusters in the 11,668 expressed genes (Fig. 4A). These geneexpression patterns were classified into functional groups, basedon their biological process as defined by the Gene Ontology (GO)annotation system. Most of the observed genes were related tophysiological processes, including apoptosis, cell cycle, develop-mental processes, and signal transduction in all clusters (Table2). These results suggest that 5-FU treatment affects the expressionof numerous genes via the alteration of several processes. Notably,expression of 58 genes of the 1439 genes in the cell cycle categorywas modulated by more than two-fold, between the control cells

and endothelial differentiated cells exposed to 5-fluorouracil(Fig. 4B, Supplementary data 1).

3.5. 5-FU induces arrest of G1/S phase in endothelial-like cells derivedfrom mouse ESCs

To confirm the effects of 5-fluorouracil (10 lM) with and with-out probucol (50 lM) in the cell cycle, the cell cycle distributionwas analysed by flow cytometry (Fig. 5A). Generally, the frequen-cies of G0/G1 phase and S phase were 62.7% and 33.5% in mouseESCs, respectively. In contrast, mouse ES-derived endothelial pre-cursor cells exposed to 5-FU (10 lM) for 24 h showed a decreaseto 31.1% in G0/G1 phase and an increase to 60.6% in S phase. Treat-ment with 50 lM probucol for 24 h showed 35.0% of cells in G0/G1

Fig. 3. PECAM expression of endothelial precursor cells, derived from mouse embryonic stem cells. Expression of PECAM in control and cells exposed to 5-fluorouracil(10 lM) with/without probucol (50 lM) for 24 h were analysed by immunocytochemistry. Cell nuclei were stained with DAPI (A). mRNA levels of PECAM and GAPDH wereevaluated by RT-PCR (B). GAPDH used as internal standard. Original magnification: 40� for (A).

724 G.D. Kim et al. / Toxicology in Vitro 23 (2009) 719–727

phase and 62.8% in S phase (Fig. 5A). These data suggest that 5-FUarrests cells at the G1/S phase boundary in endothelial precursorcells, derived from mouse ESCs. This was followed by decreasedproliferation of mouse ES-derived endothelial precursor cells, sim-ilar to that reported previously with other cell types (Vittet et al.,1996).To determine whether the expression levels of cell cycle-re-lated proteins were changed, we analysed Cyclins, CKDs, CKDIs,and p53 expression after treatment with 5-FU (10 lM), with andwithout probucol (50 lM), by Western blot analysis. The expres-sion of Cyclin E, CDK2, p21WAF1/CIP1, and p53 was up-regulated inES-derived endothelial precursor cells exposed to 5-FU, comparedto those in the control group and cells treated with 5-FU (10 lM)and probucol (50 lM) (Fig. 5B). Although the expression of CyclinD1 remained constant in control cells and mouse ES-derived endo-thelial precursor cells exposed to 5-FU, they were remarkably de-creased by probucol treatment. Otherwise, there were nodifferences in the expression of CKD4 and p27Kip1, the tumour sup-pressor and inhibitors of Cyclin E/CDK2 in mouse ES-derived endo-thelial precursor cells, according to 5-FU and probucol treatments(Fig. 5B). From these results, we suggest that 5-FU might inhibitG1-related Cyclin/CDK activities through the augmentation ofp21WAF1/CIP1 expression and by binding to Cyclin D/CDK complexes.

4. Discussion

In this study, we demonstrate that 5-fluorouracil, an anti-cancerdrug, can induce cytotoxicity in endothelial differentiation ofmouse ESCs via inhibition of processes involved in cell viability,

proliferation, and the cell cycle. During the induction of endothelialdifferentiation in cells derived from mouse ESCs, the expression ofmesodermal lineage-related genes was up-regulated at 7 days afterdifferentiation using EBs, followed by the up-regulation of endo-derm lineage-related genes at 14 days (Heo et al., 2005). Basedon these reports, we induced endothelial differentiation frommouse ESCs for 10 days using EGM-2 to maximise endothelial dif-ferentiation, and confirmed the expression patterns of Oct-4 andPECAM. The frequencies of cells undergoing endothelial differenti-ation gradually increased through subsequent differentiation days;the expression of Oct-4 was 50.5% at endothelial differentiationday 10. These findings show that the characteristics of cells atendothelial differentiation day 10 are similar to endothelial precur-sor cells and endothelial-like cells. Because of this, these modelscan be used to evaluate the cytotoxicity of 5-FU at different stagesof endothelial differentiation.

A number of anti-cancer agents have been implicated in vascu-lar toxicity and their effects have been attributed to direct toxicityto the endothelium, such as the HUVEC and C166 cell lines. 5-FUgives an increase in the permeability of endothelial monolayers,as well as inducing vascular collapse and tumour necrosis (Wattset al., 1997). 5-FU, as a cytostatic agent with a strong embryotoxicpotential, is known to have an effect in rapidly proliferating cells(Parker and Cheng, 1990; Shimizu et al., 2001). These reports arewell matched with our results. The cytotoxicity of 5-FU in mouseESCs was relieved by the addition of probucol (50 lM), an antago-nist of 5-FU. This prevented the endothelial injury usually causedby 5-FU, but the rate of recovery for damaged endothelial cells

Fig. 4. Gene expression profiling of endothelial precursor control and cells exposed to 5-fluorouracil using microarray analysis. Hierarchical clustering analysis wasconducted using control (n = 5) and 5-FU treated samples (n = 5). The black bars on right side of A illustrate the location of clusters shown. The dendrogram in B shows thesamples identified as being in the cell cycle category, between control and 5-FU treated cells. The intensity of red and green colour is proportional to the relative up-regulation(red) or down-regulation (green) of gene expression in the differentiated samples, compared to that in the undifferentiated reference. Gene names and accession numbers arefrom Unigene (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=unigene). (For interpretation of the references to colour in this figure legend, the reader is referred to theweb version of this article.)

G.D. Kim et al. / Toxicology in Vitro 23 (2009) 719–727 725

did not completely return to normal (Kita et al., 1987; Kanekoet al., 1996).

The powerful advantages of ESCs when they are applied to tox-icology come as a result of their unique properties compared to

Table 2Gene content list of the AB 1700 mouse chip.

Function (Panther classification system) Number

Angiogenesis 4Apoptosis 30Cell adhesion 22Cell cycle 58Cell proliferation and differentiation 9Cell structure and motility 36Developmental processes 80Homeostasis 7Immunity and defense 45Signal transduction 126Cell adhesion molecule 20Cell junction protein 8The others 994/1439

726 G.D. Kim et al. / Toxicology in Vitro 23 (2009) 719–727

primary cells: self-renewal, plasticity to generate various cellstypes, and that they are a readily available alternative source to re-place primary cells. Therefore, stem cell based screening systems

Channels (FL2-A-FL2-A)0 40 80 120 160

G0-G1 : 62.7 %

S : 33.5 %

G2-M : 3.7 %

0

Channels (FL2-A-FL2-A)0 40 80 120 160 20

G0-G1 : 35.0 %

S : 62.8 %

G2-M : 2.0 %

Control

5-FU 10uM, probucol50uM

A

020

040

060

080

010

00N

umbe

r0

010

020

030

040

050

0

100

200

300

400

Num

ber

Num

ber

Fig. 5. Cell cycle arrest in endothelial precursor cells after 5-fluorouracil exposure. DNAchannel (y-axis) in control and cells treated with 5-fluorouracil (10 lM) with/without pCyclins, CDKs, and CDKIs in control and cells exposed to 5-fluorouracil (10 lM) with/wi

for toxicants offer a very promising technology; it is possible to ob-tain large numbers of cells for consistent analysis and the study atthe different stages of differentiation. In addition, toxicity screen-ings using ESCs have been validated as a reliable source forin vitro developmental toxicology studies (Rohwedel et al., 2001).In order to understand the biological mechanism of target cell spe-cific toxicants, DNA microarrays have been used to analyse geneprofiling, to show the alteration of gene expression (Gunji et al.,2004; Fumoto et al., 2008; Mori et al., 2007). However, studies ofgene profiling and the mechanisms within ESCs exposed to specialtoxicants, including 5-FU, are still rare. Thus, we analysed expres-sion patterns of genes involved in inducing endothelial differenti-ation, using a DNA microarray with untreated and 5-FU-treatedcells. As shown in the DNA microarray data, a huge number ofchanges in gene expression were identified by 5-FU treatment.Among these changes, we focused on genes involved in the cell cy-cle, as our data indicated that 5-FU induces cell proliferation inhi-bition (Fig. 2). In contrast to the somatic cell cycle, ESCs have aunique feature, an abbreviated cell cycle (Becker et al., 2006). This

Channels (FL2-A-FL2-A)40 80 120 160 20

G0-G1 : 31.1 %

S : 60.6 %

G2-M : 8.2 %

5-FU 10uM

CDK-4

CDK-2

Cyclin D1

Cyclin E

p 21WAF1/CIP1

p 53

β-actin

p 27Kip1

Contro

l

5-FU

10u

M5-

FU 10

uM,

prob

ucol

50uMB

distribution histogram, using PI labelling (x-axis) and total number of cells in eachrobucol (50 lM) (A). Western blot analysis of the expression of G1/S phase-relatedthout probucol (50 lM) (B).

G.D. Kim et al. / Toxicology in Vitro 23 (2009) 719–727 727

is thought to be controlled through an unusual mechanism of CDKregulation, followed by a very short period of cell cycle. This allowsESCs to preserve their unlimited differentiation potential (Steadet al., 2002). In ESCs, the expressions of Cyclin D1 and Cyclin D3are low, and Cyclin D2 is not expressed, moreover, the expressionof CDK-4 and the CDK 4-associated kinase activity is also weak,compared to somatic cells (Burdon et al., 2002). These previous re-ports lacked a consensus with our data, which showed increasedexpression of CDK4 and Cyclin D1 during endothelial differentia-tion. This discrepancy might originate from the differences be-tween cell sources.

Until now, the mechanism that has been reported to explain thecell cycle arrest by 5-FU treatment is a G1/S phase arrest in cancercell lines, induced by blocking DNA synthesis through the inhibi-tion of thymidylate synthase (Pinedo and Peters, 1988; Hong andStambrook, 2004). When damaged by 5-FU treatment, in addition,the expression of the tumour suppressor p53 is increased p53 hasbeen identified as a participant in the DNA damage response,resulting in either cell cycle arrest or death. It activates and regu-lates the transcription of cell cycle arrest or apoptosis related genes(Lane, 1992). The expression of p21, which a member of the WAF1/CIP1 family, and activates CKI which works as a linker with thep53-dependent pathway is up-regulated (Gartel et al., 1998). Onceactivated, the p53 gene product works together with the p21WAF1/

CIP1 protein, and binds to the Cyclin D-CDK4 and Cyclin E-CDK2complexes to inhibit their kinase activities (El-Deiry et al., 1993;Sherr and Roberts, 1999,). Finally, increased expression ofp21WAF1/CIP1 and p53 is involved in the G1/S phase arrest of the cellcycle. These reports are in accordance with our data. In addition,the probucol treatment in ESCs, which blocks the cell cycle arrestby 5-FU, appears to aid the repair of damaged cells.

Taken together, 5-FU affects endothelial differentiation bydecreasing cell viability, proliferation and differentiation, as wellas inducing the G1/S phase arrest. These toxicity screening arecapable to use the mouse ESC system, therefore, mouse ESCs mightbe a useful model to use as a tool for screening the cytotoxicity ofnew compounds.

Acknowledgement

We gratefully thank Dr. Eun Jeong Choi (Duksung Women’sUniversity, Korea) for technical support in the study.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.tiv.2009.02.012.

References

Becker, K.A., Ghule, P.N., Therrien, J.A., Lian, J.B., Stein, J.L., Wijnen, A.J., Stein, G.S.,2006. Self-renewal of human embryonic stem cells is supported by a shortenedG1 cell cycle phase. J. Cell. Physiol. 209, 883–893.

Bremer, S., Hartung, T., 2004. The use of embryonic stem cells for regulatorydevelopmental toxicity testing in vitro – the current status of test development.Curr. Pharm. Des. 10, 2733–2747.

Brown, N.A., Spielmann, H., Bechter, R., Flint, O.P., Freemen, S.J., Jelinek, R.J., Koch, E.,Nau, H., Newall, D.R., Palmer, A.K., Renault, J.Y., Repetto, M.F., Vogel, R., Wiger,R., 1995. Screening chemicals for reproductive toxicity: the current alternatives.ATLA 23, 868–882.

Burdon, T., Smith, A., Savatier, P., 2002. Signalling, cell cycle and pluripotency inembryonic stem cells. Trends Cell Biol. 12, 432–438.

Dimery, I.W., Hong, W.K., 1993. Overview of combined modality therapies for headand neck cancer. J. Natl. Cancer Inst. 85, 95–111.

El-Deiry, W.S., Tokino, T., Velculescu, V.E., Levy, D.B., Parsons, R., Trent, J.M., 1993.WAF1, a potential mediator of p53 tumor suppression. Cell 75, 817–825.

Evans, M.J., Kaufman, M.H., 1981. Establishment in culture of pluripotential cellsfrom mouse embryos. Nature 292, 154–156.

Fumoto, S., Shimokuni, T., Tanimoto, K., Hiyama, K., Otani, K., Ohtaki, M., Hihara, J.,Yoshida, K., Hiyama, E., Noguchi, T., Nishiyama, M., 2008. Selection of a noveldrug-response predictor in esophageal cancer: a novel screening method usingmicroarray and identification of IFITM1 as a potent marker gene of CDDPresponse. Int. J. Oncol. 32, 413–423.

Gartel, A.L., Goufman, E., Tevosian, S.G., Shih, H., Yee, A.S., Tyner, A.L., 1998.Activation and repression of p21 (WAF1/CIP1) transcription by RB bindingproteins. Oncogene 17, 3463–3469.

Gunji, W., Kai, T., Takahashi, Y., Maki, Y., Kurihara, W., Utsugi, T., Fujimori, F.,Murakami, Y., 2004. Global analysis of the regulatory network structure of geneexpression in Saccharomyces cerevisiae. DNA Res. 11, 163–177.

Heo, J., Lee, J., Chu, I., Takahama, Y., Thorgeirsson, S.S., 2005. Spontaneousdifferentiation of mouse embryonic stem cells in vitro: characterization byglobal gene expression profiles. BBRC 332, 1061–1069.

Heuer, J., Bremer, S., Pohl, I., Spielmann, H., 1993. Development of an in vitroembryotoxicity test using murine embryonic stem cell cultures. Toxicol. In Vitro7, 551–556.

Hong, Y., Stambrook, P.J., 2004. Restoration of an absent G1 arrest and protectionfrom apoptosis in embryonic stem cells after ionizing radiation. Proc. Natl. Acad.Sci. USA 101, 14443–14448.

Kaneko, M., Hayashi, J., Saito, I., Miyasaka, N., 1996. Probucol downregulates E-selectin expression on cultured human vascular endothelial cells. Arterioscler.Thromb. Vasc. Biol. 16, 1047–1051.

Kim, G.D., Kim, G.J., Seok, J.H., Chung, H., Chee, K., Rhee, G., 2008. Differentiation ofendothelial cells derived from mouse embryoid bodies: a possible in vitrovasculogenesis model. Toxicol. Lett. 180, 166–173.

Kim, S., von Recum, H., 2008. Endothelial stem cells and precursors for tissueengineering: cell source, differentiation, selection, and application. Tissue Eng.Part B. Rev. 14, 133–147.

Kinhult, S., Albertsson, M., Eskilsson, J., Cwikiel, M., 2003. Effects of probucol onendothelial damage by 5-fluorouracil. Acta Oncol. 42, 304–308.

Kita, T., Nagano, Y., Yokode, M., Ishii, K., Kume, N., Ooshima, A., Yoshida, H.h., Kawai,C., 1987. Probucol prevents the progression of atherosclerosis in Watanabeheritable hyperlipidemic rabbit, an animal model for familialhypercholesterolemia. Proc. Natl. Acad. Sci. USA 84, 5928–5931.

Knight, A., 2007. Systematic reviews of animal experiments demonstrate poorhuman clinical and toxicological utility. Altern. Lab. Anim. 35, 641–659.

Lane, D.P., 1992. Cancer: p53, guardian of the genome. Nature 358, 15–16.Lewin, F., Skog, S., Tribukait, B., Ringborg, U., 1987. Effect of 5-fluorouracil on the cell

growth and cell cycle kinetics of a mouse ascites tumor growing in vivo. ActaOncol. 26, 125–131.

Martin, G.R., 1981. Isolation of a pluripotent cell line from early mouse embryoscultured in medium conditioned by teratocarcinoma stem cells. Proc. Natl.Acad. Sci. USA 12, 7634–7638.

Mori, N., Glunde, K., Takagi, T., Raman, V., Bhujwalla, Z.M., 2007. Choline kinasedown-regulation increases the effect of 5-fluorouracil in breast cancer cells.Cancer Res. 67, 11284–11290.

Parker, W.B., Cheng, Y.C., 1990. Metabolism and mechanism of action of 5-fluorouracil. Pharmacol. Therapeut. 48, 381–395.

Pinedo, H.M., Peters, G.F., 1988. Fluorouracil: biochemistry and pharmacology. J.Clin. Oncol. 6, 1653–1664.

Ramalho-Santos, M., Yoon, S., Matsuzaki, Y., Mulligan, R.C., Melton, D.A., 2002.‘‘Stemness” transcriptional profiling of embryonic and adult stem cells. Science298, 597–600.

Rohwedel, J., Guan, K., Hegert, C., Wobus, A.M., 2001. Embryonic stem cells as anin vitro model for mutagenicity, cytotoxicity and embryotoxicity studies:present state and future prospects. Toxicol. In Vitro 15, 741–753.

Scholz, G., Pohl, I., Genschow, E., Klemm, M., Spielmann, H., 1999. Embryotoxicityscreening using embryonic stem cells in vitro: correlation to in vivoteratogenicity. Cells Tissues Organs 165, 203–211.

Sherr, C.J., Roberts, J.M., 1999. CDK inhibitors: positive and negative regulators ofG1-phase progression. Genes Dev. 13, 1501–1512.

Shimizu, N., Aoyama, H., Hatakenaka, N., Kaneda, M., Teramoto, S., 2001. An in vitroscreening system for characterizing the cleft palate-inducing potential ofchemicals and underlying mechanisms. Reprod. Toxicol. 15, 665–672.

Spielmann, H., Pohl, I., Doering, B., Liebsch, M., Moldenhauer, F., 1997. Theembryonic stem cell test, an in vitro embryotoxicity test using twopermanent mouse cell lines: 3T3 fibroblasts and embryonic stem cells. InVitro Toxicol. 10, 119–127.

Stead, E., White, J., Faast, R., Conn, S., Goldstone, S., Rathjen, J., Dhingra, U., Rathjen,P., Walker, D., Dalton, S., 2002. Pluripotent cell division cycles are driven byectopic Cdk2, cyclin A/E and E2F activities. Oncogene 21, 8320–8333.

Steele, C.E., New, D.A., Ashford, A., Copping, G.P., 1983. Teratogenic action ofhypolipidemic agents: an in vitro study with postimplantation rat embryos.Teratology 28, 229–236.

Vittet, D., Prandini, M.H., Berthie, R., Schweitzer, A., Martin-Sisteron, H., Uzan, G.,Dejana, E., 1996. Embryonic stem cells differentiate in vitro to endothelial cellsthrough successive maturation steps. Blood 88, 3424–3431.

Watts, M.E., Woodcock, M., Arnold, S., Chaplin, D.J., 1997. Effects of novel andconventional anti-cancer agents on human endothelial permeability: influenceof tumour secreted factors. Anticancer Res. 17, 71–75.