TargetingRatBrainstemGliomaUsingHumanNeuralStemCells andHumanMesenchymalStemCells · The human umbilical cord blood–derived MSCs (UC-MSC) were purchased from Medipost Biomedical

Cancer Therapy: Preclinical

Targeting Rat Brainstem Glioma Using Human Neural Stem Cellsand Human Mesenchymal Stem CellsDo-Hun Lee,1 Yong Ahn,2 Seung U. Kim,4,5 Kyu-Chang Wang,1 Byung-Kyu Cho,1 Ji Hoon Phi,1

In Ho Park,6 Peter M. Black,7 Rona S. Carroll,7 Joonyub Lee,3 and Seung-Ki Kim1

Abstract Purpose: Brainstem gliomas are usually inoperable and have a dismal prognosis.Based on the robust tropisms of neural stem cells (NSC) and mesenchymal stem cells(MSC) to brain tumors, we compared the tumor-tropic migratory capacities of thesestem cells and evaluated the therapeutic potential of genetically engineered humanNSCs encoding cytosine deaminase (CD) and IFNβ against brainstem gliomas.Experimental Design: The directed migratory capacities of NSCs and MSCs to brain-stem glioma (F98) were evaluated both in vitro and in vivo. The human NSCs (HB1.F3) and various human MSCs, such as bone marrow–derived MSCs (HM3.B10), adi-pose tissue–derived MSCs, and umbilical cord blood–derived MSCs, were tested. Hu-man fibroblast cells (HFF-1) were used as the negative control. As a proof of concept,the bioactivity of HB1.F3-CD-IFNβ was analyzed with a cell viability assay, and animalswith brainstem gliomas were injected with HB1.F3-CD-IFNβ cells followed by systemic5-fluorocytosine treatment.Results: In an in vitro modified Transwell migration assay and in vivo stem cell injec-tion into established brainstem gliomas in rats, all the stem cells showed a significantmigratory capacity compared with that of the control (P < 0.01). Histologic analysisshowed a 59% reduction in tumor volume in the HB1.F3-CD-IFNβ–treated group (P <0.05). Apoptotic cells were increased 2.33-fold in animals treated with HB1.F3-CD-IFNβcompared with the respective control groups (P < 0.01).Conclusion: The brainstem glioma-tropic migratory capacities of MSCs from varioussources were similar to those of NSCs. Genetically engineered NSCs show therapeuticefficacy against brainstem gliomas.

Brainstem gliomas account for 10% to 15% of childhood cen-tral nervous system tumors. Diffuse intrinsic pontine glioma isthe most common of the brainstem gliomas, and it cannot be

removed by surgery (1, 2). Despite technological advances in ra-diation therapy and chemotherapy, the prognosis is poor, witha mean survival of

tumor-tropic properties of NSCs and MSCs has been made.Furthermore, the differences between migratory cells andnonmigratory cells have rarely been reported.

This is the first preclinical study assessing the potential ofstem cell–based gene therapy in the treatment of brainstem gli-omas. The purpose of this study was to compare quantitativelythe tumor-tropic migratory capacities of stem cells from differ-ent sources and to assess the different characteristics of the mi-gratory cells and nonmigratory cells. As a proof of concept, wealso evaluated the therapeutic efficacy of genetically engineeredhuman NSCs encoding cytosine deaminase (CD) suicide geneand IFNβ proinflammatory cytokine gene against this represen-tative inoperable malignant brain tumor.

Materials and Methods

Cells. Human NSCs (HB1.F3) were derived from a human fetalbrain (ventricular zone) at 15 wk of gestation with an amphotropic,

replication-incompetent retroviral vector containing v-myc (21–24).This is a well-established and well-characterized NSC line, which ismultipotent, migratory, and nontumorigenic in vivo. The NSCs werecultured in DMEM (WelGene Biopharmaceuticals) supplemented withL-glutamine, 10% fetal bovine serum (FBS), and 1% antibiotic-antimy-cotic solution (Invitrogen).

The BM-MSCs (BM3.B10) were derived from human fetal spinal ver-tebrae bone marrow at 12 to 15 wk of gestation with an amphotropic,replication-incompetent retroviral vector containing v-myc (25). Thesecells differentiate into neural cell types and restore functional deficits inmice with intracerebral hemorrhage after brain transplantation. TheBM-MSCs were suspended in α-MEM (WelGene Biopharmaceuticals)supplemented with L-glutamine, deoxyribonucleosides, ribonucleo-sides, 10% FBS, and 1% antibiotic-antimycotic solution.

Adipose tissue–derived MSCs (AT-MSC) were isolated from humanfat tissue. Adipose tissues were obtained from the abdominal fat pre-pared for sellar floor reconstruction in patients who had undergonetranssphenoidal surgery at Seoul National University Hospital. All eli-gible patients or their parents provided their written informed consent,and permission to isolate the MSCs from the fat tissues was given by theinstitutional review board of Seoul National University Hospital. Thecells were cultured to passages 2 and 3. The AT-MSCs were suspendedin MSC expansion medium (Chemicon) supplemented with 10% FBSand 1% antibiotic-antimycotic solution.

The human umbilical cord blood–derived MSCs (UC-MSC)were purchased from Medipost Biomedical Research Institute (Medi-post). The UC-MSCs were suspended in α-MEM supplemented withL-glutamine, deoxyribonucleosides, ribonucleosides, 10% FBS, and1% antibiotic-antimycotic solution.

The human fibroblast cell line HFF-1 and the rat glioma cell line F98were obtained from the American Type Culture Collection. The HFF-1and F98 cells were cultured in DMEM supplemented with L-glutamine,110 mg/mL sodium pyruvate, 4,500 mg/L D-glucose, 10% FBS, and 1%antibiotic-antimycotic solution. All cells were incubated at 37°C in anincubator in a 5% CO2/95% air atmosphere.Flow cytometric analysis. Whereas theNSCs and BM-MSCs are estab-

lished cell lines, the AT-MSCs and UC-MSCs were primary cultured cells.Therefore, the primary cell lines were characterized further. AT-MSCs andUC-MSCswere cultured in controlmedium for 72 hbefore analysis. Flowcytometry was done using a FACscan argon laser cytometer (BectonDick-inson). Briefly, the cells were harvested in 0.25% trypsin/EDTA and fixedfor 30 min in ice-cold 2% formaldehyde. The fixed cells were washed inflow cytometry buffer (PBS, 2% FBS, 0.2% Tween 20) and incubated for30 min in flow cytometry buffer containing fluorescein isothiocyanate–conjugated monoclonal antibodies directed against cluster of differenti-ation (CD) antigens (CD29, CD49d, CD90, and CD105; Chemicon) orphycoerythrin-conjugated monoclonal antibodies directed against CDantigens (CD34 and CD45; Chemicon).

Translational Relevance

Generally, brainstem gliomas are inoperable anddo not have an encouraging prognosis. These disap-pointing results in the treatment of brainstem gliomahave encouraged numerous experimental trials inthe search for a novel treatment.

Recently, stem cell–based gene therapy hasshown potential as a new treatment modality for ma-lignant brain tumor because of the strong inherenttumor-tropic properties of stem cells. Not only neu-ral stem cells but also mesenchymal stem cells cantarget brain tumors.

In this study, we quantitatively compared thetumor-tropic migratory capacities of neural stem cellsand mesenchymal stem cells from various sourcesand qualitatively described the characteristics of mi-gratory and nonmigratory stem cells. This is the firstpreclinical trial of a stem cell–based gene therapydirected against brainstem glioma in an animal mod-el. Our results provide the rationale for further evalu-ation of this strategy for human brainstem gliomas.

Fig. 1. Flow cytometric analysis of the expression of surface antigens on AT-MSCs and UC-MSCs. MSC-specific markers, including CD29, CD49d, andCD105, were expressed in AT-MSCs and UC-MSCs, whereas hematopoietic stem cell markers, including CD34, CD45, and CD90, were not.

4926Clin Cancer Res 2009;15(15) August 1, 2009 www.aacrjournals.org


Research. on June 26, 2021. © 2009 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

Published OnlineFirst July 28, 2009; DOI: 10.1158/1078-0432.CCR-08-3076

http://clincancerres.aacrjournals.org/

In vitro migration study. The directed migratory capacities of NSCsand MSCs toward F98 glioma cells were evaluated using a modifiedTranswell migration study. F98 cells (50,000/0.5 mL) were incubatedin serum-free medium for 24 h and placed in the lower well of Matri-gel-coated (BD Biosciences) 10-mm tissue culture Transwell plates(8 μm; Nunc International). The NSCs and MSCs from various sources(50,000/0.5 mL) in serum-free medium were seeded in the upper wellsof the Transwell plates. HFF-1 cells were used as the negative control formigrating cells. Serum-free medium served as negative control for trig-ger factors. After incubation for 6 h at 37°C, the nonmigratory cellswere removed from the upper side of the filter, and the filters were thenstained with the Three-Step Stain Set (Richard-Allan Scientific) to quan-tify the migratory cells. All experiments were conducted in quadrupli-cate. The nuclei of the migratory cells were counted in five high-powerfields (×200), and the values were expressed as means ± SE.

In vivo migration study. Female Fisher 344 rats (Central Lab Ani-mal), weighing 150 to 200 g, were anesthetized with an i.m. injectionof a solution of 20 mg/kg Zoletil (Virbac) and 10 mg/kg xylazine (BayerKorea). The posterior cranial region was shaved and prepared in a ster-ile fashion. A midline scalp incision of ∼2 cm was made, and a smallburr hole was created using a 22-gauge needle. To establish brainstemgliomas, F98 tumor cells were stereotactically implanted into the rightbrainstem, as previously described (26). The stereotactic coordinateswere 1.4 mm to the right of the sagittal suture and 1 mm anterior tothe lambdoid suture. A 26-gauge Hamilton needle was inserted at anangle of 5° anteflexed from vertical to a depth of 7 mm from the duramater. F98 tumor cells (50,000 cells in 3 μL) were then implanted viathe needle at an injection rate of 1 μL/min. The needle was kept in situfor 3 min and was then carefully withdrawn for 3 min. One week aftertumor cell implantation into the brainstem region, the animals wererandomized into four experimental groups and one control group.Stem cells (NSC, BM-MSC, AT-MSC, and UC-MSC cells; n = 7 in each

group) or HFF-1 cells (n = 7) were injected into the right forebrain. Thestereotactic coordinates were 2.5 mm to the right of the sagittal sutureand 1 mm anterior to the bregma, at a depth of 2.5 mm from the duramater. Each cell type (200,000 cells in 5 μL) was labeled for 30 minwith chloromethylebenzamido-DiI (CM-DiI; Molecular Probes) beforeinjection, according to the protocol of the manufacturer. Bromodeox-yuridin (BrdUrd; 10 μmol/L; BD Pharmingen) was added to thecultures of NSCs or MSCs for 72 h.

On day 21, the animals were perfused with 4% paraformaldehydeunder deep anesthesia and killed. Their brains were harvested and im-mersed in 4% paraformaldehyde solution for 24 h. After fixation, thebrains were immersed sequentially in 10%, 20%, and 30% sucrose so-lution for dehydration, embedded in optimum cutting temperaturecompound (Tissue-Tek), and stored at -80°C.

To quantify the migratory capacity of the NSCs and the MSCs fromvarious sources, the cells that had migrated to the brainstem gliomawere counted. The brains were sectioned sagittally to 20 μm thicknessacross the whole extent of the tumor using a cryostat (Microm). Wemounted the sections at intervals of ∼300 μm. Ten slides centered onthe tissue sections that had strongly fluorescent DiI-positive cells wereselected. The slides were stained with 4′,6-diamidino-2-phenylindole(DAPI; Vector Laboratories). The viable cells were quantified by BrdUrdstaining. The tissue sections were washed twice with PBS. After incuba-tion in permeabilizing buffer for 15 min, 0.3% H2O2 was added for15 min to block any endogenous peroxidase activity. The tissue sectionswere incubated in 2 N HCl for 1 h to denature the nuclear DNA andwere then washed thrice with PBS. The tissue sections were then treatedwith mouse monoclonal anti-BrdUrd antibody (Calbiochem) diluted1:100 in PBS containing 1% normal goat serum. The sections were in-cubated overnight with the primary antibody. The sections were rinsedwith PBS (0.05 mol/L, pH 7.4), treated for 30 min with 1% hydrogenperoxide in PBS to eliminate endogenous peroxidase activity and rinsed

Fig. 2. In vitromigration study. The directed migratorycapacities of NSCs and MSCs toward F98 gliomacells were evaluated using a modified Transwellmigration study. A, all stem cells showed significantdirectional migration toward F98-conditionedmedium compared with the migration of HFF-1 cells.*, P < 0.01, Kruskal-Wallis test. Themigratory capacitiesof BM-MSC and AT-MSC did not differ significantlyfrom that that of NSCs. UC-MSCs showed amigratory capacity inferior to that of NSCs. **, P < 0.05,Kruskal-Wallis test. B, representative photographs(magnification, ×200).

4927 Clin Cancer Res 2009;15(15) August 1, 2009www.aacrjournals.org

Targeting Rat Brainstem Glioma




with PBS. After rinsing, the secondary antibody, Alexa 488–conjugatedgoat anti-mouse IgG (1:100; Molecular Probes), was applied. Thesections were mounted with antifading solution containing DAPI andobserved under confocal and fluorescence microscopes (Zeiss).

The DiI- and BrdUrd-labeled cells were counted from images of high-power (×200) optical fields using a fluorescence microscope with Im-age-Pro Plus 4.5 software. Stem cell migration was calculated as follows:

ð½Stem cells in the tumor bed�=½Stem cells in the injection siteþ stem cells in the tumor bed�Þ � 100:

To evaluate the differences between the migratory and nonmigratorycells, an immunohistochemical study was done with a stem cell–specif-ic marker (nestin, 1:300; Chemicon), a neuronal marker (NeuN, 1:200;Chemicon), and a cell type–specific marker for astrocytes [glial fibrillaryacidic protein (GFAP), 1:200; Chemicon]. All sections were treated for

30 min with a solution of normal goat serum to block nonspecificbinding sites. The primary antibodies were diluted with PBS containing1% normal goat serum. The sections were incubated overnight with theprimary antibody. Sections were rinsed with PBS (0.05 mol/L, pH 7.4),treated for 30 min with 1% hydrogen peroxide in PBS to eliminate anyendogenous peroxidase activity, and rinsed with PBS. The secondary an-tibody, Alexa 488–conjugated goat anti-mouse IgG (1:100), was thenapplied. The sections were stained with DAPI and were observed withconfocal and fluorescence microscopy.

All animal studies were carried out at the animal facility of SeoulNational University Hospital in accordance with national and institu-tional guidelines.

In vitro transfection of NSCs. The clonal HB1.F3-CD line wasderived from the parental HB1.F3 cells, as previously described (10, 22,23). An expression plasmid was constructed using the retroviral

Fig. 3. In vivo migration assay. A, F98 glioma cells were implanted into the brainstems of rats (T). One week later, CM-DiI–labeled and BrdUrd-labeled stemcells were injected into the right forebrains (I). After 2 wk, the migration of the stem cells to the tumor site was evaluated. B, all the stem cells showedconspicuously greater migratory capacity, measured as percentage migration, than that of the HFF-1 cells. *, P < 0.01, Kruskal-Wallis test. There was nosignificant difference in the cell numbers or percentages of migratory cells between the NSCs and MSCs. C, BrdUrd-labeled stem cells migrated across thebrain parenchyma in the injected hemisphere (I) and populated the tumor bed (T) in the brainstem. Magnification, ×200.






pBabePuro backbone to include the Escherichia coli CD cDNA (1.5-kbfragment) transcribed from the long terminal repeat. The vectors werepackaged by cotransduction into pA317 cells of the CD puro plasmidwith the plasmid cDNA encoding MV12 envelope protein. The CDpuro retroviral supernatant was used for multiple transductions ofthe HB1.F3 cells. The HB1.F3-CD cells were selected with 3 μg/mLpuromycin for 4 wk. The clonal HB1.F3-CD-IFNβ line was derivedfrom the parental HB1.F3-CD cells. An expression plasmid was con-structed using the retroviral pLHCX backbone to include the humanIFN cDNA (0.5-kb fragment) transcribed from the long terminal repeat.To produce retroviral vectors encoding human IFNβ, PA317 packagingcells were transfected with pLHC.IFNβ using Lipofectamine (Life Tech-nologies-Bethesda Research Laboratories) and the viral producer cellclones were selected. The viral supernatants were collected from theviral producers, and the HB1.F3-CD cells were then incubated withthe viral supernatants in the presence of polybrene (8 μg/mL; Sigma)for 16 h. After incubation for 2 d in growth medium, the transduced cells(HB1.F3-CD-IFNβ) were selected with 5 μg/mL hygromycin for 4 wk.

Successful transduction of the HB1.F3-CD-IFNβ cells was confirmedby reverse transcription-PCR. The sense and antisense primers of eachprimer pair were designed to bind to different exons to avoid DNA con-tamination: CD (sense 5′-GAGTCACCGCCAGCCACACCACGGG-3′,antisense 5′-GTTTGTAATCGATGGCTTCTGGCTGC-3′, 550 bp ampli-con) and IFNβ (sense 5′-AAAGAAGCAGCAATTTTCAG-3′, antisense5′-TTTCTCCAGTTTTTCTTCCA-3′, 296 bp amplicon). Total RNA wasextracted with TRIzol reagent (Life Technologies-Bethesda Research Lab-oratories). Complementary DNA templates from each sample were pre-pared from 1 μg of total RNA primed with oligodT primer (Pharmacia)using 400 units of Moloney murine leukemia virus reverse transcriptase(Promega), followed by 30 PCR amplification cycles (94°C for 30 s,annealing at 60°C for 60 s, and extension at 72°C for 90 s). Glyceral-dehyde-3-phosphate dehydrogenase was used as the reaction standard:sense 5′-CATGACCACAGTCCATGCCATCACT-3′, antisense 5′-TGAGGTCCACCACCCTGTTGCTGTA-3′, 450 bp amplicon. Each PCRproduct (10 μL) was analyzed by 1.5% agarose gel electrophoresis. Au-thentic bands were determined by selective enzyme digestion.

In vitro therapeutic efficacy of HB1.F3-CD-IFNβ. To confirm thebioactivity of HB1.F3-CD-IFNβ, the cytotoxic effects of 5-fluorocytosineand 5-fluorouracil on HB1.F3-CD-IFNβ were analyzed with a cell via-bility assay. The therapeutic efficacy of HB1.F3-CD-IFNβ cells after 5-fluorocytosine treatment was analyzed by coculture experiments.F98 cells (4,000 per well) were plated in 96-well plates (Corning,Inc.), and the experiments were done as follows: on day 1, HB1.F3 orHB1.F3-CD-IFNβ cells (8,000 per well) were added to the tumor cellcultures; on day 2, 5-fluorocytosine (100 μg/mL) and 5-fluorouracil(100 μg/mL) were added to the mixed cell cultures; and on day 6, cellviability was quantified with colorimetric assays using the Cell Count-ing Kit-8 (Dojindo Molecular Technologies). All experiments were donein quadruplicate. The determination of viability was based on the bio-conversion of the tetrazolium compound, 2-(2-methoxy-4-nitrophe-nyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium (WST-8)into formazan, as determined by absorbance at 450 nm using a multi-well scanning spectrophotometer. Cell viability was expressed as themean (±SE) percentage relative to the viability of the control.

In vivo therapeutic efficacy of HB1.F3-CD-IFNβ. Animals werestereotactically implanted with F98 cells (50,000 in 3 μL PBS) in theright brainstem as described above. One week after tumor cell implan-tation, the animals were randomized into three groups: group 1, treatedwith an intratumoral injection of PBS (5 μL; n = 10); group 2, treatedwith an intratumoral injection of 200,000 HB1.F3 cells in 5 μL of PBS(n = 10); group 3, treated with intratumoral injection of 200,000 HB1.F3-CD-IFNβ cells in 5 μL of PBS (n = 10). In this study, the ratio of stemcells/tumor cells was 4:1. HB1.F3 and HB1.F3-CD-IFNβ were labeledwith the CM-DiI cell tracker before injection. One week after PBS,HB1.F3, or HB1.F3-CD-IFNβ injection, all groups received i.p. injec-tions of 5-fluorocytosine (900 mg/kg) twice a day for 14 d. After thelast 5-fluorocytosine treatment, the animals were perfused and their

brains were harvested and processed as described above. The brainswere sectioned coronally, and the tissue was stained with DAPI orH&E using standard protocols. The intratumoral distribution of CM-DiI–labeled NSCs was assessed by fluorescence microscopy. The tumorvolumes were estimated using the formula for ellipsoids and expressedas means ± SE, as described previously (27). Immunohistochemistrywas done with the Vectastain Elite ABC kit (Vector Laboratories). Thebrain sections were fixed in cold acetone. The primary antibodies in-cluded one directed against cleaved caspase-3 (1:100; Cell SignalingTechnology) to detect apoptosis. The sections were counterstained withhematoxylin, and negative control slides were established by omittingthe primary antibody. The apoptotic indices were defined as the per-centage of positively stained cells per 100 nuclei from five randomlyselected high-power fields.Statistical analysis. All the values were calculated as means ± SE or

were expressed as a percentage of the control ± SE. Difference in stemcell migration, tumor volume, and apoptosis index was tested withKruskal-Wallis test with post hoc analysis. Difference in tumor cellviability was determined using the Mann-Whitney U test. Values ofP < 0.05 were considered significant.

Results

Flow cytometric analysis of AT-MSCs and UC-MSCs. Fluores-cence-activated cell sorting analysis of AT-MSCs and UC-MSCsshowed that the AT-MSCs and UC-MSCs expressed MSC-specif-ic cell type markers, including CD29, CD49d, and CD105, butdid not express cell type–specific markers for hematopoieticstem cells, including CD34, CD45, and CD90 (Fig. 1).Migratory ability of stem cells in vitro. Using a Transwell mi-

gration assay, the in vitro migratory capacities of NSCs andMSCs were compared. All stem cells showed significant direc-tional migration toward F98-conditioned medium comparedwith HFF-1 cells (25.6 ± 3.7, P < 0.01; Fig. 2), whereas veryfew cells migrated toward the serum-free medium (17.5-20).The mean number of migratory cells was highest for the NSCs(142.4 ± 22.8). The migratory capacities of the MSCs, includingthe BM-MSC (96.1 ± 12.7) and AT-MSC (107.6 ± 15.2), werenot statistically different from that of the NSCs. However, themigratory capacity of the UC-MSCs (57.4 ± 5.7) was inferiorto that of the NSCs (P < 0.05; Fig. 2).Migratory capacities of stem cells in vivo and differences

between migratory and nonmigratory stem cells. To comparethe migratory capacities of NSCs and MSCs toward brainstemglioma in vivo, F98 glioma cells were implanted into the brain-stems of rats. One week later, labeled stem cells were implantedinto the right forebrains. Two weeks after labeled stem cell im-plantation, the migration of the stem cells to the tumor siteswere examined. Stem cells migrated across the brain parenchy-ma and populated the tumor bed. In contrast, the HFF-1 cellsshowed no targeted migration and dispersed randomly aroundthe implanted site. The percentage of migratory NSCs was29.7 ± 5.4%. The percentages of migratory BM-MSCs, AT-MSCs,and UC-MSCs were 22.2 ± 3.1%, 29.0 ± 5.4%, and 19.7 ± 4.0%,respectively. In terms of the percentages of migratory cells, allstem cells showed markedly greater migratory capacities thanthat of HFF-1 cells (6.6 ± 1.2%, P < 0.01; Fig. 3). There wereno statistical differences between NSCs and MSCs in their cellnumbers or percentages of migratory cells.

Differences in the migratory and nonmigratory stem cellswere also observed. With fluorescence microscopy, expressionof the stem cell marker nestin was observed at the injection site






but was not observed in the tumor bed. The expression of theastrocyte marker GFAP was scanty or not observed at the injec-tion site but was observed in the tumor bed. No expression ofthe neuronal marker NeuN was found at the injection site or inthe tumor bed (Fig. 4). These different patterns were similar inthe NSCs and MSCs.In vitro therapeutic efficacy of HB1.F3-CD-IFNβ. The

expression of the CD-IFNβ transcript was confirmed by reversetranscription-PCR. The CD-IFNβ transcript was only expressedin the HB1.F3-CD-IFNβ cells and not in the parental HB1.F3cells (Fig. 5A). To confirm the cytotoxic effects of the HB1.F3-CD-IFNβ cells, cell viability studies were done in a coculturesystem. After 5-fluorouracil treatment, the viability of theF98 cells did not differ when cocultured with HB1.F3 or HB1.

F3-CD-IFNβ cells. After 5-fluorocytosine treatment, the F98cells cocultured with the HB1.F3-CD-IFNβ cells decreased by45.5% compared with cells cocultured with HB1.F3 cells (P <0.05; Fig. 5B). These data show significant growth inhibition inthe F98 cells induced by HB1.F3-CD-IFNβ cells. Without treat-ment, F98 cells cocultured with HB1.F3 or HB1.F3-CD-IFNβcells showed no growth inhibition. This indicates that HB1.F3-CD-IFNβ cells can convert sufficient amounts of 5-fluorocyto-sine to 5-fluorouracil to effectively kill F98 cells in vitro.In vivo therapeutic efficacy of HB1.F3-CD-IFNβ. To assess

the therapeutic efficacy of HB1.F3-CD-IFNβ cells in an animalmodel of brainstem glioma, we injected CM-DiI–labeled HB1.F3-CD-IFNβ cells into tumor-bearing animals. The distribu-tion of HB1.F3-CD-IFNβ cells and the tumor volumes were

Fig. 4. Differences between migratory and nonmigratory cells. Brains were sectioned sagittally into 20-μm-thick slices and then stained with DAPI. Withfluorescence microscopy, the expression of the stem cell marker nestin was observed at the injection site but not in the tumor bed. The expression of theastrocyte marker GFAP was scanty; none was observed at the injection site but GFAP-positive cells were found in the tumor bed. No expression of theneuronal marker NeuN was observed at the injection site or in the tumor bed. Bar, 5 μm.






determined in harvested brain tissue after the last 5-fluorocy-tosine treatment. Labeled HB1.F3-CD-IFNβ cells were foundpredominantly at the border between the tumor and the nor-mal parenchyma. The survival rates of the animals were 70%(7 of 10) in the PBS-treated group, 90% (9 of 10) in the HB1.F3-treated group, and 100% (10 of 10) in the HB1.F3-CD-IFNβ–treated group. The average tumor volumes in the survivinganimals in the PBS-treated, HB1.F3-treated, and HB1.F3-CD-IFNβ–treated groups were 126.6 ± 17.5, 101.7 ± 10.7, and51.8 ± 12.1 mm3, respectively. The reduction in tumor size inthe HB1.F3-treated animals was 19.7% compared with thetumor size in the PBS-treated control. In contrast, the reductionin tumor size in the HB1.F3-CD-IFNβ–treated animals was59.1%, significantly different from that in the PBS-treated con-trol (P < 0.05; Fig. 6A and B). We next sought to determinethe biological action of CD-IFNβ on the gliomas. Histologicanalysis revealed a significant increase of 2.33-fold in apoptoticcells in the animals treated with HB1.F3-CD-IFNβ relative tothose in the various control groups (P < 0.01; Fig. 6C and D).The HB1.F3 group showed a 1.27-fold increase in apoptotic cellsrelative to that in the control group, which is not statisticallysignificant.

Discussion

In the present study, we quantitatively compared the tumor-tropic migratory capacities of NSCs and MSCs from varioussources and qualitatively described the characteristics of migra-tory and nonmigratory stem cells. This is the first preclinical tri-al of stem cell–based gene therapy directed against brainstemglioma in an animal model.

Our results show that the tumor-tropic migratory capacitiesof MSCs were statistically similar to that of NSCs. Although tu-mor tropism was first shown in NSCs (5), this property mightnot be limited to stem cells with a neuroectodermal origin.

According to the previously suggested mechanism of tumor tro-pism (28), it seems that migratory capacity is mediated by thetumor cell itself or the surrounding tissue. The kind of stem/progenitor cell is a less important issue in defining tumor tro-pism. We postulate that tumor-tropic migration is a commonfeature of stem cells, regardless of their origin. The similar tropiccapacities of MSCs of various origins including bone marrow,adipose tissue, and umbilical cord blood observed in our studysupport this hypothesis.

MSCs have recently been derived from various sources, in-cluding bone marrow, adipose tissue, and umbilical cordblood. Several studies have shown the tumor tropism and anti-tumor effects of genetically engineered BM-MSCs in animalmodels of intracranial glioma (15, 18, 19). AT-MSCs arereported to show broad multipotency with differentiation intoa number of cell lineages, including adipocytic, osteocytic, andchondrocytic lineages, like BM-MSCs (29). The applicability ofhuman AT-MSCs as cellular vehicles for targeted cancer chemo-therapy has been reported for colon cancer (30). The easy andrepeatable access to subcutaneous adipose tissue, the simpleisolation procedure, and the greater numbers (∼500-fold) offresh MSCs derived from equivalent amounts of fat than canbe derived from bone marrow are clear advantages in usingAT-MSCs over BM-MSCs (31). Isolated AT-MSCs can also beeasily cryopreserved (31). Although the inferior osteogenicand chondrogenic effects of AT-MSCs compared with those ofBM-MSCs might limit their applicability in regenerative medi-cine (32, 33), this might not be a problem when used as a cel-lular vector system for gene therapy in the neurooncologicalfield. Recently, umbilical cord blood has been used as an alter-native source of MSCs (34). The potential advantages ofUC-MSC compared with adult BM-MSCs include their readyavailability from storage, the ease and low cost of harvesting,the ability to select them according to the human leukocyteantigen type required, the reduced risk of the transmission of

Fig. 4 Continued.






transplant-related infections, and their better immunologic tol-erability (35, 36). A possible disadvantage of UC-MSCs is thefinite number of cells that can be harvested (37). Our data sug-gest that not only BM-MSC but also AT-MSCs and UC-MSCshave satisfactory tumor-tropic properties and may be ideal sub-stitutes for NSCs in stem cell–based gene therapies directedagainst brainstem glioma.

Despite their robust tumor tropism, stem cells do not al-ways move toward the tumor site. Around 30% of all stemcells migrated to the target glioma in our study. This incom-plete migration of stem cells can be explained in severalways: technical errors related to the quantification of migra-tory cells (localization of cells at the injection site and thediffuse distribution of cells within the tumor bed), failureof cell adaptation in a new microenvironment, the great dis-tance between the injection site (forebrain) and the tumorsite (brainstem), or poor survival rate of the human stemcells grafted into nonimmunosuppressed rat brain. Consider-ing the invasiveness of glioma, attempts to enhance the tu-mor-tropic migratory capacity of stem cells and to combinetherapeutic genes more effectively are interesting issues for fu-ture studies. A recent study reported that cell migration mod-ulators, such as transmembrane protein 18, can enhance thetumor tropism of NSCs (38).

The migratory cells and nonmigratory cells differently ex-pressed a stem cell marker and a differentiation marker. The“migratory” cells in the tumor bed no longer expressed mar-kers of stem/progenitor cells but expressed a marker of glialdifferentiation, regardless of their origin, which is consistentwith previous observations (38, 39). Contrary to this finding,the “nonmigratory” cells at the injection site were nestin pos-itive and GFAP negative. This may simply suggest that onlyastrocytic precursors migrate to the tumor. However, whenwe examined the stem cells inoculated into the tumor, theyonly expressed the glial differentiation marker, regardless oftheir location in the tumor, in either the core or periphery(data not shown). Therefore, the differences between the mi-gratory and nonmigratory cells are thought to be caused bythe microenvironment.

Next, we studied the antitumor effects of genetically engi-neered human NSCs directed against brainstem glioma. As aproof of concept, we tested the potential of the cellular vectorusing HB1.F3 cells as the representative stem cells, which wereengineered to secrete the prodrug activating enzyme CD andIFNβ. Our aim of adopting fusion gene (CD and IFNβ) wasto maximize the antitumor effect. The therapeutic actions ofCD and IFNβ are different; thus, we could expect a synergiceffect. CD acts as a prodrug activating enzyme (10) and IFNβcan enhance the antiangiogenic effect and the immune re-sponse (15). In addition, we would like to prepare safety bar-rier for protection from potential tumorigenesis of therapeuticstem cells. The suicidal tendency of CD can protect the hostbrain tissue by killing stem cells, as well as tumor cells. Anyuncontrolled cell division of HB1.F3-CD-IFNβ cells duringantitumor therapies will likely be kept under control by ad-ministering the 5-fluorocytosine, which leads to the killingof dividing cells, including HB1.F3-CD-IFNβ cells, whichmay undergo cell division. If necessary, 5-fluorocytosine couldbe given repeatedly to eradicate any residual HB1.F3-CD-IFNβcells. Histologic analysis showed a significant reduction in thetumor volume in the treated group. The apoptotic index of thetumor cells was also elevated in the treated group. When aspatial relationship between the stem cells and the apoptoticcells was sought using a fluorescent microscope, the apoptosisof tumor cells predominantly occurred around the implantedstem cells (data was not shown), suggesting the therapeuticeffect of genetically engineered stem cells. This study providesa rationale for the further evaluation of stem cell–based genetherapies for brainstem glioma, which might be the first can-didate cancer for stem cell–based gene therapy because of itsinoperability and dismal prognosis. Using a more effectivetherapeutic gene, these stem cells could provide a practicalantitumor modulatory effect for this inoperable tumor.

There may be some limitation to our study. Firstly, the NSCsand BM-MSCs were established cell lines, whereas the AT-MSCsand UC-MSCs were primary cells. This discrepancy may inter-fere with the study as a confounding factor. However, therewas no clear difference in the tumor-tropic migration of the celllines and the primary cells. Secondly, we could only show cel-lular differentiation in a qualitative way. A laser-captured mi-crodissection technique might be useful in effectivelyquantifying cellular differentiation. Finally, no long-term out-come, such as the survival rate of the animals after stem cell–based gene therapy, was examined. However, our short-termresults provide some clues: 3 of 10 rats in the PBS-treated group

Fig. 5. In vitro therapeutic efficacy of HB1.F3-CD-IFNβ. A, expression of theCD-IFNβ transcript was confirmed by reverse transcription-PCR.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as thepositive control. The CD-IFNβ transcript was only expressed in theHB1.F3-CD-IFNβ cells and not in the HB1.F3 cells. B, to confirm the cytotoxiceffects of the HB1.F3-CD-IFNβ cells, cell viability studies were done in acoculture system. After 5-fluorocytosine (5-FC) treatment, the number ofF98 cells cocultured with HB1.F3-CD-IFNβ cells decreased by 45.5% relativeto those cocultured with HB1.F3 cells. *, P < 0.05, Mann-Whitney U test.






and one rat in the HB1.F3-treated group died during the testperiod.

In conclusion, this study shows that not only NSCs but alsoMSCs from various sources can be used as useful vehicles forgene therapy directed against brainstem glioma, based on theirtumor-tropic capacities. Furthermore, genetically modifiedNSCs display therapeutic efficacy against brainstem glioma.

These results suggest a potential role for stem cell–based genetherapy for brainstem glioma.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

References1. Albright AL. Diffuse brainstem tumors: when isa biopsy necessary? Pediatr Neurosurg 1996;24:252–5.

2. Epstein F, Wisoff JH. Intrinsic brainstem tumorsin childhood: surgical indications. J Neurooncol1988;6:309–17.

3. Littman P, Jarrett P, Bilaniuk LT, et al. Pediatricbrain stem gliomas. Cancer 1980;45:2787–92.

4. Aghi M, Chiocca EA. Gene therapy for glioblas-toma. Neurosurg Focus 2006;20:E18.

5. Aboody KS, Brown A, Rainov NG, et al. Neuralstem cells display extensive tropism for patholo-gy in adult brain: evidence from intracranial gli-omas. Proc Natl Acad Sci 2000;97:12846–51.

6. Benedetti S, Pirola B, Pollo B, et al. Gene thera-py of experimental brain tumors using neuralprogenitor cells. Nat Med 2000;6:447–50.

7. Ehtesham M, Kabos P, Kabosova A, Neuman T,Black KL, Yu JS. The use of interleukin 12-secret-ing neural stem cells for the treatment ofintracranial glioma. Cancer Res 2002;62:5657–63.

8. Ehtesham M, Kabos P, Gutierrez MA, et al. In-duction of glioblastoma apoptosis using neuralstem cell-mediated delivery of tumor necrosisfactor-related apoptosis-inducing ligand. CancerRes 2002;62:7170–4.

9. Kim SK, Cargioli TG, Machluf M, et al. PEX-pro-ducing human neural stem cells inhibit tumorgrowth in a mouse glioma model. Clin CancerRes 2005;11:5965–70.

10. Kim SK, Kim SU, Park IH, et al. Human neuralstem cells target experimental intracranial me-dulloblastoma and deliver a therapeuticgene leading to tumor regression. Clin CancerRes 2006;12:5550–6.

11. Ehtesham M, Stevenson CB, Thompson RC.Stem cell therapies for malignant glioma.Neurosurg Focus 2005;19:E5.

12. Hamada H, Kobune M, Nakamura K, et al.Mesenchymal stem cells (MSC) as therapeuticcytoreagents for gene therapy. Cancer Sci2005;96:149–56.

13. Lee HJ, Kim KS, Kim EJ, et al. Brain transplanta-tion of humanneural stem cells promotes function-al recovery in mouse intracerebral hemorrhagestroke model. Stem Cells 2007;25:211–24.

14. Mapara KY, Stevenson CB, Thompson RC,Ehtesham M. Stem cells as vehicles for the treat-ment of brain cancer. Neurosurg Clin N Am2007;18:71–80.

15. Nakamizo A, Marini F, Amano T, et al. Humanbone marrow-derived mesenchymal stem cells

in the treatment of gliomas. Cancer Res 2005;65:3307–18.

16. Singh G. Sources of neuronal material for im-plantation. Neuropathology 2001;21:110–4.

17. Kabos P, Ehtesham M, Kabosova A, Black KL,Yu JS. Generation of neural progenitor cellsfrom whole adult bone marrow. Exp Neurol2002;178:288–93.

18. Lee J, Elkahloun AG, Messina SA, et al. Cellu-lar and genetic characterization of human adultbone marrow-derived neural stem-like cells:a potential antiglioma cellular vector. CancerRes 2003;63:8877–89.

19. Nakamura K, Ito Y, Kawano Y, et al. Antitumoreffect of genetically engineered mesenchymalstem cells in a rat glioma model. Gene Ther2004;11:1155–64.

20. Yuan X, Hu J, Belladonna ML, Black KL, Yu JS.Interleukin-23-expressing bone marrow-derivedneural stem-like cells exhibit antitumor activityagainst intracranial glioma. Cancer Res 2006;66:2630–8.

21. Cho T, Bae JH, Choi HB, et al. Human neuralstem cells: electrophysiological properties ofvoltage-gated ion channels. Neuroreport 2002;13:1447–52.

Fig. 6. In vivo therapeutic efficacy of HB1.F3-CD-IFNβ and treatment effects assessed by immunohistochemistry. A, the tumor volumes were estimated byhistologic analysis. The average tumor volume after treatment with 5-fluorocytosine (5-FC) in the surviving animals in the PBS-treated, HB1.F3-treated,and HB1.F3-CD-IFNβ–treated groups were 126.6 ± 17.5, 101.7 ± 10.7, and 51.8 ± 12.1 mm3, respectively. A reduction in tumor volume of 59.1% was observedin the HB1.F3-CD-IFNβ–treated animals relative to that in the PBS-treated animals. *, P < 0.05, Kruskal-Wallis test. B, representative histologic images(H&E staining; magnification, ×1). C, histologic analysis revealed a significant 2.33-fold increase in apoptotic cells in the animals treated with HB1.F3-CD-IFNβcompared with the various control groups (P < 0.01, Kruskal-Wallis test). D, representative histologic images. Brown-colored caspase-3–positive cellsare shown (arrows). Bar, 20 μm.






22. Kim SU. Human neural stem cells geneticallymodified for brain repair in neurological disor-ders. Neuropathology 2004;24:159–71.

23. Kim SU, Nagai A, Nakagawa E, et al. Pro-duction and characterization of immortal hu-man neural stem cell line with multipotentdifferentiation property. Methods Mol Biol2008;438:103–21.

24. Ourednik V, Ourednik J, Flax JD, et al. Segrega-tion of human neural stem cells in the developingprimate forebrain. Science 2001;293:1820–4.

25. Nagai A, Kim WK, Lee HJ, et al. Multilineagepotential of stable human mesenchymal stemcell line derived from fetal marrow. PLoS ONE2007;2:e1272.

26. Lee J, Jallo GI, Guarnieri M, Carson BS, Sr.,Penno MB. A novel brainstem tumor model:guide screw technology with functional, radio-logical, and histopathological characterization.Neurosurg Focus 2005;18:E11.

27. Schmidt KF, Ziu M, Schmidt NO, et al. Volumereconstruction techniques improve the correla-tion between histological and in vivo tumorvolume measurements in mouse models of hu-man gliomas. J Neurooncol 2004;68:207–15.

28. Xu F, Zhu JH. Stem cells tropism for malignantgliomas. Neurosci Bull 2007;23:363–9.

29. Parker AM, Katz AJ. Adipose-derived stemcells for the regeneration of damaged tissues.Expert Opin Biol Ther 2006;6:567–78.

30. KucerovaL,AltanerovaV,MatuskovaM,TyciakovaS, Altaner C. Adipose tissue-derived human mesen-chymal stem cells mediated prodrug cancer genetherapy. Cancer Res 2007;67:6304–13.

31. Black LL, Gaynor J, Gahring D, et al. Effect ofadipose-derived mesenchymal stem and regen-erative cells on lameness in dogs with chronicosteoarthritis of the coxofemoral joints: arandomized, double-blinded, multicenter, con-trolled trial. Vet Ther 2007;8:272–84.

32. Im GI, Shin YW, Lee KB. Do adipose tissue-de-rived mesenchymal stem cells have the sameosteogenic and chondrogenic potential as bonemarrow-derived cells? Osteoarthritis Cartilage2005;13:845–53.

33. Winter A, Breit S, Parsch D, et al. Cartilage-likegene expression in differentiated human stemcell spheroids: a comparison of bone marrow-derived and adipose tissue-derived stromalcells. Arthritis Rheum 2003;48:418–29.

34. Kern S, Eichler H, Stoeve J, Klüter H, Bieback K.Comparative analysis ofmesenchymal stem cellsfrom bone marrow, umbilical cord blood,or adipose tissue. Stem Cells 2006;24:1294–301.

35. Barker JN, Wagner JE. Umbilical cord bloodtransplantation: current state of the art. CurrOpin Oncol 2002;14:160–4.

36. Lee OK, Kuo TK, Chen WM, Lee KD, Hsieh SL,Chen TH. Isolation of multipotent mesenchymalstem cells from umbilical cord blood. Blood2004;103:1669–75.

37. Mavroudis D, Read E, Cottler-Fox M, et al.CD34+ cell dose predicts survival, posttrans-plant morbidity, and rate of hematologic re-covery after allogeneic marrow transplantsfor hematologic malignancies. Blood 1996;88:3223–9.

38. Jurvansuu J, Zhao Y, Leung DS, et al. Trans-membrane protein 18 enhances the tropism ofneural stem cells for glioma cells. Cancer Res2008;68:4614–22.

39. Ehtesham M, Yuan X, Kabos P, et al. Gliomatropic neural stem cells consist of astrocytic pre-cursors and their migratory capacity is mediatedby CXCR4. Neoplasia 2004;6:287–93.






Correction

Correction: Targeting Rat Brainstem GliomaUsing Human Neural Stem Cells and HumanMesenchymal Stem Cells

In this article (Clin Cancer Res 2009;15:4925–34), which was published in theAugust 1, 2009, issue of Clinical Cancer Research (1), there was an error in theflow-cytometric analysis of the expression of surface antigens on adipose-tissue-derived mesenchymal stem cells (AT-MSCs) and umbilical-cord-blood-derived MSCs (UC-MSCs).The first sentence of the Results section on page 4929 should read as follows:Fluorescence-activated cell sorting (FACS) analysis of AT-MSCs and UC-MSCs

showed that the AT-MSCs and UC-MSCs expressed MSC-specific cell-typemarkers, including CD29, CD49d, CD90, and CD105, but did not express cell-type-specific markers for hematopoietic stem cells (HSCs), including CD34 andCD45 (Fig. 1).CD90 was mischaracterized in Fig. 1 and the Fig. 1 legend on page 4926. The

corrected figure and legend appear here:

Reference1. Lee D-H, Ahn Y, Kim SU, et al. Targeting rat brainstem glioma using human neural stem cells and

human mesenchymal stem cells. Clin Cancer Res 2009;15:4925–34.

Published OnlineFirst 10/26/2010.©2010 American Association for Cancer Research.doi: 10.1158/1078-0432.CCR-10-2551

ClinicalCancer

Research

Fig. 1. Flow-cytometric analysis of the expression of surface antigens on adipose-tissue–derived mesenchymal stem cells (AT-MSCs) andumbilical-cord-blood–derived MSCs (UC-MSCs). MSC-specific markers, including CD29, CD49d, CD90, and CD105, were expressed in AT-MSCs andUC-MSCs, whereas hematopoietic stem cell markers, including CD34 and CD45, were not.

www.aacrjournals.org 5367

2009;15:4925-4934. Published OnlineFirst July 28, 2009.Clin Cancer Res Do-Hun Lee, Yong Ahn, Seung U. Kim, et al. Cells and Human Mesenchymal Stem CellsTargeting Rat Brainstem Glioma Using Human Neural Stem

Updated version

10.1158/1078-0432.CCR-08-3076doi:

Access the most recent version of this article at:

Cited articles

http://clincancerres.aacrjournals.org/content/15/15/4925.full#ref-list-1

This article cites 39 articles, 13 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/15/15/4925.full#related-urls

This article has been cited by 5 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://clincancerres.aacrjournals.org/content/15/15/4925To request permission to re-use all or part of this article, use this link


http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-08-3076http://clincancerres.aacrjournals.org/content/15/15/4925.full#ref-list-1http://clincancerres.aacrjournals.org/content/15/15/4925.full#related-urlshttp://clincancerres.aacrjournals.org/cgi/alertsmailto:[email protected]://clincancerres.aacrjournals.org/content/15/15/4925http://clincancerres.aacrjournals.org/

Documents

TargetingRatBrainstemGliomaUsingHumanNeuralStemCells andHumanMesenchymalStemCells · The human umbilical cord blood–derived MSCs (UC-MSC) were purchased from Medipost Biomedical