RANK Signaling Blockade Reduces Breast Cancer Recurrence ...1].pdf · RANK Signaling Blockade...

13
Tumor and Stem Cell Biology RANK Signaling Blockade Reduces Breast Cancer Recurrence by Inducing Tumor Cell Differentiation Guillermo Yoldi 1 , Pasquale Pellegrini 1 , Eva M. Trinidad 1 , Alex Cordero 1 , Jorge Gomez-Miragaya 1 , Jordi Serra-Musach 2 , William C. Dougall 3 , Puricaci on Mu~ noz 1 , Miguel-Angel Pujana 2 , Lourdes Planelles 4 , and Eva Gonz alez-Su arez 1 Abstract RANK expression is associated with poor prognosis in breast cancer even though its therapeutic potential remains unknown. RANKL and its receptor RANK are downstream effectors of the progesterone signaling pathway. However, RANK expression is enriched in hormone receptor negative adenocarcinomas, sug- gesting additional roles for RANK signaling beyond its hormone- dependent function. Here, to explore the role of RANK signaling once tumors have developed, we use the mouse mammary tumor virus-Polyoma Middle T (MMTV-PyMT), which mimics RANK and RANKL expression patterns seen in human breast adenocarcinomas. Complementary genetic and pharmacologic approaches demonstrate that therapeutic inhibition of RANK signaling drastically reduces the cancer stem cell pool, decreases tumor and metastasis initiation, and enhances sensitivity to chemotherapy. Mechanistically, genome-wide expression anal- yses show that anti-RANKL therapy promotes lactogenic dif- ferentiation of tumor cells. Moreover, RANK signaling in tumor cells negatively regulates the expression of Ap2 tran- scription factors, and enhances the Wnt agonist Rspo1 and the Sca1-population, enriched in tumor-initiating cells. In addi- tion, we found that expression of TFAP2B and the RANK inhibitor, OPG, in human breast cancer correlate and are associated with relapse-free tumors. These results support the use of RANKL inhibitors to reduce recurrence and metastasis in breast cancer patients based on its ability to induce tumor cell differentiation. Cancer Res; 76(19); 585769. Ó2016 AACR. Introduction Multiple lines of evidence support the existence of tumor- initiating cells (TICs) or cancer stem cells (CSC) in breast cancer (1). Recent efforts to develop CSC-related therapies explored elimination of the CSC population, removal of their self-renewal capability, and forced terminal differentiation. The rst differen- tiation agent successfully used in the clinic was all-trans retinoic acid in acute promyelocytic leukemia (2). Retinoid signaling also regulates breast CSC self renewal and differentiation (3). RANK ligand (RANKL) is expressed in progesterone receptor- positive (PR þ ) mammary epithelial cells and acts as a paracrine mediator of progesterone in mouse and human mammary epithe- lia (49). Overexpression of RANKL's receptor, RANK in mammary epithelial cells enhances proliferation, impairs lactation, and induces the accumulation of mammary stem cells (MaSC) and progenitors (912). In human adenocarcinomas, RANK is predom- inantly expressed in hormone receptor-negative (HR ) tumors, supporting a progesterone-independent role. In contrast to RANK, RANKL is rarely expressed on tumor cells, but it is expressed in tumor-inltrating lymphocytes (7, 11, 13). RANK expression in human adenocarcinomas is associated with reduced overall sur- vival (13, 14). However, the mechanisms underlying these aggres- sive tumor phenotypes and the therapeutic potential of RANKL inhibition once tumors have developed remain unexplored. The MMTV-PyMT breast cancer mouse model displays wide- spread transformation of the mammary gland and a high inci- dence of lung metastasis (15, 16). Tumor cells of invasive PyMT adenocarcinomas do not express hormone receptors or RANKL, but do express high levels of RANK (7, 9). RANKL inhibitors are currently used for the treatment of bone-related pathologies, osteoporosis, and bone metastasis. Here we demonstrate that inhibition of RANK signaling acts as a differentiation therapy in breast cancer, depleting the cancer stem cells population and reducing recurrence and metastasis. 1 Cancer Epigenetics and Biology Program, Bellvitge Biomedical Research Institute, IDIBELL, Barcelona, Spain. 2 Program Against Can- cer Therapeutic Resistance (ProCURE), Breast Cancer and Systems Biology Lab, Catalan Institute of Oncology, IDIBELL, Barcelona, Spain. 3 Therapeutic Innovation Unit, Amgen Inc., Seattle,Washington. 4 Cen- tro Nacional de Biotecnología/CSIC, UAM Cantoblanco, Madrid, Spain. Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). G. Yoldi, P. Pellegrini, and E.M. Trinidad contributed equally to this work. Current address for P. Pellegrini: Gladstone Institutes, San Francisco, California; current address for W.C. Dougall: Department of Immunology in Cancer and Infection, QIMR Berghofer Medical Research Institute, Herston, QLD, Australia; current address for A. Cordero: Department of Neurological Surgery Feinberg School of Medicine, Northwestern University Chicago, IL. Corresponding Author: Eva Gonz alez-Su arez, Bellvitge Biomedical Research Institute (IDIBELL), Av. Gran Via de L'Hospitalet, 199-203, L'Hospitalet de Llobregat, Barcelona 08908, Spain. Phone: 932607347; Fax: 932607219; E-mail: [email protected] doi: 10.1158/0008-5472.CAN-15-2745 Ó2016 American Association for Cancer Research. Cancer Research www.aacrjournals.org 5857

Transcript of RANK Signaling Blockade Reduces Breast Cancer Recurrence ...1].pdf · RANK Signaling Blockade...

Tumor and Stem Cell Biology

RANK Signaling Blockade Reduces BreastCancer Recurrence by Inducing Tumor CellDifferentiationGuillermo Yoldi1, Pasquale Pellegrini1, Eva M. Trinidad1, Alex Cordero1,Jorge Gomez-Miragaya1, Jordi Serra-Musach2,William C. Dougall3,Purificaci�on Mu~noz1, Miguel-Angel Pujana2, Lourdes Planelles4, andEva Gonz�alez-Su�arez1

Abstract

RANK expression is associated with poor prognosis in breastcancer even though its therapeutic potential remains unknown.RANKL and its receptor RANK are downstream effectors of theprogesterone signaling pathway. However, RANK expression isenriched in hormone receptor negative adenocarcinomas, sug-gesting additional roles for RANK signaling beyond its hormone-dependent function. Here, to explore the role of RANK signalingonce tumors have developed, we use the mouse mammarytumor virus-Polyoma Middle T (MMTV-PyMT), which mimicsRANK and RANKL expression patterns seen in human breastadenocarcinomas. Complementary genetic and pharmacologicapproaches demonstrate that therapeutic inhibition of RANKsignaling drastically reduces the cancer stem cell pool, decreases

tumor and metastasis initiation, and enhances sensitivity tochemotherapy. Mechanistically, genome-wide expression anal-yses show that anti-RANKL therapy promotes lactogenic dif-ferentiation of tumor cells. Moreover, RANK signaling intumor cells negatively regulates the expression of Ap2 tran-scription factors, and enhances the Wnt agonist Rspo1 and theSca1-population, enriched in tumor-initiating cells. In addi-tion, we found that expression of TFAP2B and the RANKinhibitor, OPG, in human breast cancer correlate and areassociated with relapse-free tumors. These results support theuse of RANKL inhibitors to reduce recurrence and metastasis inbreast cancer patients based on its ability to induce tumor celldifferentiation. Cancer Res; 76(19); 5857–69. �2016 AACR.

IntroductionMultiple lines of evidence support the existence of tumor-

initiating cells (TICs) or cancer stem cells (CSC) in breast cancer(1). Recent efforts to develop CSC-related therapies exploredelimination of the CSC population, removal of their self-renewalcapability, and forced terminal differentiation. The first differen-

tiation agent successfully used in the clinic was all-trans retinoicacid in acute promyelocytic leukemia (2). Retinoid signaling alsoregulates breast CSC self renewal and differentiation (3).

RANK ligand (RANKL) is expressed in progesterone receptor-positive (PRþ) mammary epithelial cells and acts as a paracrinemediator of progesterone in mouse and humanmammary epithe-lia (4–9).Overexpression of RANKL's receptor, RANK inmammaryepithelial cells enhances proliferation, impairs lactation, andinduces the accumulation of mammary stem cells (MaSC) andprogenitors (9–12). In humanadenocarcinomas, RANK is predom-inantly expressed in hormone receptor-negative (HR�) tumors,supporting a progesterone-independent role. In contrast to RANK,RANKL is rarely expressed on tumor cells, but it is expressed intumor-infiltrating lymphocytes (7, 11, 13). RANK expression inhuman adenocarcinomas is associated with reduced overall sur-vival (13, 14). However, the mechanisms underlying these aggres-sive tumor phenotypes and the therapeutic potential of RANKLinhibition once tumors have developed remain unexplored.

The MMTV-PyMT breast cancer mouse model displays wide-spread transformation of the mammary gland and a high inci-dence of lung metastasis (15, 16). Tumor cells of invasive PyMTadenocarcinomas do not express hormone receptors or RANKL,but do express high levels of RANK (7, 9). RANKL inhibitors arecurrently used for the treatment of bone-related pathologies,osteoporosis, and bone metastasis. Here we demonstrate thatinhibition of RANK signaling acts as a differentiation therapy inbreast cancer, depleting the cancer stem cells population andreducing recurrence and metastasis.

1Cancer Epigenetics and Biology Program, Bellvitge BiomedicalResearch Institute, IDIBELL, Barcelona, Spain. 2ProgramAgainst Can-cer Therapeutic Resistance (ProCURE), Breast Cancer and SystemsBiology Lab,Catalan Institute ofOncology, IDIBELL, Barcelona, Spain.3Therapeutic Innovation Unit, Amgen Inc., Seattle,Washington. 4Cen-troNacional deBiotecnología/CSIC,UAMCantoblanco,Madrid, Spain.

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

G. Yoldi, P. Pellegrini, and E.M. Trinidad contributed equally to this work.

Current address for P. Pellegrini: Gladstone Institutes, San Francisco, California;current address for W.C. Dougall: Department of Immunology in Cancer andInfection, QIMR Berghofer Medical Research Institute, Herston, QLD, Australia;current address for A. Cordero: Department of Neurological Surgery FeinbergSchool of Medicine, Northwestern University Chicago, IL.

Corresponding Author: Eva Gonz�alez-Su�arez, Bellvitge Biomedical ResearchInstitute (IDIBELL), Av. Gran Via de L'Hospitalet, 199-203, L'Hospitalet deLlobregat, Barcelona 08908, Spain. Phone: 932607347; Fax: 932607219; E-mail:[email protected]

doi: 10.1158/0008-5472.CAN-15-2745

�2016 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 5857

Materials and MethodsAnimals, RANKL, RANK-Fc, and docetaxel treatments

All research involving animals was performed at the IDI-BELL animal facility in compliance with protocols approvedby the IDIBELL Committee on Animal Care and followingNational and European Union regulations. MMTV-PyMT(FVB/N-Tg(MMTV-PyVT)634Mul) were obtained from the

Jackson Laboratory (15). MMTV-PyMT�/þ;RANK�/� mice wereobtained by backcrossing the MMTV-PyMT (FvB/N) strainwith RANKþ/� mice into the C57BL/6 background (17).RANKL (1 mg/kg, Amgen) and RANK-Fc (10 mg/kg, Amgen)were injected subcutaneously three times a week (7, 10).Docetaxel (Actavis) was administered at 25 mg/kg intraperi-toneally twice per week.

A

Hyp

erpl

asia

Car

cino

ma

RANK RANKL RANK RANKL

Rank

MMTV-PyM

T T

Draining LN

MMTV-PyM

T

Rel

ativ

e ex

pres

sion

0.0000

0.0005

0.0010

0.0015

0.0020

Rankl

*

***

WT MG

**

0.0

0.1

0.2

0.3

WT MG

MMTV-PyM

T T

Rel

ativ

e ex

pres

sion

to K

8

B

50 μm 50 μm

RANKL

2 weeks 24 h

Tumorresection

Tumor formation

E

Control

RANKL

Days

Rel

ativ

e tu

mor

vol

ume

(mm

3 )

1 2 3 4 5 6 7 8 9 10 11 120

2

4

6

8

10D

G Control ControlRANKL

200 μm

Ant

i-milk

RANKL

Secretory areas

% o

f Tot

al a

rea

0

5

10

*

F

RANKLControl

0.0

0.1

0.2

0.3

0.4

TAMTUM

CD4+0.0

0.1

0.2

0.3

0.4

0.5

0.6

RanklC

Rel

ativ

e ex

pres

sion

Rank

Rel

ativ

e ex

pres

sion

CD8+TAM

TUMCD4+

CD8+

Figure 1.

RANKL decreases MMTV-PyMT tumor cell differentiation. A, representative images of RANK and RANKL protein expression. Note that in carcinomas, RANKLis not expressed in tumor cells but in tumor-infiltrating lymphocytes. B, Rank and RanklmRNA expression relative to K8 or b-actin in seven WT, eight MMTV-PyMTtumors, and three draining lymph nodes of MMTV-PyMT tumor-bearingmice. C, Rank and RanklmRNA expression relative to Rpl38 in FACS-sorted (SupplementaryFig. S2) tumor cells (TUM) CD45�, macrophages (TAM) CD45þCD11bþF4/80þGr1�, CD4þ, and CD8þ T lymphocytes (CD45þCD11b�CD3þCD4þ or CD8) fromfour MMTV-PyMT tumors. D, schematic overview of short-term (2 weeks) RANKL treatment in MMTV-PyMT tumor-bearing females. E, tumor volume normalized tothe first day of treatment of five MMTV-PyMT tumor-bearing mice undergoing RANKL treatment and controls. F, percentage of secretory areas relative tototal tumor area in (3–5) MMTV-PyMT primary tumors after RANKL treatment. G, representative images of hematoxylin and eosin and milk protein staining inMMTV-PyMT primary tumors. B, E, F, mean, SEM, and t test probabilities are shown (� , 0.01 < P < 0.05; �� , 0.001 < P < 0.01; ��� , 0.001 < P < 0.0001).

Yoldi et al.

Cancer Res; 76(19) October 1, 2016 Cancer Research5858

Tumor cell isolation, tumor and metastasis initiation assaysTumor cells were isolated as described (18). For orthotopic

transplants and tumor-limiting dilution assays (LDA), tumor cellswere mixed 1:1 with Matrigel matrix (BD Biosciences) and ortho-topically implanted in the inguinal mammary gland of 6- to 10-week-old syngeneic females. For metastasis assays, tumor cellsresuspended in cold PBS were injected intravenously in 6- to 10-week-old Foxn1nu females.

Tissue histology and immunostainingTissue samples were fixed in formalin and embedded in par-

affin. Three-micrometer sections were cut for histologic analysis

and stained with hematoxylin and eosin. Entire lungs were stepsectioned at 100 mmand15 cuts per lungwere quantified. Antigenheat retrievalwith citratewas used for PR (DAKO), SMA-1 (Sigma-Aldrich), mRANKL (R&D Systems), Ki67 (Thermo Scientific),cleaved caspase-3 (Cell Signaling) antibodies, and rabbit anti-milk serum (kindly provided by Prof. Nancy E. Hynes). mRANK(R&D Systems) immunostaining was performed, pretreating sec-tions with Protease XXIV 5 U/mL (Sigma-Aldrich) for 15 minutesat room temperature. All antibodies were incubated overnight at4�C, detected with biotinylated secondary antibodies and strep-tavidin horseradish peroxidase (Vector), and revealed with DABsubstrate (DAKO).

*

Tumor spheres

*

Group N cells N fat pad N tumors TIC frequency

CTRL

100,000 4 41/ 206

(CI 599–70.9)10,000 6 61,000 6 5100 6 6

Post-RANK-Fc

100,000 6 61/ 466

(CI 1,138–190.7)10,000 6 61,000 6 4100 6 4

Pre-RANK-Fc

100,000 6 61/ 1,929

(CI 5,431–685.1)10,000 6 51,000 6 5100 6 2

Pre & postRANK-Fc

100,000 6 61/ 2,353

(CI 6,378–868.3)10,000 6 51,000 6 3100 6 3

GROUP P 2

CTRL vs. post RANK-Fc0.13 2.27

CTRL vs. pre RANK-Fc 2.99 E–5 17.4

CTRL vs. pre & post RANK-Fc

7.48 E–6 7.41

Pre RANK-Fc vs. pre & post RANK-Fc 0.701 0.147

χ

E

B

A

RecurrencePosttreatment(passage 2)

Occult diseaseSurgery Pretreatment

(passage 1)

24 h 2 weeks

Pre & post-RANK-Fc

Pre-RANK-Fc

Post-RANK-Fc

CTRLMock

24 h 2 weeks

RANK-Fc

24 h 2 weeks

Mock

24 h 2 weeks

RANK-Fc24 h

24 h

24 h

24 h4 weeks

4 weeks

Mock

RANK-Fc

LDA

LDA

PyMT tumor cells

Passage 1 Passage 2

Tumor growth

Days

Tum

or v

olum

e

1 2 3 4 50.0

0.1

0.2

0.3

0.4

0.5

CTRLRANK-Fc

Passage 1C

D

% C

aspa

se-3

/nuc

lei

CTRL

Post-R

ANK-Fc

Pre-RANK-F

c

Pre & po

st-RANK-F

c0

1

2

3

4

5 Apoptotic cells

CTRL

Post-R

ANK-Fc

Pre-RANK-Fc

Pre & po

st-RANK-Fc

N o

f Sph

eres

60

40

20

0

Figure 2.

Inhibition of RANK signaling depletesthe pool of MMTV-PyMT TICs. A,schematic overview of RANK-Fctreatments in orthotopic MMTV-PyMTtumors. One million cells isolated fromone MMTV-PyMT carcinoma wereorthotopically injected into syngeneicWT mice (FVB), which wererandomized 1:1 for RANK-Fc(10 mg/kg, 3 times per week, 4 weeks)or mock treatment starting 24 h later(passage 1). Cells isolated from threetumors from each treatment arm werepooled and orthotopically injected intoWT (passage 2) mice in limitingdilutions and randomized 1:1 foradditional RANK-Fc or mock treatment(2 weeks). Total number of tumors wasscored after 26weeks.B, tumor growthof passage 1 tumors. C, percentage ofpositive cleaved caspase-3 cells inpassage 2 tumors. D, TIC frequencies(with confidence intervals), x2 values,and associated probabilities. E, numberof secondary tumorspheres formed byRANK-Fc-treated MMTV-PyMT tumors.Each bar represents data from fourtumors plated in triplicates. B, C, E,mean, SEM, and t test statistics areshown (� , 0.01 < P < 0.05).

Therapeutic RANKL Inhibition Reduces Cancer Stem Cells

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 5859

Tumorsphere cultureCells isolated from primary tumors were resuspended in

serum-free DMEM F12 mammosphere medium containing20 ng/mL of EFG, 1� B27, and 4 mg/mL heparin (Sigma-Aldrich), as previously described (19) with 2% of growthfactor–reduced matrigel. Primary tumorspheres were derivedby plating 20,000 cells/mL in 2 mL of medium onto cell-suspension culture plates. After 14 days, tumorspheres wereisolated by 5 min treatment with PBS-EDTA 1 mmol/L þ 5 minof trypsin at 37�C and plated for secondary tumorsphereformation at a concentration of 5,000 cells/mL in triplicate.Individual spheres from each replicate well were counted undera microscope.

Flow cytometrySingle cells were resuspended and blocked with PBS 2% FBS

and IgG blocking reagent for 10 min on ice and incubated for 30min on ice with CD45-APC-Cy7 (30-F11), CD4-PE-Cy7 (RM4-5),CD11b-APC (M1/70), CD8-PE or CD8-FITC (53-6.7), Gr1-FITC(RB6-8C5), F4/80-PE (BM8), CD49b-Alexa 647 (1HMa2), CD45-PECy7 or –APCCy7 (30-F11), and CD31-PECy7 (390), all fromBiolegend, CD24-FITC (M1/69), CD61-FITC (2C9.G2), Sca-1-APC (Ly-6A/E) from BD Pharmingen, CD90-PE (HIS51; Biosci-ence) and CD49f-AF647 (GoH3; R&D Systems). FACS analysiswas performed using FACS Canto, FACS Aria (Becton Dickinson)and Diva software. Cells were sorted using MoFlo (BeckmanCoulter) at 25 psi (172 kPa) and a 100-mm tip.

Tm4sf1DspFggAtp1b1SycnEgfbp2Wnk2Celsr1Ptrh2Atp1b1Tspan8LplLplTm4sf1Csn2Csn1s2aCsn1s1Scgb1b27WapGlycam1Fabp3Golph3Col9a2PipSlc2a1Btn1a1GldcHomer2Wnk2Car6Ucp3Wfdc2Prol1Csn1s2bSlc39a8Isg20Gjb1Pthlh

Pregnancy

1 d 3 d 7 d 12 d 17 d 19 d 1 d 2 d 9 d Involution

Lactation

-4 4Standarized expression

A

*

**

*

Csn2 Pip

Scgb1b27 Scgb2b27

CTRL

Pre & post-RANK-Fc

Rel

ativ

e ex

pres

sion

100 μm

Rel

ativ

e ex

pres

sion

D

CTRL

Pre & post- RANK-Fc

Anti-milkC

E

Fold

cha

nge

Csn2

Pip

Scgb2

b27

-10

-8

-4

-6

-2

0

010203040

0.00.51.01.52.0

0

5

10

15

0

10

20

30

RANKL

**

*

P = 0.001

Enric

hmen

t sco

re

0.2

0

0.1

0.4

Pre & post RANK-FCOverexpressed

CTRLOverexpressed

0

0.3

Pregnancy - lactation (upregulated set)CTRL versus pre & post RANK-FC

B

KLK1B16GLYCAM1C77370SLC50A1TCHHMEIS1CSN2GDF15HLTFPPARGC1ALRRC26etc.

Figure 3.

RANKL inhibition induces differentiation of tumor cells into milk secreting cells. A, expression profile in mammary gland development of differentiallyexpressed genes between RANK-Fc-treated tumors and controls (21). Genes further validated by reverse transcription-PCR are shown in red. B, GSEA graphicaloutput for the association between lactation overexpressed genes and RANK-Fc treatment. The top genes contributing to this association are listed.C, mRNA expression levels of indicated genes relative to HPRT. Each bar is representative of three tumors. D, representative images of milk staining inRANK-Fc–treated tumors. E, fold change of mRNA expression levels of indicated genes in RANKL-treated acinar cultures of MMTV-PyMT tumor cells relativeto untreated controls. Three tumors were analyzed. B and E, mean, SEM, and t test P values are shown (� , 0.01 < P < 0.05).

Yoldi et al.

Cancer Res; 76(19) October 1, 2016 Cancer Research5860

RNA labeling and hybridization to Agilent microarraysHybridization to SurePrint G3 Mouse Gene Expression

Microarray (ID G4852A; Agilent Technologies) was conductedfollowing manufacturer's protocol (Two-Color Microarray-Based Gene Expression Analysis v. 6.5; Agilent Technologies),and dye swaps (Cy3 and Cy5) were performed for RNA ampli-fied from each sample. Microarray chips were washed andscanned using a DNA Microarray Scanner (Model G2505C;Agilent Technologies).

Microarray analysisMicroarray data were feature extracted using Feature Extraction

Software (v. 10.7) available from Agilent, using the default vari-ables. Outlier features on the arrays were flagged by the samesoftware package. Data analysis was performed using Bioconduc-tor package, under R environment. Data preprocessing and dif-ferential expression analysis was performed using limma andRankProd package, and latest gene annotations available wasused. Raw feature intensities were background corrected usingnormexp background correction algorithm.Within-array normal-ization was done using spatial and intensity-dependent loess.Aquantile normalization was used to normalize between arrays.The expression of each gene is reported as the base 2 logarithm ofratio of the value obtained of each condition relative to controls. Agene is considered differentially expressed if it displays a pfp(proportion of false positives) less than 0.05 by nonparametric

test. TheGSEAwas runusingdefault values for all parameters. Rawmicroarray data has been deposited in GEO, access numberGSE66085. The mature luminal and stem cell gene sets weretaken from the original publication (20). The differentiallyexpressed genes between lactation and pregnancy were identifiedusing the GEO GSE8191 dataset (21) and the TFAP2C-regulatedgenes in breast cancer cells using the GEO GSE8640 dataset (22).Cox proportional hazard regression analyses were applied toevaluate associations with prognosis (relapse or distant metasta-sis) at the level of microarray probes.

Statistical analysisDifferences were analyzedwith a two-tailed Student t test or an F

test, one sample t test against a reference value of 1. Two-wayanalysis of variance was used to compare tumor growth curves. TheMantel–Cox test was used for tumor-free survival studies. Frequen-cy of tumor initiating was estimated using the extreme limitingdilution software (ELDA; ref. 23). The statistical significance ofdifference between groups is indicated by asterisks (�, 0.01 < P <0.05; ��, 0.001 < P < 0.01; ���, 0.001 < P < 0.0001; ����, P < 0.0001).

ResultsRANKL stimulation promotes tumor growth in MMTV-PyMTprimary tumor cells

MMTV-PyMT preneoplasic lesions and adenocarcinomasexpressed high levels of RANK (Fig. 1A and B). RANKL expression

A

C

% M

ice

with

lung

met

asts

is

Preneoplasic lesions

Pre

neop

lasi

c le

sion

s/M

Gs

0

5

10

15

20

25

PyMT;

RANK +/+

PyMT;

RANK −/−

**

PyMT;

RANK +/+

Lung metastasis

1-1010-100more than 100

Num

ber o

f tum

ors/

mou

se

0

2

4

6

8

10

PyMT;

RANK −/−

Palpable lesions

*

0

20

40

60

80

100

B

Latency (days after birth)

% T

umor

free

PyMT;RANK+/+

PyMT;RANK−/−

PyMT;

RANK +/+

PyMT;

RANK −/−

D

*

0 50 100 150 200 2500

20

40

60

80

100

****

Figure 4.

Constitutive deletion of RANK increases tumor latency, decreases tumor incidence, and prevents lung metastasis of MMTV-PyMT tumors. A, kinetics ofpalpable tumor onset with age in 18 PyMT;RANKþ/þ and 10 PyMT;RANK�/� mice. Log-rank test (���� , P < 0.0001). B, number of palpable lesions detected atnecropsy in 17 PyMT;RANKþ/þ and 10 PyMT;RANK�/� mice. C, number of preneoplasic regions per mammary gland detected in mammary whole mounts of9 PyMT;RANKþ/þ and 5 PyMT;RANK�/� mice 13 to 22 weeks old. Each dot represents one mammary gland. D, percentage of PyMT;RANKþ/þ (n ¼ 6) andPyMT;RANK�/� (n ¼ 7) mice with lung metastasis. The total number of metastatic foci per mouse is indicated. x2 ¼ 6.96, as calculated by contingency2 � 2, P ¼ 0.01. B and C, mean, SEM, and t test P values are shown (� , 0.01 < P < 0.05).

Therapeutic RANKL Inhibition Reduces Cancer Stem Cells

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 5861

A

** ** ****

RANK+/+ Tumor in WT host

RANK+/+ Tumor in RANK–/– host

RANK–/– Tumor in RANK–/– host

RANK–/– Tumor in WT host

LATE

NC

Y (d

ays)

0

5

10

15

20

25

% N

onvi

able

are

as

RANK+/+

in WT RANK–/–

in W

T

% C

aspa

se-3

/nuc

lei

C* *

D

Apoptotic cellsNon-viable areas

# # ####### #Days

RANK+/+ in WT RANK–/– in WT RANK+/+ in WT docetaxel RANK–/– in WT docetaxel

***

B

E

FGroup N cells N cells Mice Metastasis

incidence

PyMT;RANK+/+ 100,000 4 4

10,000 4 4

1,000 4 0

100 4 1

PyMT;RANK–/– 100,000 4 3

10,000 4 3

1,000 4 1

100 4 0

GGroup N fat

padNtumors

PyMT;RANK+/+ 10,000 4 4

1,000 4 3

100 4 4

10 4 0

PyMT;RANK–/– 10,000 4 4

1,000 4 2

100 4 1

10 4 0

PyMT;RANK +/+1 / 2,952 (C.I. 9,374–929)

PyMT;RANK–/–1/ 29,420 (C.I. 93,261−9,281)

P value : 0.00262 χ2: 9.06

H

**

N o

f met

asta

tic fo

ci

RANK+/+

RANK–/–0

50

100

150

Tumor-initiating cell (TIC)frequency

Metastasis-initiating cell (MIC)frequency

PyMT;RANK +/+1/ 285 (C.I. 913−89)

PyMT;RANK–/–1/ 1,078 (C.I. 3,572−326)

P value : 0.05 χ2: 3.58

Docetaxel response

Tum

or v

olum

e

1 3 7 10 14 17 21 24 28 31 35 380.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0

10

20

30

40

50

RANK–/–

in RANK–/–

RANK+/+

in RANK–/–

0

10

20

30

RANK+/+

in WT RANK–/–

in W

T RANK–/–

in RANK–/–

RANK+/+

in RANK–/–

50 μm

N o

f Sph

eres

0

200

400

600

800

** **

RANK+/+ Tumor in WT host

RANK+/+ Tumor in RANK–/– host

RANK–/– Tumor in RANK–/– host

RANK–/– Tumor in WT host

Tumorspheres

RANK–/– Tumor in WT

RANK+/+ Tumor in WT

Yoldi et al.

Cancer Res; 76(19) October 1, 2016 Cancer Research5862

was found in nontumorigenic ducts and hyperplasias (Fig. 1A)but was lost in MMTV-PyMT adenocarcinomas, consistent withthe loss of PR positivity (Fig. 1A and Supplementary Fig. S1;ref. 24). RANKL was expressed in draining lymph nodes andtumor-infiltrating leucocytes of tumor-bearing mice (Fig. 1Aand B). Rankl mRNA was predominantly found in tumor-infil-trating CD4þ and CD8þ T lymphocytes (CD45þCD11b�CD4þ

and CD45þCD11b�CD8þ), whereas Rank mRNA was found intumor cells and macrophages (Fig. 1C and Supplementary Fig.S2), consistent with the expression of RANK and RANKL inhuman breast adenocarcinomas (7, 11, 13), highlighting therelevance of the MMTV-PyMT tumor model to the study ofhuman pathology.

Administration of RANKL resulted in increased acinar size inMMTV-PyMT tumor acini (Supplementary Fig. S3A and S3B;refs. 25, 26). No significant changes in proliferation werefound, but decreased apoptosis was observed in RANKL-treatedtumor cultures (Supplementary Fig. S3B). RANKL-treated acinishowed an "invasive-like" phenotype, with isolated cells sur-rounding the acini (Supplementary Fig. S3A). Remarkably, 2-week RANKL-treated acini gave rise to faster growing tumorsand more metastasis than untreated controls when injected inimmunodeficient mice (Supplementary Fig. S3C and S3D).These results demonstrate that activation of RANK signalingcould promote tumor growth and metastasis in MMTV-PyMTprimary tumor cells.

Inhibition of RANKL signaling decreases the frequency oftumor-initiating cells

No significant changes in tumor growth, tumor cell prolifera-tion, or apoptosis were observed after 2 weeks of RANKL treat-ment in vivo on tumor-bearing MMTV-PyMTmice (Fig. 1D–E andSupplementary Fig. S3E; ref. 10), but tumor cell density washigher in RANKL treated lesions, in contrast to control mice,where extensive areas of dilated ducts and hyperplasias full ofmilk secretions were observed (Fig. 1F–G). These results indicatethat short-term in vivo activation of RANK signaling is not suffi-cient to change the growth of established tumors, but appears toprevent secretory differentiation of tumor cells.

The putative benefit of pharmacologic RANKL inhibitionwith RANK-Fc in tumor recurrence was interrogated (Fig. 2A;ref. 7). No significant differences in tumor growth or thefrequency of apoptotic cells were observed after RANK-Fctreatments (Fig. 2B and C). However, LDAs revealed thattumor cells pretreated with RANK-Fc at passage 1 showed a10-fold decrease in tumor-initiating ability, whereas RANK-Fctreatment only at passage 2 did not significantly change tumor-initiating cell (TIC) frequency (Fig. 2D). Concomitantly, theability to form secondary tumorspheres was significantlyimpaired in cells derived from the RANK-Fc-pretreated pool(Fig. 2E), consistent with a reduction in the CSC population

(19). These results demonstrate that pretreatment with RANK-Fc reduces tumor-initiating ability, and suggests that in clinics,RANKL inhibition may reduce the risk of relapse by depletingthe population of CSCs.

Pharmacological inhibition of RANK signaling induceslactogenic differentiation of tumor cells

To investigate the molecular mechanism underlying the reduc-tion in tumor-initiating ability, we analyzed global gene expres-sion profiles from all RANK-Fc treatment arms. Genes induced byRANK-Fc included milk proteins such as Pip, caseins, Wap, or Lpl,which are expressed during differentiation of mammary cells intomilk-secreting alveoli (21) and multiple members of the secre-toglobin family (Scgb1b27, Scgb1b30, Scgb2b2), which are asso-ciatedwith differentiation and low risk of relapse in human breastcancer (Table S1 and Fig. 3A; refs. 3, 27). In fact, genes upregulatedduring lactation (21) were significantly overexpressed in theRANK-Fc-treated tumors (Fig. 3A–B). Upregulation of Csn2, Pip,Scgb1b27, and Scgb2b27 mRNA was confirmed in the RANK-Fc-treated tumors (Fig. 3C). Immunostaining with an anti-milkantibody confirmed that RANKL inhibition induced differentia-tion of late-adenocarcinoma cells into milk secreting cells (Fig.3D). Conversely, tumor acini cultured with RANKL showed lowerCsn2, Pip, and Scgb2b27 expression (Fig. 3E), in correlation withthe reduced milk protein found in vivo (Fig. 1F–G). These resultsdemonstrate that pharmacological inhibition of RANK signalingin PyMT tumor-bearing mice promotes tumor cell differentiationinto an apocrine, milk-secreting phenotype that mimics mam-mary lactogenesis, concomitantly with the reduction in tumor-initiating ability.

RANK deletion increases tumor latency, decreases tumorincidence, and impairs lung metastasis in MMTV-PyMT mice

Genetic deletion of RANK in the MMTV-PyMT backgroundsignificantly delayed tumor onset and reduced tumor incidence(Fig. 4A–B and Supplementary Fig. S4A). In accordance with theirmultifocal origin (24), PyMT;RANKþ/þ palpable lesions showedmultiple stages of tumor progression, whereas one predominantstage was found throughout the whole PyMT;RANK�/� palpablemass (Supplementary Fig. S4B-C). Accordingly, the number ofpreneoplasic lesions quantified in mammary gland wholemounts was significantly reduced in PyMT;RANK�/� comparedwith control mice (Fig. 4C). PyMT;RANK�/� lesions containedextensive areas of early and/or late carcinoma, indicating thattumors can progress to the invasive stage in the absence of RANK.However, most of PyMT;RANK�/- mice were devoid of lungmetastasis, whereas all PyMT;RANKþ/þ mice with early/late car-cinomas developed lungmetastasis, and several showed30 to 200metastatic foci per lung (Fig. 4D). Thus, RANK deletion increasestumor latency, decreases tumor incidence, and impairs lungmetastasis in the MMTV-PyMT tumor-prone model.

Figure 5.RANK-null tumors contain fewer tumor andmetastasis-initiating cells have enhanced apoptosis and aremore sensitive to docetaxel.A, latency to tumor formation ofPyMT;RANKþ/þ andPyMT;RANK�/� tumor cells orthotopically implanted inWTandRANK�/� syngeneicmice. Twenty-two tumors fromeachgroupwere quantified.B, representative pictures of cleaved caspase-3 staining in transplants. C, percentage of nonviable areas versus total tumor area and of cleaved caspase-3–positivenuclei in transplants. Each dot represents one tumor. D, relative tumor volume (length � width/100) of PyMT;RANKþ/þ and PyMT;RANK�/� tumors treated withdocetaxel (25 mg/kg) twice per week. #, docetaxel doses. E, number of tertiary tumorspheres. Each bar represents four tumors. F–G, tumor-initiating(F) and metastasis-initiating frequencies (G; with confidence intervals) and x2 values. Cells from two tumors per genotype were pooled for injections andmetastasis was scored after 8 weeks. H, absolute number of lung metastatic foci. Each dot represents the lung of one mouse. A, E, H, mean, SEM, and t test(F test for H) probabilities are shown (� , 0.01 < P < 0.05; �� , 0.001 < P < 0.01; ��� , 0.001 < P <0.0001; ����, P < 0.0001).

Therapeutic RANKL Inhibition Reduces Cancer Stem Cells

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 5863

A

P < 0.001

Enric

hmen

t sco

re

0.3

0

0.2

0.5

Pre&post RANK-FCOverexpressed

CTRLOverexpressed

0

0.1

0.4

CTRL versus pre&post RANK-FCMature luminal (upregulated subset)

P < 0.001

Enric

hmen

t sco

re

0.2

0

0.1

0.4

0

0.3

Mammary stem cell (downregulated subset)

CTRLOverexpressed

Pre&post RANK-FCOverexpressed

CTRL versus pre&post RANK-FC

SPDEFDMBT1WNK4PPP1R1BCREB3L4ATP6V0E2GABRPSORL1HOMER2LRRC26etc.

SPDEFTSPAN1WNK4ATP6V0E2MEIS1FAAHFOXA1SLC7A4ZFHX2TSPAN13etc.

D

Tfap2b Tfap2c0.0

0.1

0.2

0.3

Tfap2a

0.00

0.05

0.10

0.15

0.20

Tfap2b

0.0000

0.0001

0.0002

0.0003

0.0004

Tfap2c

WT Mammary gland

RANK−/− mammary gland

Cm

RN

A R

elat

ive

expr

essi

on

** ** *

RANKL 24 h

Fold

cha

nge

–15

–10

–5

0

Tfap2a

***

B

TNFRSF11B (204933_s_at) I

0 50 100 150

0.0

0.2

0.4

0.6

0.8

1.0

High expressionLow expression

TFAP2B (X215686_x_at)

P = 0.001

0.0

0.2

0.4

0.6

0.8

1.0

Time to relapse (months)

Pro

porti

on d

ista

nt m

etas

tasi

s fr

ee

Pro

porti

on d

ista

nt m

etas

tasi

s fr

ee

High expressionLow expression

1500 50 100

P = 0.002

Time to relapse (months)

Enric

hmen

t sco

re P = 0.0060.2

0

0.1

0

0.3

CTRLOverexpressed

Pre&post RANK-FCOverexpressed

CTRL versus pre&post RANK-FCTFAP2C Targets (upregulated sub-set)

SPDEFFAM174BPLCB4TSPAN1VAV2EPS8FBP1SIDT1ITGB4FOXA1etc.

CT

shTfap

2b-54

28

shTfap

2b-54

27

shRan

k-698

pSD69-Tfap

2b

plenti6

-Tfap2b

0

50

100

150

200

Rspo1

E RANK-/- vs. RANK+/+

**

*

*

Pre-RANK-Fc vs. non pre-RANK-Fc

Fold

cha

nge

-3

-2

-1

0

1

2

3

4

5

* * **

* *

Tfap

2b p21

Fbp1

Spd

efK

rt8G

ata3

Foxa

1E

sr1

Elf5

Rsp

o1A

xin2

Krt1

4p6

3

F

Tfap

2b p21

Fbp1

Spd

ef

Krt8

Tfap

2c

Krt1

4

p63

Rsp

o1

Fold

cha

nge

-6

-3

0

3

G

0

50

100

150Rspo1

3 14Days

*****

*

H

0

−1

−2

Tfap2b

3 14Days

**** ***

Fold

cha

nge

Fold

cha

nge

CT

shTfap2b

shRank

Tfap2b-overexpressing

Fold

cha

nge

Yoldi et al.

Cancer Res; 76(19) October 1, 2016 Cancer Research5864

RANK loss in tumor cells depletes the tumor and metastasis-initiating cell pools and increases apoptosis and sensitivity todocetaxel

To rule out the progesterone/RANKL-mediated effects acting inearly tumorigenesis (7) and the influence of the RANK-nullmicroenvironment (17), PyMT;RANK�/� and PyMT;RANKþ/þ

tumors cells, isolated from established carcinomas were ortho-topically implanted in syngeneic wild-type (WT) females. PyMT;RANK�/� tumor cells showed a significantly longer latency totumor formation than did PyMT;RANKþ/þ tumor cells, indicatinga tumor cell autonomous defect (Fig. 5A). Longer latency was alsoobserved when PyMT;RANKþ/þ tumor cells were implanted inRANK null mice compared with WT, but no synergic effect afterimplantation of PyMT;RANK�/� in RANK null hosts was found(Fig. 5A).

PyMT;RANK�/� tumors growing in WT hosts contained moreapoptotic cells and extensive nonviable areas relative to PyMT;RANKþ/þ tumors (Fig. 5B–C). No significant differences in tumorcell survival were observed when the same tumor, either PyMT;RANKþ/þ or PyMT;RANK�/�, was implanted on WT and RANKnull hosts supporting a tumor cell intrinsic mechanism (Fig. 5C).This demonstrates that tumor cell survival is impaired in theabsence of RANK, which may contribute to the delayed tumorformation observed. Moreover, the absence of RANK on tumorcells sensitized tumors to docetaxel (Fig. 5D).

Next, we aimed to determine whether loss of RANK signalingexclusively on tumor cells reduced the CSC pool as observed afterRANKL inhibition. PyMT;RANK�/� tumor cells gave rise to lesstumorspheres than controls, independently of the initial host(Fig. 5E), highlighting an extenuation of a self-renewal capabilitythat is tumor cell-autonomous. LDA in WT hosts also revealed asignificant reduction in the frequency of TICs in the absence ofRANK (Fig. 5F). PyMT;RANKþ/þ tumor cells efficiently colonizedthe lung of Foxn1numice and abundantmetastatic foci were found(Fig. 5G–H). Strikingly, in the PyMT;RANK�/� pool a 10-folddecrease in the frequency of metastasis-initiating cells (MIC) wasobserved with very fewmetastatic foci (Fig. 5G–H), implying thatRANK expression in tumor cells is determinant for metastasis.Thus, RANK loss in advanced adenocarcinomas depleted the poolof tumor and MICs, decreased survival and sensitized tumors todocetaxel.

RANKL negatively regulates the Ap2 transcription factors,drivers of luminal differentiation, and induces Rspo1

To further understand the molecular mechanism underlyingtumor cell differentiation after RANKL inhibition, we focusedon genes specifically induced in tumors that received RANK-Fc

treatment at passage 1 such as Tfap2b (Supplementary Table S1).The AP2 transcription factor family is a set of retinoic acidinducible genes that governs the luminal epithelial phenotypein mammary development and carcinogenesis (28, 29) andwhose expression is associated with survival (30, 31). Consistentwith a tumor cell-luminal differentiation phenotype, GSEAanalyses of the genes that characterize mammary differentia-tion hierarchy (20), revealed that the mature luminal upregu-lated set and the MaSC downregulated set, were overexpressedon the RANK-Fc-treated tumors (Fig. 6A). Spdef, which alsopromotes luminal differentiation (32) was the top gene in theseassociations (Fig. 6A). Genes upregulated by TFAP2C in humanbreast cancer cells (22) were significantly overexpressed inRANK-Fc-treated tumors (Fig. 6B). Moreover, an increase inTfap2a and Tfap2b was observed in the RANK null mammaryepithelia and RANKL treatment significantly reduced theTfap2a, Tfap2b, and Tfap2c mRNA expression in tumor culturesof PyMT cells (Fig. 6C–D and G).

Gene expression analysis in the pre-RANK-Fc-treated tumorsconfirmed upregulation of Tfap2b, the luminal genes Spdef andFbp1 and cdkn1a/p21 (known to be induced by Tfap2; ref. 29) anddownregulation of the basal genes p63, Krt14. No significantchanges were detected between groups in Krt8, Foxa1, Gata3, Esr1,Elf5, and Rspo1 (Fig. 6E). Higher levels of Tfap2b, Tfap2c, and p21and lower levels ofKrt14 andRspo1were found in PyMT;RANK�/�

tumor cells isolated from transplants as compared to controls(Fig. 6F). R-spondin1 (Rspo1), aWnt agonist that has been showntobe expressed on luminal progenitors andmediate RANK-drivenexpansion ofmammary progenitors in the healthy gland (33, 34),was strongly induced by RANKL on PyMT acini cultures (Fig. 6G).Tfap2b-overexpressing and knockdown PyMT tumor cells wereobtained as little is known about the specific role of Tfap2b onmammary tumors (Supplementary Fig. S5A and S5B). Geneexpression analyses confirmed that Spdefwas positively regulatedby Tfap2b (Supplementary Fig. S5C). Analyses of PyMT tumoracini cultured for 2 weeks revealed that RANKL led to an increaseon acini size irrespectively of Tfap2b expression (SupplementaryFig. S6A–B). However, in Tfap2b-overexpressing PyMT acinitreated with RANKL, Rspo1 mRNA expression was 30% lowerthan in RANKL-treated control acini. Conversely, in shTfap2bPyMT acini treatedwith RANKL,Rspo1 expressionwas 50%higherthan in controls, demonstrating that Tfap2b interfered withRANKL-driven increase in Rspo1 (Fig. 6H). Rspo1 expressiondecreased, whereas Tfap2b and Spdef expression increased inPyMT tumor cells infected with shRANK, further supportingthat RANK pathway negatively regulates luminal differentiation(Supplementary Fig. S5D).

Figure 6.RANK loss or inhibition induces the expression of AP2 transcription factors and reduces Rspo1. A and B, GSEA graphical outputs for the association betweenmammarymature luminal (upregulated genes) andmammary stem (downregulated genes) cells gene sets (A) and TFAP2C upregulated genes sets in human breastcancer (B) and RANK-Fc treatment. The top genes contributing to the association are listed. C, fold changes in mRNA expression of indicated genes in PyMT tumoracini cultures treated with RANKL for 24 h relative to untreated cultures. Each bar is representative of three tumors. D,mRNA expression levels of indicated genesrelative to Krt8 in PyMT RANKþ/þ and PyMT;RANK�/� mammary glands. Each bar is representative of three mammary glands. E, fold changes in mRNAexpression of indicated genes in RANK-Fc treated PyMT tumors at passage 1 relative to expression in the other treatment arms. Each bar is representative of sixtumors. F, fold changes in mRNA expression of indicated genes in PyMT;RANK�/� relative to expression in PyMT;RANKþ/þ sorted tumor cells. Each bar isrepresentative of three-four independent tumors. G, fold change of Rspo1 and Tfap2b mRNA expression levels in PyMT acini tumor acini cultured with RANKLfor 3 and 14 days relative to untreated cultures. H, relative induction of Rspo1 mRNA expression in Tfap2b-knockdown or -overexpressing PyMT tumoracini culturedwith RANKL for 14 days relative to the induction in RANKL-treated controls (normalized as 100%). Induction ofRspo1mRNA in Rank-Knockdown PyMTacini is included. I, association between TFAP2B and TNFRSF11B tumor expression and distant metastasis in lymph-node negative breast cancer patients (GSE2034).Graphs show the proportion of distant metastasis-free patients over time (months) and are stratified according to the first (low expression) or the third (highexpression) tertiles. C, D, E, F, mean, SEM and t-test statistics are shown. (� , 0.01 < P < 0.05; �� , 0.001 < P < 0.01; ��� , 0.001 < P < 0.0001; ����, P < 0.0001).

Therapeutic RANKL Inhibition Reduces Cancer Stem Cells

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 5865

To investigate the clinical relevance of TFAP2B, we analyzed anexpression dataset from lymph-node negative breast cancerpatients that developed distant metastasis (35). The expressionof TFAP2B was found to be significantly associated with theabsence of distant metastasis: Cox regression HR ¼ 0.25; 95%confidence interval (CI), 0.11–0.57; P ¼ 0.001 (Fig. 6I). Similarresults were observed in tumors with a luminal phenotype (ERþ):HR¼ 0.24; 95%CI, 0.09–0.63, P¼ 0.004, and the same trend for(ER�) tumors: HR ¼ 0.24; 95% CI, 0.04–1.29; P ¼ 0.09. Con-sistent with the proposed cancer-promoting role for enhancedRANK signaling, associations with relapse free were observed forTNFRSF11B (OPG), the canonical negative regulator of the RANKpathway: HR ¼ 0.49; 95% CI, 0.31 – 0.78; P ¼ 0.002 (Fig. 6I); inluminal tumors (ERþ): HR ¼ 0.33; 95% CI, 0.17–0.62; P ¼0.0006. Accordingly, TFAP2B and TNFRSF11B were found to besignificantly coexpressed (Pearson's correlation coefficient ¼0.14, P ¼ 0.018). Together these data indicate that RANKLinhibition leads to tumor differentiation, metastasis impairment,and good prognosis.

RANK signaling inhibition depletes the pool of Sca1� TICsNext, we aimed to identify the CSC population regulated by

RANK in our models, which remains elusive in the MMTV-PyMTmodel (36–38). The levels of CD49f, CD49b, CD61, and CD90within epithelial cells were comparable for all RANK-Fc treatmentarms; in contrast, Sca1þ/hi cells were more abundant in tumorspretreated with RANK-Fc, which show a lower tumor-initiatingability (Fig. 7A and B). In the normal mammary gland, Sca1þ

identifies a population enriched in ERþ/PRþ luminal mature cells(37). However, we could not detect an increase in PRþ cellsafter RANK-Fc treatment (Fig. 7C). Similarly, an increase in theSca1þ/hi population, but none of the othermarkers, was found inPyMT;RANK�/� tumors as compared to controls (Fig. 7D). Sec-ondary tumorspheres of Sca1�/lo tumor cells were larger and fivetimes more numerous as those of Sca1þ/hi cells (Fig. 7E–F).Strikingly, LDA assays revealed the TIC frequency is significantlyenhanced by 200-fold in Sca1�/lo comparedwith Sca1þ/hi tumorcells (Fig. 7G), indicating that the Sca1�/lo population is enrichedin CSCs. Altogether these results demonstrate that RANK loss orRANKL inhibition reduced the frequency of the Sca1�/lo CSCpopulation.

DiscussionThe work presented here reveals a central role of RANK

signaling promoting recurrence and metastasis in aggressivebreast tumors, providing a rationale for additional therapeuticapplications of RANK inhibitors beyond its current use for themanagement of skeletal-related events. We found that consti-tutive deletion of RANK in MMTV-PyMT mice increases tumorlatency and decreases tumor and lung metastasis incidence, asobserved in MMTV-neu mice upon RANK-Fc preventive treat-ment (7), reinforcing the role of RANK signaling in early stagesof tumorigenesis (8).

Our previous data showed that enhanced RANK activationpromotes stemness in human and mouse mammary epithelia,leading to the accumulation of MaSC and progenitors (11, 12,39). Importantly, now we demonstrate that inhibition of RANKsignaling reduces CSC in invasive mammary tumors decreasingrecurrence andmetastasis, and induces tumor cell differentiation.LDA assays aim to mimic occult disease that remains in breast

cancer patients after surgery. Our results suggest that neoadjuvantRANKL inhibition may be more efficient in reducing recurrenceandmetastasis than adjuvant treatment, as a significant reductionin the CSC population was observed on tumors treated at passage1.Alongwithour previous data demonstrating that overactivationof RANK signaling at midgestation disrupts lactogenesis (12, 39),current results suggest that RANK signaling regulates the balancebetween self-renewal and differentiation not only during mam-mary gland development but also in breast adenocarcinomas.

The impaired tumor and metastasis initiation ability observedin RANK null tumor cells growing in WT hosts demonstrates thattumor cell intrinsic mechanisms mediate the observed reductionin CSC. However, we cannot discard that tumor cell extrinsicmechanisms induced by RANK signaling inhibition in the micro-environment can also contribute to reduce recurrence (40).

Mechanistically, we demonstrate that RANK signaling nega-tively regulates the AP2 transcription factor family that canmediate retinoic acid responsiveness (41, 42). TFAP2A functionsas a tumor suppressor in several solid tumors including breastcancer (43, 44). Overexpression of Tfap2a and Tfap2c mimicsthe mammary phenotype of RANK null mice (5, 45, 46). Tfap2aand Tfap2c maintain the luminal phenotype (28, 29) and neg-atively regulate cancer stem cell markers (28). Although little isknown about TFAP2B in mammary epithelia, our results suggestthat, similarly to other members of the family, TFAP2B promotesluminal differentiation and is associated with good prognosis.The positive correlation between the RANKL inhibitor, OPG, andTFAP2B expression in human breast tumors and their associationwithmetastasis-free phenotype support the clinical implication ofour findings.

Although enhanced Tfap2b expression alone cannot preventRANKL-driven increased in acinar size, it interferes with theinduction of the Wnt agonist Rspo1. Rspo1 together with Wnt4promote MaSC self-renewal and Rspo1 rescues some of themammary developmental defects in RANK null epithelia (33,34). Similarly to our previous results onmammary epithelial cellsat midgestation where RANKL induces the expression of Rspo1,leads to the expansion of basal and bipotent cells and preventslactogenic differentiation (39), we now observe that on PyMTtumor acini, RANK pathway also enhances Rspo1 and interfereswith differentiation. Our results evidence a complex regulatoryloop between RANK, Tfap2, and Rspo1 underlying the reductionin the CSC pool observed upon RANK pathway inhibition (Fig.7H). Further experiments will be required to clarify their contri-bution to the protumorigenic role of RANK in cancer. Sca-1/Ly6Ais found in the luminal differentiated ERþ/PRþ cell cluster andaccording to our data it is likely to be induced by Tfap2, whereasRspo1 is expressedon luminal Sca1�progenitor cells (33, 34). Thedecrease in Sca1� cells upon RANK loss or inhibition and theirenhancedmammosphere-forming and tumor-initiating potentialdemonstrate that this population is enriched in CSCs in the PyMTtumors, as shown in the MMTV-wnt model (47). A negativeregulationof Sca1þbyRANKhasbeenobserved duringmammarygland development (12, 34). The relevance of Sca1/Ly6a as a CSCmarker in human luminal adenocarcinomas deserves furtherinvestigation.

Mortality in breast cancer is due to tumor recurrence andmetastasis, which is driven by surviving CSC. RANKL inhibitors,although unable to reduce tumor growth, can be used as differ-entiation therapy of CSC (Fig. 7H). Moreover, RANK null tumorcells aremore susceptible to taxanes than RANK-expressing tumor

Cancer Res; 76(19) October 1, 2016 Cancer Research5866

Yoldi et al.

F

N o

f sph

eres

Sca1+

/hi

Sca1-/

lo0

100

200

300

400

500Tumor spheres

**

100 μm

A

Sca1-/lo

Sca1+/hi

*

CTRL

Pre-RANK-Fc

Post-RANK-Fc

CD49f Sca1+/hi

CD61 CD90

E

B

CTRL

Sca1-/lo Sca1+/hi

Sca1-/lo Sca1+/hi

Sca-1

Pre-RANK-Fc

Group N cells N fat pad N tumors

Sca1+/hi

100,000 4 310,000 4 31,000 4 2100 4 3

100,000 4 410,000 4 41,000 4 4100 4 3

G

Sca1-/lo

TICs: Tumor ini�a�ng cell frequency (ELDA)Sca1+/hi: 1/ 15767.4 (CI 56,295–4,416.4)

1/ 72.6 (CI 246–21.7)

P value: 1.04E–11 χ = 46.32

Sca1-/lo:

D

RANK+/+ tumor in WT host

RANK–/– tumor in WT host

0

50

100

150

0

10

20

30

0

10

20

30

40

0

20

40

60

80

100*CD49f Sca1+/hi CD61

CD49b

0

1

2

3

4

5 CD90

C

PR

Pre & post-RANK-Fc

CTRL

50 μm

PR

Tumor cell

Cancer stem cell

RANK

Tumor cell differentiation & reduced survival

Recurrence& metastasis

RANK-ON RANK-OFF

Milk genesTfap2Sca1+/hi

RANKLCSC expansion

Recurrence& metastasis

TherapeuticRANK inhibition

HDifferentiated tumor cell

CD49b

Rspo1Sca1–/lo

100

80

60

40

20

0

80

60

40

20

0

20

15

10

5

0

% o

f Cel

ls in

Lin

− C

D24

+%

of C

ells

in L

in−

CD

24+

% o

f Cel

ls in

Lin

− C

D24

+

% C

ells

in L

in− C

D24

+%

Cel

ls in

Lin

− CD

24+

% o

f C

ells

in L

in−

CD

24+

% o

f Cel

ls in

Lin

− C

D24

+

30

20

10

0

15

10

5

0Pre & post-RANK-Fc

Figure 7.

RANK-Fc pretreatment reduces the Sca1� tumor cell population. A, frequency of the indicated populations within tumor CD45-CD31-CD24þ cells. Each barrepresents data from four mice in two independent experiments. B, representative histograms of Sca1þ/hi and Sca1�/lo populations. C, representative images ofPR immunostaining. D, frequency of the indicated populations within tumor cell transplants. Each bar is representative of 3-5 tumors. E and F, representativeimages (E) and number (F) of tumorspheres derived from FACS-sorted Sca1þ/hi and Sca1� tumor cells. Each bar is representative of two tumors quantified intriplicates. G, tumor-initiating frequency (with confidence intervals), x2 values, and associated probabilities. H, graphical abstract indicating the multiple effectsobserved after therapeutic inhibition of RANK pathway in MMTV-PyMT tumors. A, D, F, mean, SEM and t-test statistics are shown (�� , 0.001 < P < 0.01).

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 5867

Therapeutic RANKL Inhibition Reduces Cancer Stem Cells

cells, supporting the use of neoadjuvant RANKL inhibitors in theclinical setting to reduce the frequency of tumor relapse andmetastasis and to increase sensitivity to chemotherapy. FDA-approved RANKL inhibitors are currently used in clinic for themanagement of skeletal-related events, therefore patients mayquickly benefit from this therapeutic strategy to combat advancedbreast cancer.

Disclosure of Potential Conflicts of InterestW.C. Dougall has ownership interest (including patents) in a patent. No

potential conflicts of interest were disclosed by the other authors.

Authors' ContributionsConception and design: W.C. Dougall, E. Gonz�alez-Su�arezDevelopment of methodology: G. Yoldi, P. Pellegrini, E.M. Trinidad,A. Cordero, E. Gonz�alez-Su�arezAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): G. Yoldi, P. Pellegrini, E.M. Trinidad, A. Cordero,J. Gomez-Miragaya, E. Gonz�alez-Su�arezAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): G. Yoldi, P. Pellegrini, E.M. Trinidad, A. Cordero,J. Gomez-Miragaya, J. Serra-Musach, P. Mu~noz, M.-A. Pujana, E. Gonz�alez-Su�arezWriting, review, and/or revision of the manuscript: G. Yoldi, P. Pellegrini,E.M. Trinidad, A. Cordero, J. Gomez-Miragaya, W.C. Dougall, P. Mu~noz,M.-A. Pujana, L. Planelles, E. Gonz�alez-Su�arez

Administrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): P. Pellegrini, J. Serra-Musach, E. Gonz�alez-Su�arezStudy supervision: E. Gonz�alez-Su�arez

AcknowledgmentsWe thank L. Alcaraz and D. Amoros (Bioarray) for microarray analyses; N.E.

Hynes, M. Glukhova's group, M. Bentires-Alj, S. Duss, and K. Jin for sharingprotocols and reagents; G. Boigues, A. Villanueva, E Casta~no, and the IDIBELLanimal facility for their technical assistance.

Grant SupportThis work was supported by grants to E. Gonz�alez-Su�arez by MINECO

and ISCIII [SAF2011-22893 , SAF2014-55997; PIE13/00022, cofunded byFEDER funds/European Regional Development Fund (ERDF)—a way tobuild Europe-], by the Susan Komen Foundation CCR13262449, ConcernFoundation, and by Fundaci�o La Marat�o; to M.A. Pujana from ISCIII (PI12/01528) and AGAUR (SGR 2014-364), and to L. Planelles from ISCIII(PI10/01556). P Pellegrini and J. Gomez-Miragaya are recipients of anFPI grant from MINECO. The funders had no role in study design, datacollection and analysis, decision to publish, or preparation of themanuscript.

Received October 5, 2015; revised June 15, 2016; accepted July 7, 2016;published OnlineFirst August 1, 2016.

References1. Li X, Lewis MT, Huang J, Gutierrez C, Osborne CK, Wu MF, et al. Intrinsic

resistance of tumorigenic breast cancer cells to chemotherapy. J Natl CancerInst 2008;100:672–9.

2. TallmanMS, Andersen JW, Schiffer CA, Appelbaum FR, Feusner JH, OgdenA, et al. All-trans-retinoic acid in acute promyelocytic leukemia. N Engl JMed 1997;337:1021–8.

3. Dontu G, Abdallah WM, Foley JM, Jackson KW, Clarke MF,Kawamura MJ, et al. In vitro propagation and transcriptionalprofiling of human mammary stem/progenitor cells. Genes Dev2003;17:1253–70.

4. Beleut M, Rajaram RD, Caikovski M, Ayyanan A, Germano D, Choi Y,et al. Two distinct mechanisms underlie progesterone-induced prolif-eration in the mammary gland. Proc Natl Acad Sci U S A 2010;107:2989–94.

5. Fata JE, Kong YY, Li J, Sasaki T, Irie-Sasaki J, Moorehead RA, et al. Theosteoclast differentiation factor osteoprotegerin-ligand is essential formammary gland development. Cell 2000;103:41–50.

6. Tanos TSG, Echeverria PC, Ayyanan A, GutierrezM,Delaloye JF, RaffoulW,et al. Progesterone/RANKL is a major regulatory axis in the human breast.Sci Transl Med 2013;5:182ra155.

7. Gonzalez-Suarez E, Jacob AP, Jones J, Miller R, Roudier-MeyerMP, Erwert R, et al. RANK ligand mediates progestin-inducedmammary epithelial proliferation and carcinogenesis. Nature 2010;468:103–7.

8. Gonzalez-Suarez E. RANKL inhibition: a promising novel strategy forbreast cancer treatment. Clin Transl Oncol 2011;13:222–8.

9. Schramek D, Leibbrandt A, Sigl V, Kenner L, Pospisilik JA, Lee HJ, et al.Osteoclast differentiation factor RANKL controls development of proges-tin-driven mammary cancer. Nature 2010;468:98–102.

10. Gonzalez-Suarez E, Branstetter D, Armstrong A, Dinh H, Blumberg H,Dougall WC. RANK overexpression in transgenic mice with mouse mam-mary tumor virus promoter-controlled RANK increases proliferation andimpairs alveolar differentiation in the mammary epithelia and disruptslumen formation in cultured epithelial acini. Mol Cell Biol 2007;27:1442–54.

11. Palafox M, Ferrer I, Pellegrini P, Vila S, Hernandez-Ortega S, UrruticoecheaA, et al. RANK induces epithelial-mesenchymal transition and stemness inhumanmammary epithelial cells and promotes tumorigenesis andmetas-tasis. Cancer Res 2012;72:2879–88.

12. Pellegrini P, Cordero A, Gallego MI, Dougall WC, Purificacion M, PujanaMA, et al. Constitutive activation of RANK disrupts mammary cell fateleading to tumorigenesis. Stem Cells 2013;31:1954–65.

13. Pfitzner BM, Branstetter D, Loibl S, Denkert C, Lederer B, SchmittWD, et al.RANK expression as a prognostic and predictive marker in breast cancer.Breast Cancer Res Treat 2014;145:307–15.

14. Santini D, Schiavon G, Vincenzi B, Gaeta L, Pantano F, Russo A, et al.Receptor activator of NF-kB (RANK) expression in primary tumors associ-ates with bone metastasis occurrence in breast cancer patients. PLoS One2011;6:e19234.

15. Guy CT, Cardiff RD, Muller WJ. Induction of mammary tumors byexpression of polyomavirusmiddle T oncogene: a transgenicmousemodelfor metastatic disease. Mol Cell Biol 1992;12:954–61.

16. Maglione JE, Moghanaki D, Young LJ, Manner CK, Ellies LG, Joseph SO,et al. Transgenic Polyoma middle-T mice model premalignant mammarydisease. Cancer Res 2001;61:8298–305.

17. Dougall WC, Glaccum M, Charrier K, Rohrbach K, Brasel K, De Smedt T,et al. RANK is essential for osteoclast and lymph node development. GenesDev 1999;13:2412–24.

18. Smalley MJ. Isolation, culture and analysis of mouse mammary epithelialcells. Methods Mol Biol 2010;633:139–70.

19. Dontu G, Wicha MS. Survival of mammary stem cells in suspensionculture: implications for stem cell biology and neoplasia. J MammaryGland Biol Neoplasia 2005;10:75–86.

20. Lim E, Wu D, Pal B, Bouras T, Asselin-Labat ML, Vaillant F, et al. Tran-scriptome analyses of mouse and human mammary cell subpopulationsreveal multiple conserved genes and pathways. Breast Cancer Res 2010;12:R21.

21. Anderson SM, Rudolph MC, McManaman JL, Neville MC. Key stages inmammary gland development. Secretory activation in the mammary gland:it's not just about milk protein synthesis! Breast Cancer Res 2007;9:204–18.

22. Woodfield GW, Chen Y, Bair TB, Domann FE, Weigel RJ. Identification ofprimary gene targets of TFAP2C in hormone responsive breast carcinomacells. Genes Chromosomes Cancer 2010;49:948–62.

23. Hu Y, Smyth GK. ELDA: extreme limiting dilution analysis for comparingdepleted and enriched populations in stem cell and other assays. J Immu-nol Methods 2009;347:70–8.

24. Lin EY, Jones JG, Li P, Zhu L, Whitney KD, Muller WJ, et al. Progression tomalignancy in the polyoma middle T oncoprotein mouse breast cancer

Yoldi et al.

Cancer Res; 76(19) October 1, 2016 Cancer Research5868

model provides a reliable model for human diseases. Am J Pathol2003;163:2113–26.

25. Barcellos-Hoff MH, Aggeler J, Ram TG, Bissell MJ. Functionaldifferentiation and alveolar morphogenesis of primary mammarycultures on reconstituted basement membrane. Development 1989;105:223–35.

26. Lee GY, Kenny PA, Lee EH, Bissell MJ. Three-dimensional culturemodels of normal and malignant breast epithelial cells. Nat Methods2007;4:359–65.

27. Span PN, Waanders E, Manders P, Heuvel JJ, Foekens JA, Watson MA, et al.Mammaglobin is associated with low-grade, steroid receptor-positivebreast tumors from postmenopausal patients, and has independent prog-nostic value for relapse-free survival time. J Clin Oncol 2004;22:691–8.

28. Bogachek MV, Chen Y, Kulak MV, Woodfield GW, Cyr AR, Park J M, et al.Sumoylation pathway is required to maintain the basal breast cancersubtype. Cancer Cell 2014;25:748–61.

29. Cyr AR, Kulak MV, Park JM, Bogachek MV, Spanheimer PM, WoodfieldGW, et al. TFAP2C governs the luminal epithelial phenotype in mammarydevelopment and carcinogenesis. Oncogene 2014;34:436–44.

30. Bar-Eli M. Role of AP-2 in tumor growth and metastasis of humanmelanoma. Cancer Metastasis Rev 1999;18:377–85.

31. Gee JM, Robertson JF, Ellis IO, Nicholson RI, Hurst HC. Immunohisto-chemical analysis reveals a tumour suppressor-like role for the transcrip-tion factor AP-2 in invasive breast cancer. J Pathol 1999;189:514–20.

32. Buchwalter G, Hickey MM, Cromer A, Selfors LM, Gunawardane R N,Frishman J, et al. PDEF promotes luminal differentiation and acts as asurvival factor for ER-positive breast cancer cells. Cancer Cell 2013;23:753–67.

33. Cai C, Yu QC, Jiang W, Liu W, Song W, Yu H, et al. R-spondin1 is a novelhormone mediator for mammary stem cell self-renewal. Genes Dev2014;28:2205–18.

34. Joshi PA,WaterhousePD, KannanN,Narala S, FangH,DiGrappaMA, et al.RANK signaling amplifiesWNT-responsivemammary progenitors throughR-SPONDIN1. Stem Cell Reports 2015;5:31–44.

35. Wang Y, Klijn JG, Zhang Y, Sieuwerts AM, Look MP, Yang F, et al. Gene-expression profiles to predict distant metastasis of lymph-node-negativeprimary breast cancer. Lancet 2005;365:671–9.

36. Malanchi I, Santamaria Martinez A, Susanto E, Peng H, Lehr HA, DelaloyeJF, et al. Interactions between cancer stem cells and their niche governmetastatic colonization. Nature 2011;481:85–9.

37. SleemanKE, KendrickH, RobertsonD, IsackeCM,AshworthA, SmalleyMJ.Dissociation of estrogen receptor expression and in vivo stem cell activity inthe mammary gland. J Cell Biol 2007;176:19–26.

38. Vaillant F, Asselin-Labat ML, Shackleton M, Forrest NC, Lindeman GJ,Visvader JE. The mammary progenitor marker CD61/beta3 integrin iden-tifies cancer stem cells in mouse models of mammary tumorigenesis.Cancer Res 2008;68:7711–7.

39. Cordero A, Pellegrini P, Sanz-Moreno A, Trinidad EM, Serra-Musach J,Deshpande C, et al. Rankl impairs lactogenic differentiation throughinhibition of the prolactin/Stat5 pathway at midgestation. Stem Cells2016;34:1027–39.

40. Gonzalez-Suarez E, Sanz-Moreno A. RANK as a therapeutic target in cancer.Febs J 2016;283:2018–33.

41. Bosher JM, Totty NF, Hsuan JJ, Williams T, Hurst HC. A family of AP-2proteins regulates c-erbB-2 expression in mammary carcinoma. Oncogene1996;13:1701–7.

42. McPherson LA, Woodfield GW, Weigel RJ. AP2 transcription factors reg-ulate expression of CRABPII in hormone responsive breast carcinoma.J Surg Res 2007;138:71–8.

43. McPherson LA, Loktev AV, Weigel RJ. Tumor suppressor activity of AP2al-pha mediated through a direct interaction with p53. J Biol Chem2002;277:45028–33.

44. Scibetta AG,Wong PP, Chan KV, CanosaM,Hurst HC. Dual association byTFAP2A during activation of the p21cip/CDKN1A promoter. Cell Cycle2010;9:4525–32.

45. Jager R, Werling U, Rimpf S, Jacob A, Schorle H. Transcription factor AP-2gamma stimulates proliferation and apoptosis and impairs differentia-tion in a transgenic model. Mol Cancer Res 2003;1:921–9.

46. Zhang J, Brewer S, Huang J, Williams T. Overexpression of transcriptionfactor AP-2alpha suppresses mammary gland growth and morphogenesis.Dev Biol 2003;256:127–145.

47. Batts TD, Machado HL, Zhang Y, Creighton CJ, Li Y, Rosen JM. Stem cellantigen-1 (sca-1) regulates mammary tumor development and cell migra-tion. PLoS One 2011;6:e27841.

www.aacrjournals.org Cancer Res; 76(19) October 1, 2016 5869

Therapeutic RANKL Inhibition Reduces Cancer Stem Cells