Download - Brain-Derived Neurotrophic Factor Promotes …clincancerres.aacrjournals.org/content/clincanres/17/10/3123.full.pdf · Brain-Derived Neurotrophic Factor Promotes Tumorigenesis via

Human Cancer Biology

Brain-Derived Neurotrophic Factor Promotes Tumorigenesis via Inductionof Neovascularization: Implication in Hepatocellular Carcinoma

Chi-Tat Lam1, Zhen-Fan Yang1,2, Chi-Keung Lau1, Ka-Ho Tam1, Sheung-Tat Fan1,2, and Ronnie T.P. Poon1,2

AbstractPurpose: Brain-derived neurotrophic factor (BDNF) has emerged as a novel angiogenic factor, and yet

its impact on tumorigenesis is unclear. This study aimed at investigating the roles of BDNF in angiogenesis

and tumor development.

Experimental Design: BDNF was overexpressed in a mouse endothelial cell (EC) line by stable

transfection, and angiogenic properties of the transfectants were assessed. Microarray analysis was

employed to explore the molecular pathways. The impact of modulating BDNF levels in two mouse

EC lines on tumorigenic potential of a transformed mouse liver cell line was evaluated by an in vivo

cotransplantation model. BDNF and tropomyosin receptor kinase B (TrkB) protein levels were determined

in 50 pairs of human hepatocellular carcinoma (HCC) tissues by Western blotting and immunohisto-

chemistry. Survival analysis was carried out to determine their clinical significance.

Results: Overexpression of BDNF could promote EC proliferation, migration, invasion, and survival.

Microarray and molecular studies showed that RhoA, caspase-9, caspase-3, growth arrest specific 6, and

VEGF could mediate BDNF/TrkB-induced angiogenesis. The cotransplantation experiment showed that

high BDNF-expressing ECs could facilitate tumor angiogenesis and growth, whereas knockdown of BDNF

by short hairpin RNAs impaired such effects. Furthermore, examination on human HCC tissues revealed

upregulation of BDNF and TrkB protein levels in 46.0% and 33.3% of the cases studied, respectively.

Immunohistochemistry disclosed strong BDNF reactivity in both tumor and endothelial cells. High TrkB

expression was associated with shorter overall survival.

Conclusions: BDNF/TrkB systemwas crucial for tumor angiogenesis and growth, whichmay represent a

potential target for antiangiogenic therapy in HCC. Clin Cancer Res; 17(10); 3123–33. �2011 AACR.

Introduction

Antiangiogenic therapy is emerging as a therapeuticoption for cancer patients because of relatively low toxicity,low risk of drug resistance (1), and high target selectivity(2). Recent insights into the molecular mechanisms ofangiogenesis have led to identification of novel therapeutictargets, and brain-derived neurotrophic factor (BDNF)seems to be one of the candidates (3–5).BDNF belongs to a class of growth factors called neu-

rotrophins, which have well-documented functions inneural development and are linked to neoplasia (4, 6–12). The expression of BDNF and its receptor, tropomyo-

sin receptor kinase B (TrkB), has been reported in tumorcells of several human cancers (9, 13, 14). Intriguingly,overexpression of TrkB and BDNF is often associated withaggressive phenotype and poor prognosis of the disease(13), implying the oncogenic characteristics of BDNF/TrkB signaling cascades. Indeed, it has been shown thatBDNF/TrkB could promote cell proliferation and survival,and it induces metastasis by suppressing anoikis in var-ious cell types (8, 14, 15).

Several lines of evidence indicate that BDNF may play arole in angiogenesis. BDNF deficiency leads to an abnor-mal cardiac vessel system in knockout mice, whereasBDNF overexpression in mouse gestational heartsincreases capillary density (3). BDNF also promotesangiogenic behaviors of rat brain endothelial cells (EC)in vitro (16). Furthermore, BDNF is capable of recruitingTrkBþ ECs and bone marrow–derived hematopoieticprogenitor cells in an ischemic mouse model (17). Thesefindings support BDNF as a potential player in angiogen-esis. However, its detailed roles in angiogenesis andtumor development remain unclear and warrant furtherstudies.

It is known that microenvironment plays critical roles intumorigenesis (18). Tumor microenvironment comprisingextracellular matrix, ECs, and stromal cells interacts with

Authors' Affiliations: 1Department of Surgery and 2State Key Laboratoryfor Liver Research, The University of Hong Kong, Pokfulam, Hong Kong,China

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

Corresponding Author: Ronnie T. Poon, Department of Surgery, TheUniversity of Hong Kong, QueenMary Hospital, 102 Pokfulam Road, HongKong, China. Phone: 852-2255-3641; Fax: 852-2817-5475. E-mail:[email protected]

doi: 10.1158/1078-0432.CCR-10-2802

�2011 American Association for Cancer Research.

ClinicalCancer

Research

www.aacrjournals.org 3123

Research. on August 29, 2018. © 2011 American Association for Cancerclincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 18, 2011; DOI: 10.1158/1078-0432.CCR-10-2802

http://clincancerres.aacrjournals.org/

tumor cells by secreting soluble factors and by providingcell–cell and cell–matrix interactions (19, 20). Here, wehypothesize that overexpressing BDNF in ECs may pro-mote angiogenesis and provide a favorable microenviron-ment for tumor growth. To validate this hypothesis, weconducted in vitro and in vivo assays after modulating BDNFlevels in ECs. In addition, human hepatocellular carcinoma(HCC) tissues were examined for BDNF/TrkB expressionand its implication in hepatocarcinogenesis was evaluated.

Materials and Methods

Cell culture and reagentsTwo mouse EC lines, MILE SVEN 1 (MS1) and SVEC4-

10EE2 (SVEC4), and a transformed mouse liver cell line,BNL 1ME A.7R.1 (BNL), were purchased from AmericanType Culture Collection. Human umbilical vein endothe-lial cells (HUVEC) were obtained from Cascade Biologics.MS1 and BNL cells were maintained in Dulbecco’s mod-ified Eagle’s medium (DMEM) supplemented with 10%FBS, and 100 units/mL penicillin and 100 mg/mL strepto-mycin. SVEC4 cells were cultured in DMEM with 10%horse serum. HUVECs were grown in Medium 200 sup-plemented with low serum growth supplement (CascadeBiologics). All cells were grown in a humidified 5% CO2

atmosphere at 37�C.Recombinant human BDNF and Trk inhibitor, K252a,

were obtained from Calbiochem. Warfarin was purchasedfrom Sigma–Aldrich. Recombinant mouse Gas6 wasobtained from R&D Systems.

Patients and sample collectionTumor and corresponding nontumorous tissues were

collected from HCC patients undergone hepatectomy at

Queen Mary Hospital, Hong Kong, between 2003 and2008. Fifty cases were randomly recruited in the currentstudy. All patients had a diagnosis of primary HCC, andnone of them had received treatment before surgery.Pathologic diagnosis was based on the histologic exam-ination of tumor specimens by experienced pathologists.For the total of 50 patients, 34 had cirrhotic livers and 16had noncirrhotic livers (13 chronic hepatitis and 3noncirrhotic). Six normal livers (from liver donors; 2men and 4 women; aged 50–62) were included as con-trols. All tissues were obtained from consenting patientsand approved by the Institutional Review Board of theUniversity of Hong Kong. Tissues were immediatelysnap-frozen in liquid nitrogen after surgical resectionand stored at �80�C prior to analysis. Parallel sectionswere formalin-fixed and paraffin-embedded for histolo-gic and immunohistochemical studies. The clinical datafor the patients are summarized in SupplementaryTable S1.

Reverse transcription-PCR and quantitative PCRTotal RNAwas extracted from cells using RNeasyMini Kit

(QIAGEN) according to the user manual. TRIzol reagentwas used to extract total RNA frommouse tissues followingthe manufacturer’s instructions (Invitrogen). cDNA wassynthesized from total RNA by using ImProm-II ReverseTranscription System (Promega). The primer sequencesand PCR conditions are summarized in SupplementaryTable S2.

Quantitative PCR (qPCR) was carried out on a 7900HTFast Real-Time PCR system (Applied Biosystems), usingSYBRGreen PCR reagents. Relative quantificationwas doneusing DDCt method, normalizing to eukaryotic 18S rRNA.Dissociation curves were generated to evaluate PCR pro-duct specificity and purity.

Construct preparation and stable transfectionTotal RNA was purified from adult BALB/c mouse brain,

and cDNAwas synthesized as described above. After reversetranscription using oligo(dT) primers, cDNA template wasamplified by PCR using primers (Supplementary Table S2)designed to flank BDNF coding sequence (CDS; GenBankno. NM_001048139). PCR products were cloned intopGEM-T Easy Vector (Promega) and then subcloned intopcDNA3.1/Hygro(þ) vector (Invitrogen) at NotI site. Theconstructs were verified by restriction enzyme digestionand sequencing using T7 primer (50-GTAATACGACTCAC-TATAGGGC-30). BDNF construct [pcDNA3.1(þ)-BDNF]was transfected into MS1 cells by using FuGENE 6 Trans-fection Reagent (Roche) following the manufacturer’s pro-tocol. Clones were selected in medium furthersupplemented with 400 mg/mL Hygromycin B (Invitrogen)and verified by measuring BDNF at mRNA and proteinlevels, using reverse transcription-PCR (RT-PCR), Westernblotting, and ELISA, respectively.

Preparation of short hairpin RNA (shRNA) constructstargeting against BDNF was described in SupplementaryMaterials and Methods. Transfections of empty vector

Translational Relevance

Brain-derived neurotrophic factor (BDNF) is a poten-tial angiogenic factor. In this study, we showed thatoverexpression of BDNF confers angiogenic propertieson endothelial cells (EC). By establishing an in vivocotransplantation model in nude mice, we showed thathigh BDNF-expressing ECs could promote tumorgrowth, whereas its knockdown by shRNAs impairedthe tumor-promoting effect. These data suggested acritical role of BDNF/tropomyosin receptor kinase B(TrkB) system in tumorigenesis. Interestingly, expres-sion study by Western blotting by using 50 pairs ofhuman hepatocellular carcinoma (HCC) tissuesrevealed overexpression of BDNF and TrkB in tumortissues. By immunohistochemistry, aside from positivestaining in tumor cells, ECs also showed strong BDNFreactivity in HCC tissues, implicating a role of tumormicroenvironment in hepatocarcinogenesis. Patientswith TrkB overexpression in tumors had significantlyshorter overall survival, highlighting the clinical signif-icance of BDNF/TrkB pathway in HCC.

Lam et al.

Clin Cancer Res; 17(10) May 15, 2011 Clinical Cancer Research3124




(pGE-1), negative control vector (pGE-1-N; containing ascrambled shRNA sequence), and 2 shRNA constructs(pGE-1-shBDNF-1 and pGE-1-shBDNF-2) into SVEC4were carried out using the protocol described above.Clones were selected and maintained in culture mediumwith 400 mg/mL G418 (Invitrogen). The blocking efficacyof the shRNA clones was measured by Western blotting.

Western blot analysisWestern blot analysis was carried out as previously

reported (21). The following primary antibodies were used:BDNF (1:250, N-20; Santa Cruz Biotechnology), TrkB(1:200, 794; Santa Cruz Biotechnology), mouse TrkB(1:200, 47/TrkB; BD Biosciences), VEGF (1:250, catalogue#AF564; R&D Systems), caspase-3 (1:1,000, 8G10; CellSignaling Technology), caspase-9 (1:1,000, C9; Cell Signal-ing Technology), and b-actin (1:100,000, AC-15; Sigma-Aldrich). Respective secondary antibodies conjugated withperoxidase were from DAKO (1:5,000).

ELISASupernatant was collected from subconfluent culture of

each cell line after growing in serum containingmedium for2 to 3 days. Media were centrifuged to remove any detachedcells. BDNF concentration was quantified by using a BDNFEmax ImmunoAssay System ELISA kit (Promega) followingthemanufacturer’s instructions. The sampleswere loaded intriplicate. The absorbance at 450 nm was recorded with aBioRad microplate reader (model 680).

Migration and invasion assaysCells (1 � 105) were serum-starved for 6 hours before

seeding in 24-well migration Transwells (8.0-mmpore sizepolycarbonate filters; Corning) in 100 mL DMEM.DMEM/10% FBS was added to the bottom chamber ofeach well to serve as chemoattractants. After 12 hours ofincubation, the nonmotile cells at the top of the filterwere removed and the motile cells at the bottom of thefilter were fixed with ice-cold absolute methanol. Themigrating cells were visualized by standard hematoxylinand eosin staining and quantified by counting at least 6fields randomly under an inverted microscope (OlympusCK40; Olympus Optical) at 200� magnification. Imageswere acquired with a Nikon COOLPIX E950 digital cam-era (Nikon) connecting to the microscope. Invasion assaywas carried out by employing similar procedures asdescribed above, except using BD BioCoat Matrigel Inva-sion Chambers (BD Biosciences). The chambers weretransferred to the wells containing 0.75 mL DMEM/10% FBS, and cells were seeded at 1 � 105 cells/24-wellchamber in 0.5 mL DMEM/1% FBS. Invasion was scoredafter 24 hours. Each experiment was repeated for at least 3times.

Rho activation assayThe relative RhoA activity in MS1 and its BDNF trans-

fectants was measured using the Rho activation assayBiochem Kit (Cytoskeleton Inc.) according to the man-

ufacturer’s recommendations. Cells were cultured inDMEM/10% FBS for 48 hours before growing in mediumwith low serum content (1% FBS) for 24 hours. Cell lysateswere prepared and their equal amounts (600 mg) wereincubated with rhotekin-RBD beads. Bound RhoA proteinswere detected by Western blotting by using a RhoA-specificantibody.

Tumorigenicity assay: cotransplantation oftransformed liver cells and ECs

Male BALB/c nude mice, 4- to 8-week-old, were obtainedfrom Laboratory Animal Unit of the University of HongKong. One million BNL (transformed mouse liver) cellswere injected with 1 � 105 ECs (MS1, MS1/EV 1, MS1/BDNF C3, or C4 clone) subcutaneously into the right flankof nude mice. One million BNL cells were injected into theleft flank of the same animal for comparison. A minimumof 8 mice were injected for each group. Meanwhile, at least8 sites were injected with ECs (1� 105) alone to serve as anadditional control. The dimensions of developing tumorswere measured weekly with sliding calipers. Tumorvolumes were calculated in cubic millimeters using theformula: L � W � H. The kinetics of tumor formationwas compared among these coinjection combinations.Mice were sacrificed 11 weeks after injection, and tumorswere excised. One half of each tumor was frozen and theother half was fixed in formalin for subsequent histologicand immunohistochemical analyses. For the cotransplan-tation experiment using SVEC4/shRNA clones, thedescribed protocol was applied. An empty vector clone(SVEC4/EV 1) and a negative control vector clone(SVEC4/NEG 1) were included as controls.

ImmunohistochemistryImmunohistochemistry was conducted as previously

described (21), using either antihuman BDNF (1:15, cat-alogue #ab80124; Abcam) or anti-TrkB antibody (1:50, H-181; Santa Cruz Biotechnology).

Quantification of microvessel density (MVD) in mousetumor xenografts was done by using the Blood VesselStaining Kit (ECM590; Millipore) following the man-ufacturer’s protocol. The entire tumor section was scannedat 100� magnification to search for vascular hotspots (22,23). MVD of tumors was quantified by viewing at least 6hotspots of each section under the microscope at 200�magnification. Values were calculated from sections of atleast 3 different tumors.

StatisticsSPSS version 14.0 and GraphPad Prism 4.0 were used

for statistical analyses. One-way ANOVA, unpaired 2-tailed Student’s t test, the Mann–Whitney test, and thec2 test were used whenever applicable. The Kaplan–Meiermethod was applied to calculate overall survival rates,and the log-rank test was used to assess the significance ofthe differences. P value less than 0.05 was considered tobe significant and is denoted in the figures with anasterisk(s).

BDNF in Tumor Angiogenesis

www.aacrjournals.org Clin Cancer Res; 17(10) May 15, 2011 3125




Results

Overexpression of BDNF in ECs increased cell pro-liferation and protected them from apoptosis underserum starvation condition

To investigate the roles of BDNF in tumor angiogenesis,BDNF was overexpressed in MS1, a mouse EC line withundetectable BDNF level (Supplementary Fig. S1A). ThreeBDNF-overexpressing stable clones were obtained forfurther functional characterization (Fig. 1). The MTT assayshowed that all MS1/BDNF clones exhibited significantlyhigher proliferation rates than the parental and emptyvector controls (Supplementary Fig. S2A). Annexin V–binding assay showed that upregulation of BDNF couldrescue the cells from serum starvation–induced apoptosis(Supplementary Fig. S2B). The most prominent effect wasobserved in MS1/BDNF C4 clone displaying downregula-tion of active caspase-9 and caspase-3 (SupplementaryFig. S2C).

Upregulation of BDNF enhanced migration andinvasion abilities of ECs through inhibition of RhoAactivation

To assess the effect of overexpressing BDNF onMS1 cells,cell motility of the 3 BDNF clones were compared with theparental and empty vector controls by migration assay(Fig. 2A). The number of migrated cells of all clones wasincreased 10.6- to 66.7-fold (P < 0.001). Similarly, theirinvasive capacity was strikingly enhanced 14.9- to 52.7-fold(P < 0.05 or 0.001) when compared with controls.

Rho has been indicated to be a pivotal protein for theregulation of actin cytoskeleton during cell movement(24). To explore the underlying molecular mechanismsby which BDNF regulates migration and invasion pro-cesses, the activity of Rho in MS1/BDNF clones was mea-sured by Rho activation assay. Under low serum condition,levels of GTP-bound Rho in BDNF-overexpressing cloneswere suppressed when compared with the parental andempty vector controls (Fig. 2B). The results suggested thatBDNF could promote EC migration and invasion throughinhibition of RhoA activation.

We further extended the migration study to TrkB-expres-sing human ECs, HUVECs. Conditioned media from 2clones expressing the highest level of secretory BDNF(MS1/BDNF clone C3 and C4; Fig. 1B) were used aschemoattractants for HUVECs (Fig. 2C). A significantenhancement (more than 5-fold) of HUVEC migrationwas noted when compared with controls (P < 0.001;Fig. 2D), showing cross-species reactivity of BDNF. Nota-bly, this enhancement was completely abrogated by the Trkblocker K252a, indicating that BDNF increased EC migra-tion by acting through TrkB.

Enhanced BDNF expression facilitated tumor growth,whereas knockdown of BDNF abrogated tumor-promoting activity

To investigate the effect of elevated BDNF expression inMS1 cells on tumorigenesis, 2 clones showing the highestsecretory BDNF levels and migration capacities (MS1/BDNF C3 and C4) were coinjected with a chemicallytransformed mouse liver cell line, BNL. Parental MS1 cellsand empty vector clone were also coinjected with BNL cellsto serve as controls. We found that BDNF transfectantscould significantly promote tumor growth when comparedwith the controls [P < 0.05 or 0.01; Fig. 3A (i)]. At the endpoint, MVD in each tumor was quantified. Tumor xeno-grafts derived from coinjection of BNL with MS1/BDNFclone C3 or C4 showed a significant increase (6.8- to 9.8-fold) in microvessels compared with tumors from BNL orthose from coinjection withMS1 or empty vector clone [P <0.001; Fig. 3A (ii)].

On the other hand, SVEC4, a mouse tumor-derived ECline (25, 26) which expresses a high level of BDNF protein(Supplementary Fig. S1A), was employed for the cotrans-plantation experiment. Three SVEC4/shRNA clones show-ing themost effective suppressionof BDNF (SupplementaryFig. S4) were coinjected with BNL cells. SVEC4 controlvector clones (SVEC4/EV 1 and SVEC4/NEG 1) showed a

kDa25.9

14.8

Pro-BDNF

A

B

MS1/BDNF clone

MatureBDNF

Actin

300

Secretory BDNF levels

200

100

0

MS

1

BD

NF

(pg

/mL)

MS

1/E

V 1

MS

1/B

DN

F C

3

MS

1/B

DN

F C

4

MS

1/B

DN

F C

5

MS

1

MS

1/E

V 1

C1 C2 C3∗ ∗ ∗

C4 C5

Figure 1. Detection of BDNF expression in MS1 cell line and its BDNFtransfectants. A, increased cytosolic BDNF protein levels in indicatedclones by Western blotting. *, clones showed elevated BDNF levels. EV,empty vector. B, increased secretory BDNF levels in MS1/BDNF clones asdetected by ELISA. Columns, mean values of 2 experiments, each done induplicate; bars, SEM.

Lam et al.





Migration

A

B

C

D

MS1 MS1/EV 1 MS1/BDNF C3 MS1/BDNF C4 MS1/BDNF C5

MS

1/E

V 1

MS

1/E

V 1

MS1/BDNF cloneC3

Activated RhoA

Actin

HUVEC

HUVEC

Allow to migrate for 6 hTrkB

Transwell insert

Microporousmembrane

DMEM/0.5% FBS

MS1,MS1/EV 1,

MS1/BDNF C3,or

MS1/BDNF C4

HUVECConditioned

medium(+/– K252a)

Grow for 2 days

Total RhoA

C4 C5

MS

1

MS

1 MS

1/B

DN

F C

3

MS

1/B

DN

F C

4

MS

1/B

DN

F C

5

125

100

300400500600

200

100

7550

Mig

rate

d ce

ll/2

00x

field

Inva

ded

cell

/200

x fie

ld

100

0

Conditionedmedium source

300

200

Mig

rate

d H

UV

EC

s/2

00x

field

250

0

MS1+VEGF

MS1

(–)

(+)

K25

2a

MS1/EV 1 MS1/BDNF C3 MS1/BDNF C4

MS1

K252a – + – + – + – +

MS1/EV 1 MS1/BDNF C3 MS1/BDNF C4

Invasion

***

******

*

***

***

DMEM/0.5% FBS

#***

#***

1 ×105 cells

StarveO/N

Figure 2. BDNF enhanced migration/invasion ability of ECs, which was blocked by TrkB inhibitor K252a. A, MS1/BDNF clones showed significantenhancement of cell migration and invasion. Left, images of indicated clones after 12-hour migration/24-hour invasion (magnification � 200). As a positivecontrol, 10 ng/mL VEGF164 (from Sigma–Aldrich) was added to the bottom chamber of MS1 cells. Right, quantification of the assays (n ¼ 4). B, inhibition ofRhoA activation in MS1/BDNF clones. Cells were cultured under low serum condition (1% FBS) for 24 hours prior to the assay. Equal amounts (600 mg) of celllysate were incubated with rhotekin-RBD beads and bound RhoA proteins were detected by Western blotting by using a RhoA-specific antibody (top).Bottom, the amount of total RhoA and actin in cell lysates is shown. Blots are representative of 4 experiments. C, experimental outline of HUVEC migrationassay. TrkB-expressing HUVECs were starved overnight and seeded on the top chambers of 24-well migration Transwells in low serum medium(0.5% FBS). Conditioned media collected from 2-day culture of indicated cells were added to the bottom chambers in the presence or absence of200 nmol/L K252a (TrkB blocker). Cells were allowed to migrate for 6 hours. D, migration of HUVECs was induced by the conditioned media from the highBDNF-expressing clones (MS1/BDNF C3 and C4), but the increase was completely abrogated by K252a (left). Result summary of 3 independentexperiments (right). Values aremeans�SEM. *,P < 0.05; ***, P < 0.001 versus parental (MS1) cells and empty vector control (MS1/EV 1); #,P < 0.001 comparedbetween the groups with or without K252a treatment.






strong tumor-promoting ability upon coinjectionwith BNLcells [Fig. 3B (i)]. However, knockdown of BDNF in theshRNA clones could significantly abrogate such tumor-promoting effect (P < 0.05), accompanied by a remarkablereduction in the number of microvessels in tumor xeno-grafts [decreased 103- to 149-fold; P < 0.01 or 0.001; Fig. 3B(ii)].

High BDNF-expressing ECs induced TrkB and VEGFexpression in BNL cells after in vivo cotransplantationand in vitro coculture

To understand the molecular mechanism underlying theinteraction between BDNF-expressing ECs (MS1/BDNF C3and C4) and BNL, TrkB and VEGF protein levels in tumorxenografts from the cotransplantation experiments were

BNL+MS1/BDNF C42,000

1,500

Tum

or v

olum

e (m

m3 )

1,000

500

00 1 2 3 4 5 6

Weeks postinjection

BNL+SVEC4/EV 1

BNL+SVEC4/shBDNF 11BNL+MS1/BDNF C4BNL+MS1/BDNF C3BNL+MS1/EV 1 BNL+SVEC4/shBDNF 12 BNL+SVEC4/shBDNF 21

BNL+SVEC4/NEG 1BNLBNL−ve control BNL+MS1

7 8 9 10 11 12

BNL+MS1/EV 1MS1MS1/BDNF C3MS1/BDNF C4

BNL+MS1/BDNF C3BNL+MS1

BNL+SVEC4/NEG 1BNL+SVEC4/EV 1BNL+SVEC4/shBDNF 21

BNL+SVEC4/shBDNF 11BNL+SVEC4/shBDNF 12SVEC4/EV 1SVEC4/NEG 1

BNL

BNL 2,000

1,500

Tum

or v

olum

e (m

m3 )

1,000

500

00 1 2 3 4 5 6

Weeks postinjection7 8 9 10 11 12

*B

NL

BN

L+M

S1

BN

L+M

S1/

EV

1

BN

L+M

S1/

BD

NF

C3

BN

L+M

S1/

BD

NF

C4

10

20

30

Mic

rove

ssel

s/2

00x

field

0

***

***B

NL

BN

L+S

VE

C4/

EV

1

BN

L+S

VE

C4/

NE

G 1

BN

L+S

VE

C4/

shB

DN

F 1

1

BN

L+S

VE

C4/

shB

DN

F 1

2

BN

L+S

VE

C4/

shB

DN

F 2

1

10

5

15

20

Mic

rove

ssel

s/2

00x

field

0 *** ********

Ai

ii ii

iB

Figure 3. BDNF expression conferred ECs tumor-promoting ability. A, tumorigenicity study for MS1 and its derivative clones. A, i, high BDNF-expressing ECs(MS1/BDNFC3 and C4) promoted tumor growth of BNL cells in nudemice. Tumor volumes are reported asmeans� SEM (n¼ 8). *, P < 0.05; **, P < 0.01 versusparental cells and empty vector control. A, ii, analysis of MVD in tumor xenografts derived from indicated coinjection of BNL and MS1/BDNF clones.Representative images were shown (top; magnification � 200). �ve control, PBS substitution for the primary antibody; arrows, microvessels; bar, 50 mm.Bottom, quantitative analysis of vWF staining. ***, P < 0.001 versus controls. B, tumorigenicity study for SVEC4 and its shRNA clones. B, i, SVEC4 transfectingwith negative control vector (SVEC4/NEG 1) or empty vector (SVEC4/EV 1) promoted tumor growth after coinjecting with BNL cells, but knockdown of BDNF inSVEC4 cells by shRNAs could abrogate such tumor-promoting effect. *, P < 0.05 for BNL coinjecting with SVEC4/NEG 1 versus those coinjecting withshRNA clones. B, ii, MVD analysis for tumor xenografts derived from combined injection of BNL and indicated SVEC4 clones. Representative images(top; magnification � 200) and quantitative analysis are shown (bottom). **, P < 0.01; ***, P < 0.001 versus negative and empty vector controls. Countsfrom 6 fields per section and a minimum of 3 tumors from each group were included for statistical analysis.

Lam et al.





measured by Western blotting. Both levels were enhancedin tumors derived from coinjection of BNL with MS1/BDNF clones (Fig. 4A). An in vitro coculture study sup-ported the finding by showing upregulation of TrkB(increased by >150%, P < 0.05) and VEGF (elevated by>110%, P < 0.05) in BNL cells after coculture with these ECs[Fig. 4B (ii)]. The results indicated that high BDNF-expres-sing ECs could stimulate BNL cells by upregulation of TrkBand elicit an angiogenic response via inducing VEGFexpression.

Implication of growth arrest specific 6 (Gas6) gene inBDNF-elicited angiogenic responses

To explore the molecular pathways associated withBDNF-induced angiogenesis, microarray analysis(described in Supplementary Materials and Methods)was carried out to compare the gene expression profilesbetween MS1/BDNF clone C4 and the empty vector clone(MS1/EV 1). Around 1% (345 genes) of 34,000 genescovered by the Affymetrix Genechip exhibited more than2-fold change on overexpression of BDNF inMS1 cells. Thegenes were partially listed in Supplementary Table S3 andS4.

On the basis of the fold change magnitude and theirknown biological functions, 12 potential angiogenic geneswere selected. qPCR was carried out to validate microarraydata (Fig. 5A). A high consistency between the data setsfrom microarray and qPCR was observed (SupplementaryTable S5). Remarkably, we detected universal expressionpatterns of these genes in a panel of 5 BDNF transfectantsby qPCR (data not shown).

The microarray data implied that Gas6might have a rolein modulating BDNF-elicited angiogenic responses. Weinvestigated the effect of warfarin, a Gas6 inhibitor, oncell motility of the high BDNF-expressing ECs. Our resultshowed that warfarin could effectively suppress their moti-lity by 30.3-fold (P < 0.001; Fig. 5B), showing the pivotalrole of Gas6 in regulating BDNF-induced cell migration.Then, we examined the potential linkage between Gas6 andapoptosis. Treating the parental MS1 cells and empty vectortransfectant with recombinant Gas6 protein did not protectcells from serum starvation–induced apoptosis, and block-ing Gas6 by warfarin in the MS1/BDNF clone did notinduce apoptosis (Fig. 5C). The results indicated thatGas6 did not account for the antiapoptotic effect of BDNFon MS1 cells.

Clinical significance of TrkB expression in humanHCC

To investigate the potential implication of BDNF/TrkB inhepatocarcinogenesis, their protein expression levels inhuman HCC tissues were detected by Western blotting(Fig. 6A). BDNF was elevated in 46.0% (23 of 50) andTrkB was upregulated in 33.3% (15 of 45) of HCC tissueswhen compared with the adjacent nontumorous livertissues and normal liver controls. Immunohistochemicalstaining showed strong BDNF reactivity in both tumor andECs of HCC tissues, whereas TrkB expression was detectedonly in tumor cells (Fig. 6B). Association analysis of BDNFlevel with TrkB expression showed that concurrent upre-gulation of both proteins in HCC is infrequent [6.7% (3 of45); P < 0.05]. Kaplan–Meier survival analysis revealed thatpatients with tumors exhibiting high TrkB expression hadsignificantly shorter overall survival (P ¼ 0.0268; Fig. 6C).

Discussion

Accumulating evidence suggests that neurotrophinsincluding BDNF may serve as angiogenic factors (5, 27)

Transwell insert

DMEM/0.5% FBS

BNL

BNL

Starve O/N

MS1/EV 1,MS1/BDNF C3

orMS1/BDNF C4

Western blotting

Protein extraction

BNL

Tumor from coinjection of BNL with

BNL cocultured withEV 1 BDNF C3 BDNF C4

TrkB

* *4.0

3.0

Rel

ativ

e ex

pres

sion

2.0

1.0

0.0

**VEGF

2.5

2.0

Rel

ativ

e ex

pres

sion

BNL coculture withEV 1 BDNF C3 BDNF C4

1.5

1.0

0.0

0.5

EV 1

TrkB

A

Bi

ii

VEGF

Actin

TrkB

VEGF

Actin

BDNF C3 BDNF C4

Figure 4. Induction of TrkB and VEGF expression by high BDNF-expressing ECs in both in vivo and in vitro settings. A, protein levels ofTrkB and VEGF were elevated in 6 representative tumor xenograftsderived from coinjection of BNL with indicated MS1/BDNF clones whencompared with those derived from coinjection with empty vectorcontrol. B, i, schematic illustration of the in vitro coculture experiment.BNL (3 � 105) cells were seeded onto a 6-well plate and serum-starvedovernight. MS1/EV 1 or MS1/BDNF C3 or C4 (1.5 � 105) cells wereseeded on the top chambers of Transwells (0.4-mm pore size). Cellswere cocultured in low serum medium (0.5% FBS) for 8 to 24 hours.Expression levels of TrkB and VEGF in BNL cells were measured byWestern blotting. B, ii, TrkB and VEGF protein levels in BNL cells wereelevated after coculture with the high BDNF-expressing clones (MS1/BDNF C3 and C4). The blots are representative of 3 independentexperiments. Right, quantification of blots; *, P < 0.05 versus emptyvector control.






and may play a pathologic role in tumorigenesis (8). Theimplication of BDNF in tumor angiogenesis, however,remains obscure. This study attempts to identify the rolesof BDNF in angiogenesis and tumor development. Toexamine the effects of BDNF/TrkB system on EC functions,we generated stable BDNF-overexpressing clones in themouse EC line MS1. The clones displayed enhanced cellproliferation under low serum condition, indicating thatthe ECs became less growth factor dependent on upregula-tion of BDNF. About cell survival, we reported that BDNFupregulation could protect the ECs from serum starvation–induced apoptosis, which correlated with the decrease incaspase-9 and caspase-3 activation. The results suggestedthat BDNF exerted its antiapoptotic effect on MS1 cells byregulating the levels of the active caspases.

Next, we showed significant increases in cell motility andinvasiveness of MS1 cells on overexpression of BDNF.Provided that Rho regulates signaling pathways linked tocytoskeletal remodeling (24) and is a critical player in cellmigration and invasion (28, 29), its activity in the highBDNF-expressing clones was measured. Consistent withprevious reports, we found that activated RhoA levels wereconsistently reduced in all clones showing enhanced cellmotility. In fact, inhibition of Rho activation has beenreported to induce migration in several cell types, such asmesenchymal stromal cells (30), murine hematopoieticprogenitor cells (31), and mouse embryonic fibroblasts(32). Although the detailed signal transduction pathway isnot fully understood, it has been suggested that by regulat-ing cytoskeletal rearrangement, Rho inhibition can pro-mote migration (30). In addition, we showed that

conditioned media of high BDNF-expressing clones couldsubstantially increase migration of HUVECs and the effectwas abrogated by the TrkB blocker. Taken together, wesuggested that BDNF, by acting through TrkB, could pro-mote cell migration/invasion of MS1 ECs by inhibition ofRho activation.

We then provided functional evidence for the role ofBDNF in tumorigenesis by using a cotransplantationmodelin athymic nude mice. MS1/BDNF transfectants were coin-jected with BNL, the transformed mouse liver cell line thatshowed low tumorigenic potential over a 3-month period.We found that BDNF-overexpressing ECs could signifi-cantly enhance tumor growth of BNL cells on coinjection,accompanied by a prominent increase in the number ofmicrovessels in the corresponding tumors. On the otherhand, when BDNF expression in SVEC4, the tumor-derivedEC line, was stably knocked down by shRNAs, its tumor-promoting effect was abrogated. Interestingly, we observeda significant reduction in MVD in the tumor xenograftsobtained. The source of the ECs that form microvessels intumors, however, remains uncertain. The injected ECs werelikely to participate in vessel formation, though we cannotpreclude the possibility that vascular ECs were recruitedfrom the host animals. Yet, these findings provide compel-ling evidence that high expression of BDNF in ECs couldpromote tumor growth in vivo through enhancing neovas-cularization in tumors.

To further understand the molecular mechanismsthrough which BDNF regulates tumorigenesis, the expres-sion levels of TrkB and VEGF in BNL cells were determinedin both in vivo and in vitro settings. Their levels in BNL cells

100

A

C

B

806040201050

Fol

d ch

ange

Gas

6

CX

CL1

0

TF

DP

1

Cul

4A

VC

AM

-1

Fos

MC

AM

Lam

p1

JunB

ICA

M-1

DU

SP

9

Ndn

–5–10–15–20

–20

1,500***

*** ***

1,000

Mig

rate

d ce

lls/2

00x

field

War

farin

(–)

MS1 MS1/EV 1 MS1/BDNF C4

(+)

500

0

Warfarin

15

10

Apo

ptot

ic in

dex

(%)

5

0MS1

Gas6Warfarin

– + – +– +

MS1/EV 1 MS1/BDNF C4

MS1– + – + – +

MS1/EV 1 MS1/BDNF C4

Figure 5. Gas6 modulated migration process initiated by BDNF. A, validation of microarray data by qPCR. Fold changes of 12 potential angiogenicgenes in MS1/BDNF C4 clone by using MS1/EV 1 as a reference were shown. Detailed calculation formula is provided in Supplementary MaterialsandMethods. B, warfarin inhibitedmigration of theMS1/BDNF clone. ECswere seeded on the top chambers of Transwells in serum-freemedium, DMEM/10%FBS was added to the bottom chambers in the presence (þ) or absence (�) of 10 mmol/L of warfarin. Top, representative images after 24-hourmigration. Bottom, quantification of the assay. Values are means � SEM; n ¼ 3. ***, P < 0.001. C, Gas6 treatment of MS1 ECs did not reduce apoptosis,and blocking of Gas6 by warfarin in MS1/BDNF C4 clone did not induce apoptosis.

Lam et al.





were elevated after coinjecting with the high BDNF-expres-sing ECs in vivo or after coculture with these ECs in vitro.Collectively, these findings suggested that high BDNF-expressing ECs could stimulate BNL cells by upregulation

of TrkB and induced neovascularization via VEGF path-ways, which eventually facilitated tumor growth.

Verification of the microarray data by qPCR disclosedsimilar expression patterns of 12 angiogenic genes in a

Figure 6. Overexpression ofBDNF and TrkB in human HCCtissues. A, by Western blotting,BDNF protein level wasupregulated in 46.0% (23 of 50) ofHCC tissues (T) when comparedwith adjacent nontumorous livertissues (NT) and normal livertissues. TrkB level was elevated in33.3% (15 of 45) of HCC tissues.B, immunohistochemical stainingof BDNF and TrkB in human HCCtissues. Representative imagesfrom selected cases are shown.Top, both tumor (short arrows) andendothelial cells (long arrow)showed BDNF immunoreactivity.Bottom, TrkB was stainedpositively on both plasmamembrane (solid arrows) andcytoplasm (open arrows) of tumorcells. Magnification � 200; inset,�400; bar, 50 mm. C, the Kaplan–Meier overall survival plotaccording to TrkB expressionlevels. Patients with more than 2-fold increase in TrkB expression intumor compared withnontumorous tissue wereclassified into the high expressiongroup. High TrkB expression wassignificantly associated with poorsurvival of HCC patients (log-ranktest, P ¼ 0.0268).

HCC patient Liver donorHT35

BDNF

A

B

C

Actin

899 970 980 988 1057NT T NT T NT T NT T NT T NT T

1 2 3

HCC patient Liver donorHT59

TrkB

Actin

NT

BDNF

TrkB

1.2P = 0.0268

Low TrkB expression (n = 29)

High TrkB expression(n = 15)

1.0

.8

.6

.4

.2

–.2–10 0 10 20 30

Overall survival (mo)

Cum

ulat

ive

surv

ival

40 50 60 70

0.0

HT35

HT67

T

HT72 HT75 HT62 HT67NT T NT T NT T NT T NT T

1 2

145 kDa

95 kDa

3






panel of 5 MS1/BDNF clones. The finding reflected auniversal regulation of these genes by BDNF in the studiedclones and thus suggested their linkage with BDNF-induced angiogenesis. Among these genes, Gas6 exhibitedthe most dramatic upregulation in MS1/BDNF clones. It isa vitamin K–dependent growth factor (33), primarilyacting through Axl receptor (34) and is involved in angio-genesis and tumorigenesis (35). We linked Gas6 to BDNF-induced angiogenesis by showing that the migration-pro-moting effect of BDNF on the MS1/BDNF clone could beabolished by the Gas6 inhibitor. On the other hand, Gas6showed no role in regulating the antiapoptotic effect ofBDNF.

As a follow-up to previous studies on examining expres-sion of BDNF and TrkB in humanmalignancies (9, 13, 14),we conducted a detailed BDNF and TrkB expression studyin human HCC by Western blotting and immunohisto-chemistry. We showed upregulation of BDNF and TrkB inHCC tissues. Association analysis revealed that concurrentupregulation of both BDNF and TrkB is infrequent, imply-ing that elevation of either one of these proteins may besufficient to promote hepatocarcinogenesis. Interestingly,immunohistochemical analysis disclosed the presence ofhigh BDNF-expressing ECs in HCC, suggesting that tumormicroenvironment comprising ECs may represent animportant source of BDNF and contribute to tumorigen-esis. From survival analysis, we noted that patients withhigh TrkB levels had poor overall survival, unveiling theclinical significance of BDNF/TrkB pathway in HCC.

On the basis of current findings, we suggested 2 mechan-isms by which BDNF-expressing ECs could promote angio-

genesis and hence facilitate tumor growth. First, by actingvia TrkBs on ECs in an autocrine manner, BDNF inducedangiogenesis through regulation of themolecules includingRhoA, caspase-3, caspase-9, and Gas6. Second, BDNF fromECs could stimulate tumor cells by upregulationof TrkB andmay elicit angiogenic responses through VEGF pathways.

In conclusion, we showed that BDNF is a potent angio-genic factor that facilitates tumor growth via promotingangiogenesis. To our knowledge, we are the first group toreport the association of TrkB upregulation in HCC withoverall survival, suggesting a critical role of BDNF/TrkBpathway in HCC. We also showed overexpression of BDNFby ECs in HCC, implicating a role for the tumor micro-environment in hepatocarcinogenesis. Future studies arewarranted to evaluate the BDNF/TrkB pathway as a poten-tial target for antiangiogenic therapy in HCC.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Grant Support

This study was supported by Small Project Funding of the University ofHong Kong and by the Collaborative Research Fund (HKU5/CRF/08) of theResearch Grant Council Hong Kong.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received October 19, 2010; revised February 17, 2011; accepted March10, 2011; published OnlineFirst March 18, 2011.

References1. Folkman J. Endogenous angiogenesis inhibitors. APMIS 2004;

112:496–507.2. Neri D, Bicknell R. Tumour vascular targeting. Nat Rev Cancer

2005;5:436–46.3. Donovan MJ, Lin MI, Wiegn P, Ringstedt T, Kraemer R, Hahn R, et al.

Brain derived neurotrophic factor is an endothelial cell survival factorrequired for intramyocardial vessel stabilization. Development2000;127:4531–40.

4. Hu Y, Wang YD, Guo T, Wei WN, Sun CY, Zhang L, et al. Identificationof brain-derived neurotrophic factor as a novel angiogenic protein inmultiple myeloma. Cancer Genet Cytogenet 2007;178:1–10.

5. Kermani P, Hempstead B. Brain-derived neurotrophic factor: a newlydescribed mediator of angiogenesis. Trends Cardiovasc Med 2007;17:140–3.

6. Adriaenssens E, Vanhecke E, Saule P, Mougel A, Page A, Romon R,et al. Nerve growth factor is a potential therapeutic target in breastcancer. Cancer Res 2008;68:346–51.

7. Knezevich SR, McFadden DE, Tao W, Lim JF, Sorensen PH. A novelETV6-NTRK3 gene fusion in congenital fibrosarcoma. Nat Genet1998;18:184–7.

8. Kruttgen A, Schneider I, Weis J. The dark side of the NGF family:neurotrophins in neoplasias. Brain Pathol 2006;16:304–10.

9. Pearse RN, Swendeman SL, Li Y, Rafii D, Hempstead BL. A neuro-trophin axis in myeloma: TrkB and BDNF promote tumor-cell survival.Blood 2005;105:4429–36.

10. Ricci A, Greco S, Mariotta S, Felici L, Bronzetti E, Cavazzana A, et al.Neurotrophins and neurotrophin receptors in human lung cancer. AmJ Respir Cell Mol Biol 2001;25:439–46.

11. Sakamoto Y, Kitajima Y, Edakuni G, Sasatomi E, Mori M, Kitahara K,et al. Expression of Trk tyrosine kinase receptor is a biologicmarker forcell proliferation and perineural invasion of human pancreatic ductaladenocarcinoma. Oncol Rep 2001;8:477–84.

12. Walch ET, Albino AP, Marchetti D. Correlation of overexpression of thelow-affinity p75 neurotrophin receptor with augmented invasion andheparanase production in human malignant melanoma cells. Int JCancer 1999;82:112–20.

13. Rubin JB, Segal RA. Growth, survival and migration: the Trk to cancer.Cancer Treat Res 2003;115:1–18.

14. Yang ZF, Ho DW, Lam CT, Luk JM, Lum CT, Yu WC, et al. Identifica-tion of brain-derived neurotrophic factor as a novel functional proteinin hepatocellular carcinoma. Cancer Res 2005;65:219–25.

15. Douma S, Van Laar T, Zevenhoven J, Meuwissen R, Van Garderen E,Peeper DS. Suppression of anoikis and induction of metastasis by theneurotrophic receptor TrkB. Nature 2004;430:1034–9.

16. Kim H, Li Q, Hempstead BL, Madri JA. Paracrine and autocrinefunctions of brain-derived neurotrophic factor (BDNF) and nervegrowth factor (NGF) in brain-derived endothelial cells. J Biol Chem2004;279:33538–46.

17. Kermani P, Rafii D, Jin DK, Whitlock P, Schaffer W, Chiang A, et al.Neurotrophins promote revascularization by local recruitment ofTrkBþ endothelial cells and systemic mobilization of hematopoieticprogenitors. J Clin Invest 2005;115:653–63.

18. Tuxhorn JA, McAlhany SJ, Dang TD, Ayala GE, Rowley DR. Stromalcells promote angiogenesis and growth of human prostate tumors in adifferential reactive stroma (DRS) xenograft model. Cancer Res2002;62:3298–307.

Lam et al.





19. Khodarev NN, Yu J, Labay E, Darga T, Brown CK, Mauceri HJ, et al.Tumour-endothelium interactions in co-culture: coordinated changesof gene expression profiles and phenotypic properties of endothelialcells. J Cell Sci 2003;116:1013–22.

20. Shchors K, Evan G. Tumor angiogenesis: cause or consequence ofcancer? Cancer Res 2007;67:7059–61.

21. Luk JM, Lam CT, Siu AF, Lam BY, Ng IO, Hu MY, et al. Proteomicprofiling of hepatocellular carcinoma in Chinese cohort revealsheat-shock proteins (Hsp27, Hsp70, GRP78) up-regulation andtheir associated prognostic values. Proteomics 2006;6:1049–57.

22. Hasan J, Byers R, Jayson GC. Intra-tumoural microvessel density inhuman solid tumours. Br J Cancer 2002;86:1566–77.

23. Vermeulen PB, Libura M, Libura J, O’Neill PJ, van Dam P, Van MarckE, et al. Influence of investigator experience and microscopic field sizeon microvessel density in node-negative breast carcinoma. BreastCancer Res Treat 1997;42:165–72.

24. Etienne-Manneville S, Hall A. Rho GTPases in cell biology. Nature2002;420:629–35.

25. O’Connell K, Landman G, Farmer E, Edidin M. Endothelial cellstransformed by SV40 T antigen cause Kaposi's sarcoma like tumorsin nude mice. Am J Pathol 1991;139:743–9.

26. Walter-Yohrling J, Morgenbesser S, Rouleau C, Bagley R, CallahanM,Weber W, et al. Murine endothelial cell lines as models of tumorendothelial cells. Clin Cancer Res 2004;10:2179–89.

27. Nico B, Mangieri D, Benagiano V, Crivellato E, Ribatti D. Nerve growthfactor as an angiogenic factor. Microvasc Res 2008;75:135–41.

28. Lamalice L, Le Boeuf F, Huot J. Endothelial cell migration duringangiogenesis. Circ Res 2007;100:782–94.

29. Price LS, Collard JG. Regulation of the cytoskeleton by Rho-familyGTPases: implications for tumour cell invasion. Semin Cancer Biol2001;11:167–73.

30. Jaganathan BG, Ruester B, Dressel L, Stein S, GrezM, Seifried E, et al.Rho inhibition induces migration of mesenchymal stromal cells. StemCells 2007;25:1966–74.

31. Gottig S, Mobest D, Ruster B, Grace B, Winter S, Seifried E, et al. Roleof the monomeric GTPase Rho in hematopoietic progenitor cellmigration and transplantation. Eur J Immunol 2006;36:180–9.

32. Besson A, Gurian-West M, Schmidt A, Hall A, Roberts JM. p27Kip1modulates cell migration through the regulation of RhoA activation.Genes Dev 2004;18:862–76.

33. Yanagita M. Gas6, warfarin, and kidney diseases. Clin Exp Nephrol2004;8:304–9.

34. Nagata K, Ohashi K, Nakano T, Arita H, Zong C, Hanafusa H, et al.Identification of the product of growth arrest-specific gene 6 as acommon ligand for Axl, Sky, and Mer receptor tyrosine kinases. J BiolChem 1996;271:30022–7.

35. Holland SJ, Powell MJ, Franci C, Chan EW, Friera AM , Atchison RE,et al. Multiple roles for the receptor tyrosine kinase Axl in tumorformation. Cancer Res 2005;65:9294–303.






2011;17:3123-3133. Published OnlineFirst March 18, 2011.Clin Cancer Res Chi-Tat Lam, Zhen-Fan Yang, Chi-Keung Lau, et al. Hepatocellular Carcinomavia Induction of Neovascularization: Implication in Brain-Derived Neurotrophic Factor Promotes Tumorigenesis

Updated version

10.1158/1078-0432.CCR-10-2802doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2011/03/21/1078-0432.CCR-10-2802.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/17/10/3123.full#ref-list-1

This article cites 35 articles, 13 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/17/10/3123.full#related-urls

This article has been cited by 1 HighWire-hosted articles. Access the articles at:

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

SubscriptionsReprints 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/17/10/3123To 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-10-2802

http://clincancerres.aacrjournals.org/content/suppl/2011/03/21/1078-0432.CCR-10-2802.DC1

http://clincancerres.aacrjournals.org/content/17/10/3123.full#ref-list-1

http://clincancerres.aacrjournals.org/content/17/10/3123.full#related-urls

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

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/17/10/3123


Download - Brain-Derived Neurotrophic Factor Promotes …clincancerres.aacrjournals.org/content/clincanres/17/10/3123.full.pdf · Brain-Derived Neurotrophic Factor Promotes Tumorigenesis via

Top Related