Endoglin Regulates Cancer Stromal Cell Interactions in ... · Microenvironment and Immunology...

Microenvironment and Immunology

Endoglin Regulates Cancer–Stromal Cell Interactionsin Prostate Tumors

Diana Romero1,4, Christine O’Neill1, Aleksandra Terzic1, Liangru Contois1, Kira Young1,2,Barbara A. Conley1, Raymond C. Bergan3, Peter C. Brooks1,2, and Calvin P.H. Vary1,2

AbstractEndoglin is an accessory receptor for TGF-b that has been implicated in prostate cancer cell detachment,

migration, and invasiveness. However, the pathophysiologic significance of endoglin with respect to prostatetumorigenesis has yet to be fully established. In this study, we addressed this question by investigation ofendoglin-dependent prostate cancer progression in a TRAMP (transgenic adenocarcinoma mouse prostate)mouse model where endoglin was genetically deleted. In this model, endoglin was haploinsufficient such that itsallelic deletion slightly increased the frequency of tumorigenesis, yet produced smaller, less vascularized, andless metastatic tumors than TRAMP control tumors. Most strikingly, TRAMP:engþ/�-derived tumors lacked thepronounced infiltration of carcinoma-associated fibroblasts (CAF) that characterize TRAMP prostate tumors.Studies in human primary prostate-derived stromal cells (PrSC) confirmed that suppressing endoglin expressiondecreased cell proliferation, the ability to recruit endothelial cells, and the ability to migrate in response to tumorcell–conditioned medium. We found increased levels of secreted insulin-like growth factor–binding proteins(IGFBP) in the conditioned medium from endoglin-deficient PrSCs and that endoglin-dependent regulation ofIGFBP-4 secretion was crucial for stromal cell–conditioned media to stimulate prostate tumor cell growth.Together, our results firmly establish the pathophysiologic involvement of endoglin in prostate cancer progres-sion; furthermore, they show how endoglin acts to support the viability of tumor-infiltrating CAFs in the tumormicroenvironment to promote neovascularization and growth. Cancer Res; 71(10); 3482–93. �2011 AACR.

Introduction

Prostate cancer is the second leading cause of male cancerdeath in the United States, mainly because of metastaticdisease (1). Endoglin expression is altered in prostate cancer(2) and high endoglin levels are associated with decreasedsurvival in patients with tumor Gleason scores between 6 and7 (3). We have shown that endoglin, a TGF-b coreceptor, isinvolved in prostate cancer cell migration and invasion.Importantly, endoglin expression is lost in human metastaticprostate cancer cells (4). When restored, endoglin inhibits cellmigration in vitro via modulation of both Smad-dependentand Smad-independent signaling mechanisms (5, 6). Further-

more, endoglin expression in human prostate cancer cellsrepresses their tumorigenicity in SCID (severe combinedimmunodeficient) immunosuppressed mice (6) and metasta-sis in an orthotopic mouse model of prostate cancer (7). Thesestudies, however, did not address the mechanisms underlyingendoglin function with regard to stromal cell support of tumorvascularization and growth.

Solid tumors are a heterogeneous population of malignantand nonmalignant cell types. The latter include inflammatorycells, stem cells, fibroblasts, and endothelial cells (8). Thesecell populations constitute the tumor stroma, which provideskey regulatory determinants for tumor progression andmetastasis (9). We have previously described the effects ofendoglin expression in prostate tumor cells in vitro (4–6) andin vivo (6, 7). However, the in vivo role of endoglin expression inother tumor cell types is unknown. To address this question,we developed a genetic model of prostate cancer that com-bined endoglin haploinsufficiency (engþ/�; ref. 10) with theTRAMP (transgenic adenocarcinoma mouse prostate) mouse,a well-characterized transgenic model for the study of pros-tate cancer (11). TRAMP mice express the SV40 large Tantigen, under the control of the prostate epithelium-specificprobasin promoter, and develop prostate cancer from hyper-plasia through more aggressive and metastatic stages (11, 12).In this model, the resulting level of endoglin in all engþ/�

mouse-derived tissues is deficient as compared with engþ/þ

tissues (10). Our results show that endoglin is essential for the

Authors' Affiliations: 1Center for Molecular Medicine, Maine MedicalCenter Research Institute, Scarborough, Maine; 2The Graduate Schoolof Biomedical Sciences, University of Maine, Orono, Maine; 3Divisionof Hematology/Oncology, Department of Medicine, Northwestern Uni-versity Medical School, Chicago, Illinois; and 4The Institute of Repro-ductive and Developmental Biology, Imperial College London, London,United Kingdom

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

Corresponding Author: Calvin P.H. Vary, Center for Molecular Medicine,Maine Medical Center Research Institute, 81 Research Drive, Scarbor-ough, ME 04074. Phone: 207-396-8148; Fax: 207-396-8179; E-mail:[email protected]

doi: 10.1158/0008-5472.CAN-10-2665

�2011 American Association for Cancer Research.

CancerResearch

Cancer Res; 71(10) May 15, 20113482

Research. on May 31, 2020. © 2011 American Association for Cancercancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 28, 2011; DOI: 10.1158/0008-5472.CAN-10-2665





http://cancerres.aacrjournals.org/



presence of carcinoma-associated fibroblasts (CAF) in pros-tate tumors. Furthermore, data suggest that the prostatetumor CAFs impaired by endoglin deficiency in the TRAMPmodel are myogenic in origin and that endoglin suppressionimpairs CAF-mediated endothelial cell recruitment and CAFmigratory response to tumor-derived factors. Finally, datasupport the hypothesis that endoglin downregulation in can-cer affects CAF insulin-like growth factor–binding protein-4(IGFBP-4) expression, supporting a novel mechanism of can-cer–stromal cell cross-talk mediated through endoglin.

Materials and Methods

Mouse strainsEndoglin-targeted mice were screened for the presence of a

neocassette in the truncated, engineered, endoglin allele, aspreviously described (10). TRAMP mice (The Jackson Labora-tory) were screened for the presence of the SV40 large Tantigen, as described on The Jackson Laboratory Web site(research.jax.org). Both TRAMP and endoglin heterozygousmice were maintained in the C57BL/6 mouse strain back-ground. Mice were bred, maintained, and experimentationwas conducted according to the NIH standards established inthe Guidelines for the Care and Use of Experimental Animals.

Necropsy and analysis of mouse tissuesMice were weighed and euthanized at 21 or 25 weeks of age.

All mice were genotyped twice: after birth and followingsacrifice. The internal organs were examined and dissectedin accordance with established guidelines (13), andmetastaseswere determined as previously described (14). Harvestedtumors and prostates were fixed in 4% paraformaldehydefor 48 hours and embedded in paraffin. Hematoxylin and eosin(H&E), Masson's trichrome, and platelet/endothelial cell adhe-sion molecule 1 (PECAM-1) staining were done as described(15). Antibodies used for immunohistochemistry included anti-endoglin antibodyMJ7/18 (Developmental Studies HybridomaBank, The University of Iowa, Iowa City, IA); anti-stromal-derived factor 1 (SDF-1), anti-smooth muscle actin (aSMA),anti-IGF-I, and anti-IGF-IR (Abcam); anti-Ki67 (DAKO); andanti-IGFBP-4 (R&D Systems). Terminal deoxynucleotidyltransferase-mediated dUTP nick-end labeling (TUNEL) stain-ing was done with the In Situ Cell Death Detection Kit (Roche)according to manufacturer's instructions.For immunofluorescence analysis, anti-FSP-1 (S100A4 Ab-8

from NeoMarkers; 1:50 dilution), anti-SM22a (Abcam; 1:200dilution), and anti-IGFBP-4 (R&D Systems; 1:50 dilution) wereused as previously described (16, 17).The slides were examined with a Carl Zeiss Axioskop

microscope. Imaging was carried out using the Scion Imagesoftware program, and processed with Adobe Photoshopsoftware as previously described (18).

Protein analysisThe tumors were ground and homogenized in lysis buffer

(150mmol/L NaCl, 300 mmol/L sucrose, 1% Triton X-100, 0.5%sodium deoxycholate, 50 mmol/L Tris–HCl, pH 7.5) contain-ing a cocktail of protease (Roche) and phosphatase (Calbio-

chem-EMD) inhibitors. Immunoprecipitation and Westernblot analysis were carried out with anti-endoglin (BectonDickinson Transduction Laboratories), and anti-b-actin(Sigma) as previously described (16, 19).

Cell culture, gene silencing, and growth factortreatment

Human primary prostate stromal cells (PrSC; Clonetics)were grown in stromal cell growthmedium (SCGM; Clonetics).PrSCs were used between passages 5 to 10. PC3-M-C and PC3-M-FL cells were grown by the procedure described by Romeroand coworkers (6). Human primary umbilical vein endothelialcells (HUVEC, passages 3–6) were cultured as previouslydescribed (19). TRAMP-C2 cells were obtained from theAmerican Type Culture Collection (Manassas, VA), and main-tained as described in the Supplementary Data and by Carterand coworkers (20).

siRNA for human endoglin interference was cloned inpSilencer 5.1 (Ambion). A pSilencer control vector (nonspe-cific) was purchased from the same company. The cells weretransfected using Effectene (QIAGEN). RNA isolation andreverse transcriptase (RT)-PCR for endoglin and GAPDH weredone as previously described (6). Alternatively, constructsexpressing 21-nucleotide endoglin-specific short hairpin RNAs(shRNA) targeting human endoglin [shENG(1–3)] or nontar-geting control (shSC; Sigma; SHC002) were obtained fromSigma–Aldrich. Constructs were packaged into lentiviruspseudotyped with the vesicular stomatitis virus glycoprotein(VSV-G). Transduction was conducted by incubating PrSCswith lentivirus, and stably transduced cells were subsequentlyused for studies without drug-marker selection (see Supple-mentary Data and Supplementary Table S1). All cell lines wereverified by morphology, mouse and human endoglin-specificPCR, certified mycoplasma-negative by PCR (Lonza),and primary cell cultures used within the indicated passagenumbers. Human recombinant IGF-I, IGFBP-4, and IGFBP-6proteins, and the neutralizing anti-IGFBP-4 were obtainedfrom R&D Systems.

Cell migrationMigration assays were conducted as described (21). Briefly,

5� 105 cells (HUVECs or PrSCs) were suspended in migrationbuffer [stromal cell basal medium (SCBM) containing 1mmol/L MgCl2, 0.2 mmol/L MnCl2, and 0.5% BSA], plated in theupper chamber of Transwell migration chambers (8.0 mm;CoStar), and allowed to invade through a polycarbonatemembrane toward conditioned medium for 4 to 8 hours at37�C. Cells remaining on the upper surface of the membranewere removed and cells that had migrated to the undersidewere stained with crystal violet. Cell migration was quantifiedin at least 3 independent experiments using triplicates, eitherby counting or by extraction of crystal violet and quantifyingabsorbance at 600 nm.

Analysis of conditioned mediaA total of 1.2� 106 PrSCs were plated onto culture plates 10

centimeters in diameter. Forty-eight hours later, they wererinsed 3 times in SCBM (Clonetics), and 5 mL per plate of fresh

Endoglin Regulates Tumor–Stromal Cell Cross-Talk

www.aacrjournals.org Cancer Res; 71(10) May 15, 2011 3483




SCBM was added. Forty-eight hours later, the conditionedmedia were filtered (0.2-mm-pore-size mesh), concentrated,and stored at �20�C until further analysis.

For isotope-coded affinity tag (ICAT) tandemmass spectro-metry (MS/MS), the conditioned media were concentrated byultracentrifugation, labeled, and purified using the CleavableICAT Reagent Kit for Protein Labeling (Applied Biosystems),and analyzed with a tandem quadrupole time-of-flight massspectrometer (QSTAR; MDS-SCIEX) as described by Kolevaand coworkers (19). Analysis of mass spectrometric data wasconducted using ProteinPilot software program (Life Tech-nologies). Detailed description of methods is provided inSupplementary Data.

Results

TRAMP:engþ/� mice have a greater number of tumors,which are smaller and less metastatic than those inTRAMP:engþ/þ mice

To generate TRAMP:engþ/þ and TRAMP:engþ/� transgenicmice, we crossed endoglin heterozygous (engþ/�) males (10)with TRAMP females (12). We analyzed tumor formation inTRAMP:engþ/þ and TRAMP:engþ/� 21-week-old (n ¼ 12), and25-week-old males (n ¼ 10), obtaining similar results.

Western blot analysis indicated that TRAMP:engþ/� tumorsshowed lower levels of endoglin than TRAMP:engþ/þ tumors,although heterogeneity was observed as expected (ref. 10;Fig. 1A). Quantitative analysis indicated that endoglin protein

expression in TRAMP:engþ/� tumors was approximately one-third of the levels detected in TRAMP:engþ/þ tumors (Fig. 1B).The stromal cells and most of the cancer cells within TRAMP:engþ/þ-derived tumors expressed endoglin, which was sig-nificantly reduced in TRAMP:engþ/�-derived tumors (Fig. 1C).Image analysis (Fig. 1D) suggested that this reduction wasconsistent [30%–40% of wild type (WT)] with the data shownin Figure 1B. Normal prostate tissue sections exhibited onlydiffuse background staining using anti-endoglin antibody.However, the stromal cells within TRAMP:engþ/þ-derivedtumors expressed endoglin, with significantly reduced endo-glin expression in TRAMP:engþ/�-derived tumors (Fig. 1C).

The frequency of prostate tumorigenesis was slightly higherin TRAMP:engþ/� mice than TRAMP:engþ/þ mice (Fig. 2A).Two-thirds of the TRAMP:engþ/�-derived tumors were non-metastatic, whereas all of the TRAMP:engþ/þ-derived tumorswere metastatic (Fig. 2A). Metastases were observed in lungand lymph nodes with similar frequencies in TRAMP:engþ/þ

and TRAMP:engþ/� mice: 50% of the metastases occurred inlocal lymph nodes and 50% in lungs.

Tumors in TRAMP:engþ/� mice were smaller than those inTRAMP:engþ/þ mice (Fig. 2B). Quantification of the percen-tage of cells positive for the proliferation marker Ki67 andTUNEL staining indicated that proliferation and apoptoticrates were similar in the tumor cells of TRAMP:engþ/þ andTRAMP:engþ/� mice (data not shown), suggesting that thetumormicroenvironment promotedmore sustained growth ofTRAMP:engþ/þ-derived tumors over time.

ENG

A

C D

B

ENG

Normal prostate TRAMP tumor

WtLung TRAMP:eng+/+

eng+/+

eng+/–

TRAMP:eng+/–

TRAMP:eng+/–n = 3

TRAMP:eng+/+n = 3

TRAMP:eng+/–n = 5

TRAMP:eng+/+n = 5

1.4

P < 0.05

P < 0.025

1.2

1.0

0.8

Ave

rage

den

sity

Ave

rage

pix

el d

ensi

ty

0.6

0.4

0.2

0.0

350.0300.0250.0200.0150.0100.0

50.00.0

β-Actin Figure 1. Endoglin expression isreduced in TRAMP:engþ/þ andTRAMP:engþ/�-derived tumors.A, Western blot for endoglin andb-actin (control) in tumors derivedfrom TRAMP:engþ/þ and TRAMP:engþ/� 21-week-old mice (n ¼ 3).B, quantitation of Western blots byimage densitometry using ScionImage Analysis Software (averagediffuse optical density � SD).C, immunohistochemistry forendoglin in tumors derived fromnormal prostate and tumors fromTRAMP:engþ/þ and TRAMP:engþ/� 21-week-old mice. Theslides were counterstained withhematoxylin. Bars, 300 mm.D, quantitation of TRAMP tumorendoglin staining using ScionImage Analysis Software (averagepixel density � SD; ref. 16).

Romero et al.

Cancer Res; 71(10) May 15, 2011 Cancer Research3484




TRAMP:engþ/þ tumors are more vascularized thanTRAMP:engþ/� tumorsEndoglin is a marker of tumor neoangiogenesis [reviewed in

(22)]. To investigate differences in tumor vascularization, theendothelial cell marker PECAM-1 and endoglin (Fig. 2C) wereused toquantify themicrovascular density (Fig. 2D). Thenumberof PECAM-1-positive vessels was 5-fold higher in TRAMP:engþ/þ-derived tumors versus TRAMP:engþ/�-derived tumors,whereas endoglin positive vessels were 25% to 30% higher inTRAMP:engþ/þ-derived tumors versus TRAMP:engþ/�-derivedtumors, suggesting that TRAMP:engþ/þ-derived tumors benefitfrom higher amounts of metabolites and oxygen.

Endoglin is associated with CAF investment of TRAMP:engþ/þ-derived tumorsH&E and Masson's trichrome staining revealed that

TRAMP:engþ/þ and TRAMP:engþ/�-derived tumors were

poorly differentiated adenocarcinomas, with a predominantsolid mass of epithelial-derived cells and very rare glandformation, as defined in (13). Furthermore, we observed thatTRAMP:engþ/þ-derived tumors contained areas enriched infibroblast-like cells. In contrast, all the TRAMP:engþ/�-derivedtumors analyzed were nonfibrotic, indicating the absence ofstromal fibroblasts (Fig. 3A and B). Image analysis confirmedthat the average area occupied by epithelial-like cells wasapproximately 75% in TRAMP:engþ/þ-derived tumors versus99% in TRAMP:engþ/�-derived tumors.

CAFs are a major and heterogeneous constituent of thetumor stroma (23). CAFs are characterized by the expressionof aSMA and SDF-1 (8), which were both detected in theTRAMP:engþ/þ CAFs but not in TRAMP:engþ/� CAFs(Fig. 3C).

One of the cellular components of CAFs is the SM22a-positive myofibroblast (24), which plays an important role in

Figure 2. Prostate tumorigenesisand tumor angiogenesis arealtered in TRAMP:engþ/þ versusTRAMP:engþ/� mice. A,frequency of prostatetumorigenesis and metastasis inTRAMP:engþ/þ and TRAMP:engþ/� 21-week-old mice (n¼ 12).B, tumor size in TRAMP:engþ/þ

(n ¼ 4) and TRAMP:engþ/� (n ¼ 5)21-week-old mice (average weight� SD). C, immunohistochemistryfor PECAM-1 and endoglin(arrows) in TRAMP:engþ/þ andTRAMP:engþ/� tumors from21-week-old mice, counterstainedwith hematoxylin. Bars, 300 mm.D, the number of microvesselsstained for PECAM-1 andendoglin, determined in at least 8fields per sample (n ¼ 4). *,P < 0.05 (Student's t test).

TRAMPeng +/+

n = 12

A

C

D

B

TRAMP:eng +/+

n = 4

TRAMP:eng +/+

TRAMP:eng +/–

Tota

l pos

itive

ves

sels

/fiel

d

PECAM-1 ENG

TRAMPeng +/+

n = 4

TRAMPeng +/–

n = 12

TRAMP:eng +/–

n = 4

*

TRAMPeng +/–

n = 5

P = 0.12

Metastatic

Nonmetastatic

PECAM-1+

ENG+

10020

15

10

5

0

4567

123

0

80

Tum

or fr

eque

ncy

(%)

Pro

stat

e / b

ody

wei

ght (

%)

60

40

20

0






tumor behavior (25). SM22a was restricted to the TRAMP:engþ/þ stromal fibroblast; however, it was largely absent fromTRAMP:engþ/�-derived tumors (Fig. 3D). Immunofluores-cence staining for fibroblast-specific protein 1 (FSP-1) wasmore pronounced in TRAMP:engþ/þ-derived tumors confirm-ing the identity of prostate-associated fibroblasts (17). How-ever, double immunofluorescence analysis using anti-endoglinand anti-SM22a or anti-FSP-1 antibodies revealed that endo-glin expression was associated with SM22a-positive cells butnot FSP-1-positive cells (Fig. 4A and B). These results indicatethat TRAMP:engþ/þ-derived tumors are largely comprised ofendoglin-expressing myofibroblast-derived CAFs.

Endoglin expression is necessary for the viability ofcultured prostate stromal cells

We attempted to establish primary cultures of CAFs derivedfrom TRAMP:engþ/þ and TRAMP:engþ/�-derived tumors.However, although we were able to propagate TRAMP:engþ/þ-derived CAFs in culture, the TRAMP:engþ/�-derivedCAFs were not viable under a variety of culture conditions(data not shown). To overcome this limitation, we usedhuman PrSCs. Consistent with TRAMP immunohistochemis-

try, human PrSCs robustly expressed endoglin, as detectedby RT-PCR (Fig. 5A, left panel). Endoglin expression wastransiently knocked down in PrSCs with a specific interfer-ing RNA construct, siENG. The efficiency of endoglin RNAsilencing was approximately 60%, as detected by both RT-PCR and immunoprecipitation (Fig. 5A, right panel). Thisreduction of endoglin protein level approximated the dif-ference seen in tumors (Fig. 1A) and was sufficient tosignificantly impair PrSC growth in vitro (Fig. 5B, left panel),suggesting that endoglin expression promotes CAF prolif-eration in prostate tumor.

Because growth factor secretion is a recognized CAF func-tion (23), we analyzed the effect of the conditioned mediumfrom PrSCs in their proliferation. PrSC growth was stimulatedwhen they were cultivated in their own conditioned medium.Moreover, PrSC-conditioned medium partially rescued theinhibitory effect of endoglin knockdown in PrSCs. The con-ditioned medium from endoglin knockdown in PrSCs failed tostimulate PrSC growth or to rescue the inhibitory effect ofdecreased endoglin levels (Fig. 5B, middle panel). These resultssuggest that endoglin influences prostate stromal cell viabilityvia secretion of soluble factors.

TRAMP:eng +/+

TRAMP:eng +/+

TRAMP:eng +/+

n = 4

NonfibroticFibrotic

TRAMP:eng +/–

n = 4

H&E

SMA SDF-1

SM22:DAPI FSP-1:DAPI

A

C

D

BTrichrome

Tum

or c

lass

ifica

tion

(%)

10080604020

0TRAMP:eng +/–

TRAMP:eng +/–

TRAMP:eng +/+

TRAMP:eng +/–

Figure 3. TRAMP:engþ/� tumorslack carcinoma-associatedfibroblasts. A, H&E and Masson'strichrome staining in TRAMP:engþ/þ and TRAMP:engþ/�-derived tumors revealedfibroblast-enriched areas inTRAMP:engþ/þ-derived tumors(arrow). Bars: 300 mm. B,frequency of fibrotic andnonfibrotic tumors in TRAMP:engþ/þ and TRAMP:engþ/� 21-week-old mice. At least 3 sectionsper tumor were analyzed (n ¼ 4).C, immunohistochemistry forstromal markers SMA and SDF-1in TRAMP:engþ/þ and TRAMP:engþ/� tumors counterstainedwith hematoxylin. Bars, 300 mm.D, immunofluorescence forSM22a and FSP-1 in TRAMP:engþ/þ and TRAMP:engþ/�-derived tumors. The nucleiwere stained with DAPI. Bars,SM22a, 300 mm; FSP-1, 200 mm.

Romero et al.





Stromal fibroblasts stimulate the proliferation ofprostate cancer cells through an endoglin-dependentmechanismCAFs contribute to tumor development, in part, because

they stimulate tumor cell proliferation (8). To further inves-tigate the link between endoglin expression in PrSCs andprostate cancer cell proliferation, we used PC3-M cells that

did not express endoglin (4, 6; PC3-M-C, control) or that stablyoverexpressed endoglin (PC3-M-FL, full-length; ref. 6). PC3-M-C and PC3-M-FL cells were grown in the presence of basalmedium, or in the presence of conditioned medium fromPrSCs transfected with an interfering RNA against endoglin ornontargeting control. Control PrSC-conditioned mediumstrongly stimulated the proliferation of both PC3-M-C and

Figure 4. Endoglin is associatedwith tumor myofibroblasts. A,double immunofluorescence forendoglin, SM22a, and FSP-1 inTRAMP:engþ/þ tumors. The nucleiwere stained with DAPI. Arrows,endoglin (ENG) and SM22adouble-positive cells. Bars,200 mm. B, the number of ENG:SM22a and ENG:FSP-1 double-positive cells were counted in atleast 5 fields per sample (average� SD; ref. 18).

ENG+:SM22αα+

18

ENG ENG:FSP-1:DAPIFSP-1

ENG

TRAMP:eng +/+A

B

ENG:SM22:DAPISM22

161412

Cel

ls p

er 1

00-m

m2 v

isu

al f

ield

1086420

P < 0.002

ENG+:FSP-1+






PC3-M-FL cells. The conditioned medium from endoglinknockdown in PrSCs had a lower stimulatory effect in PC3-M-C cells, and no effect in PC3-M-FL cells (Fig. 5B, right panel).Considered together, these results are consistent with the viewthat endoglin expression in stromal cells is necessary tostimulate cancer cell proliferation via a mechanism thatinvolves secreted factors.

Endoglin deficiency in PrSCs impairs endothelial cellmigration and tumor cell recruitment

To further suppress endoglin expression in PrSCs, 3 sepa-rate shRNA constructs were delivered using a lentivirus (26).PrSC shENG(1–3) shRNAs resulted in either partial or com-plete suppression of endoglin protein levels, respectively(Fig. 5C, inset). Conditioned medium collected from shENG1,

ENG ENG

IgG

HC

Mw, kDa

100

75

50

PrSC

PrSC

A

B

C D

siSC

shSC shENG(2)

siENGcond.media

cond.media

Basal media ++–

– ––– +

– ++–

– ––– +

–PrSC

1.5*

*

*1.0

Cel

l pro

lifer

atio

n (r

)

0.5

0.0

2.5

**** **

****

2.01.51.00.50.0C

ell p

rolif

erat

ion

(r )

PrSC siSC

siSC PrSCsiENG+

+–

– ––– +

– ++–

– ––– +

–PC3-M-C PC3-M-FL

PrSC siENG

Basal mediaPrSC siSCPrSC siENG

PrSC

PrSC TRAMP-C2

PrSC

ShSC

ShSC

ShENG(1

)

ShENG(1

)

ShENG(2

)

ShENG(2

)

ShENG(3

)

ShENG(3

)Cell

s + C

MCell

s + B

M

Mw, kDa

Endoglin

β-Actin

1007550

*

** ***

**

*

HuVEC

1.2

1.0

0.8

0.6

Cel

l pro

lifer

atio

n (r

)

0.40.20.0

0.3

706050403020100

0.25

Ave

rage

opt

ical

den

sity

, 600

nm

Ave

rage

cel

l num

ber0.2

0.1

0.15

0.05

0

PC3-M PrSC

GAPDH

siS

C

siS

C

C FLsi

EN

G

siE

NG

-RT

Figure 5. Endoglin knockdown reduces PrSC proliferation and affects PrSC-dependent modulation of PC3-M cell proliferation. PrSCs were transfected withsiRNAs directed against endoglin (siENG) or a control scrambled sequence (siSC) for 48 hours. Endoglin expression was analyzed by RT-PCR andimmunoprecipitation (A). PC3-M-C and endoglin-expressing PC3-M-FL cells (6): negative and positive controls, respectively. HC, immunoglobulin heavychain. B, left, PrSC proliferation following siENG transfection: Two independent experiments using triplicates were done. r, ratio of siENG- versus siSC-treatedcells. Middle, PrSC siENG- or siSC siRNA–derived conditioned SCBM was used to treat new cultures of PrSCs. r, number of cells divided by the number ofsiSC cells in basal media. Right, PC3-M-C or PC3-M-FL cells were prepared in SCBM or PrSC-conditioned medium. The number of cells per well wasdetermined 48 hours later as described above. r, number of cells divided by the number of cells in basal media. C, PrSCs were transduced with shRNAconstructs targeting human endoglin (left, inset). Endoglin Western blot of PrSCs. Left, HUVECs were tested for ability to migrate toward basal PrSC shSC orshENG(1–3) medium (25 mg protein). CM and BM, conditioned and basal media, respectively. D, TRAMP-C2 cells were used to prepare conditionedmedium asdescribed above. Following shRNA transduction, PrSCs were used for migration assays as above. * P < 0.05 and ** P < 0.005 (Student's t test). SeeSupplementary Data for detailed methods.

Romero et al.





shENG2, and shENG3 all reduced the ability of HUVECmigration, reflecting the degree of endoglin suppression.Furthermore, TRAMP-C2-conditioned medium was testedfor its ability to recruit PrSCs. Endoglin-deficient PrSCs weresignificantly impaired in their capacity to migrate in responseto tumor cell-conditioned medium (Fig. 5D). TRAMP-C2endoglin knockdown did not affect cell recruitment (datanot shown), suggesting that endoglin is required for CAF-dependent recruitment of endothelial cells and their responseto tumor cell factors.

Endoglin-dependent modulation of IGFBP-4 secretionby PrSCs is involved in the regulation of tumor cellgrowthTo identify peptides secreted by PrSCs, we carried out

ICAT-MS/MS (27) to compare the conditioned media fromcontrol and endoglin knockdown PrSCs. Among the proteinsoverexpressed by endoglin knockdown in PrSCs were (i) tissueinhibitors of metalloproteinases 1 and 2 (TIMP1, TIMP2); (ii)sulfhydryl oxidase 1; (iii) SPARC, and (iv) 2 members from theIGFBP family—IGFBP-4 and IGFBP-6 (Table 1). These pro-teins are implicated in the induction of cell growth arrest andin cell invasiveness (15, 28–31).Mass spectrometric sequencing of the putative IGFBP-4 and

IGFBP-6 peptides confirmed their identities and corroborated

the quantitative data indicating their upregulation in endo-glin-deficient PrSCs (Supplementary Figs. S1–S7). IGFBPs playimportant roles in neoplastic processes and prostate cancer(32, 33), and TGF-b signaling regulates tumor–stromal inter-actions via IGF-I (34). Therefore, we quantified the cell growthof PC3-M-C cells in response to recombinant IGF-I, IGFBP-4,and IGFBP-6 treatment. IGF-I and IGFBP-6 stimulated PC3-M-C proliferation (Fig. 6A). IGFBP-4 alone did not affect cellproliferation; however, in combination with IGF-I, it inhibitedIGF-I–dependent stimulation of cell proliferation (Fig. 6A).When these treatments were carried out in PrSC-conditionedmedium, the growth-stimulation effect of IGF-I and IGFBP-6was enhanced, and, surprisingly, IGFBP-4 alone inhibited cellproliferation. These effects were likely due to the presence ofPrSC-derived IGF-I in the medium (35). It is reasonable topostulate that IGFBP-4 inhibits PC3-M proliferation throughan IGF-dependent mechanism because PC3 cells express IGFsignaling components (36). A similar response was detected inPC3-M-FL cells (data not shown). The use of a blockingantibody for IGFBP-4 partially prevented its inhibition ofPC3-M-C cell proliferation when the treatment was carriedout in control PrSC-conditioned medium (Fig. 6B). Whenadded in the presence of endoglin knockdown PrSC-condi-tioned medium, the neutralizing antibody had the samepartial blocking effect on IGFBP-4–dependent inhibition of

Table 1. Summary of proteins identified and quantified by ICAT-MS/MS in PrSC-conditioned media:siENG:siSC PrSCs

Accession no. Name H vs. La Nb Function

P16035 Tissue inhibitor of metalloproteinases2 (TIMP-2)

3.026 18 ECM degradation

P09382 Galectin-1 1.947 6 Cell–ECM interactionP00338 L-Lactate dehydrogenase A chain 1.794 6 MetabolismP07355 Annexin A2 1.750 8 Signal transductionP01033 Tissue inhibitor of metalloproteinases

1 (TIMP-1)1.633 1 ECM degradation

P14618 Pyruvate kinase isozymes M1/M2 1.615 21 MetabolismO00391 Sulfhydryl oxidase 1 1.559 5 Induced in quiescent fibroblastsO76061 Stanniocalcin-2 1.554 11 Calcium homeostasisP27797 Calreticulin 1.448 14 Calcium homeostasisP22692 Insulin-like growth factor-binding protein 4c 1.439 3 IGFBP familyP12109 Collagen alpha-1(VI) chain 1.347 12 ECMP24592 Insulin-like growth factor-binding protein 6c 1.357 4 IGFBP familyP60174 Triosephosphate isomerase 1.289 16 MetabolismQ99497 DJ-1 1.247 18 ChaperoneP23142 Fibulin-1 1.220 7 ECMP08123 Collagen alpha-2(I) chain 1.190 8 ECMP29400 Collagen alpha-5(IV) chain 1.184 14 ECMP09486 SPARC 1.126 24 Inhibition of cancer cell proliferationQ12841 Follistatin-related protein 1 0.849 27 Actin binding

Abbreviation: ECM, extracellular matrix.aHeavy isotope (siENG)-tagged peptide versus light isotope (siSC)-tagged peptide, average of confirmed samples.bNumber of peptides identified and quantified by more than 95% CI.cSee Supplementary Data.






PC3-M-C cell growth (Fig. 6B). This experimental approachconfirmed the presence of functional IGFBP-4 in endoglinknockdown PrSC-derived medium, which is consistent withthe reduced size of TRAMP:engþ/�-derived tumors.

TRAMP:eng-derived tumor sections were stained for theseIGF signaling components. In TRAMP:engþ/þ-derived tumors,IGFBP-4 was detected in both fibroblast-like and epithelial-derived cancer cells. The epithelial staining appeared to be

peripheral, suggesting that most of the IGFBP-4 detected wasassociated with the stromal compartment (Fig. 6C, arrow).TRAMP:engþ/�-derived tumors showed minimal staining forIGFBP-4 (Fig. 6C) due to the lack of CAFs. IGF-I and IGF-IRwere detected mainly in the nonstromal compartment(Fig. 6C).

Immunofluorescence analysis of TRAMP:engþ/þ-derivedtumor showed more myofibroblast incursion (SM22a-positive

6.0Untreated

Untreated

Basal media PrSC siSC cond. media PrSC siENG cond. media

Untreated Untreated

+ IGF-1

+ IGFBP-4

+ IGFBP-6

+ IGF-I+ IGFBP-4

+ IGFBP-4 + IGFBP-4+ α-IGFBP-4

+ α-IGFBP-4

IGF-IRIGF-IIGFBP-4

IGFBP-4 SM22:IGFBP-4:DAPISM22

TRAMP:eng +/+

TRAMP:eng +/–

TRAMP:eng +/+

TRAMP:eng +/–

5.0

4.0

3.0

2.0

1.0Cel

l pro

lifer

atio

n (r

)C

ell p

rolif

erat

ion

(r )

0.0Basal media PrSC cond. media

2.5 *

A

B

C

D

**

* *

**

** * *

****2.0

1.5

1.0

0.5

0.0

Figure 6. IGF-I signaling andPrSC-dependent modulation ofPC3-M cell proliferation. A, PC3-M-C cells were prepared in SCBMor PrSC-conditioned medium,with or without 50 ng/mL IGF-I, 50ng/mL IGFBP-4, and 50 ng/mLIGFBP-6. A total of 19,000 cellsper well were plated in 24-wellplates. Forty-eight hours later, thenumber of cells per well wasdetermined. Two independentexperiments using triplicates wereconducted. r, number of cellsdivided by the number ofuntreated cells in basal media.*, P < 0.05; **, P < 0.005 (Student'st test). Asterisk-tagged barstatistics are referenced to lane 1.B, PC3-M-C cells were trypsinizedand resuspended in SCBM orPrSC-conditioned medium, withor without 50 ng/mL IGFBP-4, and100 ng/mL anti-IGFBP-4antibody. Proliferation assay wasconducted as described for A. C,immunohistochemistry for IGFBP-4, IGF-I, and IGF-IR in tumorsderived from TRAMP:engþ/þ andTRAMP:engþ/� 21-week-oldmice, counterstained withhematoxylin. Bars, 300 mm. D,immunofluorescence for SM22aand IGFBP-4 in tumors derivedfrom TRAMP:engþ/þ and TRAMP:engþ/� 21-week-oldmice. Arrows,IGFBP-4 staining. Bars, 300 mm.

Romero et al.





cells), and less IGFBP-4 staining, which predominantlycolocalized with SM22a staining. In contrast, TRAMP:engþ/�-derived tumor showed less SM22a-positive areas butmore prominent IGFBP-4 staining that was localized in theextracellular space adjacent to SM22a-positive cells (Fig. 6D).Thus, the expression pattern of IGFBP-4 in these tumors isconsistent with endoglin-dependent modulation of IGFBP-4availability and it affects stromal investment in prostate tumors.

Discussion

The role of endoglin in tumorigenesis in vivo has beenprincipally studied using tumor cell xenografts. Such studiesindicate that endoglin expression represses migration andinvasiveness of prostate cancer cells (4, 5), and that it attenu-ates their tumorigenicity (6). However, more accurate animalmodels are needed to elucidate the behavior of particulartumor types in their microenvironment. The present work isthe first to study the effect of endoglin haploinsufficiency in anautologous model of cancer. This bigenic model is based onthe TRAMP mouse, which develops in situ and invasivecarcinoma of the prostate (11) and, ultimately, late-stagemetastatic cancer (37).Endoglin expression inhibits prostate cancer cell migration

in vitro (4, 5), but, surprisingly, the frequency of metastasis inour in vivo model was higher in TRAMP:engþ/þ mice than inTRAMP:engþ/� mice. The increased vascularization ofTRAMP:engþ/þ tumors is likely the reason for this difference,as the intravasation of tumor cells into the blood stream is thefirst step in the establishment of distant-site metastaticlesions (9).Histologic and immunohistochemical examination of

TRAMP:engþ/þ versus TRAMP:engþ/� tumors showed thatendoglin was required for the presence of CAFs in the tumor.This phenotype is much more profound than expected fromendothelial cell haploinsufficiency (50% reduction in endoglinlevel) or the asymptomatic reduction of endoglin systemically.Interestingly, studies of the effect of endoglin haploinsuffi-ciency on xenografted Lewis lung carcinoma 3LL cell–derivedtumors showed no such CAF phenotype (38). Moreover,endoglin expression in endothelial cells of engþ/þ versusengþ/� mice cause relatively small effects (compared withthe tumor CAF phenotype) in the context of skin carcinogen-esis (39). These observations suggest that the endoglin-depen-dent CAF phenotype is specific to the prostate tumor stroma.The origin of CAFs is unclear. Candidate CAF precursors

include activated quiescent local fibroblasts (8) and circulat-ing bone marrow mesenchymal stem cells (40). Moreover,recent work suggests the intriguing possibility that CAFsoccur as a consequence of endothelial cells undergoingendothelial–mesenchymal transition (41). Our studies suggestthat endoglin is required for continuous tumor CAF invest-ment. Furthermore, CAFs are compared with myofibroblasts,defined as activated fibroblasts involved in processes such aswound healing (23). Endoglin is a marker of myofibroblasts(42) and its expression is increased in these cell type duringatherosclerosis-related and vascular TGF-b–dependent myo-genic differentiation (43) and cell migration (44). The current

data suggest that endoglin is primarily associated with myofi-broblast-related SM22a-positive fibroblasts. Based on ourprevious studies (45), we propose that endoglin expressionis required for the viability or the lineage specification of themyofibroblast-related CAF precursors.

To study the role of endoglin in CAF function, we isolatedCAFs from TRAMP:engþ/þ and TRAMP:engþ/� tumors. How-ever, we were not able to establish cell cultures of TRAMP:engþ/�-derived CAFs. Human PrSCs were utilized as an alter-native. Two studies showed that co-injection of PrSCs withprostate cancer cells in mice enhances tumor incidence andgrowth (35, 46). We showed that endoglin is expressed inPrSCs and found that PrSC growth is impaired in conditions ofreduced endoglin expression. In addition, reduction of endo-glin expression in human prostate stromal cells reduced theirability to recruit endothelial cells and their capacity to migratein response to tumor-secreted factors. These results suggestthat endoglin is required for multiple aspects of CAF functionincluding viability, endothelial cell recruitment, and tumor-induced migration.

CAFs recruit several cell types to the tumor area via growthfactor secretion (8, 23). Therefore, decreased tumor angiogen-esis in TRAMP:engþ/� mice may be directly related to theabsence of CAFs needed to recruit endothelial cell precursors.However, the signals that CAFs use to communicate withadjacent tissue are poorly understood.

Quantitative isotope peptide tagging methods suggestedthat endoglin regulated PrSC secretion of several potentiallyimportant secreted proteins involved in cell recruitment. Forexample, endoglin knockdown resulted in increased TIMP1and TIMP2 detected in PrSC-conditioned medium (Table 1).Previous studies implicate tumor–stromal interactions in theregulation of TIMP expression and its role in prostate cancerprogression (30), which is consistent with the view thatreduced endoglin expression raised TIMP levels, impairingCAF invasion of the tumor.

Mass spectrometric data suggested that the IGF signalingsystem is an important mediator of endoglin-dependent can-cer cell–stromal cell interactions. This hypothesis is supportedby studies showing that IGF-I stimulates cancer cell prolifera-tion (33) and promotes cell growth in several cancer cell linesincluding PC3, the precursors of PC-3-M cells (47). PrSCssecrete IGF-I, promoting the proliferation of human prostatecancer cells (35). IGFBP-4 and IGFBP-6 are modifiers of IGFpathway signaling. IGFBP-4 antagonizes the growth-stimula-tory effect of IGF-I (31) and inhibits the proliferation andtumorigenicity of human prostate cancer cells (48). Addition-ally, inhibition of IGFBP-6 expression promotes colon cancercell proliferation (49). In this article, we provide evidencesuggesting that PrSCs secrete IGFBP-4 and -6 in responseto decreased endoglin expression, which may repress tumorgrowth. In our experimental model, IGFBP-4 inhibits IGF-I–dependent stimulation of prostate cancer cell growth. Ourinterpretation is that PrSCs secrete IGF-I and several modu-lators of its activity. Under WT conditions of endoglin expres-sion (engþ/þ), the balance is switched toward the stimulationof prostate cancer cell proliferation. Therefore, we suggestendoglin expression is necessary for PrSC/IGF-dependent






modulation of tumor growth, potentially by regulation of TGF-b signaling in CAFs (34). ICAT studies did not reveal endoglin-dependent contributions from other secreted factors includingWnt family members. Further studies are needed to elucidatethe mechanisms underlying endoglin-dependent modulationof IGFBP secretion.

The present study supports the view that endoglin plays acritical role in prostate cancer stromal cell function in themicroenvironment. Experiments in the TRAMP:eng mousemodel, combined with conditional transgenic approaches(16) will help elucidate the effect of systemic endoglin levelson stromal investment at several stages of tumorigenesis.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors thank Kathleen Carrier (Maine Medical Center ResearchInstitute, Scarborough, ME) for her excellent technical assistance,Dr. Michael Jones (Department of Pathology, Maine Medical Center) for analysisof TRAMP tumor pathology, and Norma Albrecht for critical review.

Grant Support

This work was supported by the Maine Cancer Foundation and the NIHNational Center for Research Resources P20-RR-15555 (to C.P.H. Vary, P.C.Brooks); NIH grants HL083151 (to C.P.H. Vary), CA91645 (to P.C. Brooks),CA122985, and Prostate SPORE CA90386 (to R.C. Bergan).

The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received August 03, 2010; revised February 23, 2011; accepted March 22,2011; published OnlineFirst March 28, 2011.

References1. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics,

2009. CA Cancer J Clin 2009;59:225–49.2. Kassouf W, Ismail HR, Aprikian AG, Chevalier S. Whole-mount pros-

tate sections reveal differential endoglin expression in stromal, epithe-lial, and endothelial cells with the development of prostate cancer.Prostate Cancer Prostatic Dis 2004;7:105–10.

3. Josefsson A, Wikstrom P, Granfors T, Egevad L, Karlberg L, StattinP, et al. Tumor size, vascular density and proliferation as prognosticmarkers in GS 6 and GS 7 prostate tumors in patients withlong follow-up and non-curative treatment. Eur Urol 2005;48:577–83.

4. Liu Y, Jovanovic B, Pins M, Lee C, Bergan RC. Over expression ofendoglin in human prostate cancer suppresses cell detachment,migration and invasion. Oncogene 2002;21:8272–81.

5. Craft CS, Romero D, Vary CP, Bergan RC. Endoglin inhibits prostatecancer motility via activation of the ALK2-Smad1 1 pathway. Onco-gene . 2007;26:7240–50.

6. Romero D, Terzic A, Conley BA, Craft C, Jovanovic B, Bergan R,et al. Endoglin phosphorylation by ALK2 2 contributes to theregulation of prostate cancer cell migration. Carcinogenesis 2009;31:359–66.

7. Lakshman M, Huang X, Ananthanarayanan V, Jovanovic B, Liu Y,Craft CS, et al. Endoglin suppresses human prostate cancer metas-tasis. Clin Exp Metastasis 2011;28:39–53.

8. Orimo A, Weinberg RA. Stromal fibroblasts in cancer: a novel tumor-promoting cell type. Cell Cycle 2006;5:1597–601.

9. Joyce JA, Pollard JW. Microenvironmental regulation of metastasis.Nat Rev Cancer 2009;9:239–52.

10. Arthur HM, Ure J, Smith AJ, Renforth G, Wilson DI, Torsney E, et al.Endoglin, an ancillary TGFbeta receptor, is required for extraembryo-nic angiogenesis and plays a key role in heart development. Dev Biol2000;217:42–53.

11. Gingrich JR, Barrios RJ, Morton RA, Boyce BF, DeMayo FJ, FinegoldMJ, et al. Metastatic prostate cancer in a transgenic mouse. CancerRes 1996;56:4096–102.

12. Greenberg NM, DeMayo F, FinegoldMJ,Medina D, TilleyWD, AspinallJO, et al. Prostate cancer in a transgenic mouse. Proc Natl Acad Sci US A 1995;92:3439–43.

13. Shappell SB, Thomas GV, Roberts RL, Herbert R, Ittmann MM,Rubin MA, et al. Prostate pathology of genetically engineeredmice: definitions and classification. The consensus report fromthe Bar Harbor meeting of the Mouse Models of Human CancerConsortium Prostate Pathology Committee. Cancer Res 2004;64:2270–305.

14. Lakshman M, Xu L, Ananthanarayanan V, Cooper J, Takimoto CH,Helenowski I, et al. Dietary genistein inhibits metastasis of humanprostate cancer in mice. Cancer Res 2008;68:2024–32.

15. Said N, Frierson HF Jr., Chernauskas D, Conaway M, Motamed K,Theodorescu D. The role of SPARC in the TRAMP model of prostatecarcinogenesis and progression. Oncogene 2009;28:3487–98.

16. Mancini ML, Terzic A, Conley BA, Oxburgh LH, Nicola T, Vary CP.Endoglin plays distinct roles in vascular smooth muscle cell recruit-ment and regulation of arteriovenous identity during angiogenesis.Dev Dyn 2009;238:2479–93.

17. Bhowmick NA, Chytil A, Plieth D, Gorska AE, Dumont N, Shappell S,et al. TGF-beta signaling in fibroblasts modulates the oncogenicpotential of adjacent epithelia. Science 2004;303:848–51.

18. Mouta-Bellum C, Kirov A, Miceli-Libby L, Mancini ML, Petrova TV,Liaw L, et al. Organ-specific lymphangiectasia, arrested lymphaticsprouting, andmaturation defects resulting from gene-targeting of thePI3K regulatory isoforms p85alpha, p55alpha, and p50alpha. Dev Dyn2009;238:2670–9.

19. Koleva RI, Conley BA, Romero D, Riley KS, Marto JA, Lux A, et al.Endoglin structure and function: determinants of endoglin phosphor-ylation by transforming growth factor-beta receptors. J Biol Chem2006;281:25110–23.

20. Carter N, Nakamoto T, Hirai H, Hunter T. EphrinA1-induced cytoske-letal re-organization requires FAK and p130(cas). Nat Cell Biol2002;4:565–73.

21. Akalu A, Roth JM, Caunt M, Policarpio D, Liebes L, Brooks PC.Inhibition of angiogenesis and tumor metastasis by targeting a matriximmobilized cryptic extracellular matrix epitope in laminin. Cancer Res2007;67:4353–63.

22. Bernabeu C, Lopez-Novoa JM, Quintanilla M. The emerging role ofTGF-beta superfamily coreceptors in cancer. Biochim Biophys Acta2009;1792:954–73.

23. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer2006;6:392–401.

24. Faggin E, Puato M, Zardo L, Franch R, Millino C, Sarinella F, et al.Smooth muscle-specific SM22 protein is expressed in the adventitialcells of balloon-injured rabbit carotid artery. Arterioscler Thromb VascBiol 1999;19:1393–404.

25. Desmouliere A, Guyot C, Gabbiani G. The stroma reaction myofibro-blast: a key player in the control of tumor cell behavior. Int J Dev Biol2004;48:509–17.

26. Dull T, Zufferey R, Kelly M, Mandel RJ, Nguyen M, Trono D, et al. Athird-generation lentivirus vector with a conditional packaging sys-tem. J Virol 1998;72:8463–71.

27. Shiio Y, Donohoe S, Yi EC, Goodlett DR, Aebersold R, Eisenman RN.Quantitative proteomic analysis of Myc oncoprotein function. EMBO J2002;21:5088–96.

28. Coppock DL, Cina-Poppe D, Gilleran S. The quiescin Q6 gene(QSCN6) is a fusion of two ancient gene families: thioredoxin andERV1. Genomics 1998;54:460–8.

Romero et al.





29. Koyama N, Zhang J, Huqun , Miyazawa H, Tanaka T, Su X, et al.Identification of IGFBP-6 6 as an effector of the tumor suppressoractivity of SEMA3B. Oncogene 2008;27:6581–9.

30. Dong Z, Nemeth JA, Cher ML, Palmer KC, Bright RC, Fridman R.Differential regulation of matrix metalloproteinase-9, tissue inhibitorof metalloproteinase-1 1 (TIMP-1) and TIMP-2 expression in co-cultures of prostate cancer and stromal cells. Int J Cancer 2001;93:507–15.

31. Durai R, Davies M, Yang W, Yang SY, Seifalian A, Goldspink G, et al.Biology of insulin-like growth factor binding protein-4 4 and its role incancer (review). Int J Oncol 2006;28:1317–25.

32. Tennant MK, Thrasher JB, Twomey PA, Birnbaum RS, Plymate SR.Insulin-like growth factor-binding proteins (IGFBP)-4, -5, and -6 6 inthe benign and malignant human prostate: IGFBP-5 messengerribonucleic acid localization differs from IGFBP-5 protein localization.J Clin Endocrinol Metab 1996;81:3783–92.

33. Pollak M. Insulin and insulin-like growth factor signalling in neoplasia.Nat Rev Cancer 2008;8:915–28.

34. Kawada M, Inoue H, Arakawa M, Ikeda D. Transforming growthfactor-beta1 1 modulates tumor-stromal cell interactions of pros-tate cancer through insulin-like growth factor-I. Anticancer Res2008;28:721–30.

35. Kawada M, Inoue H, Masuda T, Ikeda D. Insulin-like growth factor Isecreted from prostate stromal cells mediates tumor-stromal cellinteractions of prostate cancer. Cancer Res 2006;66:4419–25.

36. Kimura G, Kasuya J, Giannini S, Honda Y, Mohan S, Kawachi M, et al.Insulin-like growth factor (IGF) system components in human pro-static cancer cell-lines: LNCaP, DU145, and PC-3 3 cells. Int J Urol1996;3:39–46.

37. Singh SV, Warin R, Xiao D, Powolny AA, Stan SD, Arlotti JA, et al.Sulforaphane inhibits prostate carcinogenesis and pulmonary metas-tasis in TRAMP mice in association with increased cytotoxicity ofnatural killer cells. Cancer Res 2009;69:2117–25.

38. Duwel A, Eleno N, Jerkic M, Arevalo M, Bolanos JP, Bernabeu C, et al.Reduced tumor growth and angiogenesis in endoglin-haploinsuffi-cient mice. Tumour Biol 2007;28:1–8.

39. Perez-Gomez E, Villa-Morales M, Santos J, Fernandez-Piqueras J,Gamallo C, Dotor J, et al. A role for endoglin as a suppressor of

malignancy during mouse skin carcinogenesis. Cancer Res 2007;67:10268–77.

40. Mishra PJ, Humeniuk R, Medina DJ, Alexe G, Mesirov JP, Ganesan S,et al. Carcinoma-associated fibroblast-like differentiation of humanmesenchymal stem cells. Cancer Res 2008;68:4331–9.

41. Zeisberg EM, Potenta S, Xie L, Zeisberg M, Kalluri R. Discovery ofendothelial to mesenchymal transition as a source for carcinoma-associated fibroblasts. Cancer Res 2007;67:10123–8.

42. Conley BA, Smith JD, Guerrero-Esteo M, Bernabeu C, Vary CP.Endoglin, a TGF-beta receptor-associated protein, is expressed bysmooth muscle cells in human atherosclerotic plaques. Atherosclero-sis 2000;153:323–35.

43. Hirschi KK, Rohovsky SA, D’Amore PA. PDGF, TGF-beta, and het-erotypic cell–cell interactions mediate endothelial cell-inducedrecruitment of 10T1/2 cells and their differentiation to a smoothmuscle fate. J Cell Biol 1998;141:805–14.

44. Ma X, Labinaz M, Goldstein J, Miller H, Keon WJ, Letarte M, et al.Endoglin is overexpressed after arterial injury and is required fortransforming growth Factor-SS-induced inhibition of smooth musclecell migration. Arterioscler Thromb Vasc Biol 2000;20:2546–52.

45. Mancini ML, Verdi JM, Conley BA, Nicola T, Spicer DB, Oxburgh LH,et al. Endoglin is required for myogenic differentiation potential ofneural crest stem cells. Dev Biol 2007;308:520–33.

46. Verona EV, Elkahloun AG, Yang J, Bandyopadhyay A, Yeh IT, Sun LZ.Transforming growth factor-beta signaling in prostate stromal cellssupports prostate carcinoma growth by up-regulating stromal genesrelated to tissue remodeling. Cancer Res 2007;67:5737–46.

47. Iwamura M, Sluss PM, Casamento JB, Cockett AT. Insulin-like growthfactor I: action and receptor characterization in human prostatecancer cell lines. Prostate 1993;22:243–52.

48. Damon SE, Maddison L, Ware JL, Plymate SR. Overexpression of aninhibitory insulin-like growth factor binding protein (IGFBP), IGFBP-4,delays onset of prostate tumor formation. Endocrinology 1998;139:3456–64.

49. Barni T, Vannelli BG, Sadri R, Pupilli C, Ghiandi P, Rizzo M, et al.Insulin-like growth factor-I (IGF-I) and its binding protein IGFBP-4 4 inhuman prostatic hyperplastic tissue: gene expression and its cellularlocalization. J Clin Endocrinol Metab 1994;78:778–83.






Correction

Correction: Endoglin Regulates Cancer–StromalCell Interactions in Prostate Tumors

In this article (Cancer Res 2011;71:3482–93), which was published in the May 15,2011, issue of Cancer Research (1), a funding source was left out of the originalstatement. The additional funding source is provided below. The authors regret thiserror.

Grant Support

Additional support was provided by the Maine Medical Center Research Institute Histopathology CoreFacility as supported by NIH P20RR18789 (D.M. Wojchowski, PI).

Reference1. Romero D, O'Neill C, Terzic A, Contois L, Young K, Conley BA, et al. Endoglin regulates cancer–

stromal cell interactions in prostate tumors. Cancer Res 2011;71:3482–93.

Published OnlineFirst January 25, 2012.doi: 10.1158/0008-5472.CAN-11-3979�2012 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 1037

2011;71:3482-3493. Published OnlineFirst March 28, 2011.Cancer Res Diana Romero, Christine O'Neill, Aleksandra Terzic, et al. Prostate Tumors

Stromal Cell Interactions in−Endoglin Regulates Cancer

Updated version

10.1158/0008-5472.CAN-10-2665doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2011/03/28/0008-5472.CAN-10-2665.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/71/10/3482.full#ref-list-1

This article cites 49 articles, 19 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/71/10/3482.full#related-urls

This article has been cited by 8 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://cancerres.aacrjournals.org/content/71/10/3482To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-10-2665

http://cancerres.aacrjournals.org/content/suppl/2011/03/28/0008-5472.CAN-10-2665.DC1

http://cancerres.aacrjournals.org/content/71/10/3482.full#ref-list-1

http://cancerres.aacrjournals.org/content/71/10/3482.full#related-urls

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

mailto:[email protected]

http://cancerres.aacrjournals.org/content/71/10/3482


Endoglin Regulates Cancer Stromal Cell Interactions in ... · Microenvironment and Immunology...

Documents

Transcript of Endoglin Regulates Cancer Stromal Cell Interactions in ... · Microenvironment and Immunology...