autocrine endothelin-3/endothelin receptor b signaling ... · sirna (ambion), nonsilencing control...

Cancer Genes and Genomics

Autocrine Endothelin-3/Endothelin Receptor B SignalingMaintains Cellular and Molecular Properties ofGlioblastoma Stem Cells

Yue Liu1, Fei Ye1,9, Kazunari Yamada1, Jonathan L. Tso1, Yibei Zhang1, David H. Nguyen1, Qinghua Dong1,10,Horacio Soto2, Jinny Choe1, Anna Dembo1, HayleyWheeler1, Ascia Eskin3, Ingrid Schmid4,WilliamH. Yong5,8,Paul S. Mischel5,8, Timothy F. Cloughesy6,8, Harley I. Kornblum7,8, Stanley F. Nelson3,8, Linda M. Liau2,8, andCho-Lea Tso1,8

AbstractGlioblastoma stem cells (GSC) express both radial glial cell and neural crest cell (NCC)-associated genes. We

report that endothelin 3 (EDN3), an essential mitogen forNCCdevelopment andmigration, is highly produced byGSCs. Serum-induced proliferative differentiation rapidly decreased EDN3 production and downregulated theexpression of stemness-associated genes, and reciprocally, two glioblastoma markers, EDN1 and YKL-40transcripts, were induced. Correspondingly, patient glioblastoma tissues express low levels of EDN3 mRNA andhigh levels of EDN1 and YKL-40 mRNA. Blocking EDN3/EDN receptor B (EDNRB) signaling by an EDNRBantagonist (BQ788), or EDN3 RNA interference (siRNA), leads to cell apoptosis and functional impairment oftumor sphere formation and cell spreading/migration in culture and loss of tumorigenic capacity in animals. Usingexogenous EDN3 as the sole mitogen in culture does not support GSC propagation, but it can rescue GSCs fromundergoing cell apoptosis. Molecular analysis by gene expression profiling revealed that most genes downregulatedbyEDN3/EDNRBblockadewere those involved in cytoskeleton organization, pause of growth and differentiation,and DNA damage response, implicating the involvement of EDN3/EDNRB signaling in maintaining GSCmigration, undifferentiation, and survival. These data suggest that autocrine EDN3/EDNRB signaling is essentialfor maintaining GSCs. Incorporating END3/EDNRB-targeted therapies into conventional cancer treatments mayhave clinical implication for the prevention of tumor recurrence.Mol Cancer Res; 9(12); 1668–85.�2011 AACR.

Introduction

Glioblastoma (WHO grade IV) is the most common andmost aggressive type of primary brain tumor in humans. Itremains virtually incurable despite extensive surgical excisionand postoperative adjuvant radiotherapy and chemotherapy(1). Glioblastoma stem cells (GSC) have been recentlyisolated from glioblastoma tumors of patients and were

characterized as a small subset of stem-like tumor cellscapable of initiating and sustaining tumor growth whengrafted into mice (2–6). Although CD133/prominin, anormal neural stem cell (NSC) marker, is not an obligatorymarker for GSCs (6, 7), CD133 was first applied as a surfacemarker for isolation and enrichment of GSCs (2–5, 7–9).Tumors initiated fromCD133þGSCs often recapitulate thehistopathologic features of the patient tumors from whichthe cells were derived, indicating the ability to self-renew andreproduce the cellular heterogeneity found in human glio-blastoma tumors (2, 4–6). Studies showed that CD133þ

tumor-initiating cells possess marked resistance to radio-chemotherapy (10, 11) and thus are now suggested to beresponsible for posttreatment failure and tumor recurrence.The molecular profiles of GSCs revealed characteristics of

neuroectodermal-like cells, expressing both neural and mes-enchymal developmental genes, and portraying an undif-ferentiated, migratory, astrogliogenic, and chondrogenicphenotype (8). This suggests that a subset of GSCs mayinherit neural crest cell (NCC)-like developmental pathwaysto initiate a tumor. In particular, endothelin 3 (EDN3), apotent mitogen for NCC and its derived lineage precursorcells (12, 13), was identified as one of the top genes highlyexpressed in tumorigenic GSCs (8). EDN3 is a member of

Authors' Affiliations: Departments of 1Surgery/Surgical Oncology, 2Neu-rosurgery, 3Human Genetics, 4Medicine/Hematology-Oncology, 5Pathol-ogy and Laboratory Medicine, 6Neurology, and 7Psychiatry, David GeffenSchool ofMedicine; 8JonssonComprehensiveCancerCenter, University ofCalifornia Los Angeles, Los Angeles, California; 9Department of Neurosur-gery, Tongji Hospital, Tongji Medical College, Huazhong University ofScience and Technology, Wuhan, Hubei; and 10Cancer Institute, ZhejiangUniversity, Hangzhou, Zhejiang, People's Republic of China

Note: Supplementary data for this article are available at Molecular CancerResearch Online (http://mcr.aacrjournals.org/).

Corresponding Author: Cho-Lea Tso, Department of Surgery/SurgicalOncology, Factor Building, Rm 13-260, 10833 Le Conte Avenue, DavidGeffen School of Medicine at UCLA, Los Angeles, CA 90095. Phone: 310-825-1066; Fax: 310-825-6192; E-mail: [email protected]

doi: 10.1158/1541-7786.MCR-10-0563

�2011 American Association for Cancer Research.

MolecularCancer

Research

Mol Cancer Res; 9(12) December 20111668

on December 5, 2020. © 2011 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst October 19, 2011; DOI: 10.1158/1541-7786.MCR-10-0563

http://mcr.aacrjournals.org/

the endothelin (EDN) family, which consists of a group ofvasoactive peptides referred to as EDN1, EDN2, and EDN3(14). EDNs are synthesized initially as inactive larger pre-cursor molecules and then posttranslationally cleaved byEDN-converting enzyme (ECE1) to yield the biologicallyactive 21-amino acid form (15). The effects of EDNs aremediated by 2 distinct but highly homologous G-protein–coupled receptors, EDN receptor A (EDNRA) and EDNreceptor B (EDNRB), in an autocrine and paracrine manner(16, 17). The EDNRA predominantly binds to EDN1 andEDN2 with similar affinities, and EDN3 with 1,000- to2,000-fold lower affinity, whereas EDNRB has similaraffinities for all 3 isopeptides (18). Mutations in EDN3 orEDNRB can lead to abnormal development of the entericnervous system (ENS) and melanocytes and are known toaccount for the majority of patients with Waardenburgsyndrome (WS) type IV, who exhibit both pigmentationand megacolon phenotypes (19, 20). EDN3 and EDNRBmRNA expression has been reported in fetal human entericmesenchyme and NCCs (21), and EDN3/EDNRB signal-ing is known to influence NCC proliferation, differentia-tion, and migration during ENS development (22). Theexpression of the EDN system genes has been shown in thebrain and in human glioblastoma (23, 24), as well as a broadrange of other types of human cancers (25). The role of theEDN-axis, especially in EDN1-axis, has been implicated inpromoting tumor progression. Blocking EDN receptors hasbeen suggested as a novel strategy in cancer therapy (25).In this study, we provide the first comprehensive analysis

of the expression and function of the EDN system in patient-derived GSCs. We found an essential role of autocrineEDN3/EDNRB system in GSCs. Blocking either EDNRBfunction or EDN3 production leads to GSC apoptosis andloss of migration, self-renewal, and tumorigenesis. Further-more, genome-wide expression array analyses have elucidat-edmolecular pathways and gene network connections to thisessential signaling for maintaining GSCs. This findingimplicates that the EDN3/EDNRB signaling pathway mayserve a novel therapeutic target for the development ofpotentiallymore effective treatment protocols for preventingGSC-mediated tumor recurrence.

Materials and Methods

Glioblastoma sphere cultureGlioblastoma tumor specimens were obtained from

patients who underwent surgery at Ronald Reagan UCLAMedical Center. All samples were collected under protocolsapproved by the UCLA Institutional Review Board. Disso-ciated tumor cells from fresh tumor tissue were cultured in aserum-free media containing Dulbecco's modified Eagle'smedium (DMEM)/Ham's F-12 (Mediatech) supplementedwith 20 ng/mL human recombinant epidermal growthfactor (EGF; Sigma), 20 ng/mL basic fibroblast growthfactor (FGF; Millipore), 10 ng/mL leukemia inhibitoryfactor (LIF; Millipore), and 1� B27 without vitamin A(Invitrogen) and is designated as stem cell growth factor(SGF) media. The tumorigenicity of GSCs was verified in

xenograft animals (8). In parallel, autologous glioblastomacell lines were established by propagating cells in DMEM/Ham's F-12 media supplemented with 10% FBS.

Fluorescence-activated cell sorting analyses andpurification of CD133þ GSCsA total of 2 � 105 to 5 � 105 dissociated cells from

glioblastoma tumor sphere cultures were stained with anti–CD133-APC (Miltenyi Biotech) for 20minutes at 4�C.Theanalyses for CD133 positive cells were done on a FACSCa-libur flow cytometer (Becton Dickinson) and 10,000 ormore events were collected in each analysis. To purifyCD133þ cells, dissociated glioblastomas sphere cultures orenzyme digested tumor xenografts were immunostainedwith anti-CD133 APC (allophycocyanin) under sterile con-ditions. The CD133þ and CD133� cells were sorted andcollected on a BD FACSAria II cell sorter at 70 psi using a70-mm nozzle.

Quantitative reverse transcriptase PCR analysisTotal RNAwas extracted using a RNeasy Kit (QIAGEN).

Twomicrograms of RNA from each sample were transcribedto cDNA using a TaqMan RT Reagent Kit (Applied Bio-systems). PCR was done using 5 mL cDNA equivalents to100 ng total RNA and was carried out by using SYBRGreenPCR Core Reagents (Applied Biosystems). After amplifica-tion, PCR products (5 mL) were electrophoresed on 2%agarose gel. The primer sequences and expected size ofamplified PCR products are described in SupplementaryInformation.

Immunocytochemical, histopathologic, andimmunohistochemical analysisGSCs were fixed in 4% paraformaldehyde and then singly

or doubly stained with anti-CD133 (1:100; Abcam), anti-nestin (1:200; Chemicon), or anti-EDN3 (1:200; SantaCruz Biotech) antibodies. After washing, cells were incu-bated with rhodamine red–conjugated anti-mouse or AlexaFluor 488–conjugated anti-rabbit antibodies (Invitrogen).Nuclei were counterstained with Hoechst 33342 (Invitro-gen). Histopathologic analyses were done on frozen sectionor paraffin slides stained with hematoxylin and eosin (H&E)staining as per standard technique. Immunohistochemicalstaining was done on paraffin slides. Slides were subject to a1-hour blocking step followed by the application of primaryantibody or control antibody for 1 hour at room temper-ature. The following primary antibodies were used: CD133(1:100; Abcam), EDN3 (1:100; Epitomics), EDN1 (1:250;Abcam), EDNRB (1:100; Epitomics), and EDNRA (1:200;ENZO life Science). The immunodetection was done usingVectastain ABC Standard Kit andVectorNovaRED (VectorLaboratories).

ELISATo determine the amount of EDN3 released from GSCs,

dissociated GSCswere plated in 24-well plates in triplicate atthe density of 5� 104 cells per mL per well. Four hours later,stem cell culture media was replaced with fresh SGF media

EDN3 is a Survival Factor for Glioblastoma Stem Cells

www.aacrjournals.org Mol Cancer Res; 9(12) December 2011 1669




with or without ECE1 inhibitor, plain media or serum-containing media. The respective conditioned media werecollected at various time points and were stored at �80�C.The amount of EDN3 peptides in condition media wasdetermined by an ELISA Kit specific for EDN3 (Immuno-Biological Laboratories). The absorbance was read at 450 nmby Epoch Microplate Reader (BioTek Instruments).

Flow cytometric analysis for cell-cycle distribution andcell apoptosisTo determine the effects of BQ788, BQ123, or combi-

nation treatment on cell-cycle distribution and cell apopto-sis, 2 � 105 to 5 � 105 treated and untreated GSCs werestained with propidium iodide (PI)-based hypotonic DNAstaining buffer. Cell-cycle distribution for approximately10,000 cells was analyzed using a FACScan (Becton Dick-inson). Apoptotic cells with degraded DNAwere detected asa hypodiploid or "sub-G1" peak in a DNA histogram.

Proliferation assay for GSCsThe effects of BQ788 and/or BQ123 treatment on the

proliferative activity of GSCs were determined by a MTS/PMS (phenazine methosulfate) colorimetric assay accordingto the manufacturer's instructions (Promega). Cells wereseeded into 96-well tissue culture plates at a density of 2,000cells per well in the presence or absence of EDNRA orEDNRB antagonists and incubated for 72 hours. Theabsorbance was measured at 490 nm after a 4-hour incu-bation with MTS/PMS reagents.

Quantitative measurement of cell apoptosis and cellviabilityTo determine whether EDN3 alone can support GSC

survival, cell apoptosis assays were done by using a CellDeath Detection ELISAPLUS kits (Roche), which quanti-tatively detects the amount of cleaved DNA/histone com-plexes (nucleosomes) in a given sample. The GSCs wereseeded at the concentration of 2,000 cells/well/100 mL intriplicates. The culture media was changed and switched toplain media that contain various concentrations of EDN3.Cell apoptosis was measured after 72 hours incubation byreading the absorbance at 405 nm. The same method wasapplied to the measurement of BQ788 and EDN3siRNAtreatment-induced cell apoptosis. The viability of GSCs wasdetermined by Trypan Blue staining (Invitrogen).

Knockdown of EDN3 by siRNA transfectionA reverse transfection protocol was done to deliver EDN3

siRNA (Ambion), nonsilencing control siRNA (Ambion)or glyceraldehyde-3-phosphate dehydrogenase (GAPDH)siRNA (Dharmacon) into GSCs. Briefly, a transfectioncomplex was prepared by diluting siRNA in 10 mLOPTI-MEMI (Invitrogen), then adding 10 mL OPTI-MEMI containing 0.3 mL Lipofectamine RNAiMAX trans-fection reagent (Invitrogen). This complex was then addedinto each well in 96-well plate followed by seeding 6,000GSCs in 100 mL SGF media to give a final siRNA concen-tration of 30 nmol/L in each well. EDN3 gene silencing was

determined 72 hours after transfection by quantitativereverse transcriptase PCR (qRT-PCR), using Power SYBRGreen Cells-to-CT Kit (Ambion).

Intracranial tumor formation assay and histopathologicanalysisThe role of EDN3/EDNRB signaling in maintaining the

tumorigenic capacity of GSCs was examined in Beige/severecombined immunodeficient mice (SCID) mice. A total of104 live GSCs treated with and without 50 mmol/L BQ788for 24 hours were injected in a total volume of 3 mL intobrains of mice under a UCLA Institutional Animal ResearchCommittee–approved protocol. Mice were maintained for25 weeks or until development of neurologic signs. Brains ofeuthanized mice were collected, fixed in 10% formalin,paraffin-embedded, and sectioned. Alternatively, brain tis-sue was placed in O.C.T. embedding medium (Tissue-Tekm), dipped in liquid nitrogen, and sectioned in a�20�Ccryostat. Histopathologic analyses were done on frozensection or paraffin slides stained with H&E staining as perstandard technique.

Microarray procedures, data analysis, and geneannotationMolecular profiling of 6 untreated (n¼ 3, duplicate) and 3

BQ788-treated GSCs (n ¼ 3) were carried out usingstandard Affymetrix protocols and hybridized to AffymetrixGeneChip U133 Plus 2.0 Array as described previously (8).TheDCP files were globally normalized, and gene expressionvalues were generated using the dChip implementation ofperfect-match minus mismatch model-based expressionindex. The group comparisons were done in dChip andsamples were permuted to assess the false discovery rate(FDR). To avoid inclusion of low level and unreliable signals,the difference of the means needed to exceed 100 and becalledpresent byMAS5.0 ingreater than20%of the samples.In addition, the Gene Set Enrichment Analysis (GSEA) toolwas done to identify significant gene ontology groups thatwere enriched (26). Functional annotation of individual genewas obtained from NCBI/Entrez Gene (http://www.ncbi.nlm.nih.gov/sites/entrez), UniProt (http://www.uniprot.org/), information hyperlinked over protein (http://www.ihop-net.org/), and the published literature in PubMedCentral (http://www.ncbi.nlm.nih.gov/pubmed).

Statistical analysisEach experiment was set up in triplicate and repeated at

least twice.Data were expressed asmeans� SD and analyzedusing 1-way ANOVA tests, depending on homogeneity ofvariances. All P values were 2-sided, and values less than 0.05were considered significant. SPSS v13.0 for Windows soft-ware was used for all statistical analysis.

Results

Expression of the EDN system components in GSCsThrough genome-wide expressionmicroarray analysis, we

previously identified EDN3 as one of themost overexpressed

Liu et al.

Mol Cancer Res; 9(12) December 2011 Molecular Cancer Research1670




transcripts in tumorigenic GSCs, when compared withautologous, differentiated glioblastoma cells cultured inserum-containing media (8). We thus hypothesized thatEDN3 may contribute to maintaining GSC properties. Totest this hypothesis, we first investigated the expression of theEDN system genes in GSC cultures. Three patient-derivedtumorigenic GSC lines (D431, S496, and E445), whichcontain 39.6%, 9.6%, and 1.5% CD133þ cells, respective-ly, were employed for this study (Fig. 1A). GSC, which formsemiadherent tumor spheres with highly motile cells, spon-taneously migrate outward from sphere bodies (Fig. 1A).Wetested both unsorted and sorted CD133þ and CD133� cellsfromD431 and S496GSC cultures, as well as the autologousCD133� glioblastoma cell lines passaged in serum-contain-ing media. Because E445 GSC cultures contain only 1.5%CD133þ cells, we only tested the unsorted cell population.By using qRT-PCR analysis, mRNA expression of EDN1,EDNRA, EDNRB, and ECE1 was detected in all glioblas-toma cell samples tested. Notably, the expressions level ofEDN1 and EDNRA mRNAs were particularly upregulatedin all glioblastoma cell lines cultured in 10% serum, whereashigh levels of EDN3mRNAwas only detected in GSC linesestablished in SGF media, unsorted GSCs, or sortedCD133þ or CD133� cells (Fig. 1B). EDN3 mRNA seemsto not be a SGF responsive gene or direct target of SGFbecause switching serum-cultured glioblastoma cells to SGFfor 6, 24, and 48 hours (Fig. 1B, not shown) did notstimulate comparable EDN3 mRNA expression in glioblas-toma cells (Fig. 1B). These data indicate that EDN3, but not

EDN1, is a molecular/cellular signature for GSCs.Although both CD133þ and CD133� cells sorted fromglioblastoma sphere cultures express EDN3, differentiatedCD133� glioblastoma cell lines cultured in serum do notexpress EDN3, indicating that EDN3 is only expressed inGSCs and their immediately differentiated CD133�

daughter cells maintained in stem cell culture condition.The data also suggest that GSC lines represent differentcell populations from autologous glioblastoma cell lines,which cannot be subsequently reversed following transi-tion to SGF conditions (27).

Upregulation of EDN3 mRNA coupled withdownregulation of EDN1 mRNA expressioncharacterizes stemness state of GSCsEDN1 has been implicated in the growth and progres-

sion of a wide range of human tumors (25), whereasrecently, EDN3 has been suggested as a tumor suppressorgene in female malignances (28, 29). Because we detecteda reciprocal relationship in the expression of EDN1 andEDN3 mRNA in glioblastoma cells cultured under dif-ferent conditions, we examined whether there is a recip-rocal regulation of EDN1 and EDN3 in GSCs. FBS,which contains a variety of cytokines and growth factors,was used as a surrogate for inflammatory mediators toforce GSC cultures to undergo proliferative differentiationas transit-amplifying-like cells capable of proliferation anddifferentiation, which eventually can be propagated asglioblastoma cell lines (Fig. 2A). Progressively diminished

D431D431 S496S496 D431D431 S496S4960.1% D431

A Bcell line cell line cell line cell line

D43139.6%

EDN3

(110 bp)

EDN1

(250 bp)- EDN1

- EDN3

Ce

ll lin

e (

+ 6

h S

GF

)

Un

so

rte

d s

ph

ere

s

So

rte

d C

D1

33

+ c

ells

So

rte

d C

D1

33

– c

ells

Fib

rob

last

Ce

ll lin

e (

se

rum

)

Ce

ll lin

e (

+ 6

h S

GF

)

Un

so

rte

d s

ph

ere

s

So

rte

d C

D1

33

+ c

ells

So

rte

d C

D1

33

– c

ells

Ce

ll lin

e (

se

rum

)

MW

Ce

ll lin

e (

+ 6

h S

GF

)

Un

so

rte

d s

ph

ere

s

So

rte

d C

D1

33

+ c

ells

So

rte

d C

D1

33

– c

ells

Fib

rob

last

Ce

ll lin

e (

se

rum

)

Ce

ll lin

e (

+ 6

h S

GF

)

Un

so

rte

d s

ph

ere

s

So

rte

d C

D1

33

+ c

ells

So

rte

d C

D1

33

– c

ells

Ce

ll lin

e (

se

rum

)

E4

45

ce

ll lin

e (

se

rum

)

E4

45

ce

ll lin

e (

+ 6

h S

GF

)

E4

45

9.6% S496

D43125 μm

β-Actin (152 bp)ECE1

(184 bp)

EDNRA

(119 bp)

EDNRB

(197 bp)

- β-Actin

- EDNRA

- EDNRB

- ECE1

E4451.5%

S496

200 bp100 bp50 bpβ-Actintitration

1

E445

Anti–C

D133-A

PC

Anti–C

D133-A

PC

Anti–C

D133-A

PC

Anti–Ig

γ1-A

PC

Figure 1. Presence of EDN system components in GSCs. A, tumorigenic GSC cultures, D431, S496, and E445, were established from patient-derivedglioblastoma tumors. Percentages of CD133þ cells in each culture were determined by flow cytometry analysis using anti-CD133 directly conjugated to APC.GSC cultures form semiadherent tumor spheres with cells spontaneously migrating out of spheres. Scale bar¼ 25 mm. B, total RNA from the indicated GSCscultured in SGFmedia, autologous glioblastoma cell lines cultured in serum-containingmedia and switched to SGFmedia for 6 hours and fibroblasts culturedin serum-containingmediawere extracted. ThemRNAexpression levels of EDN1, EDN3, EDNRA,EDNRB, andECE1were analyzedbyqRT-PCRwith specificprimers. b-Actin was used as an internal control gene.






EDN3 mRNA expression coupled with marked inductionof EDN1 mRNA was seen in serum-stimulated GSCcultures (Fig. 2B). Serum-induced GSC differentiationwas further characterized by the codownregulation ofseveral radial glial cells (RGC)- and NSC-associated genesexpressed by GSCs (ref. 8; Fig. 2B), including fatty acid-binding protein 7 (FABP7; ref. 30), CD133 (31), sex-determining region Y-box 2 (SOX2; ref. 32), inhibitor ofDNA binding 4 (ID4; ref. 33), and GAP43 (34), suggest-ing that EDN3 is a marker for undifferentiated, stem-likeglioblastoma cells. In contrast, the expressions of EDN1(glioblastoma tumor marker; ref. 23) and YKL-40 mRNA(glioblastoma prognostic marker; ref. 35, only expressed inD431 and S496), were induced as a result of proliferativedifferentiation (Fig. 2B). The mRNA level of topoisom-erase (DNA) II alpha (TOP2A), a maker that is associatedwith cell growth, was unchanged. This data supports thenotion that the reciprocal regulation of EDN3 and EDN1expression in GSCs may modulate and determine tumordevelopment (36, 37). Indeed, the transcripts of EDNsystem genes were determined in patient-derived glioblas-toma tumors, and elevated expression of EDN1 mRNAwas determined in all 11 tumors analyzed, whereas EDN3mRNA was expressed at a lower level in the majority oftested primary tumors when compared with EDN1mRNA (Fig. 2C). Likewise, YKL-40 mRNA wasexpressed in most glioblastoma tumors (Fig. 2C).

GSCs are actively secreting EDN3 peptides whenmaintained in SGF conditionsBecause EDN3 mRNA is uniquely expressed in GSCs,

but not in more differentiated autologous glioblastoma cellscultured in the serum, we then tested whether EDN3 wasexpressed at the protein level and secreted by GSCs main-tained in SGFmedia. Immunostaining analysis revealed thatEDN3 was only expressed in GSCs, but not in autologousglioblastoma cells cultured in serum-containing media (Fig.3A). Moreover, sorted CD133þ GSCs coexpressed EDN3and CD133 or nestin, whereas glioblastoma cell lines onlyexpressed nestin, not EDN3 or CD133. Either CD133þ orCD133� cells in all 3 tested sphere cultures express EDN3 asdetermined by 2-color flow cytometric analysis (Supple-mentary Fig. S1). ELISA assays further confirmed that theconditioned media collected from GSC cultures containedhigh concentration of EDN3, although no EDN3 peptideswere measured in conditioned media collected from glio-blastoma cell line cultures (Fig. 3B). The production ofEDN3 peptides by GSC cultures was markedly decreasedwhen cultures were switched to plain media or serum-containing media (1 week), or treated with the ECE1inhibitor SM-19721, by which GSCwere forced to undergostarvation, proliferative differentiation, and acute EDN3deficiency (Fig. 3B). These data confirmed that bioactiveEDN3 peptides were only actively secreted by GSCs main-tained in SGF media. Moreover, under light microscope,

A Patient tumor # C

200 bp100 bp

EDN3 (110 bp)

1 3 6 4 8 5 9 7 10 112 D431 E445S496

a c e 25 μm

D431 S496 E445

50 bp

EDN1 (250 bp)fdb +serum+serum +serum

B

-EDN3

EDNRB (197 bp)

EDNRA (119 bp)SG

F

+seru

m-1

wk

+seru

m-2

wk

D431 c

ell

line

S496 c

ell

line

SG

F

+seru

m-1

wk

+seru

m-2

wk

E445 c

ell

line

SG

F

+seru

m-1

wk

+seru

m-2

wk

-FABP7

-CD133

-EDN1

-β-Actin ECE1 (184 bp)

-ID4

-YKL-40

-GAP43

-SOX2

ββ-Actin (152 bp)

YKL-40 (206 bp)

-TOP2A

-GAPDH

Figure 2. Reciprocal regulation andexpression of EDN1 and EDN3mRNA in GSCs and glioblastomatumors. A, light-microscopicmorphology of indicated GSCcultures before (a, c, e) and after2-week serum-induced proliferativedifferentiation (b, d, f). Scale bar ¼25 mm. B, the indicated GSCcultures were switched to serum-containing media for the timeshown. The mRNA expressionlevels of EDN1 and EDN3 and theindicated NSC (CD133, SOX2, ID4),RGC (FABP7), and glioblastoma-associated genes (YKL-40, TOP2A)were analyzed by qRT-PCR withspecific primers. b-Actin andGAPDH were used as internalcontrol genes. C, total RNA frompatient-derived glioblastomatumors (n¼ 11) were extracted. ThemRNA expression levels of theindicated EDN system componentswere analyzed by qRT-PCR withspecific primers. YKL-40represents a glioblastomaprogression-associated gene, andb-actin was used as an internalcontrol gene.

Liu et al.





acute cell death was observed when GSCs were treated withthe ECE1 inhibitor (Fig. 3C), suggesting that autocrineEDN3 may be an essential survival factor only when GSCsare maintained in SGF conditions, not when GSCs undergocell differentiation.

Blockade of EDNRB, not EDNRA, signaling inducesGSC apoptosis without prominent cell-cycle arrestBecause proteolytic processing of EDN precursors to

biologically active peptides by ECE1 is not specific toEDN3, we then analyzed and compared the effects ofEDNRB and EDNRA antagonists on GSCs. GSCs weretreated for 72 hours with various doses of the EDNRBantagonist BQ788 or the EDNRA antagonist BQ123.

Evidently, structural/morphologic alterations and cell apo-ptosis were only detected in GSCs treated with BQ788, notBQ123 or vehicle control (ethanol), as examined micro-scopically, and BQ788-induced morphologic changes andcell apoptosis were significantly enhanced in a dose-depen-dent manner (Fig. 4A). Correspondingly, the increase inapoptotic cells (2.5- to 4-fold) appeared in a sub-G1 peak inDNA histograms analyzed by flow cytometric analysis andwas only found in GSCs treated with BQ788 (50 mmol/L),not BQ123 (50 mmol/L), in all 3 GSC cultures tested (Fig.4B). Notably, no major changes were found in the cell-cycledistribution in BQ788-treated GSCs when compared withuntreated control cells. Only slightly increased numbers ofcells in G1-phase (11%), S-phase (8%), and G2/M phases

A

CD133/nestinDAPI

Fibroblast S496 cell lineSorted CD133+

S496

CD133/DAPIDAPI

D431 cell line

DAPI DAPIDAPI

Sorted CD133+

D431 Sorted CD133+

D431

Sorted CD133+

D431

Sorted CD133+

S496

EDN3EDN3 EDN3 EDN3/DAPICD133CD133/nestin

DAPI

25 μm

MergeMerge MergeCD133/EDN3

DAPIMerge

CD133/nestinDAPI

B CD431 S496 FibroblastE445

25 μm

400

300

500

600

a

b

untreated d

e

g

h

j

k

untreated untreated untreated

10 μmol/LSM-19721

10 μmol/LSM-19721

10 μmol/LSM-19721

10 μmol/LSM-19721

ED

N3

(pg

/mL

)

0

100

200

c f i l25 μmol/LSM-19721

25 μmol/LSM-19721

25 μmol/LSM-19721

25 μmol/LSM-19721

D43

1 (d

ay 1

)D

431

(day

2)

+E

CE

1 in

hib

ito

r +

pla

in m

edia

+ 1

wk

seru

mD

431

cell

line

S49

6 (d

ay 1

)S

496

(day

2)

+ E

CE

1 in

hib

ito

r +

pla

in m

edia

+ 1

wk

seru

mS

496

cell

line

E44

5 (d

ay 1

)E

445

(day

2)

+ E

CE

1 in

hib

ito

r +

pla

in m

edia

+ 1

wk

seru

mE

445

cell

line

Med

ium

(n

o c

ells

)

Figure 3. GSCs express and secret EDN3. A, the indicatedGSC lines and autologous glioblastoma cell lines were subjected to single or dual color-fluorescentimmunocytochemical analysis for the expression of EDN3, CD133, and nestin. The fluorescent stain 40, 6-diamidino-2-phenylindole (DAPI) was usedto counterstain nuclei. Scale bar ¼ 25 mm. B, the amounts of EDN3 protein peptides secreted into the conditioned media by GSCs (5 � 104 cells/mL/well)cultured under the indicated conditions were determined by an anti-EDN3 quantitative ELISA kit and were presented as picograms per milliliter (pg/mL). Theindicated glioblastoma cell lines were cultured in serum-containing media. Data represent mean values � SD of triplicate measurements in triplicateexperiments.C, light-microscopicmorphologyofGSCcultures andfibroblasts (5�104 cells/mL/well) treatedwith andwithoutECE1 inhibitor (SM-1972) at theindicated concentrations for 7 days. Scale bar ¼ 25 mm.






C

A

a

Solvent

D431D431

untreated untreated untreated

S496 E445

a f k

020406080

100120140160

BQ788 BQ123

%G

row

th

b

(alcohol)

q

p25 μmol/L

BQ788

50 μmol/LBQ788

50 μmol/LBQ788

50 μmol/LBQ788

25 μmol/LBQ788

25 μmol/LBQ788b

c

d

g

h

i

l

m

n

50 μm

01007550250100755025010075502501007550250

1007550250100755025010075502501007550250

1007550250 1007550250

D431 cell lineSorted CD133- D431Sorted CD133+D431D431

020406080

100120140160

BQ788 BQ123

%G

row

th

μm μm μm μm

μm

μm μm

μm μm μm

Untreated

45.1% 60.3% 47.2%

UntreatedUntreated

D431 E445S496

c

B

re j o

25 μm

S496 Sorted CD133+S496 Sorted CD133-S496 S496 cell line

%G

row

th

020406080

100120140160

E445S496D431

untreated alcohol BQ123 BQ788 BQ123+BQ788

%G

row

th

0

20

40

60

80

100

120

140

160BQ788 BQ123

d e

%G

row

th

0

20

40

60

80

100

120

140

160

Fibroblast

BQ788 BQ123

E445

15.0% 42.0%13.0%

40.3%

56.3%

29.4%14.4%

6.1%24.2%

15.5%

25.5%

48.2%

32.1% 19.7%

5.8%46.5%

6.3%

21.3%

35.0%

48.2%

16.8%

BQ78850 μmol/L

BQ78850 μmol/L

BQ78850 μmol/L

BQ12350 μmol/L

13.3%

54.6%

33.7%11.7%

6.1%

61.1%

23.4%15.5%

6.7%

48.2%

46.4%

5.5%

75 μmol/LBQ788

100 μmol/LBQ788

100 μmol/LBQ788

100 μmol/LBQ788

100 μmol/LBQ788

100 μmol/LBQ123

75 μmol/LBQ788

75 μmol/LBQ788 BQ123

50 μmol/LBQ123

50 μmol/L

Figure 4. Blockade of EDNRB not EDNRA induces GSC apoptosis independent from cell-cycle arrest. A, light-microscopic morphology of GSC culturestreated with the indicated concentrations of BQ788, BQ123, or vehicle control (ethanol alcohol) for 3 days. Scale bar ¼ 25 or 50 mm. B, the indicated GSClines were treated with 50 mmol/L BQ788 or BQ123 for 3 days, and cell-cycle distributions including sub-G1 fraction were determined by flow cytometricanalysis.C, the sortedCD133þandCD133�cells andunsortedGSC lines, autologousglioblastomacell lines, andfibroblastswere treatedwithBQ788,BQ123(a–d), or combinations (e) at the indicated concentrations for 3 days. The cell growth was determined by MTS/PMS cell proliferation assays. Datarepresent mean values � SD of triplicate measurements in triplicate experiments.

Liu et al.





(10%) were detected in D431, S496, and E445 GSCs,respectively, after BQ788 treatment (Fig. 4B).To gain further insight into the effects of the blockade of

EDNRB on GSC growth, GSCs were treated for 72 hourswith various doses of BQ788 or BQ123 (Fig. 4C) andassayed for cell proliferation. Inhibition of proliferation wasnot seen when GSCs (sorted or unsorted GSC) were treatedwith 25 to 100 mmol/L BQ123, although treatment with 75or 100 mmol/L BQ788 reduced cell proliferation by 30% to80% in D431, 30% to 60% in S496 and 40% to 60% inE445 (Fig. 4C, a–c). The combination of BQ788 andBQ123 had no additive effect on suppressing cell growth,although a notable growth stimulation was observed inunsorted D431 and S496 by treatment with 50 mmol/LBQ123 (Fig. 4, e). Of note, both fibroblasts and autologousglioblastoma cell lines passaged in serum-containing mediathat do not produce EDN3 were resistant to BQ788treatment (Fig. 4C, a–d). In contrast, both CD133þ andCD133� daughter cells sorted from sphere culture (D431

and S496) are sensitive to BQ788 treatment. BQ788 treat-ment also decreased clonogenic efficiency of both CD133þ

and CD133� daughter cells sorted from GSC cultures thatwere initiated from purified CD133þ GSCs (Fig. 4D).Notably, untreated CD133� daughter cells are fast-growingcells, which exhibited higher clonogenic efficiency comparedwith those of untreated CD133þ daughter cells on day 7after cell seeding.To verify that the observed decreased proliferative activity

and clonogenic efficiency caused by BQ788 is mainly due tocell apoptosis, not cell-cycle arrest, the induction of GSCapoptosis by the EDNRB antagonist was further determinedusing an ELISA-based cell death assay, which quantitativelydetects the amount of cleaved DNA/histone complexes.Significant cell apoptosis were detected in all 3 GSC linestreated with 50 to 100 mmol/L BQ788 when compared withuntreated cells (Fig. 4E). Thus, cell morphology observa-tions, clonogenic assays, cell-cycle analyses, and cell apopto-sis assays suggest that EDN3/EDNRB signaling is a survivalpathway for GSCs.

Blockade of EDN3/EDNRB signaling leads to the loss ofcell spreading/migration, self-renewal, and tumorinitiationNext, we determined whether blockade of EDN3/

EDNRB signaling would weaken GSC self-renewal capa-bility. Significant cell apoptosis and impaired GSC migra-tion were detected in all 3 GSC lines treated with 50 mmol/LBQ788 when compared with untreated cells (Fig. 5A, a–f).More importantly, these BQ788-treated GSCs failed toregenerate spheres when live cells were reseeded at clonaldensity (Fig. 5A, h, j, l), indicating that the self-renewalcapacity of GSCs was lost. The impairment of cell migrationand self-renewal by BQ788 treatment was also shown inpurified CD133þ GSCs (Supplementary Fig. S2). To studywhether blocking the production of endogenous EDN3 inGSCs will have impact on these GSC functions, we carriedout an on-target knockdown of EDN3. Treatment withEDN3 siRNA, not negative control siRNA, not only effi-ciently knocked down EDN3 mRNA expression (Fig. 5B),but also abrogated the EDN3 peptide release (Fig. 5C).More importantly, prominent cell apoptosis was only deter-mined in GSCs treated with EDN3 siRNA (Fig. 5D).Microscopic changes in cellular morphology were observed,and in contrast to negative control siRNA-treated GSCs(Fig. 5E, c, g, k), loss of endogenous EDN3 by EDN3siRNA treatment resulted in loss of self-renewal potentialand cell migration capacity (Fig. 5E, d, h, l). By contrast,GSCs in negative control siRNA-treated cultures spontane-ously migrated/spread out of tumor spheres and formed thesurrounding monolayer (Fig. 5E, q–s), implicating a role ofEDN3/EDNRB signaling in GSC self-renewal andmigration.On the basis of the requirement of EDN3/EDNRB

signaling for self-renewal, migration, and survival of GSCsin culture, we test whether the EDN3/EDNRB signaling isrequired for maintaining the tumorigenic potential ofGSCs. Unsorted GSCs or sorted CD133þ GSCs were

D

30

40

50500 250 120

60 30 15(4.8

%)

(4.4

%)

(1.7

%)

(0.0

%)

(0.0

%)

(1.6

%)

(0.8

%)

(0.0

%)

(0.0

%)

(0.0

%)

6 %

).8

%)

5 %

).0

%)

.0 %

)

.4 %

).2

%)

.0 %

).0

%)

.0 %

)

0

10

20

50 µmol/L BQ788Untreated50 µmol/L BQ788Untreated

Co

lony

co

un

ts

(8.

(2.

(2.

(0.

(0.

sorted CD133+ cells sorted CD133- cells

(2.

(1.

(0.

(0.

(0.

0.4

0.5

0.6

0.7

0.8

0.9

1

*

*

*

*

*

*

0.4

0.5

0.6

0.7

0.8

0.9

1

*

*

*

*

*

*

of

cyto

pla

smic

nu

cleo

som

esb

ance

405

nm

)

E

0

0.1

0.2

0.3

D431 S496 E445

0

0.1

0.2

0.3

D431 S496 E445

Am

ou

nt

olig

o(a

bso

r

BQ788 0 µmol/L

BQ788 25 µmol/L BQ788 100 µmol/L

BQ788 50 µmol/L

Figure 4. (Continued ) D, CD133þ and CD133� daughter cells sortedfrom sphere cultures that initiated from purified CD133þ GSCs wereseeded at the indicated cell density per well in 96-well plate by limitingdilution in 100 mL SGFmedia. The clonogenic efficiency was determinedunder the presence or absence of 50 mmol/L BQ788 treatment. Coloniescontaining more than 20 cells were counted at day 7 after cell seeding.Data represent mean values� SD of triplicate measurements in triplicateexperiments. % clonogenic efficiency was indicated. E, the indicatedGSCcultureswere treatedwithBQ788at the indicatedconcentrations for3 days, and cell apoptosis were determined by a Cell Death DetectionELISAPLUS kit, which quantitatively detects the amount of cytoplasmicoligonucleosomes. Data represent mean values � SD of triplicatemeasurements in triplicate experiments. �, P < 0.05 versus untreatedcells.






treated with or without 50 mmol/L BQ788 for 24 hours.The viability of the inhibitor-treated cells, which rangedfrom 80% to 87%, was determined by trypan blue staining.Following washing, 104 live cells were stereotacticallyinjected into the brains of SCIDmice. Mice (15/16) whichreceived untreated GSCs developed neurologic signs atweeks 14 to 22 postinjection. The H&E staining of tumorxenografts identified histologic hallmarks of human glio-blastomas (Fig. 5F, a–l). In particular, the hypercellularzones surrounding necrotic foci similar to histopathologicfeatures of pseudopalisading necrosis was observed (Fig. 5F,a–d). The infiltrative tumor exhibit hypercellularity, hyper-chromatism, pleomorphism, mitosis, vascular endothelialhyperplasia, and oligodendroglial components were alsoidentified (Fig. 5F, e–l). On the other hand, injection ofGSCs pretreated with 50 mmol/L BQ788 failed to developa tumor (0/14) at week 25. The area of tumor injection siteshows no evidence of neoplastic cells (Fig. 5F, m–r) andshow no significant changes when compared with normalbrain tissue (Fig. 5F, s–v), except a scar-like feature isoccasionally observed near the injection site (Fig. 5F, r).These data thus suggest the potential involvement ofautocrine EDN3/EDNRB signaling in maintaining thetumorigenic capacity of GSCs in culture.To investigate whether in vitro cultured GSCs change the

expression of the EDN ligands and receptors, we analyzedand compared the expression of EDN system genes in tumorxenografts (initiated from purified CD133þ GSCs), andCD133þ and CD133� cells directly sorted from them, byRT-PCR (Fig. 5G, a). CD133, EDN1, EDN3, andEDNRAmRNA were determined in tumor xenografts (Fig.5G, a), whereas a lower level of EDN3 mRNA was detectedcompared with that of EDN1 mRNA. However, althoughEDNRB mRNA was detected in CD133þ GSCs used forinjection, it could not be detected in tumor xenografts afterthe 30 cycles of PCR. Similar expression patterns of theEDN system genes were also shown in tumors by immu-nohistochemical staining (Fig. 5G, b). Likewise, EDN1,EDN2, EDN3, and EDNRA mRNA were also determinedin both CD133þ and CD133� cells sorted from tumorxenografts, whereas CD133þ cells express a lower level ofEDN1 and a higher level of EDN3 mRNA when comparedwith those of CD133� cells (Fig. 5G, c). YKL-40 mRNAcould only be detected in tumor xenografts and sortedCD133� cells (Fig. 5G, a, c). When CD133þ cells wererecultured in SGF media for 7 days, the expression ofEDNRB mRNA could be redetected (Fig. 5G, d). On theother hand, tumor xenografts which exhibit a higher level ofEDN3 mRNA and lower levels of EDN1 and EDN2mRNA were also identified (Supplementary Fig. S3). Nev-ertheless, a higher level of EDNRA mRNA is alwaysexpressed in tumor xenografts and sorted CD133þ andCD133� cells, whereas a low level of EDNRBmRNA couldonly be detected in tumor xenografts. These data maysuggest that the initiation of tumor growth in vivo fromGSCs is accompanied with the activation/upregulation ofEDNRA signaling and downregulation of EDNRBsignaling.

Exogenous EDN3 alone can maintain GSC survival butnot promote GSC expansionTo explore the possibility that EDN3 alone might be a

growth factor or a survival factor for GSCs, we removed stemcell growth factors (FGF, EGF, LIF) from the media andreplaced them with various doses of recombinant EDN3.Apoptotic cells were observed in all tested GSCs maintainedin plain media (40%–50% viable cells) and the addition ofEDN3 (1–100 nmol/L) improved the viability of GSCs asdetermined by the trypan blue exclusion test (50%–80%viability; P < 0.05). However, reduced viability was observedwhen high doses of EDN3 (300 and 1,000 nmol/L) wereadded (Fig. 6A). Rescue of GSCs from apoptosis by addingexogenous EDN3 was further evidenced by a quantitativeapoptosis assay coupled withmorphologic examination (Fig.6B and C). In general, 30 nmol/L EDN3 alone lowered thedegree of cell apoptosis to a level close to that of SGFcompared with plain media. Furthermore, low proliferativestimulation was seen with the addition of exogenous EDN3(�1.5-fold; Fig. 6A), yet exogenous EDN3 alone was notable to propagate GSCs in culture as that of SGF media,suggesting that the significance of "growth stimulation" (P <0.05) is likely due to significantly fewer apoptotic cellsmaintained when compared with cells cultured in plainmedia. This notion was further tested by investigating theeffects of exogenous EDN3 on sorted CD133þ andCD133� daughter cells. We found that EDN3 alone couldenhance the viability of both types of cells in a dose–responsefashion as determined by trypan blue staining (Fig. 6D).However, in contrast to cells cultured in SGF, sphereformation and cell expansion were not achievable by EDN3alone, even when EDN3 were repetitively added daily(Supplementary Fig. S4). These data therefore suggest thatEDN3 alone is not a potent mitogen for GSCs but mayprovide cell survival benefits.

Genes regulated by EDN3/EDNRB signalingTo gain further insight into the molecular mechanisms

underlying the requirement of EDN3/EDNRB signaling inmaintaining GSCs, we carried out a large-scale gene expres-sion comparison of GSCs treated with and without EDNRBinhibitor (BQ788). Probe set signals on the expression arraythat were equal to less than 1.5-fold lower in 3 treated (n¼ 3patients) versus 6 untreated GSC samples (50 mmol/L, 24hours; n¼ 3 patients, sample duplicate) with a pairwise t test(P < 0.05) were selected. Samples were permuted 100 timesby dChip for a FDR of 11% (median), and 61 significantgenes were obtained (Table 1). The major downregulatedgenes identified among 61 genes are those involved inorganizing cytoskeleton structure and function (e.g.,EFEMP1, IQGAP1, CADL1, WASF2, S100A6, MYO10,MYO6, CCDC88A, RHOQ, SRGAP2, PALLD, CTSB,WASL, FRMD4B, HIP1, RAB31, and VPS35), mitoticspindle assembly, mitotic checkpoint (e.g., ASPM,NUSAP1, ASPM, and USAP1), and DNA repair or anti-apoptosis (e.g., UHMK1, NIPBL, KRAS, G2E3, DHX9,GTSE1, RIF1, and NAA15), reflecting disruptions in cellstructure, cell polarity, cell movement, intracellular

Liu et al.





BQ788treatment

Posttreatmentreseeding

A

F

B

C

D

E

a e i m q s

b f j n

c g k o

d h l p

r t

u

v

D43

1

400

350

300

250

200

150

100

50

0

0.90.80.70.60.50.40.30.20.1

0Am

ou

nt

of

cyto

pla

sm

ico

ligo

nu

cle

oso

me

s(a

bso

rba

nce

40

5 n

m)

Control siRNA

ED

N3

(pg

/mL

) EDN3 siRNA

Control siRNA

EDN3 siRNA-1

D431 S496 E445 Fibroblast

D431 S496 E445 Fibroblast

Co

ntr

ol

Rea

gen

tsC

on

tro

l siR

NA

ED

N3

siR

NA

-1E

DN

3 si

RN

A-2

No

cD

NA

M.W

.C

on

tro

lR

eag

ents

Co

ntr

ol s

iRN

AE

DN

3 si

RN

A-1

ED

N3

siR

NA

-2N

o c

DN

A

S49

6E

445

D43

1

50 μmol/L

S49

6E

445

Figure 5. Blockade of EDN3/EDNRB signaling impairs GSC self-renewal, cell migration, and tumorigenic capacity. A, light-microscopic morphology of GSCcultures. The indicatedGSC cultureswere treatedwith andwithout BQ788 at the concentration of 50 mmol/L for 3 days and the pictureswere taken (a–f). Cellswere harvested and were reseed at clonal density, which prevent cells from forming aggregates. The cultures were incubated for 3 days and pictures weretaken (g–l). Scale bar¼25mm.B, the indicatedGSCcultureswere transfectedwith negative control siRNA, EDN3siRNA (clone#1andclone#2), or treatedwithor without transfection reagents. Cells were incubated for 3 days and the levels of EDN3 mRNA were assayed by qRT-PCR. The GAPDH was used as aninternal control gene. Fibroblasts serve as a negative control cell line, which does not express EDN3 (data not shown). C, EDN3 protein peptidesconcentrations secreted into the conditionedmedia by transfected cells were assayed by anti-EDN3 ELISA kit. Data representmean values�SDof triplicate.






transports, cell division, and cell survival pathway.Concomitantly, a series of genes that provides negativeregulation of development, cell growth, and differentiationwere identified (e.g., IFI27, SOCS3, MTUS1, SCML1,IFI16, LUZP1, IL6ST, JMJD1C, TEAD1, LINGO1,C9orf86, PSD3, andUSP10), suggesting the stemness prop-erties and cellular quiescence in GSCs were disturbed. Thedistinctive gene expression profiles of randomly selectedgenes were verified by qRT-PCR (Supplementary Fig.

S5). Correspondingly, GSEA identified 97 significant geneontology (GO) clusters (Supplementary Table S1), mostlyassociated with microtubule cytoskeleton organization, celldivision, motor activity, spindle, DNA packaging, leadingedge, transcription repressor activity, negative regulation ofDNA binding, detection of chemical stimulus, cell-cyclearrest, and neuron development, offering a potential molec-ular explanation for how loss of EDN3/EDNRB signalingmay impact GSC migration, self-renewal, cell survival, and

G

200 bp

ab c

d

100 bp

200 bp100 bp

200 bp

100 bp

M.W

.

β-ac

tin

(15

2 b

p)

CD

133

(226

bp

)

ED

N1

(250

bp

)

ED

N3

(110

bp

)

ED

NR

A (

119

bp

)

ED

NR

B (

197

bp

)

YK

L-4

0 (2

06 b

p)

M.W

.

No

cD

NA

β-ac

tin

(15

2 b

p)

ED

N1

(250

bp

)

ED

N2

(158

bp

)

ED

N3

(110

bp

)

ED

NR

A (

119

bp

)

ED

NR

B (

197

bp

)

β-actin (152 bp)

CD133 (226 bp)

EDN1 (250 bp)

EDN2 (158 bp)

EDN3 (110 bp)

EDNRA (119 bp)

EDNRB (197 bp)

YKL-40 (206 bp)

Figure 5. (Continued ) measurements in triplicate experiments. �, P < 0.05, ��, P < 0.001 versus control siRNA transfected cells. D, cell apoptosis in negativecontrol siRNA or EDN3 siRNA transfected cells was determined by a cell death detection ELISA kit. Data represent mean values � SD of triplicatemeasurements in triplicate experiments. �, P < 0.05 versus control siRNA-treated cells. E, light-microscopic morphology of indicated cells treated with assaybuffer (a, e, i, m), transfection reagents (b, f, j, n), negative control siRNA (c, g, k, o), and EDN3 siRNA-1 (d, h, l, p) for 72 hours, aswell as negative control siRNAfor 7 days (q–s). Scale bar¼25or 50mm. F, representative photographs of hematoxylin and eosin (H&E) staining ofmouse brain tissue. Brain tissues frommiceinjectedwith untreatedGSCs display invasive growth of gliomaswith diffuse infiltration into the surrounding tissue and vessels (a–l). Infiltrating tumor exhibitshypercellular zones surrounding necrotic foci and forms a "pseudo-palisading" necrosis pattern (S496, frozen section; a–d). Other important histopathologicfeatures of glioblastoma are also seen in tumor lesions, including hypercellularity (e), hyperchromatism (f, g), pleomorphism and mitosis (h–j), vascularendothelial hyperplasia (k), andoligodendroglial component (l; D431,S496, E445).Mousebrains injectedwithGSCspretreatedwith50mmol/LBQ788 showedno evidence of tumor development (m–r). Mouse brain injectedwith stem cell media serves as control for normal brain tissue (s–u). Magnification, 4� (o, s), 8�(a, m), 100� (e, f, p, t, u), 200� (b, c, g, n, q, r, v), 400� (d, h–l). G, presence of EDN system components in glioma xenografts. The expression of the indicatedEDN system components in both subcutaneous and intracranial glioma xenografts were analyzed by qRT-PCRwith specific primers. b-Actin was used as aninternal control gene (a). Representative immunohistochemical stainings of EDNcomponents in intracranial tumor xenografts (b). ThemRNA levels of the EDNsystem components in uncultured CD133þ and CD133� cells directly sorted from the glioma xenografts were analyzed by qRT-PCR (c). The expressions ofEDN system transcripts in CD133þ cells recultured in SGF media for 7 days (d).

Liu et al.





EDN30 nmol/L

EDN30 nmol/L

EDN30 nmol/L

D431 S496 E445A C

40

60

80

100

1

1.5

2P < 0.05

P < 0.05

0

20

0

40

60

80

100

1

1.5

2

Gro

w (f

old

)G

row

(fo

ld)

Gro

w (f

old

)

D431

P < 0.05

P < 0.05

40

60

80

100

1

1.5

2

0

20

0

0.5

0.5

0.5

X

XX

Xp>0.05

S496

% V

iab

ility

% V

iab

ility

% V

iab

ility

0

20

0

E445

EDN3 (nmol/L)1,00030010030310 10

SGF SGFSGF25 μm

1,000nmol/L

300nmol/L

100nmol/L

30nmol/L

10nmol/L

10nmol/L

10nmol/L

3nmol/L

3nmol/L

3nmol/L

1nmol/L

1nmol/L

1nmol/L

30nmol/L

30nmol/L

100nmol/L

100nmol/L

300nmol/L

300nmol/L

1,000nmol/L

1,000nmol/L

B

1

1.2

1.40 1 30 100 1,000 SGF

*

*

EDN3 (nmol/L)

Am

ou

nt

of

cyto

pla

smic

o

ligo

nu

cleo

soes

(ab

sorb

ance

40

nm

)

0

0.2

0.4

0.6

0.8

E445S496D431

++

+

D

40

60

80

100

**

**

Δ ΔΔ

ΔΔ

0

20

SGF1,000100300

% V

iab

ility

CD133+ D431

CD133 – D431

EDN3 (nmol/L)

Figure 6. EDN3 is a survival factor forGSCs. A, the indicatedGSC lineswere cultured in the plainmedia supplementedwith variousdoses of EDN3as indicated.Cellswere incubated for 3 days and viability was determined bymanually counting the trypan blue-stained samples in a hemacytometer chamber and presentas percentage of viable cells in cultures (D). The cell growth was determined by MTS/PMS cell proliferation assays and present as fold increase comparedwith nontreated cells (&). Data representmean values�SDof triplicatemeasurements in triplicate experiments. XP > 0.05 versus cells cultured in plainmedia.B, cell apoptosis in (A) was assayed by a cell death detection ELISA kit. Data representmean values�SDof triplicatemeasurements in triplicate experiments.�,P<0.05 versusD431andS496GSCs treatedwith 1, 30, 100, 1000nmol/LEDN3andSGFmedia.~,P<0.05 versusE445cells treatedwith 1, 30, 100nmol/LEDN3 and SGF media. þ, P < 0.05 versus D431, S496, and E445 cultured in SGF media. C, light-microscopic morphology of cell cultures in (A). Scalebar ¼ 25 mm. D, cell survival was assayed in sorted CD133þ and CD133� GSC cells cultured in the SGF, plain media, or plain media supplementedwith various doses of EDN3 as indicated. Cells were incubated for 7 days and viability was determined by trypan blue staining. Data represent meanvalues � SD of triplicate measurements in triplicate experiments. �, P < 0.05, ��, P < 0.01 (CD133þ cells), D, P < 0.05, DD, P < 0.01 (CD133� cells) versuscells cultured in plain media.






tumorigenic capacity. We used the default parameters ofGSEA and Gene Ontology gene sets were permuted 1,000times.We called a gene set significant if the FDRQ value wasunder 0.15. BQ788 works by potently and competitivelyinhibiting EDN3 binding to EDNRB; therefore, it mayexplain that downregulation of EDNRB transcripts byBQ788 treatment was not determined either by expressionmicroarray or RT-PCR (data not shown). However, severalGSC genes (e.g., EFEMP, IQGAP1, SOCS3, WAVE2,MYO10, and K-RAS) downregulated by BQ788 treatmenthave been previously reported to be linked to activation ofmitogen-activated protein kinase (MAPK) pathways (Sup-plementary Table S2), which are the downstream effectorpathways of EDNRB signaling pathway (38–43), suggestingthat the identified genes characterizing stemness, migration,and survival of GSCs are regulated by EDN3/EDNRBsignaling.

Discussion

The finding of EDN3 overexpression in GSCs throughthe expression microarray analysis led to our novel investi-gation of the EDN system in cultured GSCs. In normalNCC cultures, EDN3/EDNRB signaling prevents the pre-mature differentiation of crest-derived precursors (22, 44).In vivo, EDN3/EDNRB signaling is required for NCCmigration during ENS development (22). Mutations inEDN3 or EDNRB can lead to the abnormal developmentof ENS and melanocytes, implying that the loss of functionin these 2 genes may interfere with tumor development byEDN3/EDNRB signaling-dependent GSCs. Indeed, block-ade of EDN3/EDNRB signaling of GSCs leads to the loss oftumorigenic potential in our GSC model. We found thatcultured GSCs express both neural and mesenchymal sig-natures characterizing NCC-like cells and, therefore,hypothesized that autocrine EDN3/EDNRB signaling maybe a survival/maintenance factor for GSCs. In fact, whenGSCs undergo serum-induced proliferative differentiation, asignificant loss of EDN3 production accompanied bycodownregulation of a series of transcripts related to NCCs,RGCs, and NSCs was observed. For instance, SOX2 is amarker of NSCs and is essential for maintaining the plu-ripotent, self-renewal, and undifferentiated phenotypes ofembryonic stem cells.Moreover, silencing of SOX2 inGSCscan cause the loss of tumorigenicity (45). FABP7, a glial-specificmarker, is a direct target ofNotch signaling inRGCs,and the involvement of Notch signaling in the maintenanceof the tumorigenic potential of GSCs has been reported (46).Thus, autocrine EDN3/EDNRB signaling may preventGSCs from differentiating prematurely.Direct evidence of the differentiation-inhibiting effect of

EDN3 was shown in clonogenic cultures of ENS progeni-tors supplemented with EDN3, by which neuronal andglial differentiation pathways were blocked and the multi-potent state of progenitors was maintained (44, 47). Theseobservations thus support the view that autocrine EDN3maintains the undifferentiated state of GSCs. Moreover,EDN3 and EDN1mRNAs are altered reciprocally inGSCs

in response to differentiation, indicating that EDN3 andEDN1 genes can be epigenetically regulated, and thatincreased EDN1 may be associated with the onset oftumorigenesis. Indeed, we detected an increased expressionof EDN1, but not EDN3, mRNA in patient glioblastomatumors (compared withGSCs), suggesting that EDN3mayplay a tumor suppressor-like role to maintain the quies-cence and tumorigenic potential of GSCs. In agreementwith this, a frequent loss of EDN3 expression in breasttumors due to epigenetic inactivation has previously beensuggested (29). Likewise, there were decreased levels ofEDN3 and increased levels of EDN1 and EDN2mRNA incancerous cervical epithelial cells compared with normalcervical epithelial cells (28). Thus, the reduction of EDN3levels (e.g., induced by exogenous cues) may implicate theinitiation of proliferative differentiation from quiescentGSCs toward more differentiated progeny, by which theproliferative progeny gradually overpopulate a tumor. Itseems that sole EDN3/EDNRB signaling alone cannotgrant tumorigenic potential to GSCs, yet the removal ofEDN3/EDNRB signaling contributed undifferentiated,antiapoptotic, and migratory properties to GSCs, whichmay weaken or eliminate their tumorigenic capacity. Auto-crine EDN3 in EDNRBþGSCs, may therefore, play a rolein antidevelopmental, antidifferentiating, antiapoptotic,and antitumorigenic functions, thereby help in maintain-ing a continuous GSCs pool.We found that loss of EDN3 production in GSCs

through serum-induced differentiation did not cause cellapoptosis, whereas directly blocking EDN3/EDNRB sig-naling causes severe GSC apoptosis. These observationsseem to support the notion that autocrine EDN3/EDNRBsignaling only provides survival benefits when GSCs aremaintained in SGF condition. A previous study has shownthat stimulation of the EDNRB in astrocyte induces cAMPresponse element-binding protein (CREB) and c-fosexpression via multiple MAPK signaling pathways, includ-ing the extracellular regulated kinase (ERK) 2, c-Jun N-terminal kinase (JNK1), and p38 kinase (38). EDN3stimulation also rapidly increased phosphorylation ofERK2 in neural progenitor cells (39) and activated IkappaBand MAPK in the human colonic epithelial cells (40).Likewise, EDN3 treatment resulted in the activation ofMAPK-p90 ribosomal S6 kinase-CREB and cAMP-pro-tein kinase A-CREB pathways in melanocyte culture (41,42) and induction of ERK1/2 and focal adhesion kinasephosphorylation in melanoma cells (43). One importantnote is that the activation of EDNRB by EDN3 also leadsto loss of expression of the cell adhesion molecule E-cadherin and associated catenin proteins, while increasingSnail andN-cadherin expression (43). Thus, the expressionprofiles of BQ788-treated GSCs supports the notion thatthese pathways are the downstream effector pathways ofEDN3/EDNRB signaling pathway that contribute GSCsurvival and migration (Supplementary Table S2). Theusage of an inducible knockdown of EDNRB wouldstrengthen the role of EDNRB signaling in GSCs. More-over, because ex vivo treatment with BQ788 may already

Liu et al.





Table 1. Genes expressed at lower levels in BQ788-treated GSCs compared with untreated GSC

Gene GeneID

Symbol Foldchange

P Functional involvement

IFNa-inducible protein 27 3429 IFI27 4.30 0.029619 Antigrowth, antidifferentiation, apoptosisHeparan sulfate (glucosamine)3-O-sulfotransferase 1

9957 HS3ST1 4.00 0.003801 Synthesis of anticoagulant heparin,antitumor

EGF-containing fibulin-likeextracellular matrix protein 1

2202 EFEMP1 2.89 0.040817 Cell adhesion and migration

Asp (abnormal spindle)-like,microcephaly associated

259266 ASPM 2.61 0.004877 Mitotic spindle regulation

IQ motif containing GTPaseactivating protein 1

8826 IQGAP1 2.45 0.000015 Modulation of cell architecture

Suppressor of cytokine signaling 3 9021 SOCS3 2.44 0.000182 Negative regulators of cytokinesignaling, tumor suppressor

Caldesmon 1 800 CALD1 2.41 0.038236 Smooth muscle and nonmusclecontraction

Mitochondrial tumor suppressor 1 57509 MTUS1 2.36 0.002229 Tumor suppressor and antimitosisSex comb on midleg-like 1(Drosophila)

6322 SCML1 2.28 0.000178 Control of embryonic development

Mex-3 homolog A 92312 MEX3A 2.26 0.002964 Involvement in posttranscriptionalregulatory mechanisms

IFNg-inducible protein 16 3428 IFI16 2.25 0.008661 Antigrowth, cellular quiescence, apoptosisWAS protein family, member 2 10163 WASF2 2.22 0.005534 Cell shape and motilityTopoisomerase (DNA) I 7150 TOP1 2.20 0.002516 Altering the topologic states of DNA

during transcriptionU2AF homology motif(UHM) kinase 1

127933 UHMK1 2.20 0.002059 Prevent senescence in G0/G1;cell migration

S100 calcium bindingprotein A6 (calcyclin)

6277 S100A6 2.19 0.032921 Cell migration/motility, antisenescence

Nipped-B homolog (Drosophila) 25836 NIPBL 2.17 0.000182 Tissue development, sisterchromatid cohesion, DNA repair

Synaptotagmin I 6857 SYT1 2.16 0.026320 Vesicular trafficking and exocytosisMyosin X 4651 MYO10 2.16 0.012113 Filopodia formationMyosin VI 4646 MYO6 2.15 0.002580 Intracellular vesicle and organelle transportLow density lipoprotein receptor 3949 LDLR 2.14 0.012743 Cholesterol synthesisCCDC88A coiled-coil domaincontaining 88A

55704 CCDC88A 2.12 0.006578 Actin organization, cell motility/migration

Leucine zipper protein 1 7798 LUZP1 2.12 0.015862 Embryonic development of brainv-Ki-ras2 Kirsten rat sarcomaviral oncogene homolog

3845 KRAS 2.10 0.001864 Possessing intrinsic GTPaseactivity, antiapoptosis

Ectodermal-neural cortex(with BTB-like domain)

8507 ENC1 2.06 0.020856 Actin-binding protein; oxidativestress response

RNA binding motif protein 8A 9939 RBM8A 2.06 0.004386 Coupling pre- and post-mRNAsplicing events

DEP domain containing 1 55635 DEPDC1 2.05 0.007013 Cancer/testis antigenG2E3 G2/M-phase–specificE3 ubiquitin protein ligase

55632 G2E3 2.04 0.004153 Prevention of apoptosis inearly embryogenesis

ADP-ribosylation factor-like 7 10123 ARL4C 2.02 0.022123 Cholesterol transportSecretory carrier membraneprotein 1

9522 SCAMP1 2.02 0.011002 Carriers to cell surfacein post-golgi recycling pathways

Formin-like 2 114793 FMNL2 1.99 0.002077 Morphogenesis, cytokinesis,and cell polarity

Ras homolog gene family,member Q

23433 RHOQ 1.96 0.005201 Myofibril assembly, exocytic vesiclefusion, glucose uptake

(Continued on the following page)






Table 1.Genes expressed at lower levels in BQ788-treated GSCs compared with untreated GSC (Cont'd )

Gene GeneID

Symbol Foldchange

P Functional involvement

Interleukin 6 signaltransducer (gp130)

3572 IL6ST 1.94 0.011686 Embryonic development

DEAH (Asp-Glu-Ala-His)box polypeptide 9

1660 DHX9 1.93 0.006616 DNA repair, genome stability

Oxysterol binding protein-like 8 114882 OSBPL8 1.93 0.001205 Intracellular lipid receptorsTHO complex 2 57187 THOC2 1.92 0.000834 Binds to spliced mRNAs to

facilitate mRNA exportZinc finger protein 533 151126 ZNF385B 1.92 0.009225 Neuronal function; RNA

maturation and stabilityG-2 and S-phase expressed 1 51512 GTSE1 1.91 0.007537 Antiapoptosis, control DNA damageNucleolar and spindleassociated protein 1

51203 NUSAP1 1.90 0.001876 Spindle microtubule organization

Jumonji domain containing 1C 221037 JMJD1C 1.90 0.001287 Reactivation of silenced genesin undifferentiated ES cells

TEA domain family member 1 7003 TEAD1 1.88 0.001749 Transcriptional repressor; antidifferentiationSLIT-ROBO Rho GTPaseactivating protein 2

23380 SRGAP2 1.87 0.000640 Controlling cell spreadingand cell migration

Adaptor-related protein complex 1,sigma 2 subunit

8905 AP1S2 1.86 0.009349 Recognition of sorting signals

Palladin 23022 PALLD 1.85 0.003542 Control of cell shape, adhesion,and contraction

Leucine rich repeat neuronal 6A 84894 LINGO1 1.85 0.002910 Negative regulator ofoligodentrocyte differentiation

Cathepsin B 1508 CTSB 1.82 0.007249 Invasion, migrationWiskott–Aldrich syndrome-like 8976 WASL 1.81 0.009275 Induces actin polymerization

and redistributionChromosome 9 open readingframe 86

55684 C9orf86 1.81 0.001019 Cell growth regulation

HIV TAT–specific factor 1 27336 HTATSF1 1.79 0.007012 Process of transcriptional elongationPleckstrin and Sec7domain containing 3

23362 PSD3 1.79 0.003251 Tumor suppressor

FERM domain containing 4B 23150 FRMD4B 1.79 0.004509 Cytoskeletal signals andmembrane dynamics

RAP1 interacting factorhomolog (yeast)

55183 RIF1 1.76 0.006310 DNA repair

Discoidin, CUB and LCCLdomain containing 1

285761 DCBLD1 1.76 0.011326 Integral to membrane, cell adhesion

Glutamyl-prolyl-tRNA synthetase 2058 EPRS 1.75 0.000371 Catalyzes the aminoacylationMale sterility domain containing 2 84188 FAR1 1.74 0.005190 Controls ether

glycerophospholipid synthesisHuntingtin interacting protein 1 3092 HIP1 1.70 0.002967 Membrane-cytoskeletal integrityNMDA receptor regulated 1 80155 NAA15 1.68 0.003107 N-acetyltransferase, antiapoptosisUbiquitin-specific peptidase 10 9100 USP10 1.62 0.001657 Cleaves ubiquitin, a regulator

of p53, antitumorCoiled-coil domain containing 75 253635 CCDC75 1.62 0.000894 UnknownRAB31, member RASoncogene family

11031 RAB31 1.62 0.001205 Vesicle and granule targeting

Hepatitis B virus x associated protein 51773 RSF1 1.57 0.002448 Chromatin remodelingVacuolar protein sorting 35 (yeast) 55737 VPS35 1.54 0.000354 Subunit of the retromer, vesicle transport

NOTE: Probe set signals on the expression array that were <1.5-fold lower in BQ788 treated (n ¼ 3 patients) versus untreated GSCsamples (n¼ 3 patients, sample duplicate)with a pairwise t test (P < 0.05) were selected. Sampleswere permutated 100 times by dChipand 61 genes with median FDR ¼ 11% were obtained.

Liu et al.





damage the capabilities of self-renewal and proliferativedifferentiation in GSCs, inducible knockdown of EDNRBor EDN3 using viral-delivered hairpin RNA would allowmore in depth functional and in vivo studies.It has been shown that EDN3 first stimulates expression

of EDNRB, thenwhen under prolonged exposure to EDN3,EDNRB expression decreases (12). This may explain whyEDNRB are expressed at lower levels in the sphere cultures,which consistently produce EDN3. We found that EDN3alone is not a potent mitogen for GSCs and similar findingwas also reported (44). Interestingly, it was shown thatEDNRB antagonists reduced the viability and proliferationof glioma cells, which do not express EDN3 (48), implyingthe possibility of blocking EDN1/EDNRB signaling. Thedecreased glioma cell viability by EDNRB antagonists inde-pendent of their cognate receptor was also reported (49).Wespeculate that low levels of EDNRBmRNA being expressedin cultured CD133þ GSCs used for injection, but not insome tumor xenografts, could be due to tumor initiationfrom CD133þ GSCs being accompanied by activation/upregulation of EDNRA signaling and downregulation ofEDNRB signaling in our model system. Alternatively, it ispossible that EDNRB may be only expressed by a smallsubset of quiescent GSCs that express the tumor-suppressivephenotype (e.g., CD133þ/EDNRBþ cells) for maintainingGSCs pool in vivo, which can be retrieved and enriched bySGF culturing. On the other hand, CD133þ cells sortedfrom xenografts may mostly contain activated GSCs (e.g.,CD133þ/EDNRB�/EDNRAþ cells) that have entered thepathway to tumorigenesis. Interestingly, some tumor xeno-grafts express high levels of EDN3 and EDNRAmRNA, butlow levels of EDN1 and EDN2 mRNA, with a negligiblelevel of EDNRB mRNA, suggesting that the growth ofEDN3þ tumor may not be EDNRB, but EDNRA signalingdependent. Promoter hypermethylation of the EDNRB genein various human malignancies has been reported andsuggested that EDNRB may be a candidate tumor suppres-sor gene (50).Finally, our genome-wide expression analysis has provided

some molecular explanation for the requirement of EDN3/EDNRB signaling in maintaining GSC sphere cultures.Apparently, EDN3/EDNRB blockade mostly impacts cellstructure, cell movement, self-renewal/cell division, and cellsurvival, rendering GSCs nontumorigenic. We previouslyshowed that CD133þ cells, not CD133� cells, sorted from

same sphere cultures contain enriched tumorigenic cells, yetthey express tumor suppressor phenotype (8). Thus, trueGSCs may be quiescent before undergoing asymmetric celldivision to simultaneously self-renew and generate moredifferentiated progeny. It seems that differentiated progenyare active-growing cells that make up the most population intumor spheres and are likely the effector progeny which canundergo proliferative differentiation to produce more dif-ferentiated progeny that expresses hyperproliferative andhyperangiogenic phenotype, leading to tumor formation inanimals. Therefore, identifying essential genes (e.g., EDN3)which are shared by both GSCs and their immediatedifferentiated progeny/daughter cells would enhance itsclinical value because both populations can be targeted.Thus, treatment with BQ788 not only abolishes the self-renewal capacity of quiescent GSCs but also prevents dif-ferentiated progeny from populating tumor spheres andforming tumor in animals.Depletion of EDN3þ cells will likely prevent either

CD133þ or CD133� GSCs from regenerating a tumor.Futures studies should determine whether addition ofEDNRB antagonist can sensitize cells to radiochemotherapyin treatment of established glioblastoma tumors in animals.Our data support the view that the prevention of GSC-mediated tumor recurrence may need to focus on targetingactive stem cell pathways in GSCs (such as EDN3/EDNRBpathway), not proliferative pathways.Obviously, the cure forbrain cancer requires eliminating both GSC and non-GSCpopulations; therefore, it is important to evaluate the syn-ergistic benefits of incorporating GSC-targeted therapiesinto conventional cancer treatments.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Grant Support

This work was supported by grants from the American Cancer Society (RSG-07-109-01-CCE), National Cancer Institute (1 R21 CA140912-01), NIH(1DP2OD006444-01), and The Bradley Zankel Foundation to C-L Tso.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be herebymarked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

Received December 12, 2010; revised September 21, 2011; accepted October 2,2011; published OnlineFirst October 19, 2011.

References1. Maher EA, Furnari FB, Bachoo RM, Rowitch DH, Louis DN, Cavenee

WK, et al. Malignant glioma: genetics and biology of a grave matter.Genes Dev 2001;15:1311–33.

2. Beier D, Hau P, Proescholdt M, Lohmeier A, Wischhusen J, Oefner PJ,et al. CD133(þ) and CD133(-) glioblastoma-derived cancer stem cellsshowdifferential growth characteristics andmolecular profiles.CancerRes 2007;67:4010–5.

3. G€unther HS, Schmidt NO, Phillips HS, Kemming D, Kharbanda S,Soriano R, et al. Glioblastoma-derived stem cell-enriched culturesform distinct subgroups according to molecular and phenotypic cri-teria. Oncogene 2008;27:2897–909.

4. Hemmati HD, Nakano I, Lazareff JA, Masterman-Smith M, Gesch-wind DH, Bronner-Fraser M, et al. Cancerous stem cells can arisefrom pediatric brain tumors. Proc Natl Acad Sci USA 2003;100:15178–83.

5. Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, et al.Identification of human brain tumour initiating cells. Nature 2004;432:396–401.

6. Wang J, Sakariassen P�, Tsinkalovsky O, Immervoll H, Bøe SO,Svendsen A, et al. CD133 negative glioma cells form tumors in nuderats and give rise to CD133 positive cells. Int J Cancer 2008;122:761–8.






7. Galli R, Binda E, Orfanelli U, Cipelletti B, Gritti A, De Vitis S, et al.Isolation and characterization of tumorigenic, stem-like neuralprecursors from human glioblastoma. Cancer Res 2004;64:7011–21.

8. Liu Q, Nguyen DH, Dong Q, Shitaku P, Chung K, Liu OY, et al.Molecular properties of CD133þ glioblastoma stem cells derivedfrom treatment-refractory recurrent brain tumors. J Neurooncol2009;94:1–19.

9. Yuan X, Curtin J, Xiong Y, Liu G, Waschsmann-Hogiu S, Farkas DL,et al. Isolation of cancer stem cells from adult glioblastomamultiforme.Oncogene 2004;23:9392–400.

10. BaoS,WuQ,McLendonRE,HaoY,ShiQ,HjelmelandAB, et al. Gliomastem cells promote radioresistance by preferential activation of theDNA damage response. Nature 2006;444:756–60.

11. Murat A, Migliavacca E, Gorlia T, Lambiv WL, Shay T, Hamou MF,et al. Stem cell-related "self-renewal" signature and high epidermalgrowth factor receptor expression associated with resistance toconcomitant chemoradiotherapy in glioblastoma. J Clin Oncol2008;26:3015–24.

12. Lahav R, Dupin E, Lecoin L, Glavieux C, Champeval D, Ziller C, et al.Endothelin 3 selectively promotes survival and proliferation of neuralcrest-derived glial and melanocytic precursors in vitro. Proc Natl AcadSci USA 1998;95:14214–9.

13. Stone JG, Spirling LI, Richardson MK. The neural crest populationresponding to endothelin-3 in vitro includesmultipotent cells. JCell Sci1997;110:1673–82.

14. Rubanyi GM, Polokoff MA. Endothelins: molecular biology, biochem-istry, pharmacology, physiology, and pathophysiology. PharmacolRev 1994;46:325–415.

15. Shimada K,Matsushita Y,Wakabayashi K, TakahashiM,Matsubara A,Iijima Y, et al. Cloning and functional expression of human endothelin-converting enzyme cDNA. Biochem Biophys Res Commun 1995;207:807–12.

16. Bagnato A, Catt KJ. Endothelins as autocrine regulators of tumor cellgrowth. Trends Endocrinol Metab 1998;9:378–83.

17. LeiblMA, Ota T,WoodwardMN,Kenny SE, LloydDA, Vaillant CR, et al.Expression of endothelin 3 bymesenchymal cells of embryonicmousecaecum. Gut 1999;44:246–52.

18. Levin ER. Endothelins. N Engl J Med 1995;333:356–63.19. Edery P, Atti�e T, Amiel J, Pelet A, Eng C, Hofstra RM, et al. Mutation of

the endothelin-3 gene in the Waardenburg-Hirschsprung disease(Shah-Waardenburg syndrome). Nat Genet 1996;12:442–4.

20. McCallion AS,Chakravarti A. EDNRB/EDN3andHirschsprungdiseasetype II. Pigment Cell Res 2001;14:161–9.

21. BrandM, LeMoullec JM,Corvol P,Gasc JM.Ontogenyof endothelins-1 and -3, their receptors, and endothelin converting enzyme-1 in theearly human embryo. J Clin Invest 1998;101:549–59.

22. Nagy N, Goldstein AM. Endothelin-3 regulates neural crest cell pro-liferation and differentiation in the hindgut enteric nervous system. DevBiol 2006;293:203–17.

23. Egidy G, Eberl LP, Valdenaire O, Irmler M, Majdi R, Diserens AC, et al.The endothelin system in human glioblastoma. Lab Invest 2000;80:1681–9.

24. Tsutsumi K, Niwa M, Kitagawa N, Yamaga S, Anda T, Himeno A, et al.Enhanced expression of an endothelin ETA receptor in capillaries fromhuman glioblastoma: a quantitative receptor autoradiographic analy-sis using a radioluminographic imaging plate system. J Neurochem1994;63:2240–7.

25. Bhalla A, Haque S, Taylor I,WinsletM, LoizidouM. Endothelin receptorantagonism and cancer. Eur J Clin Invest 39 Suppl 2009;2:74–77.

26. SubramanianA, TamayoP,Mootha VK,Mukherjee S, Ebert BL,GilletteMA, et al. Gene set enrichment analysis: a knowledge-based approachfor interpreting genome-wide expression profiles. Proc Natl Acad SciU S A 2005;102:15545–50.

27. Lee J, KotliarovaS, Kotliarov Y, Li A, SuQ,DoninNM, et al. Tumor stemcells derived from glioblastomas cultured in bFGF and EGF moreclosely mirror the phenotype and genotype of primary tumors thando serum-cultured cell lines. Cancer Cell 2006;9:391–403.

28. Sun de J, Liu Y, Lu DC, Kim W, Lee JH, Maynard J, et al.Endothelin-3 growth factor levels decreased in cervical cancer

compared with normal cervical epithelial cells. Hum Pathol 2007;38:1047–56.

29. Wiesmann F, Veeck J, Galm O, Hartmann A, Esteller M, Kn€uchel R,et al. Frequent loss of endothelin-3 (EDN3) expression due to epige-netic inactivation in human breast cancer. Breast Cancer Res 2009;11:R34.

30. Kurtz A, Zimmer A, Schn€utgen F, Br€uning G, Spener F, M€uller T, et al.The expression pattern of a novel gene encoding brain-fatty acidbinding protein correlates with neuronal and glial cell development.Development 1994;120:2637–49.

31. UchidaN,BuckDW,HeD, ReitsmaMJ,MasekM,PhanTV, et al. Directisolation of human central nervous system stem cells. Proc Natl AcadSci USA 2000;97:14720–5.

32. Ellis P, FaganBM,MagnessST,HuttonS, TaranovaO,Hayashi S, et al.SOX2, a persistent marker for multipotential neural stem cells derivedfrom embryonic stem cells, the embryo or the adult. Dev Neurosci2004;26:148–65.

33. Yun K, Mantani A, Garel S, Rubenstein J, Israel MA. Id4 regulatesneural progenitor proliferation and differentiation in vivo. Development2004;131:5441–8.

34. ShenY,Mani S,Meiri KF. Failure to expressGAP-43 leads todisruptionof a multipotent precursor and inhibits astrocyte differentiation. MolCell Neurosci 2004;26:390–405.

35. Pelloski CE, Mahajan A, Maor M, Chang EL, Woo S, Gilbert M, et al.YKL-40 expression is associatedwith poorer response to radiation andshorter overall survival in glioblastoma. Clin Cancer Res 2005;11:3326–34.

36. Mitaka C, Hirata Y, Ichikawa K, Yokoyama K, Emori T, Kanno K, et al.Effects of TNF-alpha on hemodynamic changes and circulating endo-thelium-derived vasoactive factors in dogs. Am J Physiol 1994;267:H1530–6.

37. Thomson E, Kumarathasan P, Vincent R. Pulmonary expression ofpreproET-1 and preproET-3 mRNAs is altered reciprocally in ratsafter inhalation of air pollutants. Exp Biol Med (Maywood) 2006;231:979–84.

38. Schinelli S, Zanassi P, Paolillo M, Wang H, Feliciello A, Gallo V, et al.Stimulation of endothelin B receptors in astrocytes induces cAMPresponse element-binding protein phosphorylation and c-fos expres-sion viamultiple mitogen-activated protein kinase signaling pathways.J Neurosci 2001;21:8842–53.

39. ShinoharaH, Udagawa J,Morishita R,UedaH,Otani H, SembaR, et al.Gi2 signaling enhances proliferation of neural progenitor cells in thedeveloping brain. J Biol Chem 2004;279:41141–8.

40. Kalabis J, Li G, Fukunaga-KalabisM, Rustgi AK, HerlynM. Endothelin-3 stimulates survival of goblet cells in organotypic cultures of fetalhuman colonic epithelium. Am J Physiol Gastrointest Liver Physiol2008;295:G1182–9.

41. B€ohm M, Moellmann G, Cheng E, Alvarez-Franco M, Wagner S,Sassone-Corsi P, et al. Identification of p90RSK as the probableCREB-Ser133 kinase in human melanocytes. Cell Growth Differ 1995;6:291–302.

42. Sato-Jin K, Nishimura EK, Akasaka E, Huber W, Nakano H, Miller A,et al. Epistatic connections between microphthalmia-associated tran-scription factor and endothelin signaling in Waardenburg syndromeand other pigmentary disorders. FASEB J 2008;22:1155–68.

43. Bagnato A, Rosan�o L, Spinella F, Di Castro V, Tecce R, Natali PG, et al.Endothelin B receptor blockade inhibits dynamics of cell interactionsand communications in melanoma cell progression. Cancer Res2004;64:1436–43.

44. WuJJ, Chen JX, Rothman TP,GershonMD. Inhibition of in vitro entericneuronal development by endothelin-3: mediation by endothelin Breceptors. Development 1999;126:1161–73.

45. Gangemi RM, Griffero F, Marubbi D, Perera M, Capra MC, MalatestaP, et al. SOX2 silencing in glioblastoma tumor-initiating cells causesstop of proliferation and loss of tumorigenicity. Stem Cells 2009;27:40–8.

46. Fan X, Khaki L, Zhu TS, Soules ME, Talsma CE, Gul N, et al. NOTCHpathway blockade depletes CD133-positive glioblastoma cells andinhibits growth of tumor neurospheres and xenografts. Stem Cells2010;28:5–16.

Liu et al.





47. Bondurand N, Natarajan D, Barlow A, Thapar N, Pachnis V. Main-tenance of mammalian enteric nervous system progenitorsby SOX10 and endothelin 3 signalling. Development 2006;133:2075–86.

48. Paolillo M, Russo MA, Curti D, Lanni C, Schinelli S. Endothelin Breceptor antagonists block proliferation and induce apoptosis in gli-oma cells. Pharmacol Res 2010;61:306–15.

49. Montgomery JP, Patterson PH. Endothelin receptor B antagonistsdecrease glioma cell viability independently of their cognate receptor.BMC Cancer 2008;8:354–64.

50. Pao MM, Tsutsumi M, Liang G, Uzvolgyi E, Gonzales FA, Jones PA,et al. The endothelin receptor B (EDNRB) promoter displays hetero-geneous, site specific methylation patterns in normal and tumor cells.Hum Mol Genet 2001;10:903–10.






2011;9:1668-1685. Published OnlineFirst October 19, 2011.Mol Cancer Res Yue Liu, Fei Ye, Kazunari Yamada, et al. Cellular and Molecular Properties of Glioblastoma Stem CellsAutocrine Endothelin-3/Endothelin Receptor B Signaling Maintains

Updated version

10.1158/1541-7786.MCR-10-0563doi:

Access the most recent version of this article at:

Material

Supplementary

http://mcr.aacrjournals.org/content/suppl/2011/12/06/1541-7786.MCR-10-0563.DC1

Access the most recent supplemental material at:

Cited articles

http://mcr.aacrjournals.org/content/9/12/1668.full#ref-list-1

This article cites 50 articles, 19 of which you can access for free at:

Citing articles

http://mcr.aacrjournals.org/content/9/12/1668.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]

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

Permissions

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

.http://mcr.aacrjournals.org/content/9/12/1668To request permission to re-use all or part of this article, use this link



http://mcr.aacrjournals.org/lookup/doi/10.1158/1541-7786.MCR-10-0563

http://mcr.aacrjournals.org/content/suppl/2011/12/06/1541-7786.MCR-10-0563.DC1

http://mcr.aacrjournals.org/content/9/12/1668.full#ref-list-1

http://mcr.aacrjournals.org/content/9/12/1668.full#related-urls

http://mcr.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://mcr.aacrjournals.org/content/9/12/1668


autocrine endothelin-3/endothelin receptor b signaling ... · sirna (ambion), nonsilencing control...

Documents