E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te /yexc r

Research Article

Targeting eradication of malignant cells derived from humanbone marrow mesenchymal stromal cells

Yingbin Yanga,b,1, Shaoxi Caia,⁎,2, Li Yanga,g,3, Shuhui Yua,c,3,4, Jiahuan Jianga,3,5,Xiaoqing Yana,4, Haoxing Zhangb,6, Lan Liud,7, Qun Liue,8, Jun Duf,9,Shaohui Caig,10, K.L. Paul Sunga,h,3,11

aKey Laboratory of Biorheological Science and Technology, Ministry of Education, Bioengineering College, Chongqing University,Chongqing 400044, ChinabSchool of Life Science, Southwest University, Chongqing 400715, ChinacLibrary of Southwest University, Chongqing 400715, ChinadDepartment of Laboratory of Medicine, Children's Hospital of Chongqin Medical University, Chongqing 400014, ChinaeCollege of Life Science and Technology, Southwest University for Nationalities, Chengdu 610041, ChinafCenter of Microbiology, Biochemistry, and Pharmacology, School of Pharmaceutical Science, Sun Yat-Sen University, Guangzhou 510080, ChinagCollege of Pharmacy, Jinan University, Guangzhou 510632, ChinahDepartments of Orthopaedic Surgery and Bioengineering, University of California, SD 0412, USA

A R T I C L E I N F O R M A T I O N

⁎ Corresponding author. College of BioengineeChina. Fax: +86 023 65111633.

E-mail address: sxcai@cqu.edu.cn (S. Cai).1 Conception and design, major performance2 Conception and design, administrative supp3 Financial support and administrative suppor4 Collection and assembly of data.5 Administrative support.6 Performance of experiments.7 Provision of study material.8 Financial support and performance of exper9 Conception and design support.

10 Collection of data and performance of exper11 Conception and design, final approval of ma

0014-4827/$ – see front matter © 2010 Elseviedoi:10.1016/j.yexcr.2010.02.014

A B S T R A C T

Article Chronology:

Received 27 July 2009Revised version received26 November 2009Accepted 9 February 2010Available online 23 February 2010

Human bone marrow mesenchymal stromal cells (hBMSC) have been shown to participate inmalignant transformation. However, hampered by the low frequency of malignant transformationof hBMSC, we do not yet know how to prevent malignant transformation of implanted hBMSC. Inthis study, in order to establish a model for the eradication of hBMSC-derived malignant cells, agene fusion consisting of a human telomerase (hTERT) promoter modified with both c-Myc andmyeloid zinc finger protein2 (MZF-2) binding elements and followed by the E. coli cytosine

ring, University of Chongqing. No.174, Shazheng RD, Shapingba District, Chongqing City 400044,

of experiments, manuscript writing.ort.t.

iments.

iments.nuscript.

r Inc. All rights reserved.

Keywords:

Targeting eradicationMalignant transformation

hBMSChTERT promote

3330 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

deaminase (CD) and luciferase genes was stably transferred into hBMSC via lentiviraltransduction; n-phosphonacelyl-L-aspartic acid (PALA) selection was used to generatemalignant cell colonies derived from transduced hBMSC after treatment with the carcinogenicreagent BPDE. Cells that were amplified after PALA selection were used for transplantation and 5-

FC pro-drug cytotoxicity tests. The results showed that PALA-resistant malignant cells could begenerated from hBMSC co-induced with lentiviral transduction and treatment with Benzo(a)pyrene Diol Epoxide (BPDE); the modification of c-Myc and MZF-2 binding elements couldremarkably enhance the transcriptional activities of the hTERT promoter in malignant cells,whereas transcriptional activity was depressed in normal hBMSC; malignant cells stablyexpressing CD under the control of the modified hTERT promoter could be eliminated by 5-FCadministration. This study has provided a method for targeted eradication of malignant cellsderived from hBMSC.

© 2010 Elsevier Inc. All rights reserved.

Introduction

With the recent FDA approval to test the use of human embryonicstem cells for spinal cord injury treatment, application of pluripotentstem cells into the clinic is rapidly becoming a reality [1]. However,as one of stromal cells, hBMSC have the ability to transformspontaneously or to be transformed by carcinogenic reagents, viralinfections, etc., and present the potential danger of developing intotumor cells [2–6], which could represent a major limitation in theirtherapeutic uses. As a new kind of therapeutic delivery tool,substantial attention has been focused on hBMSC transplant safety.However, until now, a reliablemethodhasnot been found topreventneoplastic transformation of engrafted hBMSC.

Newly developed techniques in tumor gene therapymay providea feasible solution to this problem. Studies have demonstrated thatcytosine deaminase (CD) can convert the non-toxic pro-drug 5-fluorocytosine (5-FC) into toxic 5-fluorouracil (5-FU). Mammaliancells do not produce this enzyme; it is made by a variety of bacteriaand fungi [7]. It was reported that transduction of the CD genemadehuman cells highly sensitive to 5-FC at concentrations much lowerthan the safe serum concentration. An in vitro study alsodemonstrated that cell proliferation was inhibited by exposure to5-FU in a dose-dependentmanner; however, up to 1000 μg/mL of 5-FC had no effect on cell proliferation. It was reported that malignantcell proliferation was inhibited by 5-FC exposure in a time-dependentmanner, with induction of cytosine deaminase followinginfection by recombinant adenovirus [8]. The development oftumor-targeting transcriptional research may also provide furtherinformation to guide the targeted eradication of hBMSC-derivedmalignant cells. In healthy cells, telomerase activity is generallyregulated by transcriptional control of the hTERT promoter.Although hTERT transcription can be regulated by alternativesplicing during fetal development, its expression is repressed infully differentiated somatic tissues [9]. Thus, thehTERTpromoter hasbeen widely used in tumor targeting gene therapy research [10].Unfortunately, in our previous experiments, we found that thetumor targeting transcriptional activity of the core hTERT promoterwas insufficient in hBMSC-derived malignant cells.

Cis transcriptional regulation is a powerful driving force in theevolution of function and form [11,12]. Studies have demonstratedthat the core hTERT promoter contains two putative c-Myc bindingsites (E-box), positioned at the−242 and−34 positions (from theATG start site). c-Myc binding to the -34 E-box could stimulate theactivity of the hTERT promoter [12,13]. It has been confirmed that

extra repeats of c-Myc binding elements at the 3′ end of the corehTERT promoter sequence could enhance the binding potentialbetween factors and elements, ultimately resulting in theimprovement of tumor-targeting transcriptional activity in malig-nant cells [14]. However, results also showed that the transcrip-tional activity of such a modified hTERT promoter was still presentat basal expression levels in primary human fibroblasts [14]. Thismeans that patients with normal transgenic BMSC grafts could beinjured after administration of pro-drugs such as 5-FC. It has beenreported that MZF-2 is a negative regulator of the hTERT promoterand is only expressed in a few cell types, including myeloid cells[15]. Although there is no evidence to show that MZF-2 is highlyexpressed in hBMSC and at lower levels in hBMSC-derivedmalignant cells, studies have demonstrated that hBMSC do notpossess telomere maintenance mechanisms [16] and malignantcells are associated with P53 mutations [17,18], chromosomalabnormalities, and increased levels of telomerase activity, Bcl2expression and c-Myc expression [19].

Therefore, we hypothesize that MZF-2 is a negative regulator ofhTERT, and is most likely highly expressed in normal hBMSC, but itsexpression is decreasedwhenhBMSC transform intomalignant cells.Enrichmentof c-Mycbindingelement repeats in the hTERTpromoterwould result in improved binding potential between c-Myc factorsand E-boxes, ultimately leading to an improvement in transcriptionalactivity inmalignant cells. By contrast, enrichment ofMZF-2 bindingelement repeats in the hTERT promoter would lead to suppresstranscriptional activity in normal hBMSC. To test this hypothesis anderadicate hBMSC-derivedmalignant cells, we designed this study todeterminewhether co-modificationof the c-Myc andMZF-2 bindingelements in the hTERT promoter would result in an enhancement ofthe transcriptional activity of the hTERT promoter in malignant cellsand transcriptional repression in normal hBMSC.We also wanted todetermine whether hBMSC-derived malignant cells could beeliminated with the CD/5-FC pro-drug system under the guidanceof such a modified hTERT promoter.

Materials and methods

DNA clone

PromotersThe DNA fragments of the WT-hTERT promoter, Mod1-hTERTpromoter and Mod2-hTERT promoter (DNA motifs shown in

3331E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

Fig. 1A) were generated by chemical synthesis and subcloned intothe PUC-57 plasmid vector (Clontech). The newly generatedplasmids were named as PUC57-WThTERT, PUC57-Mod1hTERTand PUC57-Mod2hTERT, respectively.

CD coding sequenceThe CDwas amplified with PCR as described previously [12]. A pairof primers (sense primer: 5′-TAAACTAGTAATGTCGAATAACGC-3′;antisense primer: 5′-TATCTCGAGGAATGAATCACGGTAG-3′) wasdesigned according to the published sequence of the CD gene. ThePCR profile was 30 cycles of 94 °C (45 s), 56 °C (30 s), 72 °C (2min).

EGFP coding sequenceFor PCR amplification of enhanced green fluorescent protein(EGFP) from the pEGFP-C1 plasmid (Clontech), a pair of primers(sense primer: 5′-CACCATGGTGAGCAAGGGC-3′; antisense prim-er: 5′-ACAACACTCAACCCTATCTCGGTCT-3′) was used. The PCRprofile was 30 cycles of 94 °C (45 s), 52 °C (30 s), 72 °C (2 min).

Vector construction

The DNA fragment of -luciferase-IRES- came from digesting Plasmidsof pTelev-Coo2 (kindly provided by Dr. Guanghua Yang, http://televector.com/english/index.asp). Fragments of WT-hTERT, Mod1-hTERT,Mod2hTERT, EGFP, CD, –luciferase-IRES-, and lentivirus vector ofpLenti6/V5-D-TOPO vector (Invitrogen)were used to construct vectorsof pLenti-CMV-EGFP, pLenti-WThTERT-EGFP, pLenti-Mod1hTERT-EGFP,pLenti-Mod2hTERT-EGFP, pLenti-CMV-luciferase-IRES-CD, pLenti-WThTERT-luciferase-IRES-CD, pLenti-Mod1hTERT-luciferase-IRES-CD,pLenti-Mod2hTERT-luciferase-IRES-CD and pLenti-no promoter-luciferase-IRES-CD, according to the directions of Fig. 2.

Packaging of replication-defective lentivirusExperimental protocols were approved by the Shanghai TELEBIOBIOMEDICAL CO., LTD (www.televector.com). Briefly, for allabove constructed vectors, particles pseudotyped with thevesicular stomatitis virus G glycoprotein (VSV-G) were producedin a three-plasmid expression system by transient transfection ofhuman 293T cells with a defective packaging construct, aplasmid with the VSV-G coding region driven by the HIV LTR,

Fig. 1 – Sequences of modified hTERT promoters: The sequence fromWT-hTERT promoter; the mod1hTERT promoter consisted of the Wthe Mod2hTERT promoter consisted of the Mod1-hTERT promoter apromoter.

and a HIV-1-based vector construct. Five microgams of DNA ofeach plasmid was co-transduced into 293T cells by using calciumphosphate precipitation method. Cells were seeded into six-wellplates 24 to 30 h prior to transfection. Chloroquine (25 μM, finalconcentration) was added to the cells immediately beforetransfection, and the medium was replaced with 2 mL (perwell) of fresh DMEM supplemented with 10% FBS 12 to 14 h later.The virus was harvested 60 to 65 h later, filtered through aMillipore Millex-HA 0.45 μm filter unit, aliquoted, and frozen at80 °C [20]. The viral titer was tested by transducing HeLa cells(provided by Dr. Guanghua Yang) and genomic DNA fromtransgenic HeLa cells was extracted with the Mammaliangenomic DNA extraction kit (Invitrogen). Vector titer ofreplication-competent virus was derived from quantitativereal-time PCR analysis by using a transfected HeLa cell line. Apair of CD-specific primers (sense primer: 5′-AGTGTCGAA-TAACGC-3′, antisense primer: 5′-GAATGAATCACGGTAG-3′) anda DNA probe (FAM-ATGCCCTCCCCCATGCCATCCTGCGT-TAMRA)were used for CD amplification from HeLa cell genomic DNA. RT-PCR was performed with 30 cycles of 95 °C (15 s), 46 °C (25 s),72 °C (1 min).

Cell culture

Cell lineAn hTERT-positive cell line, 293FT (Invitrogen), was cultured inDMEM/Ham's nutrient mixture F-12 (GIBCO, with a ratio of 1:1)and 10% FCS (GIBCO), 100 U/mL streptomycin (Hyclone), and100 μg/mL penicillin (Hyclone) in 5% CO2 environment at 37 °C. Theportion of proliferating cells that were used for lentiviral infectionand the blasticidin-resistant infected cells was named as BR-293FT,and the rest of the cells were used for quantitative PCR assays.

hBMSC cultureBone marrow cells were harvested, under local or generalanesthesia, from a healthy hematopoietic stem cell donors (21years old), after obtaining written informed consent. 20 millilitersof heparinized BM were used for MSC generation and expansion.The institutional review board of Children's Hospital of ChongqingMedical University approved the design of this study.

−267 to−8 site of the hTERT gene promoter was defined as theT-hTERT promoter and an insertion containing three E-boxes;nd an inserted sequence from bp −764 to −584 of the hTERT

Fig. 2 – Schematic diagram of lentiviral vector backbone: (A) pLenti-CMV-EGFP; (B) pLenti-WThTERT-EGFP; (C)pLenti-Mod1hTERT-EGFP; (D)pLenti-Mod2hTERT-EGFP; (E)pLenti-CMV-luciferase-IRES-CD; (F)pLenti-WThTERT-luciferase-IRES-CD;(G) pLenti-Mod1hTERT-luciferase-IRES-CD; (H) pLenti-Mod2hTERT-luciferase-IRES-CD; (I) pLenti-no promoter-luciferase-IRES-CD.

3332 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

Cell culturehBMSC culture was performed as described previously [16];passage 3 cells were used for assays of phenotype, characteriza-tion, differentiation, western blots, quantitative PCR, PALA selec-tion, and lentiviral transduction.

Phenotype and characterization analysisFlow cytometry assays were used to phenotypically characterizehBMSC and to define their purity, which were performed asdescribed as reference [21] with a minor alteration. Briefly,passage 3 (P3) of hBMSC was treated with 0.25% trypsin-EDTA,harvested, and washed twice with PBS. The cells were incubatedon ice with labeled mouse anti-human antibodies (PE-conjugat-

ed) for CD14, CD34, CD45, CD71, CD73, CD90, CD105 and HLA-DR(BD PharMingen). The labeled cells were analyzed by flowcytometry.

Evaluation of cells differentiation capacity

The experiments were performed as describe in ref. [3] with aminor alteration. Briefly, for adipogenic differentiation, cellswere cultured in MEM plus 10% FCS, 0.5 mmol/L 3-isobutyl-1-methylxanthine, 0.5 mmol/L hydrocortisone, 1 mmol/L dexa-methasone, 200 mmol/L indomethacin, and 50 μg/mL gentamicinfor 2 weeks. Differentiated cells were stained with Sudan Blank B(Sigma) for evaluation of adipogenic differentiation. Passage 3 of

3333E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

hBMSC was cultured in MEM plus 10% FCS, 0.1 mmol/Ldexamethasone, 50 mmol/L ascorbate-2-phosphate, 10 mmol/Lβ-glycerophosphate, and 50 μg/mL gentamicin for 2 weeks.Differentiated cells were stained with alizarin red (Sigma) forevaluation of osteogenic differentiation.

BPDE treatment and lentiviral transfection

Passage 2 or 3 hBMSCwere treated with 1.0 μMBPDE (Sigma) for 1h [22]. The medium was replaced with BPDE-free medium and aportion of the cells was used for PALA selection. The rest of the cellswere transferred to 12-well plates for 48 h prior to lentiviraltransduction. Lentiviral transduction was performed as describedby Damme et al. [23].

Blasticidin selection

After 48 h of lentiviral transduction, cells were plated in 6-wellplates at a density of 2.5×105 cells per well, and incubated at 37 °Cfor 48 h. At this time, themediumwas replacedwith freshmediumcontaining 10 μg/mL blasticidin (Invitrogen). One to two weekslater, the blasticidin-resistant hBMSC (BR-hBMSC or BPDE/BR-hBMSC) clones had formed.

Selection of PALA

PALA was obtained from Sigma. Selections were performed asdescribed by Perry et al. [24]. Briefly, hBMSC (with BPDEtreatment), BR-hBMSC (no treatment with BPDE), and BR-hBMSC (with lentiviral transduction and BPDE treatment) wereallowed to proliferate on 10 cm dishes (5×104 cells/dish) andexposed to a low, nonselective concentration of PALA (10 μM).After 72 h, the amount of PALA was increased to a selectiveconcentration of 30 μM, and the cells were kept in this selectivecondition for 5 to 6 weeks until PALA-resistant hBMSC (PBR-hBMSC) colonies formed. The PBR-hBMSC were expanded to5×1010 and used for western blotting, quantitative PCR, luciferaseassays, and 5-FC cytotoxicity detections.

Quantitative PCR assay for c-Myc, Bcl2, P53, andMZF-2 expression

Total RNA extractionRNA was extracted from hBMSC (treated with BPDE), BR-hBMSC(with BPDE treatment), PBR-hBMSC and HepGII by using TRIZOLreagent (Invitrogen) according to the manufacturer's instructions.The total RNA was identified by agarose gel electrophoresis (AGE).

Synthesis of first strand cDNATotal RNA (2 μg) was transcribed with reverse transcriptase(Promega). The first strand cDNA was used as the template forquantitative PCR analysis.

Quantitative PCRThe 236 bp c-Myc cDNA was amplified with the primers 5′-TGGTCTTCCCCTACCCTCTCA-3′ and 5′-TTCTTCCTCATCTTCTT-GTTCC-3′. The 188 bp P53 cDNA was amplified with the primers5′-GATGACAGAAACACTTTTCGACATAG-3′ and 5′-CTGTTCCGTCC-CAGTAGATTACC-3′. The 255 bp Bcl2 cDNA was amplified withthe primers 5′-CCAGATCCCAGAGTTTGAGCC-3′ and 5′-CCC-

ATCCCTTCGTCGTCCT-3′. The 394 bp MZF-2 cDNA was amplifiedwith the primers 5′-GGACTCTGAGGAGGAGGGTGA-3′ and 5′-AGG-TAGGGGCTGGAAACTGGA-3′. The 236 bp cDNA of the internalcontrolGAPDHwas amplifiedwith the primers 5′-ATGGGGAAGGT-GAAGGTCGG-3′ and 5′-TCCTGGAAGATGGTGATGGG-3′. GAPDHcDNA was used as an internal control to normalize variances.

Results were expressed as the mean±SEM. Statistical analysiswas performed using the one-way analysis of variance (ANOVA). Aprobability of the null hypothesis <5% (P<.05) was consideredstatistically significant.

Western blot assay for c-Myc, Bcl2, P53, andMZF-2 expression

To further determine the expression levels of cell cycle-regulatedproteins in hBMSC, BR-hBMSC, PBR-hBMSC, and 293FT (amalignant positive control), protein levels were detected bywestern blotting with antibodies against c-Myc, Bcl2, P53, andMZF-2. Western blotting assays were performed as described byMunna et al. [25].

Observation of expressions of EGFP

Cells containing hBMSC/WThTERT-EGFP, hBMSC/Mod1hTERT-EGFP,hBMSC/Mod2hTERT-EGFP , hBMSC/CMV-EGFP , BR-hBMSC/WThTERT-EGFP, BR-hBMSC/Mod1hTERT-EGFP, BR-hBMSC/Mod2-hTERT-EGFP, BR-hBMSC/CMV-EGFP, PBR-hBMSC/WThTERT-EGFP,PBR-hBMSC/Mod1hTERT-EGFP, PBR-hBMSC/Mod2hTERT-EGFP, andPBR-hBMSC/CMV-EGFP were seeded in 96-well flat-bottomedmicro-plates at 2000 cells per well. Each well was observed byfluorescence microscopy at 488 to 507 nm wavelengths afterallowing the cells to grow to 70 to 80% confluence.

Luciferase activity analysis

The Luciferase Assay System (Promega) was used for the luciferaseactivity analysis in this study. Lentivirally transfected hBMSC, BR-hBMSC, PBR-hBMSC and 293FT cells were plated at 2000 cells perwell into 96-well flat-bottomed micro-plates prior to the 48-h assay. Assays were performed according to the manufacturers'instructions. Luciferase activity values were calculated as de-scribed previously [14].

Cytotoxicity measurements

In vitro cytotoxicity measurementsCells containing PBR-hBMSC/-luciferase-IRES-CD (no promoter),PBR-hBMSC/WThTERT-luciferase-IRES-CD, PBR-hBMSC/Mod1-hTERT-luciferase-IRES-CD, PBR-hBMSC/Mod2hTERT-luciferase-IRES-CD, and PBR-hBMSC/CMV-hTERT-luciferase-IRES-CD wereplated at 2000 cells per well into 96-well plates. In vitrocytotoxicity measurements were performed as described by Kaiet al. [8].

Eradicating malignant cells in vivoBalb/c athymic mice were purchased from the Center ofExperimental Animals of Third Military Medical University. Cellimplantation and 5-FC administration were performed asdescribed by Kai et al. [8] with minor modifications. Briefly, 60male mice were randomly divided into A, B and C groups, witheach group consisting of 20 mice. Group A mice were injected

3334 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

with PBR-hBMSC/WThTERT-luciferase-IRES-CD, group B mice wereinjected with PBR-hBMSC/Mod2hTERT-luciferase, and group Cmice were injected with PBR-hBMSC/CMV-promoter-luciferase-IRES-CD. Prior to injection, cells were dispersed and washed 3times in PBS. Approximately 5×106 cells, suspended in 200 μL ofserum-free IMDM, were injected subcutaneously into themouse's armpit. Two weeks after cell implantation, all micewere observed with whole-animal bioluminescence imagingafter intraperitoneal injection (i.p.) of D-luciferin (Promega) at375 mg/kg. Mice with surviving implanted cells in each groupwere used for in vivo cytotoxicity test. For in vivo cytotoxicitytest, 5-FC was diluted in 0.9% NaCl solution to a concentration of10 mg/mL. Each mouse was given an i.p. injection of 0.5 mL of 5-FC once a day. The signals of surviving engrafted cells wereobserved once a week; the maximum signal was used as the finalexperimental result.

Fig. 3 – Characteristics of hBMSC: (A) Human bone marrow-derived(b). (B) FCM analysis for surface antigens: hBMSCwere positive for Cand HLA-DR. (C) Cell differentiation capacity. hBMSC were stainedadipocyte, and stained with alizarin red S after induced differentia

Results

Phenotype analysis

Bonemarrowmesenchymal stromal cells were prepared by percollgradient centrifugation technique which consisted of two majortypes of cells, mesenchymal stem cells and hematopoietic stemcells. Hematopoietic stem cells did not attach to the culture dish,and were washed away by the changes of culture medium. ThehBMSC were spindle-shaped, attached to the culture dish tightly,and proliferated in the culture medium (Fig. 3A); hBMSC showedactive proliferative capacity in vitro with primary and passageculture. After in vitro proliferative culture to passage 3, cells wereused for phenotypic analysis with FCM. The results showed thatthe isolated cells were positive for CD71 (93.2%), CD73 (66.1%),

mesenchymal cells in vitro culture for 2 weeks (a) and 4 weeksD71, CD73, CD90, and CD105, and negative for CD14, CD19, CD34,with Sudan black B (a) after induced differentiation intotion into osteoblast (b).

3335E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

CD90 (52.9%), CD105 (91%), and negative for CD14 (14.2%), CD19(22.5%), CD34 (8.1%), and HLA-DR (15.2%) (Fig. 3B). To furtherevaluate the isolated cell's differentiation capacities, hBMSC wereinduced into adipocyte and osteoblast, respectively. The results ofdifferentiation showed that the blank lipid droplets (after staining

Fig. 4 – PALA selection for malignant cell colony formation: Coloniunderwent co-inducement with lentiviral transduction and BPDE tinhibition after 288 h (A); however, this characteristic was not fouBPDE treatment alone (C) throughout PALA selection.

with Sudan blank B) were found in isolated cells treated withadipocyte-induced medium (Fig. 3C-a), and the cells cultured inosteoblast-induced medium were stained into orange afterstaining with alizarin red S (Fig. 3C-b). These results indicatedthat characterizations of stem cells were found in the isolated cells.

es of malignant cells were formed in the hBMSC group thatreatment after PALA selection for 144 h, and lost contactnd in cells that underwent lentiviral transduction alone (B) or

Fig. 5 –Malignant transformation assays: (A) Statistical data analysis of cell cycle regulatory genes c-Myc (n=18), Bcl2 (n=18), P53(n=6) and MZF-2 (n=18). (B) AGE of quantitative PCR product. (C) Cell-morphous of 293FT, hBMSC, BR-hBMSC and PBR-hBMSC.(D) Morphous of cells before using for western blot assays. (E) Western blot assays to analyze expression of cell cycle regulatorygenes.

3336 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

3337E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

Formation of PALA-resistant colonies

To generate colonies of malignant transformed cells, malignanttransformation of hBMSC was co-induced by BPDE treatment andlentiviral transduction. PALA selection was used for isolation andproliferation of the hBMSC-derived malignant cells. Resultsindicated that malignant cell colonies formed in the hBMSCgroup with co-induction of lentiviral transduction and BPDEtreatment after PALA selection for 144 h, and exhibited loss ofcontact inhibition after incubation for 288 h in PALA selectionconditions. However, we have not seen the formation of malignant

Fig. 6 – Transcriptional activity of modified hTERT promoters: ExpreCMV promoters in cell groups (A) hBMSC, (B) BR-hBMSC, and (C) PBactivity under the control of theWThTERT promoter (n=18),Mod1different cell types.

cell colonies in cells treated with either BPDE treatment orlentiviral transduction alone (Fig. 4). This result also revealedthat lentiviral transduction does not cause the biosafety problemsof transgenic hBMSC for clinical implantation.

Transformation assays

For transformation assays of malignant cells, quantitative PCR andwestern blotting were used for detection of c-Myc, Bcl-2, P53 andMZF-2 expression in 293FT, hBMSC, BR-hBMSC and PBR-hBMSC.The quantitative PCR results revealed that the expression levels of

ssion of EGFP under the control ofMod1hTERT, Mod2hTERT andR-hBMSC, respectively. (D) Statistical data analysis of luciferasehTERT promoter (n=18) andMod2hTERT promoter (n=18), in

3338 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

P53 and MZF-2 were low in 293FT and PBR-hBMSC, whereas theexpression levels of c-Myc and Bcl2 increased remarkably in BPR-hBMSC (P<0.01), but no significant difference was found betweenthe hBMSC and BR-hBMSC groups (Fig. 4 and Supplemental Tables1–4). This result was further supported by western blot assays(Fig. 5). These findings have revealed that hBMSC could betransformed into malignant cells by co-induction of lentiviraltransduction and chemical carcinogen treatment in vitro.

Observation of EGFP expression

hBMSC do not express telomerase and have been shown to be atarget for malignant transformation after co-induction withcarcinogenic reagents and viral transduction. To detect whetherthe expression of hTERT activities was changed in hBMSC-derivedmalignant cells, the expression of the EGFP reporter in the cells wasobserved by fluorescence contrast phase microscopy. The resultsshowed that EGFP expression under the control of the Mod1hTERTand Mod2hTERT promoters could be seen in PBR-hBMSC; EGFPunder the control of the CMV promoter could also be detected;however, we could not detect EGFP expression under the control ofthe WT-hTERT promoter in any cell group (data not shown);expression could not be detected in hBMSC (with BPDE treatment)and BR-hBMSC when EGFP was under the control of theMod1hTERT and Mod2hTERT promoters (Fig. 6). These results notonly demonstrated that hTERT activity was enhanced in hBMSC-

Fig. 7 – Cytotoxicity measurements. (A) MTT assay for cytotoxicity(n=6) and CMV (n=6) promoters. (B) Bioluminescent imaging mPBR-hBMSC/WThTERT-luciferase-IRES-CD (n=5); PBR-hBMSC/Modluciferase-IRES-CD (n=4). (C) Bioluminescence imaging of mice wpoints of 5-FC administration.

derived malignant cells, but also revealed that transcriptionalactivity of the core hTERT promoter could be enhanced by co-modification of c-Myc binding elements in hBMSC-derivedmalignant cells.

Functional evaluation of modified hTERT promoters

To further evaluate the function of the hTERT promoter co-modified with c-Myc and MZF-2 binding elements, expression ofa luciferase reporter under the control of different hTERTpromoters was detected by a luciferase activity assay. The resultsshowed that the transcriptional activity of the Mod1hTERTpromoter was about 3.2-fold higher than that of the WThTERTpromoter in PBR-hBMSC and 293FT malignant cells, but nosignificant difference was found in hBMSC or BR-hBMSC.Transcriptional activity of the Mod2hTERT promoter was about4.7-fold lower than that of the Mod1hTERT promoter in hBMSCand 5.4-fold in BR-hBMSC, but no significant difference was foundin 293FT or PBR-hBMSC malignant cells (Fig. 6). These resultsindicated that 293FT and hBMSC-derived malignant cells had highlevels of telomerase activity. These results are consistent with thestudy by Sommer et al. [18]. Additionally, these results suggestedthat the c-Myc binding sites could enhance transcriptional activityin hBMSC-derived malignant cells, whereas MZF-2 bindingelements could decrease transcriptional activity in hTERT-negativehBMSC (Fig. 6).

mediated by WThTERT (n=6), Mod1hTERT (n=6), Mod2hTERTeasuring eradication of engrafted malignant cells containing2hTERT-luciferase-IRES-CD (n=4); and PBR-hBMSC/CMV-ith surviving engrafted malignant cells after different time

3339E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

In vitro cytotoxicity measurements

To evaluate the efficiency of specific killing of hBMSC-derivedmalignant cells by the pro-drug system CD/5-FC under the controlof different promoters in vitro, MTT assay was used for testing thesurvival rates in PBR-hBMSC after 5-FC administration. The relativesurvival rates of cells were only 5.79% in the Mod1hTERT group,6.53% in theMod2hTERT group, and 1.97% in the CMV group, whentreated with 100 μM 5-FC. Under the same conditions, survivalrates were 96.27% in the group lacking a promoter and 40.13% inthe WThTERT group. We also found that the group lacking apromoter had 81.63% and 51.53% survival rates after treatmentwith 1000 and 10000 μM of 5-FC, respectively (Fig. 7).

Transplantation and eradication of malignant cells in vivo

Twoweeks after implantation, there were five mice with survivingimplanted cells in group A, five in group B and four in group C.Bioluminescent assays were used to measure inhibition ofsurviving implanted cells in each group. About 15 to 20 minafter D-luciferin administration, a strong bioluminescent signalwas observed, and the maximum bioluminescent signal wasdetected between 30 and 50 min after administration. The resultshowed that maximum signal intensities of mice with PBR-hBMSC/WT-hTERT-luciferase-IRES-CDwere 4.33-fold lower than mice withPBR-hBMSC/mod2-luciferase-IRES-CD and 5.29-fold lowerthan mice with PBR-hBMSC/CMV-luciferase-IRES-CD after 5-FCadministration at the zero week time point, but no significantdifference was found after 2 to 3 weeks of 5-FC administration. Inmice with PBR-hBMSC/mod2-luciferase-IRES-CD and PBR-hBMSC/CMV-luciferase-IRES-CD, the bioluminescence signal vanished afterabout 4 to 5 weeks of 5-FC administration; however, signals werestill found in mice with PBR-hBMSC/WT-hTERT-luciferase-IRES-CD,even after 5-FC administration for 5 weeks, (Fig. 7).

Discussion

For stem cell therapy, themost serious problem is tumor formation.It is the responsibility of the community of stem cell scientists andclinicians to develop strategies to control the therapies aftertransplantation [1]. However, this strategy must face manychallenges, including high efficiency of tumor-targeting promoterdesign of mechanisms, obtaining a number of hBMSC-derivedmalignant cells for functional analysis of promoters, and lowsurvival rate of malignant cells after heterogenic implantation, etc.

Although, malignant transformations of stem cells have beenwidely reported in rodents [26,27], but malignant transformationof hBMSC is very rare [24,25,28] and no study has been undertakento study prevention of malignant transformation and eradicationof malignant transformed cells after implantation. This study doesnot show that the chemical carcinogen PBDE directly induces theup-regulation of the cad (carbamyl phosphate synthetase/aspar-tate transcarbamylase/dihydro-orotase) gene and ultimatelyresults in generation of PALA-resistant cells, but research hasshown that treatment with chemical carcinogens (e.g., BPDE) [22]and viral infection [20] may lead to P53 genetic mutation andinactivation, which in turn leads to the up-regulation of cad genesand c-Myc and Bcl2 transcription factors [19]. This ultimatelyresults in mutation and produces PLAL resistance [24]. Our

experimental results showed that treatment of hBMSC withlentiviral infection and PBDE induction led to loss of P53expression, which enabled the hBMSC with defects in DNAsynthesis to pass the P53-controlled cell cycle checkpoint andenter the cell division cycle, allowing cell proliferation [26]. Thenumber of cells being malignantly transformed by mutationaccounted for a very small portion of the total hBMSC population,so conventional methods like quantitative PCR and west blottingfailed to detect the expression of relevant genes in BR-hBMSC.Thus, quantitative PCR and western blotting experiments showedthat P53 expression had been down-regulated in the 293FT cellline and PALA-resistant PBR-hBMSC, whereas no difference ofexpression of c-Myc and Bcl2 was found in hBMSC and BR-hBMSCin the absence of PALA treatment.

Apart from loss of P53 and enhanced expression levels ofcarcinogenic genes (including c-Myc), another marker of malignanttransformation is an increase in telomerase expression. As an up-regulator of the hTERT promoter, increased telomerase expression inmalignant cells can be attributed to the increased levels ofcarcinogenic gene expression, such as that of c-Myc and Bcl2 [19].It has been shown in our previous studies that the core WT-hTERTpromoter may improve the hTERT promoter's transcriptionaltargeting activities in tumor cells after enrichment with extra c-Myc binding site elements [14]. In this study, EGFP expressionobservations demonstrated that EGFP under the control of theMod1hTERT or Mod2hTERT promoters was expressed at high levelsin PBR-hBMSC,while nohigh-level expression of EGFPwas observedin normal hBMSC or in BR-hBMSC that have not undergone PALAselection. This result further proved that this hBMSC-derived PBR-hBMSC had increased c-Myc, Bcl-2 and telomerase expression. Thereason that EGFP expression under the control of the Mod2hTERTpromoter was not detected in BR-hBMSC may be due to the smallnumbers of malignantly transformed cells in BR-hBMSC. Theluciferase activity analysis demonstrated that, after being modifiedby c-Myc binding site elements, the transcriptional activity of thehTERTpromoter did not change in hBMSCbut increased significantlyin 293FT tumor cells and malignant PBR-hBMSC.

In addition, we found that MZF-2 expression was high inhBMSC but was undetectable in PBR-hBMSC, suggesting that MZF-2 expression decreased or even disappeared upon malignanttransformation of hBMSC. As a myeloid-specific negative regulatorfor hTERT promoter transcription, MZF-2 inhibits hTERT transcrip-tion by binding to hTERT promoter DNA-specific elements [15]. Allof these results support the model that the Mod1hTERT promoterretains high transcriptional activity in PBR-hBMSC in the absenceof MZF-2 factors after the modulation of MZF-2 binding siteelements. Promoter transcriptional activity was inhibited innormal hBMSC because they contained a large number of MZF-2factors. Previous studies indicate that, unlike other promoters, thehTERT promoter is methylase-dependent and one explanationmaybe that after methylation of CpG island sequences near the MZF-2binding site elements, the altered DNA structure prevents bindingbetween MZF-2 and its cis-elements. This could lead to the releaseof MZF-2's inhibition on hTERT promoter transcription andactivation of hTERT promoter transcriptional activity. Though it isuncertain whether inhibition of the core WThTERT promoter inhBMSC was related to the methylation of CpG island sequences,the luciferase activity analysis has shown that the core hTERTpromoters with enrichment in c-Myc or MZF-2 binding siteelements had high transcriptional activity in PBR-hBMSC and

3340 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

293FT malignant cells, but had low activity in normal hBMSC andBR-hBMSC grown in the absence of PALA selection (Fig. 5).

Limited by the lowsurvival rate (approximately 25% inour study)of malignant cells after heterogenic implantation, whole-animal invivo bioluminescence imaging was used for noninvasive monitoringof elimination efficiency of the CD/5-FC system in PBR-hBMSC. Fordetection of elimination ofmalignant cells, themost commonly usedmethod to determine engrafted cell load in the abdominal cavity (orin tegument on animals) is to kill the animals and score the size andnumber of new growths at various times after drug treatment[29,30]. It is impossible to evaluate the growth of engraftedmalignant cells and response to 5-FC treatment in such a fewanimals using such an invasive procedure. It has been reported thatintravital imaging of cancer stem cells could be very valuable fordeterminingprognosis, aswell as formonitoring therapeutic efficacyand influencing therapeutic protocols. Cancer stem cells represent arare population of cells, as low as 0.1% of cellswithin a human tumor,and thephenotypeof isolatedcancer stemcells is easily alteredwhentransferred to conditions [31]. Whole-animal imaging enables muchsimpler ongoing tracking of engrafted malignant hBMSC involved intime-dependent processes, such as development, growth andapoptosis in individual animals because noise from inter-individualvariation is eliminated, and animals are used more efficiently [31].

In summary, we have demonstrated that co-inducementwith viral transfection and treatment with a carcinogenic agentcould lead to cad gene amplification and generation of PALA-resistant hBMSC-derived malignant cells (despite the lack ofdetection in this study), indicating that the Mod2hTERTpromoter presents a potential way for gene therapy to preventthe malignant transformation of engrafted hBMSC, and suggeststhat whole-animal in vivo bioluminescence imaging could makethe study of the eradication of engrafted malignant cells moreefficient.

Acknowledgments

This study was supported by the foundation of 111 project, No:B06023, China, Nature Science Foundation of China (NSFC), No.10872224, China, and 973 Project Foundation, No. 2005CB522703,China.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.yexcr.2010.02.014.

R E F E R E N C E S

[1] M. Kiuru, J.L. Boyer, T.P. O'Connor, R.G. Crystal, Genetic Controlof wayward pluripotent stem cells and their progeny aftertransplantation, Cell Stem Cell 4 (2009) 289–300.

[2] J.M. Houghton, C. Stoicov, S. Nomura, A.B. Rogers, J. Carlson, H.Li, X. Cai, J.G. Fox, J.R. Goldenring, T.C. Wang, Gastric canceroriginating from bone marrow-derived cells, Science 306 (2004)1568–1571.

[3] D. Rubio, J. Garcia-Castro, M.C. Martín, Martín, R. de la Fuente,A.C. Lloyd, A. Bernad, Spontaneous human adult stem celltransformation, Cancer Res. 65 (2005) 3035–3039.

[4] D.J. Rossi, I.L. Weissman, Pten, tumorigenesis and stem cellself-renewal, Cell 125 (2006) 229–231.

[5] I. Avital, A.L. Moreira, D.S. Klimstra, M. Leversha, E.B.Papadopoulos, M. Brennan, R.J. Downey, Donor-derived humanbone marrow cells contribute to solid organ cancers developingafter bone marrow transplantation, Stem Cells 25 (2007)2903–2909.

[6] R. Izadpanah, D. Kaushal, C. Kriedt, F. Tsien, B. Patel, J. Dufour,B.A. Bunnell, Long-term in vitro expansion alters the biologyof adult mesenchymal stem cells, Cancer Res. 68 (2008)4229–4238.

[7] M. Akimoto, T. Miyahara, J. Arai, H. Hamada, Y. Yoshida, Y.Yoshida, N. Yoshimura, A new delivery system for 5-fluorouracilusing prodrug and converting enzyme, Br. J. Ophthalmol. 86(2002) 581–586.

[8] K. Ge, L. Xu, Z. Zheng, D. Xu, L. Sun, X. Liu, Transduction of cytosinedeaminase gene makes rat glioma cells highly sensitive to5-fluorocytosine, Int. J. Cancer 71 (1997) 675–679.

[9] S. Carroll, Evolution at two levels: on genes and form, PLoS Biol. 3(2005) 1159–1166.

[10] A. Kinkhabwala, C.C. Guet, Uncovering cis regulatory codes usingsynthetic promoter shuffling, PLoS ONE 3 (2008) e2030, doi:10.1371/journal.pone.0002030.

[11] N.W. Kim, M.A. Piatyszek, K.R. Prowse, C.B. Harley, M.D. West, P.L.Ho, G.M. Coviello, W.E. Wright, S.L. Weinrich, J.W. Shay, Specificassociation of human telomerase activity with immortal cells andcancer, Science 266 (1994) 2011–2015.

[12] T. Komata, S. Koga, S. Hirohata, M. Takakura, I.M. Germano, M.Inoue, S. Kyo, S. Kondo, Y. Kondo, A novel treatment of humanmalignant gliomas in vitro and in vivo: FADD gene transfer underthe control of the human telomerase reverse transcriptase genepromoter, Int. J. Oncol. 19 (2001) 1015–1020.

[13] A. Sommer, K. Bousset, E. Kremmer, M. Austen, B. Luscher,Identification and characterization of specific DNA-bindingcomplexes containing members of the Myc/Max/Mad networkof transcriptional regulators, J. Biol. Chem. 273 (1998)6632–6642.

[14] Y.-B. Yang, S.-X. Cai, S.-H. Yu, L. Yang, J. Li, S.-T. Yu, X.-Z. Dai, X.-X.Wang, X.-Q. Yan, X. Xin, K.L.P. Sung, hTERT core promotermodified by E-boxes element and transcriptional activity assay invitro, Chin. J. Biochem. Mol. Biol. 25 (2009) 510–514.

[15] K. Fujimoto, S. Kyo, M. Takakura, T. Kanaya, Y. Kitagawa, H. Itoh,M. Takahashi, M. Inoue, Identification and characterization ofnegative regulatory elements of the human telomerase catalyticsubunit (hTERT) gene promoter: possible role of MZF-2 intranscriptional repression of hTERT, Nucleic Acids Res. 28 (2000)2557–2562.

[16] M.E. Bernardo, N. Zaffaroni, F. Novara, A.M. Cometa, M.A.Avanzini, A. Moretta, D. Montagna, R. Maccario, R. Villa, M.G.Daidone, O. Zuffardi, F. Locatelli, Human bone marrow derivedmesenchymal stem cells do not undergo transformation afterlong-term in vitro culture and do not exhibit telomeremaintenance mechanisms, Cancer Res. 67 (2007) 9142–9149.

[17] T. Else, A. Trovato, A.C. Kim, Y. Wu, D.O. Ferguson, R.D. Kuick, P.C.Lucas, G.D. Hammer, Genetic p53 deficiency partially rescues theadrenocortical dysplasia phenotype at the expense of increasedtumorigenesis, Cancer Cell 15 (2009) 465–476.

[18] Y. Wang, J. Yang, H. Zheng, G.J. Tomasek, P. Zhang, P.E. McKeever,E.Y.-H.P. Lee, Y. Zhu, Expression of mutant p53 proteinsimplicates a lineage relationship between neural stem cells andmalignant astrocytic glioma in a murine model, Cancer Cell 15(2009) 514–526.

[19] Z. Jin, W.S. May, F. Gao, T. Flagg, X. Deng, Bcl2 suppresses DNArepair by enhancing c-Myc transcriptional activity, J. Biol. Chem.281 (2006) 14446–14456.

[20] R. Foddis, A. De Rienzo, D. Broccoli, M. Bocchetta, E. Stekala, P.Rizzo, A. Tosolini, J.V. Grobelny, S.C. Jhanwar, H.I. Pass, J.R. Testa,M. Carbone, SV40 infection induces telomerase activity in humanmesothelial cells, Oncogene 21 (2002) 1434–1442.

3341E X P E R I M E N T A L C E L L R E S E A R C H 3 1 6 ( 2 0 1 0 ) 3 3 2 9 – 3 3 4 1

[21] W. Xu, X. Zhang, H. Qian, W. Zhu, X. Sun, J. Hu, H. Zhou, Y. Chen,Mesenchymal stem cells from adult human bone marrowdifferentiate into a cardiomyocyte phenotype in vitro, Exp. Biol.Med. 229 (2004) 623–631.

[22] S. Perwez Hussain, P. Amstad, K. Raja, M. Sawyer, L. Hofseth,P.G. Shields, A. Hewer, D.H. Phillips, D. Ryberg, A. Haugen,C.C. Harris, Mutability of p53 hotspot codons tobenzo(a)pyrene diol epoxide (BPDE) and the frequency of p53mutations in nontumorous human lung, Cancer Res. 61 (2001)6350–6355.

[23] T. Dull, R. Zufferey, M. Kelly, R.J. Mandel, M. Nguyen, D. Trono, L.Naldini, A third-generation lentivirus vector with a conditionalpackaging system, J. Virol. 72 (1998) 8463–8471.

[24] O.B. Chernova, M.V. Chernov, Y. Ishizaka, M.L. Agarwal,G.R. Stark, MYC abrogates p53-mediated cell cycle arrest inN-(Phosphonacetyl)-L-aspartate-treated cells, permitting CADgene amplification, Mol. Cell. Biol. 18 (1998) 536–545.

[25] M.L. Agarwal, A. Agarwal, O. Chernova, Y. Sharma, G.R. Stark, Ap53-dependent s-phase checkpoint helps to protect cells fromDNA damage in response to starvation for pyrimidinenucleotides, PNAS 95 (1998) 14775–14780.

[26] M. Miura, Y. Miura, H.M. Padilla-Nash, A.A. Molinolo, B. Fu, V.Patel, B.M. Seo, W. Sonoyama, J.J. Zheng, C.C. Baker, W. Chen, T.Ried, S. Shi, Accumulated chromosomal instability in murinebone marrow mesenchymal stem cells leads to malignanttransformation, Stem Cells 24 (2006) 1095–1103.

[27] M. Kujawski, M. Kortylewski, H. Lee, A. Herrmann, H. Kay, H. Yu,Stat3 mediates myeloid cell-dependent tumor angiogenesis inmice, J. Clin. Invest. 118 (2008) 3367–3377.

[28] T. Kuroki, N.-H. Huh, Why are human cells resistant to malignantcell transformation in vitro? Jpn. J. Cancer Res. 84 (1993)1091–1100.

[29] S. Zeamari, G. Rumping, B. Floot, S. Lyons, F.A. Stewart, In vivobioluminescence imaging of locally disseminated coloncarcinoma in rats, Br. J. Cancer 90 (2004) 1259–1264.

[30] L.S. Hart, W.S. El-Deiry, Invincible, but not invisible: imagingapproaches toward in vivo detection of cancer stem cells, JCO 26(2008) 2901–2910.

[31] G.S. Rose, L.M. Tocco, G.A. Granger, P.J. Disaia, T.C. Hamilton,A.D. Santin, J.C. Hiserodt, Development and characterization of aclinically useful animal model of epithelial ovarian cancer in theFischer 344 rat, Am. J. Obstet. Gynecol. 175 (1996) 593–599.