the transcriptional modulator bcl6 as a molecular target for breast cancer therapy

ORIGINAL ARTICLE

The transcriptional modulator BCL6 as a molecular targetfor breast cancer therapySR Walker12 S Liu12 M Xiang1 M Nicolais1 K Hatzi3 E Giannopoulou4 O Elemento4 L Cerchietti3 A Melnick3 and DA Frank12

Inappropriate expression or activation of transcription factors can drive patterns of gene expression leading to the malignantbehavior of breast cancer cells We have found that the transcriptional repressor BCL6 is highly expressed in breast cancer cell linesand its locus is amplified in about half of primary breast cancers To understand how BCL6 regulates gene expression in breastcancer cells we used chromatin immunoprecipitation followed by deep sequencing to identify the BCL6 binding sites on agenomic scale This revealed that BCL6 regulates a unique cohort of genes in breast cancer cell lines compared with B-celllymphomas Furthermore BCL6 expression promotes the survival of breast cancer cells and targeting BCL6 with a peptidomimeticinhibitor leads to apoptosis of these cells Finally combining a BCL6 inhibitor and a signal transducer and activator of transcription3inhibitor provided enhanced cell killing in triple-negative breast cancer cell lines suggesting that combination therapy may beparticularly useful Thus targeting BCL6 alone or in conjunction with other signaling pathways may be a useful therapeutic strategyfor treating breast cancer

Oncogene advance online publication 24 March 2014 doi101038onc201461

INTRODUCTIONOne hallmark of cancer cells is aberrant gene expression leadingto phenotypes promoting malignancy such as proliferationevasion of apoptosis self-renewal and metastasis Inappropriateexpression or activation of transcription factors drives these geneexpression patterns and cancer cells are often dependent on theircontinued activation Understanding how these oncogenictranscription factors regulate gene expression and phenotypiccharacteristics of cancer cells can shed light on tumor pathogen-esis and provide insight into strategies for rationally designedtargeted therapiesSeveral lines of evidence suggest that BCL6 a transcriptional

repressor that is oncogenic in B-cell lymphomas may have a rolein the pathogenesis of breast cancer In addition to maintainingthe survival and blocking the terminal differentiation oflymphoma cells BCL6 has been shown to prevent differentiationof mammary cells1 Through its effects on gene regulation BCL6controls cellular processes including cell cycle progression DNAdamage sensing and protein ubiquitination23 BCL6 is directlyregulated by the oncogenic transcription factor STAT3 which isinappropriately activated in a majority of breast cancers45

Although BCL6 is rarely detected by immunohistochemistry innormal mammary epithelium it is commonly expressed in breastcancers and has been found in 68 of high-grade ductalcarcinomas16 Whether BCL6 regulates gene expression andcontributes to the survival and biological properties of breastcancer cells remain unknownBCL6 binds to canonical DNA sequences in the regulatory

region of target genes Once bound it recruits corepressorcomplexes that introduce repressive chromatin marks therebyrepressing transcription7 For example BCL6 can recruit SMRTNCOR and BCOR through its BTB domain These repressor

complexes are then able to repress a subset of BCL6 target genesinvolved in cellular proliferation and survival8ndash12 BCL6 can alsointeract with MTA3 through its second repression domain (RD2)Through MTA3 BCL6 represses genes involved in terminaldifferentiation1314 Finally BCL6 also recruits CtBP through whichit can repress its own expression1516 Thus BCL6 can mediatecomplex and diverse effects on gene expressionGiven these distinct interactions the activity of BCL6 can be

inhibited in specific ways A peptidomimetic inhibitor RI-BPIcan prevent the recruitment of SMRT NCoR and BCoR by bindingto the BTB domain of BCL6917 Following treatment with RI-BPIBCL6 is still able to bind to DNA but it can no longer repressexpression of genes normally regulated by those complexesImportantly RI-BPI kills diffuse large B-cell lymphoma cell lines andpatient samples917 demonstrating that genes derepressed by RI-BPI are important for the survival of diffuse large B-celllymphoma cells In addition RI-BPI treatment is effective in B-celllymphoma models in vivo while having no apparent toxicity inmice17 These findings suggest that BCL6 is an appealingtherapeutic targetGiven the potential role of BCL6 in the biology of breast cancer

cells we analyzed the genomic binding patterns of BCL6 in breastcancer cell lines We also determined the role of BCL6 on geneexpression and survival of breast cancer cells and assessed itspotential as a therapeutic target in breast cancer

RESULTSBCL6 is amplified in primary breast cancersGiven the suggestion that BCL6 might have a role in breast cancerpathogenesis we wanted to determine if the BCL6 locus wasamplified in primary human breast tumors Using data derived

1Department of Medical Oncology Dana-Farber Cancer Institute Boston MA USA 2Department of Medicine Brigham and Womenrsquos Hospital and Harvard Medical SchoolBoston MA USA 3Departments of Medicine and Pharmacology Weill Cornell Medical College New York NY USA and 4HRH Prince Alwaleed Bin Talal Bin Abdulaziz AlsaudInstitute for Computational Biomedicine and Department of Physiology and Biophysics Weill Cornell Medical College New York NY USA Correspondence Dr DA FrankDepartment of Medical Oncology Dana-Farber Cancer Institute 450 Brookline Avenue M522B Boston 02215 MA USAE-mail david_frankdfciharvardeduReceived 24 July 2013 revised 16 January 2014 accepted 19 January 2014

Oncogene (2014) 1ndash10copy 2014 Macmillan Publishers Limited All rights reserved 0950-923214

wwwnaturecomonc

from The Cancer Genome Atlas we found that 51 of the 790breast tumors analyzed showed amplification of the BCL6 locuswhereas only a small subset (11) showed a reduction of thislocus (Figure 1a) The skewing toward increased copy numbershowed a nominal P-value of 0056 compared with 1000 randomlychosen sites This finding raised the possibility that overexpressionof BCL6 may be important in breast cancer pathogenesis Next weused data from the Cancer Cell Line Encyclopedia (CCLE)18 toanalyze BCL6 copy number and mRNA expression in breast cancercell lines We found that the majority of breast cancer cell linesalso displayed amplification of this locus (P= 001) (Figure 1b) andthis was evident in lines derived from all three subtypes of breastcancer (Supplementary Figure 1a) In addition gt75 of the celllines showed enhanced expression of BCL6 Of the small numberof cell lines showing a loss at this locus most were from triple-negative breast tumors Interestingly however even among thisgroup the majority of lines showed increased BCL6 mRNAexpression suggesting that there is a strong selective pressurefor expression of this gene (Supplementary Figure 1a) To betterunderstand the role BCL6 has in breast cancer we chose a panelof breast cancer cells containing both amplifications and deletionsof this locus and which represented the three major classes ofbreast cancer hormone receptor positive (ER+PR+) Her2ErbB2

amplified (Her2+) and ERminus PRminus Her2minus (triple negative)(Figure 1c) We then analyzed the expression of BCL6 mRNA(Figure 1d) and protein (Supplementary Figure 1b) in these linescompared with non-tumorigenic mammary epithelial lines Wefound that BCL6 was expressed to varying degrees in 10 of the 11breast cancer cells lines analyzed and that there was generally alow expression of BCL6 in non-tumorigenic breast lines comparedwith the breast cancer cell lines (Supplementary Figure 1b)Expression levels did not correlate with a specific breast cancersubtype We further analyzed the levels of nuclear BCL6expression and found that BCL6 was present to varying degreesin the nucleus of all 10 breast cancer cell lines tested (Figure 1e)BCL6 becomes post-translationally modified19 and runs across arange of molecular weights To ensure analysis of the correct bandfrom protein extracts we treated cells with small interfering RNA(siRNA) to BCL6 or control and performed immunoblots to BCL6on nuclear extracts This confirmed that BCL6 is expressed in alleight of the breast cancer cell lines analyzed (Figure 1f) ThusBCL6 is expressed in nearly all breast cancer cell lines tested andthe locus is amplified in approximately half of primary humanbreast tumors and the majority of breast cancer cell lines Wechose to focus on T-47D SK-BR-3 and MDA-MB-468 cells forfurther study as they represent the three different subtypes of

-06

-04

-02

MC

F7

T47

D

ZR

751

MD

AM

B45

3

BT

474

SK

BR

3

BT

549

HS

578T

MD

AM

B23

1

MD

AM

B46

8

0

02

04

06

08

1

12

Co

py

nu

mb

er (

log

2-1)

Amplification51

no change38

Deletion11

TumorNormal

200

180

150

140

120

100

Cou

nt

80

60

40

20

0

MC

F-7

T-4

7D

ZR

-75

MD

A-M

B-4

53

SK

BR

3

BT

-549

Hs5

78T

SU

M-1

59pt

MD

A-M

B-2

31

MD

A-M

B-4

68

Ram

os

ER+PR+ Her2+ Triple negative

BCL6

PARP

siBCL6 - + - + - + - + - + - + - + - +

BCL6

PARP

T47DMCF7

-1

-05

0

05

1

15

2

25

Nor

mal

ized

BC

L6ex

pres

sion

-1

-05

0

05

1

15

Cop

y nu

mbe

r

BCL6

-1 -08 08 1

Copy number (log2 - 1)

-06 06-04 04-02 020

RANDOM LOCI(AVERAGE)

MC

F7

T47

D

ZR

751

BT

474

MD

AM

B45

3

SK

BR

3

BT

549

HS

578T

MD

AM

B23

1

MD

AM

B46

8

MDA231Hs578tBT549ZR75 MDA468SKBR3

Figure 1 BCL6 is expressed in breast cancer Copy number for BCL6 was analyzed using data from (a) the Cancer Genome Atlas for breasttumors (inset contains percentages) or (b) from the CCLE for breast cancer cell lines Copy number (c) and BCL6 gene expression (d) from theCCLE for a subset of breast cancer cell lines is shown (e) Nuclear lysates were generated from the indicated breast cancer cell lines andanalyzed for BCL6 expression Whole-cell extracts of Ramos cells served as a positive control for BCL6 expression PARP served as a control fornuclear protein (f) Nuclear lysates were generated from the indicated cell lines after 72 h of transfection with siRNA to BCL6 (or control) PARPserved as a control for nuclear protein

BCL6 in breast cancerSR Walker et al

2

Oncogene (2014) 1 ndash 10 copy 2014 Macmillan Publishers Limited

breast cancer and also reflect three different levels of BCL6expression

BCL6 binds to unique genomic sites in breast cancer cellsTo determine the function of BCL6 in breast cancer cellswe next sought to identify its target genes in this tumor typeAlthough binding sites for BCL6 have been identified in B-celllymphoma cells23 it was not known whether BCL6 binding siteswould be similar in breast cancer cells We performed chromatinimmunoprecipitation (ChIP) in SK-BR-3 T-47D and MDA-MB-468breast cancer cell lines and first analyzed previously character-ized BCL6 binding sites We found that BCL6 binds to thewell-characterized BCL6 binding site within the BCL6 gene itselfthe so-called region B20 as well as the promoter of the CIS gene(Figure 2a and data not shown) Having confirmed BCL6localization to these sites we then analyzed 12 BCL6 bindingsites that had been defined in lymphoma cells23 We found thatonly two of these sites were bound by BCL6 in only one of threebreast cancer cell lines tested (Figure 2a and data not shown)This suggested that while some of the BCL6 binding sites inbreast cancer cells overlap with those identified in lymphomacells there was likely to be a unique pattern of binding of BCL6in breast cancerTo analyze BCL6 binding in breast cancer on a genomic scale

we performed ChIP followed by deep sequencing (ChIP-seq)Given that we had validated strong binding of BCL6 to four sites inSK-BR-3 cells we chose to perform ChIP-seq in this cell line Usinga P-value of 10minus 6 as a cutoff we identified 4118 BCL6 bindingsites Analysis of the sequences within the peaks using a de novomotif search identified the BCL6 motif (Figure 2b) In addition

analysis of known motifs through the TRANSFAC database alsoidentified the BCL6 binding site as one of the top binding sitesfound within these sequences (P-value = 933 times 10minus 18) Takentogether this suggests that this approach identified bona fideBCL6 binding sites Analysis of BCL6 localization revealed that 42of the binding sites are within introns of genes and 47 of thesites are located in regions that are gt3000 bp from thetranscription start site Only 7 of the sites are within the first3000 bp 5prime from the start site of genes the traditional promoterregions (Figure 2c) Similar distribution of binding has beendescribed for other transcriptional modulators2122 QuantitativeChIP analysis was performed on 20 of the most highly bound sitesidentified by ChIP-seq (with a false discovery rate of 0) All 20 ofthese sites (18 new sites and two known sites) were bound byBCL6 in SK-BR-3 cells with at least twofold binding overbackground (ie genomic regions known not to be bound byBCL6) (Figure 2d) In addition comparison of binding of these sitesin cell lines representing the three major classes of breast cancertypes determined that most but not all of the sites were boundby BCL6 in all three cell lines although to varying levels(Figure 2d) This suggests that some differences in BCL6 functionmay occur in the different subtypes of breast cancer cellsTo determine whether BCL6 target genes in breast cancer cells

were distinct from those in B-cell lymphomas we performedChIP-seq for BCL6 in the OCI-LY1 diffuse large B-cell lymphomacell line We then compared the genes containing at least onepeak in each cell line From this analysis we determined that one-sixth of the genes bound by BCL6 in lymphoma cells also showBCL6 binding in breast cancer cells (Figure 2e) Conversely abouthalf of the potential BCL6 target genes in breast cancer cells arebound by BCL6 in lymphoma cells and half are unique to the

Figure 2 BCL6 binds to specific genomic sites in breast cancer cells (a) ChIP was performed in SK-BR-3 cells for BCL6 binding to target sitesidentified in lymphoma cells Data are expressed as BCL6 binding relative to background (at a non-binding region) (b) De novo motif analysisof genomic sequences associated with BCL6 binding sites in breast cancer cells identified the BCL6 binding motif (c) Genomic distribution ofBCL6 binding sites relative to genes in breast cancer cells (d) BCL6 ChIP was performed in three different breast cancer cell lines and bindingof BCL6 to binding sites was analyzed by quantitative polymerase chain reaction (e) Genes containing at least one BCL6 binding site werecompared with the breast cancer and lymphoma ChIP-seq data sets


3

copy 2014 Macmillan Publishers Limited Oncogene (2014) 1 ndash 10

breast cancer cells These findings suggest that BCL6 has bothsimilar and distinct functions in breast cancer and lymphomaTo validate that these sites are bona fide functional BCL6

binding sites important for gene regulation we knocked downBCL6 with siRNA (Figure 3a) Focusing initially on the BCL6 bindingsite within the BCL6 locus (region B) we determined that siRNA toBCL6 reduced BCL6 binding to this site by approximately 75(Figure 3a) In addition knocking down BCL6 also resulted in anincrease of RNA polymerase II binding and an increase inacetylated histone H4 consistent with the loss of a repressiveeffect of BCL6 on transcription at this site Similar results wereseen in MDA-MB-468 cells and there was a more modest increasein RNA polymerase II binding in T-47D cells (SupplementaryFigure 2a) Analysis of the newly identified BCL6 binding sitesdemonstrated that binding of BCL6 was reduced at 17 of the 18sites upon introduction of siRNA to BCL6 (Figure 3b) In additionRNA polymerase was recruited to most but not all of the sites withdepletion of BCL6 These findings provide additional evidence thatthe sites identified by ChIP-seq in breast cancer cells are likely tobe functional BCL6 binding sites

BCL6 binding is associated with modulation of gene expressionTo measure directly the effect of BCL6 binding on geneexpression in breast cancer cells we first transfected

a BCL6-responsive luciferase reporter into SK-BR-3 MDA-MB-468 andT-47D breast cancer cells (Figure 3c and Supplementary Figure 2b)Ectopic expression of BCL6 reduced expression of luciferasedemonstrating that this reporter is repressed by BCL6 Converselyknocking down BCL6 with siRNA resulted in increased luciferaseexpression Finally forced re-expression of BCL6 in the presenceof siRNA to BCL6 again led to repression of luciferase activityIn addition the small-molecule BCL6 inhibitor 79-6 23 also resultedin increased luciferase activity in all three cell lines (SupplementaryFigure 2c) further demonstrating that BCL6 is active in these cellsand that inhibiting BCL6 results in derepression of this reporterconstructHaving determined that BCL6 represses expression of a

BCL6-responsive reporter in breast cancer cells we next deter-mined the effects of BCL6 on endogenous genes We used siRNAto deplete cells of BCL6 and then analyzed the expression oftarget genes identified by ChIP-seq PRDM1 a well-described BCL6target gene served as a control (Figure 3d) Analysis of four of thenewly identified BCL6 target genes demonstrated that three of thegenes are repressed by BCL6 (HERC5 KLF6 and SH3PXD2B)whereas the fourth MED24 is upregulated by BCL6 in SK-BR-3cells (Figure 3d) confirming these as bona fide BCL6-regulatedgenes in breast cancer cells Although BCL6 is generally thought tobe a transcriptional repressor BCL6 may also be involved inupregulating a subset of genes such as MED24 In eight additional

0

10

20

30

40

sicontrol sibcl6

Rel

ativ

e P

ol II

bin

ding

0

1

2

3

4

sicontrol sibcl6R

elat

ive

A-H

4

0

05

1

15

2

25

3

35

SK-BR-3

MDA-M

B-468

Rel

ativ

e lu

cife

rase

act

ivity

sicontrol vector

sicontrol bcl6

sibcl6 vector

sibcl6 bcl6

BCL6

PARP

SiBCL6 - +

0

2

4

6

8

10

sicontrol siBCL6

Rel

ativ

e B

CL6

bin

ding

Rel

ativ

e P

ol I

I bin

din

g

regio

n B 17 32 84 17824

225

529

234

241

446

347

549

251

452

754

668

582

288

60

5

10

15

20

30

40

50

60 sicontrolsiBCL6

Rel

ativ

e B

CL

6 b

ind

ing

regio

n B 17 32 8417824

225

529

234

241

446

347

549

251

452

754

668

582

288

60

5

10

15

2020

40

60

80

100 sicontrolsiBCL6

0

1

2

3

4

5

6

BCL6

Prdm

1

Herc5

KLF6

Med

24

SH3PXD2B

Rel

ativ

e m

RN

A e

xpre

ssio

n

siControlSiBCL61siBCL62

Figure 3 BCL6 modulates gene expression in breast cancer cells (a) SK-BR-3 cells were transfected with siRNA to BCL6 (or control) Extractswere then analyzed for BCL6 expression by immunoblot (left) or by ChIP for binding of BCL6 RNA polymerase II and acetylated histone H4 toregion B of the BCL6 gene (right) (b) SK-BR-3 cells treated with siRNA to BCL6 (or control) were analyzed by ChIP for BCL6 (top) and RNApolymerase II (bottom) binding at the indicated sites Binding is expressed relative to a negative control region (c) Breast cancer cells treatedwith the indicated expression vectors and siRNA constructs were transfected with a BCL6-responsive luciferase plasmid and analyzed forluciferase activity (d) SK-BR-3 breast cancer cells treated with siRNA to BCL6 or control were analyzed by quantitative real-time polymerasechain reaction for expression of the indicated genes relative to glyceraldehyde 3-phosphate dehydrogenase


4


diverse breast cancer cell lines depletion of BCL6 led to decreasedexpression of MED24 in all and generally increased expression ofthe other three newly identified target genes (SupplementaryFigure 3)

BCL6 is required for survival of breast cancer cellsHaving found that BCL6 has unique effects on gene expression inbreast cancer cells we wanted to determine the biological effectsof BCL6 in these cells We found that reducing the levels of BCL6

by RNA interference (RNAi) caused a distinct morphologic changein MCF-7 and MDA-MB-468 breast cancer cells compared with thecontrol (Supplementary Figure 4a) The cells appeared to be morevacuolated and displayed less cellndashcell adhesion We nextexamined viable cell number over time in culture Depletion ofBCL6 by RNAi led to a lower population doubling rate and a lowerplateau density compared with cells treated with control siRNA(Figure 4a) Analysis of a variety of breast cancer cell lines uponBCL6 depletion demonstrated reduced cell number comparedwith control cells (Figure 4b) To further evaluate the role of BCL6

BCL6

STAT3

STAT5

siControl siBCL6

Plasmid

0

05

1

15

2

25

3

35

Day 1 Day 2 Day 3 Day 4

Rel

ativ

e gr

owth

siControlsiBCL6

0

05

1

15

2

25

3

Day 1 Day 2 Day 3 Day4

Rel

ativ

e gr

owth

sicontrol

sibcl61

0

02

04

06

08

1

12

14

sicontrol sibcl6

Rel

ativ

e ce

ll vi

abili

ty

0

02

04

06

08

1

12

14

16

siCon

trol2

siCon

trol3

siBCL6

1

siBCL6

2

siSTAT3

Rel

ativ

e vi

abili

ty

BCL6

HSP90

siBCL6 - + - + - + - + - + - +

T47D

0

05

1

15

2

25


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

02

04

06

08

1

12

T47D SKBR3 Hs578T BT549 SUM159 MDA231

Rel

ativ

e vi

abili

ty

sicontrolsiBCL6

BCL6

Tubulin

siRNA C2 C3 B1 B2 S3

STAT3

BCL6

Tubulin


STAT3

SK-BR-3

MCF-7 ZR-75

BT-549

MDA231SUM159BT549Hs578tSKBR3

V B S3 V B S3

sicontrolbcl6

sicontrolstat3

sibcl6bcl6

sibcl6stat3

Figure 4 BCL6 depletion inhibits breast cancer growth and survival (a) Breast cancer cells were transfected with siRNA to BCL6 and viable cellnumber was analyzed by measuring ATP-dependent luminescence (b) The indicated breast cancer cell lines were transfected with siRNA toBCL6 or control BCL6 knockdown was confirmed by immunoblot (top) Viable cell number was measured at 96 h after transfection (bottom)(c) MDA-MB-468 cells were transfected with the indicated siRNAs and BCL6 expression (top) and viable cell number (bottom) were measured48 h after transfection (d) MDA-MB-468 cells treated with control or BCL6-targeting siRNA were transfected with expression vectors for BCL6or STAT3 Expression of the indicated proteins was measured by immunoblot (left) and viable cell number was measured 48 h aftertransfection (right) Although the coding sequence of the BCL6 expression construct was wild type high-efficiency transfection allowedexpression in the presence of the siRNA to BCL6


5


on breast cancer cell survival and proliferation we performedreplating assays Cells in which BCL6 had been depleted displayeda significant reduction in plating efficiency (SupplementaryFigure 4b) providing further evidence for the important role ofthis proteinBCL6 has a cognate binding site that overlaps with that of the

STAT family of transcription factors including STAT3 a transcrip-tion factor also known to have an oncogenic role in breast cancerWe have found that STAT3 is activated in the majority of triple-negative breast cancer cell lines but not in lines derived from theother subtypes24 Therefore we considered the possibility thatBCL6 and STAT3 played complementary roles in triple-negativebreast cancer cellular function In MDA-MB-468 breast cancercells which contain activated STAT3 (Supplementary Figure 5)we knocked down BCL6 or STAT3 by RNAi and measured thecell viability Knockdown of either BCL6 or STAT3 led to areduction in the relative viability in these cells compared withcells treated with a control siRNA (Figure 4c) Forced over-expression of BCL6 rescued most of the loss of viability caused byknocking down BCL6 (Figure 4d) By contrast ectopic expressionof STAT3 had no effect on the viability of cells in which BCL6 hadbeen knocked down Taken together these findings demonstratethat BCL6 has a unique role in promoting breast cancer cellsurvival

Pharmacologic targeting of BCL6 decreases the viability of breastcancer cellsAs BCL6 promotes breast cancer cell survival we wantedto determine if targeting BCL6 using the peptidomimetic inhibitorRI-BPI would have therapeutic benefit17 To determine if thismolecule was exerting a mechanism-specific effect we firstanalyzed the expression of BCL6 target genes after RI-BPItreatment As RI-BPI disrupts the interaction of corepressors thatbind the BTB domain of BCL6 only genes that are repressedthrough this mechanism should be affected by this compoundTherefore we treated MDA-MB-468 cells with RI-BPI and analyzedgene expression We found that RI-BPI but not a control peptideincreased the expression of PRDM1 KLF6 and SH3PXD2B(Figure 5a) Furthermore MED24 which is upregulated by BCL6showed decreased expression following RI-BPI treatment Similarresults were seen in T-47D and SK-BR-3 cells (SupplementaryFigure 6a) These findings demonstrate that as expected RI-BPIinhibits some but not all BCL6 transcriptional function in breastcancer cellsWe next treated cells derived from each major subtype of breast

cancer with RI-BPI and measured the cell viability All of the linesshowed a similar dose-dependent loss of viability in response toRI-BPI with an inhibitory concentration 50 of approximately 10 μMwhereas treatment with a control peptide had no effect(Figure 5b) In addition treatment of breast cancer cells with RI-BPI led to apoptosis as measured by poly (ADP-ribose)polymerase (PARP) cleavage (Figure 5c and SupplementaryFigure 6b) and annexin VPI positivity (Figure 5d) Importantlyafter knocking down BCL6 by siRNA RI-BPI caused no additionalloss of viability This supports the hypothesis that the effect ofRI-BPI is mediated solely through BCL6 (Figure 5e) Treatment ofadditional triple-negative breast cancer cells demonstratedthat they were all sensitive to RI-BPI to varying degrees(Supplementary Figure 6c) The degree of sensitivity to RI-BPIdid not always correlate with the levels of nuclear BCL6 expression(Figure 1e) and may relate to other cell-specific factors such as thebinding of BCL6 to specific binding sites differences inintracellular accumulation of the inhibitors or disparate distribu-tions of BCL6 among the cells in a given cell line The small-molecule BCL6 inhibitor 79-6 also reduced the cell viability of avariety of breast cancer cell lines indicating that this is amechanism-specific effect (Supplementary Figure 7)

About 70 of breast tumors contain activated STAT3 which isknown to promote survival proliferation and self-renewal ofbreast cancer cells42425 Therefore we determined if combina-tions of a STAT3 inhibitor and the BCL6 inhibitor RI-BPI showedenhanced effects in killing breast cancer cells Treatment ofMDA-MB-468 breast cancer cells with the STAT3 inhibitornifuroxazide26 resulted in a decrease in STAT3 target geneexpression (Supplementary Figure 8a) and a decrease in the cellviability (Supplementary Figure 8b) Combining nifuroxazide withRI-BPI resulted in an additive loss of cell viability (SupplementaryFigure 8b) In addition reducing the levels of BCL6 with siRNA alsosensitized breast cancer cells to nifuroxazide providing furtherevidence that inhibiting STAT3 and BCL6 may be an effectivecombination for breast cancer therapy particularly in triple-negative tumors (Supplementary Figure 8c) Importantly thiseffect was seen for the triple-negative cell line MDA-MB-468 butthere was no effect for the ERPR+ cell line T-47D and only a slightenhancement for Her2+ SK-BR-3 cells which parallels the STAT3phosphorylation status of these cells24

The activation of STAT3 in breast cancer cells is commonlydependent on Jak2 which is also a potential target in breastcancer therapy Thus we determined if depleting Jak2 also had aneffect on the cell viability in combination with BCL6 inhibition Wefound that reducing the levels of Jak2 in MDA-MB-468 cells withsiRNA inhibited STAT3 target gene expression (SupplementaryFigure 8d) and led to an approximately 50 decrease in viable cellnumber (Figure 5f) Dual knockdown of BCL6 and Jak2 lead toenhanced loss of viability compared with either knockdown alone(Figure 5f) In addition combining knockdown of Jak2 with theBCL6 inhibitor RI-BPI led to approximately an 80 reduction inviable cell number (Figure 5g) Next we assessed the effects of theJak2 inhibitor TG101348 2728 with BCL6 inhibition As expectedTG101348 inhibited STAT3 target gene expression in MDA-MB-468cells (Supplementary Figure 9a) Treatment of breast cancer cellstransfected with siRNA to BCL6 with TG101348 resulted inenhanced loss of cell viability of MDA-MB-468 cells (Figure 5h)and SK-BR-3 cells (Supplementary Figure 9b) Similarly combiningTG101348 with RI-BPI resulted in a significant loss of viabilitycompared with either treatment alone (Figure 5i) As expected theJak inhibitor TG101348 had little effect on T-47D cells which lackconstitutive STAT3 phosphorylation (Supplementary Figure 9b)Taken together these findings demonstrate that BCL6 isimportant for breast cancer cell survival and that inhibition ofBCL6 with RI-BPI alone or in combination with other therapiesmay be an important therapeutic approach for breast cancer

DISCUSSIONWe have demonstrated that BCL6 is expressed in most breastcancer cells lines and that its genetic locus is amplified inapproximately 50 of breast tumor samples and the majority ofbreast cancer cell lines In addition BCL6 is important for breastcancer cell survival Targeting BCL6 with peptidomimetic or small-molecule inhibitors kills breast cancer cells alone and incombination with STAT3 or Jak2 inhibitorsPerforming whole genome ChIP-seq allowed us to identify BCL6

binding sites in breast cancer cells When the genes associatedwith these sites were compared with the genes bound by BCL6 inB-cell lymphomas there was only limited overlap between thetwo groups (Figure 2e) This suggests that BCL6 has bothoverlapping and distinct roles in these two cell types Why BCL6does not bind to the same sites in the two different cell typesraises several interesting hypotheses and clearly indicates thatDNA sequence alone is not sufficient to specify transcription factorbinding in the context of chromatin Other sources of tissue-specific differences may include alterations in DNA itself such asmethylation or modifications of histones by methylation oracetylation Further analysis of these factors may provide


6


important information about the mechanism by which BCL6regulates gene expression differently in distinct tissuesThe interaction of BCL6 with corepressors may also be distinct

in breast cancer compared with other tissue types The well-characterized BCL6 target gene PRDM1 has been shown to berepressed by BCL6 interaction with MTA3 in lymphoma cells13

MTA3 is also expressed in breast cancer cells and is important forregulating invasiveness29 However in breast cancer cells we havefound that the BCL6 inhibitor RI-BPI which prevents theinteraction with SMRT leads to the upregulation of PRDM1(Figure 5a and Supplementary Figure 6a) This suggests that SMRTis involved in the regulation of PRDM1 in breast cancer cells

Blimp1 the protein encoded by the PRDM1 gene has been shownto be involved in breast cancer and represses ERα30 whereas in Bcells Blimp1 promotes plasma cell differentiation13 The reason fordifference in the regulation of PRDM1 between these two celltypes is unclear and warrants further studyWe have found that BCL6 is expressed in 10 of 11 breast cancer

cell lines tested (Supplementary Figure 1b) Interestingly expres-sion of BCL6 was not restricted to any one breast cancer cell typeTherefore we analyzed BCL6 binding in all three types of breastcancer cells and determined that while many of the sitesare shared between the three major classes of breast cancer thereare also distinct differences between the cells In addition the

0

02

04

06

08

1

12

cp 5 cp10

cp15

cp20

bpi5

bpi10

bpi15

bpi20

Rel

ativ

e vi

abili

ty

T-47d

SKBR3

MDA-468

0

1

2

3

4

5

6

PRDM1 Herc5 KLF6 Med24 SH3PXD2B

Rel

ativ

e m

RN

A e

xpre

ssio

n media 12hcontrol peptideRI-BPI (15uM)

PARP

RI-BPI - +

HSP90

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleRI-BPI

002040608

112

sicontrol sijak2

Rel

ativ

e vi

abili

ty vehicleRI-BPI

vehicle

RI-BPI

002040608

112

0 2 3 4uM TG101348

Rel

ativ

e vi

abili

ty vehicleBPI

002040608

112

sicontrol sibcl6 sijak2 sibcl6sijak2

Rel

ativ

e vi

abili

ty

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleTG101348

05 41

12941

520 45

62 373

Figure 5 BCL6 inhibition decreases the viability of breast cancer cells (a) MDA-MB-468 breast cancer cells were treated with RI-BPI orcontrol peptide for 12 h and then expression of BCL6 target genes were analyzed by quantitative real-time polymerase chain reaction(b) Breast cancer cells were treated with the indicated doses of RI-BPI or control peptide (μM) for 48 h after which viable cell number wasdetermined (c) MDA-MB-468 breast cancer cells were treated with 15 μM RI-BPI for 48 h and then initiation of apoptosis was determined byimmunoblot for PARP cleavage HSP90 served as a loading control (d) MDA-MB-468 cells were treated with vehicle or 15 μM RI-BPI for 48 hafter which apoptosis was determined by annexin VPI staining and flow cytometry (e) MDA-MB-468 cells treated with control orBCL6-targeted siRNA were then treated with vehicle or RI-BPI for 48 h and viable cell number was measured (f) MDA-MB-468 cellswere transfected with siRNA to BCL6 andor Jak2 Relative cell viability was measured after 72 h (g) MDA-MB-468 cells treated withcontrol or Jak2-targeted siRNA were treated with 10 μM RI-BPI for 48 h and viable cell number was determined (h) MDA-MB-468 cellstransfected with siRNA to BCL6 were treated with the Jak2 inhibitor TG101348 (3 μM) for 48 h and then the cell viability was measured(i) MDA-MB-468 cells were treated with 10 μM RI-BPI (BPI) and the indicated doses of TG101348 for 48 h and then the cell viabilitywas measured


7


responses to siRNA and drug treatment also had some distinctdifferences with the triple-negative MDA-MB-468 and BT-549 cellsresponding more than the other cell types This raises thepossibility that triple-negative breast tumors may be moresensitive to targeting BCL6 and therefore may benefit morestrongly from treatment with a BCL6 inhibitor This may beparticularly important as these tumors often become resistant tochemotherapyTreating breast cancer cells with RI-BPI leads to the loss of cell

viability and induction of apoptosis (Figures 5c and d andSupplementary Figure 6b) This suggests that targeting BCL6 maybe an important new strategy for cancer therapy In addition thecombination of RI-BPI with inhibition of another transcriptionfactor STAT3 also showed promise (Supplementary Figure 8b)Over 70 of breast tumors have activation of STAT3425 andSTAT3 activation is highly associated with triple-negative breasttumors24 Therefore triple-negative breast tumors may particularlybenefit from the combination of STAT3 and BCL6 inhibitors STAT3can also directly increase expression of BCL6 in breast cancercells4 Thus this combination may not only show efficacy byinhibiting the function of each transcription factor alone butinhibiting STAT3 may also reduce the levels of BCL6 expressionproviding a further level of inhibition of BCL6 in these triple-negative breast cancer cells In addition Jak inhibitors are beingtested in clinical trials for triple-negative breast cancer We havefound that the combination of the Jak2 inhibitor TG101348 withBCL6 inhibition reduces the viability of breast cancer cells morethan either approach alone (Figure 5h and i) Thereforecombinations of Jak inhibitors and RI-BPI or other BCL6 inhibitorsmay also be an effective treatment strategy for breast cancerUltimately it will be important to confirm these findings in in vivosystemsIn conclusion we have demonstrated that BCL6 is an important

transcriptional regulator in breast cancer and that targeting BCL6induces apoptosis in these cells About half of the genes to whichBCL6 binds in breast cancer cells are unique when compared withlymphoma cells indicating that BCL6 likely mediates some uniqueeffects in breast cancer As BCL6 is expressed in all three majorclasses of breast cancer developing targeted therapies to BCL6may be an important new avenue for breast cancer therapy

MATERIALS AND METHODSCells and treatmentsMDA-MB-468 cells (kindly provided by Myles Brown Dana-Farber CancerInstitute Boston MA USA) T-47D cells (ATCC Manassas VA USA) andMCF-7 cells (kindly provided by Francis Kern Southern Research InstituteBirmingham AL USA) were maintained in Dulbeccos modified Eaglesmedium with 10 fetal bovine serum SK-BR-3 cells (kindly provided byLyndsay Harris Yale CT USA) were maintained in RPMI with 10 fetalbovine serum These cells were authenticated before manuscript submis-sion by STR DNA analyses at Genetica DNA Laboratories Inc (CincinnatiOH USA) Immortalized mammary epithelial cells (kindly provided byMyles Brown) were maintained as described31 Additional breast cancercells were obtained and maintained as described24 Ramos cells weremaintained in RPMI with 10 fetal bovine serum Breast cancer cells weretreated with control peptide or RI-BPI at the indicated doses17 or withvehicle or the indicated doses of 79-6 (EMD Millipore Billerica MA USA)Cells were treated with nifuroxazide (Chembridge San Diego CA USA) orTG101348 (VWR Radnor PA USA) at the indicated doses

Copy number analysisLevel 3 Affymetrix Genome-Wide SNP Array 60 data for breast invasivecarcinoma was downloaded from The Cancer Genome Atlas (httpcancergenomenihgov) The data spanned 790 tumor samples and786 matched normal breast samples Copy number (log 2ndash1) at the BCL6locus was determined for each sample by identifying the value in columnlsquosegmeanrsquo at chromosome 3 position 188 929 714 corresponding to thecenter of the BCL6 gene The extent of rightward skewness of the BCL6

copy number distribution in tumor was quantified using Pearsonrsquos medianskewness coefficient defined as

3 acute ethmeanmedianTHORNstandard deviation

The statistical significance of the skewness coefficient was calculatedby determining the skewness at 1000 randomly selected locations in thegenome An empirical P-value was derived by determining thenumber of locations with skewness equal to or greater than that atthe BCL6 locus Similar analyses were performed using data from theCCLE18 for breast cancer cell lines For breast-specific subtypes the celllines were matched to subtype based on expression of ER PR andHer2 expression reported from32ndash34 The statistical significance of theskewness coefficient was calculated as above using 100 randomlyselected locations in the genome Gene expression was also obtainedfrom the CCLE

RNA interferenceBreast cancer cells were reverse transfected with siRNAs using LipofectamineRNAiMax (Invitrogen Carlsbad CA USA) at 10 nM unless otherwiseindicated for the indicated times RNAi included siControl (D-001210-03Dharmacon Lafayette CO USA) siBCL6 no 1 (HSS100968 Invitrogen)siBCL6 no 2 (HSS100966 Invitrogen) siSTAT3 (M-003544-02 Dharma-con) and siJak2 (M-003146-02-0005 Dharmacon) Rescue experimentswere performed with the combination transfection of siRNA andexpression plasmids for pRcCMV STAT3 or pCDNA3-BCL6

Chromatin immunoprecipitationChIP was performed essentially as described35 Cells were fixed in 1formaldehyde for 10min sonicated and lysates were immunoprecipitatedovernight with 2 μg of antibodies to BCL6 (sc-858) and RNA polymerase II(sc-9001) from Santa Cruz Biotechnology (Santa Cruz CA USA) or acetyl H4from Cell Signaling Technology (Danvers MA USA) ChIP product wasmeasured using Quant-iT dsDNA HS Assay (Invitrogen) and equal amountsof product and input were analyzed by quantitative polymerase chainreaction Data are expressed as fold binding over background (using bothinput and genomic regions known not to be bound by BCL6 including therhodopsin gene and a previously defined control region found within theBCL6 gene20) For siRNA ChIPs 75 times 106 SK-BR-3 T-47D or MDA-MB-468cells were plated and ChIP was performed 72 h after reverse transfectionChIP product was analyzed by quantitative polymerase chain reactionusing the indicated primers (Supplementary Table 1 or as described3)OCI-Ly1 ChIP was performed essentially as above except that cells werelysed in the RIPA buffer (150 mM NaCl 1 (vv) Nonidet P-40 05 (wv)deoxycholate 01 (wv) sodium dodecyl sulfate 50 mM Tris (pH 8) and5 mM ethylenediaminetetraacetic acid)

ChIP-seqLibraries for ChIP-seq were created from 10 ng of BCL6 ChIP and 10 ng ofcorresponding input in SK-BR-3 cells according to the manufacturerrsquosprotocol (Illumina San Diego CA USA) with size selection of 250ndash400 bpChIP-seq was performed with a SR36bp run Sequences were aligned tothe human genome (UCSC hg18) using ELAND Peaks were defined usingMACS36 Motifs were identified using both the de novo motifs search andthe TRANSFAC database using the SeqPos motif tool37 Binding siteanalysis and annotation was performed using CEAS38 Analysis tools werefrom the GalaxyCistrome website (http cistromeorg) ChIP-seq wasperformed essentially as above for the OCI-LY1 cells For comparisonsbetween BCL6 binding in breast cancer and lymphoma cells onlysequences mapping uniquely to the genome with not more than twomismatches were used and peaks were assigned using ChIPseeqer39 forboth data sets Each data set was compared to its own input control

Luciferase reporter analysisBreast cancer cells were reverse transfected with siRNA The following day(BCL6)4-TK-Luc was transfected into breast cancer cells in combinationwith Renilla luciferase using Lipofectamine 2000 (Invitrogen) These werealone or with vector or pCDNA3-BCL6 Luciferase was analyzed asdescribed20 For cells treated with 79-6 breast cancer cells weretransfected with BCL6-Luc for 6 h and then treated with 300 μM 79-6 for16 h


8


Reverse transcriptionndashpolymerase chain reactionReverse transcriptionndashpolymerase chain reaction was performed asdescribed4 mRNA primer sequences are as indicated (SupplementaryTable 2)

Immunoblots and nuclear extractsImmunoblotting was performed as described26 using the followingantibodies PARP (9542 Cell Signaling) tubulin (T-5168 Sigma) (St LouisMO USA) BCL6 (sc-858 Santa Cruz Biotechnology) HSP90 (sc-13119 SantaCruz Biotechnology) P-STAT3 (9131 Cell Signaling Technology) STAT3(sc- 482 Santa Cruz Biotechnology) and actin (Sigma) Nuclear extractswere generated using the Nuclear Extract Kit (Active Motif Carlsbad CAUSA) according to the manufacturerrsquos protocol

Viable cell number quantitation and clonogenic assaysBreast cancer cells were plated in 384- or 96-well plates for viability induplicate or quadruplicate Cells were treated with drug for the indicatedtime and viable cell number was measured using an ATP-dependentluciferase assay (CellTiterGlo Promega Madison WI USA) and expressedrelative to control or vehicle-treated cells For siRNA knockdown breastcancer cells were reverse transfected and then analyzed on the indicatedday For growth assays cell number was measured daily by CellTiterGloand normalized to day 1 values For clonogenic assays cells were reversetransfected with siRNA The following day 500 cells were seeded in 6-welldishes After 10ndash14 days colonies were visualized by staining with a 6glutaraldehyde and 05 crystal violet solution

Apoptosis assayBreast cancer cells were treated with vehicle or 15 μM BPI for 48 hand analyzed for annexinPI staining (BD Biosciences San Jose CA USA)by flow cytometry

CONFLICT OF INTERESTThe authors declare no conflict of interest

ACKNOWLEDGEMENTSThis work was supported by a BCRF-AACR grant for Translational Breast CancerResearch and grants from the National Cancer Institute (R01-CA160979) SusanG Komen for the Cure the Brent Leahey Fund and Friends of the Dana-Farber CancerInstitute

REFERENCES1 Logarajah S Hunter P Kraman M Steele D Lakhani S Bobrow L et al BCL-6 is

expressed in breast cancer and prevents mammary epithelial differentiaitonOncogene 2003 22 5572ndash5578

2 Polo JM Juszczynski P Monti S Cerchietti L Ye K Greally JM et al Transcriptionalsignature with differential expression of BCL6 target genes accurately identifiesBCL6-dependent diffuse large B cell lymphomas Proc Natl Acad Sci USA 2007104 3207ndash3212

3 Ci W Polo JM Cerchietti L Shaknovich R Wang L Yang SN et alThe BCL6 transcriptional program features repression of multiple oncogenesin primary B cells and is deregulated in DLBCL Blood 2009 1135536ndash5548

4 Walker SR Nelson EA Zou L Chaudhury M Signoretti S Richardson A et alReciprocal effects of STAT5 and STAT3 in breast cancer Mol Cancer Res 2009 7966ndash976

5 Walker SR Nelson EA Yeh JE Pinello L Yuan GC Frank DA STAT5 outcompetesSTAT3 to regulate the expression of the oncogenic transcriptionalmodulator BCL6 Mol Cell Biol 2013 33 2879ndash2890

6 Bos R van Diest PJ van der Groep P Greijer AE Hermsen MA Heijnen I et alProtein expression of B-cell lymphoma gene 6 (BCL-6) in invasive breast cancer isassociated with cyclin D1 and hypoxia-inducible factor-1alpha (HIF-1alpha)Oncogene 2003 22 8948ndash8951

7 Polo JM Ci W Licht JD Melnick A Reversible disruption of BCL6 repressioncomplexes by CD40 signaling in normal and malignant B cells Blood 2008 112644ndash651

8 Ahmad KF Melnick A Lax S Bouchard D Liu J Kiang CL et al Mechanismof SMRT corepressor recruitment by the BCL6 BTB domain Mol Cell 2003 121551ndash1564

9 Polo JM DellOso T Ranuncolo SM Cerchietti L Beck D Da Silva GF et alSpecific peptide interference reveals BCL6 transcriptional and oncogenicmechanisms in B-cell lymphoma cells Nat Med 2004 10 1329ndash1335

10 Huynh KD Fischle W Verdin E BCoR Bardwell VJ a novel corepressor involved inBCL-6 repression Genes Dev 2000 14 1810ndash1823

11 Huynh KD Bardwell VJ The BCL-6 POZ domain and other POZ domainsinteract with the co-repressors N-CoR and SMRT Oncogene 1998 172473ndash2484

12 Ghetu AF Corcoran CM Cerchietti L Bardwell VJ Melnick A Prive GG Structure ofa BCOR corepressor peptide in complex with the BCL6 BTB domain dimer Mol Cell2008 29 384ndash391

13 Parekh S Polo JM Shaknovich R Juszczynski P Lev P Ranuncolo SM et alBCL6 programs lymphoma cells for survival and differentiation through distinctbiochemical mechanisms Blood 2007 110 2067ndash2074

14 Fujita N Jaye DL Geigerman C Akyildiz A Mooney MR Boss JM et al MTA3 andthe Mi-2NuRD complex regulate cell fate during B lymphocyte differentiationCell 2004 119 75ndash86

15 Mendez LM Polo JM Yu JJ Krupski M Ding BB Melnick A et alCtBP is an essential corepressor for BCL6 autoregulation Mol Cell Biol 2008 282175ndash2186

16 Wang X Li Z Naganuma A Ye BH Negative autoregulation of BCL-6 is bypassedby genetic alterations in diffuse large B cell lymphomas Proc Natl Acad Sci USA2002 99 15018ndash15023

17 Cerchietti LC Yang SN Shaknovich R Hatzi K Polo JM Chadburn A et alA peptomimetic inhibitor of BCL6 with potent antilymphoma effects in vitro andin vivo Blood 2009 113 3397ndash3405

18 Barretina J Caponigro G Stransky N Venkatesan K Margolin AA Kim S et alThe Cancer Cell Line Encyclopediaenables predictive modelling of anticancerdrug sensitivity Nature 2012 483 603ndash607

19 Bereshchenko OR Gu W Dalla-Favera R Acetylation inactivates the transcriptionalrepressor BCL6 Nat Genet 2002 32 606ndash613

20 Walker SR Nelson EA Frank DA STAT5 represses BCL6 expression by binding to aregulatory region frequently mutated in lymphomas Oncogene 2007 26224ndash233

21 Welboren W-J van Driel MA Janssen-Megens EM van Heeringen SJ FCGJ SweepSpan PN et al ChIP-Seq of ER[alpha] and RNA polymerase II defines genesdifferentially responding to ligands EMBO J 2009 28 1418ndash1428

22 Kwon H Thierry-Mieg D Thierry-Mieg J Kim H-P Oh J Tunyaplin C et alAnalysis of interleukin-21-induced Prdm1 gene regulation reveals functionalcooperation of STAT3 and IRF4 transcription factors Immunity 2009 31941ndash952

23 Cerchietti LC Ghetu AF Zhu X Da Silva GF Zhong S Matthews M et alA small-molecule inhibitor of BCL6 kills DLBCL cells in vitro and in vivo Cancer Cell2010 17 400ndash411

24 Marotta LL Almendro V Marusyk A Shipitsin M Schemme JWalker SR et alThe JAK2STAT3 signaling pathway is required for growth of CD44+CD24minus stemcell-like breast cancer cells in human tumors J Clin Invest 2011 1212723ndash2735

25 Alvarez JV Febbo PG Ramaswamy S Loda M Richardson A Frank DAIdentification of a genetic signature of activated signal transducer and activator oftranscription3 in human tumors Cancer Res 2005 65 5054ndash5062

26 Nelson EA Walker SR Kepich A Gashin LB Hideshima T Ikeda H et alNifuroxazide inhibits survival of multiple myeloma cells by directlyinhibiting STAT3 Blood 2008 112 5095ndash5102

27 Wernig G Kharas MG Okabe R Moore SA Leeman DS Cullen DE et alEfficacy of TG101348 a selective JAK2 inhibitor in treatment of a murinemodel of JAK2V617F-induced polycythemia vera Cancer Cell 2008 13311ndash320

28 Lasho TL Tefferi A Hood JD Verstovsek S Gilliland DG Pardanani A TG101348 aJAK2-selective antagonist inhibits primary hematopoietic cells derived frommyeloproliferative disorder patients with JAK2V617F MPLW515K or JAK2exon 12 mutations as well as mutation negative patients Leukemia 2008 221790ndash1792

29 Fujita N Jaye DL Kajita M Geigerman C Moreno CS Wade PA MTA3 aMi-2NuRD complex subunit regulates an invasive growth pathway inbreast cancer Cell 2003 113 207ndash219

30 Wang X Belguise K ONeill CF Sanchez-Morgan N Romagnoli M Eddy SF et alRelB NF-kappaB represses estrogen receptor alpha expression via induction of thezinc finger protein Blimp1 Mol Cell Biol 2009 29 3832ndash3844

31 DiRenzo J Signoretti S Nakamura N Rivera-Gonzalez R Sellers W Loda M et alGrowth factor requirements and basal phenotype of an immortalized mammaryepithelial cell line Cancer Res 2002 62 89ndash98

32 Kao J Salari K Bocanegra M Choi YL Girard L Gandhi J et al Molecular profilingof breast cancer cell lines defines relevant tumor models and provides a resourcefor cancer gene discovery PLoS One 2009 4 e6146


9


33 Lehmann BD Bauer JA Chen X Sanders ME Chakravarthy AB Shyr Y et alIdentification of human triple-negative breast cancer subtypes and preclinicalmodels for selection of targeted therapies J Clin Invest 2011 121 2750ndash2767

34 Neve RM Chin K Fridlyand J Yeh J Baehner FL Fevr T et al A collection of breastcancer cell lines for the study of functionally distinct cancer subtypes Cancer Cell2006 10 515ndash527

35 Nelson EA Walker SR Alvarez JV Frank DA Isolation of unique STAT5 targets bychromatin immunoprecipitation-based gene identification J Biol Chem 2004 27954724ndash54730

36 Zhang Y Liu T Meyer CA Eeckhoute J Johnson DS Bernstein BE et alModel-based analysis of ChIP-Seq (MACS) Genome Biol 2008 9 R137

37 He HH Meyer CA Shin H Bailey ST Wei G Wang Q et al Nucleosomedynamics define transcriptional enhancers Nat Genet 2010 42343ndash347

38 Shin H Liu T Manrai AK Liu XS CEAS cis-regulatory element annotation systemBioinformatics (Oxford UK) 2009 25 2605ndash2606

39 Giannopoulou EG Elemento O An integrated ChIP-seq analysis platform withcustomizable workflows BMC Bioinform 2011 12 277

Supplementary Information accompanies this paper on the Oncogene website (httpwwwnaturecomonc)


10


from The Cancer Genome Atlas we found that 51 of the 790breast tumors analyzed showed amplification of the BCL6 locuswhereas only a small subset (11) showed a reduction of thislocus (Figure 1a) The skewing toward increased copy numbershowed a nominal P-value of 0056 compared with 1000 randomlychosen sites This finding raised the possibility that overexpressionof BCL6 may be important in breast cancer pathogenesis Next weused data from the Cancer Cell Line Encyclopedia (CCLE)18 toanalyze BCL6 copy number and mRNA expression in breast cancercell lines We found that the majority of breast cancer cell linesalso displayed amplification of this locus (P= 001) (Figure 1b) andthis was evident in lines derived from all three subtypes of breastcancer (Supplementary Figure 1a) In addition gt75 of the celllines showed enhanced expression of BCL6 Of the small numberof cell lines showing a loss at this locus most were from triple-negative breast tumors Interestingly however even among thisgroup the majority of lines showed increased BCL6 mRNAexpression suggesting that there is a strong selective pressurefor expression of this gene (Supplementary Figure 1a) To betterunderstand the role BCL6 has in breast cancer we chose a panelof breast cancer cells containing both amplifications and deletionsof this locus and which represented the three major classes ofbreast cancer hormone receptor positive (ER+PR+) Her2ErbB2

amplified (Her2+) and ERminus PRminus Her2minus (triple negative)(Figure 1c) We then analyzed the expression of BCL6 mRNA(Figure 1d) and protein (Supplementary Figure 1b) in these linescompared with non-tumorigenic mammary epithelial lines Wefound that BCL6 was expressed to varying degrees in 10 of the 11breast cancer cells lines analyzed and that there was generally alow expression of BCL6 in non-tumorigenic breast lines comparedwith the breast cancer cell lines (Supplementary Figure 1b)Expression levels did not correlate with a specific breast cancersubtype We further analyzed the levels of nuclear BCL6expression and found that BCL6 was present to varying degreesin the nucleus of all 10 breast cancer cell lines tested (Figure 1e)BCL6 becomes post-translationally modified19 and runs across arange of molecular weights To ensure analysis of the correct bandfrom protein extracts we treated cells with small interfering RNA(siRNA) to BCL6 or control and performed immunoblots to BCL6on nuclear extracts This confirmed that BCL6 is expressed in alleight of the breast cancer cell lines analyzed (Figure 1f) ThusBCL6 is expressed in nearly all breast cancer cell lines tested andthe locus is amplified in approximately half of primary humanbreast tumors and the majority of breast cancer cell lines Wechose to focus on T-47D SK-BR-3 and MDA-MB-468 cells forfurther study as they represent the three different subtypes of

-06

-04

-02

MC

F7

T47

D

ZR

751

MD

AM

B45

3

BT

474

SK

BR

3

BT

549

HS

578T

MD

AM

B23

1

MD

AM

B46

8

0

02

04

06

08

1

12

Co

py

nu

mb

er (

log

2-1)

Amplification51

no change38

Deletion11

TumorNormal

200

180

150

140

120

100

Cou

nt

80

60

40

20

0

MC

F-7

T-4

7D

ZR

-75

MD

A-M

B-4

53

SK

BR

3

BT

-549

Hs5

78T

SU

M-1

59pt

MD

A-M

B-2

31

MD

A-M

B-4

68

Ram

os

ER+PR+ Her2+ Triple negative

BCL6

PARP

siBCL6 - + - + - + - + - + - + - + - +

BCL6

PARP

T47DMCF7

-1

-05

0

05

1

15

2

25

Nor

mal

ized

BC

L6ex

pres

sion

-1

-05

0

05

1

15

Cop

y nu

mbe

r

BCL6

-1 -08 08 1

Copy number (log2 - 1)

-06 06-04 04-02 020

RANDOM LOCI(AVERAGE)

MC

F7

T47

D

ZR

751

BT

474

MD

AM

B45

3

SK

BR

3

BT

549

HS

578T

MD

AM

B23

1

MD

AM

B46

8

MDA231Hs578tBT549ZR75 MDA468SKBR3

Figure 1 BCL6 is expressed in breast cancer Copy number for BCL6 was analyzed using data from (a) the Cancer Genome Atlas for breasttumors (inset contains percentages) or (b) from the CCLE for breast cancer cell lines Copy number (c) and BCL6 gene expression (d) from theCCLE for a subset of breast cancer cell lines is shown (e) Nuclear lysates were generated from the indicated breast cancer cell lines andanalyzed for BCL6 expression Whole-cell extracts of Ramos cells served as a positive control for BCL6 expression PARP served as a control fornuclear protein (f) Nuclear lysates were generated from the indicated cell lines after 72 h of transfection with siRNA to BCL6 (or control) PARPserved as a control for nuclear protein


2









3







0

10

20

30

40

sicontrol sibcl6

Rel

ativ

e P

ol II

bin

ding

0

1

2

3

4

sicontrol sibcl6R

elat

ive

A-H

4

0

05

1

15

2

25

3

35

SK-BR-3

MDA-M

B-468

Rel

ativ

e lu

cife

rase

act

ivity

sicontrol vector

sicontrol bcl6

sibcl6 vector

sibcl6 bcl6

BCL6

PARP

SiBCL6 - +

0

2

4

6

8

10

sicontrol siBCL6

Rel

ativ

e B

CL6

bin

ding

Rel

ativ

e P

ol I

I bin

din

g

regio

n B 17 32 84 17824

225

529

234

241

446

347

549

251

452

754

668

582

288

60

5

10

15

20

30

40

50

60 sicontrolsiBCL6

Rel

ativ

e B

CL

6 b

ind

ing

regio

n B 17 32 8417824

225

529

234

241

446

347

549

251

452

754

668

582

288

60

5

10

15

2020

40

60

80

100 sicontrolsiBCL6

0

1

2

3

4

5

6

BCL6

Prdm

1

Herc5

KLF6

Med

24

SH3PXD2B

Rel

ativ

e m

RN

A e

xpre

ssio

n




4





BCL6

STAT3

STAT5

siControl siBCL6

Plasmid

0

05

1

15

2

25

3

35


Rel

ativ

e gr

owth

siControlsiBCL6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl61

0

02

04

06

08

1

12

14

sicontrol sibcl6

Rel

ativ

e ce

ll vi

abili

ty

0

02

04

06

08

1

12

14

16

siCon

trol2

siCon

trol3

siBCL6

1

siBCL6

2

siSTAT3

Rel

ativ

e vi

abili

ty

BCL6

HSP90

siBCL6 - + - + - + - + - + - +

T47D

0

05

1

15

2

25


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

02

04

06

08

1

12


Rel

ativ

e vi

abili

ty

sicontrolsiBCL6

BCL6

Tubulin


STAT3

BCL6

Tubulin


STAT3

SK-BR-3

MCF-7 ZR-75

BT-549


V B S3 V B S3

sicontrolbcl6

sicontrolstat3

sibcl6bcl6

sibcl6stat3



5











6







0

02

04

06

08

1

12

cp 5 cp10

cp15

cp20

bpi5

bpi10

bpi15

bpi20

Rel

ativ

e vi

abili

ty

T-47d

SKBR3

MDA-468

0

1

2

3

4

5

6


Rel

ativ

e m

RN

A e

xpre

ssio


PARP

RI-BPI - +

HSP90

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleRI-BPI

002040608

112

sicontrol sijak2

Rel

ativ

e vi

abili

ty vehicleRI-BPI

vehicle

RI-BPI

002040608

112

0 2 3 4uM TG101348

Rel

ativ

e vi

abili

ty vehicleBPI

002040608

112


Rel

ativ

e vi

abili

ty

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleTG101348

05 41

12941

520 45

62 373



7















8










































9











10









3







0

10

20

30

40

sicontrol sibcl6

Rel

ativ

e P

ol II

bin

ding

0

1

2

3

4

sicontrol sibcl6R

elat

ive

A-H

4

0

05

1

15

2

25

3

35

SK-BR-3

MDA-M

B-468

Rel

ativ

e lu

cife

rase

act

ivity

sicontrol vector

sicontrol bcl6

sibcl6 vector

sibcl6 bcl6

BCL6

PARP

SiBCL6 - +

0

2

4

6

8

10

sicontrol siBCL6

Rel

ativ

e B

CL6

bin

ding

Rel

ativ

e P

ol I

I bin

din

g

regio

n B 17 32 84 17824

225

529

234

241

446

347

549

251

452

754

668

582

288

60

5

10

15

20

30

40

50

60 sicontrolsiBCL6

Rel

ativ

e B

CL

6 b

ind

ing

regio

n B 17 32 8417824

225

529

234

241

446

347

549

251

452

754

668

582

288

60

5

10

15

2020

40

60

80

100 sicontrolsiBCL6

0

1

2

3

4

5

6

BCL6

Prdm

1

Herc5

KLF6

Med

24

SH3PXD2B

Rel

ativ

e m

RN

A e

xpre

ssio

n




4





BCL6

STAT3

STAT5

siControl siBCL6

Plasmid

0

05

1

15

2

25

3

35


Rel

ativ

e gr

owth

siControlsiBCL6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl61

0

02

04

06

08

1

12

14

sicontrol sibcl6

Rel

ativ

e ce

ll vi

abili

ty

0

02

04

06

08

1

12

14

16

siCon

trol2

siCon

trol3

siBCL6

1

siBCL6

2

siSTAT3

Rel

ativ

e vi

abili

ty

BCL6

HSP90

siBCL6 - + - + - + - + - + - +

T47D

0

05

1

15

2

25


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

02

04

06

08

1

12


Rel

ativ

e vi

abili

ty

sicontrolsiBCL6

BCL6

Tubulin


STAT3

BCL6

Tubulin


STAT3

SK-BR-3

MCF-7 ZR-75

BT-549


V B S3 V B S3

sicontrolbcl6

sicontrolstat3

sibcl6bcl6

sibcl6stat3



5











6







0

02

04

06

08

1

12

cp 5 cp10

cp15

cp20

bpi5

bpi10

bpi15

bpi20

Rel

ativ

e vi

abili

ty

T-47d

SKBR3

MDA-468

0

1

2

3

4

5

6


Rel

ativ

e m

RN

A e

xpre

ssio


PARP

RI-BPI - +

HSP90

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleRI-BPI

002040608

112

sicontrol sijak2

Rel

ativ

e vi

abili

ty vehicleRI-BPI

vehicle

RI-BPI

002040608

112

0 2 3 4uM TG101348

Rel

ativ

e vi

abili

ty vehicleBPI

002040608

112


Rel

ativ

e vi

abili

ty

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleTG101348

05 41

12941

520 45

62 373



7















8










































9











10







0

10

20

30

40

sicontrol sibcl6

Rel

ativ

e P

ol II

bin

ding

0

1

2

3

4

sicontrol sibcl6R

elat

ive

A-H

4

0

05

1

15

2

25

3

35

SK-BR-3

MDA-M

B-468

Rel

ativ

e lu

cife

rase

act

ivity

sicontrol vector

sicontrol bcl6

sibcl6 vector

sibcl6 bcl6

BCL6

PARP

SiBCL6 - +

0

2

4

6

8

10

sicontrol siBCL6

Rel

ativ

e B

CL6

bin

ding

Rel

ativ

e P

ol I

I bin

din

g

regio

n B 17 32 84 17824

225

529

234

241

446

347

549

251

452

754

668

582

288

60

5

10

15

20

30

40

50

60 sicontrolsiBCL6

Rel

ativ

e B

CL

6 b

ind

ing

regio

n B 17 32 8417824

225

529

234

241

446

347

549

251

452

754

668

582

288

60

5

10

15

2020

40

60

80

100 sicontrolsiBCL6

0

1

2

3

4

5

6

BCL6

Prdm

1

Herc5

KLF6

Med

24

SH3PXD2B

Rel

ativ

e m

RN

A e

xpre

ssio

n




4





BCL6

STAT3

STAT5

siControl siBCL6

Plasmid

0

05

1

15

2

25

3

35


Rel

ativ

e gr

owth

siControlsiBCL6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl61

0

02

04

06

08

1

12

14

sicontrol sibcl6

Rel

ativ

e ce

ll vi

abili

ty

0

02

04

06

08

1

12

14

16

siCon

trol2

siCon

trol3

siBCL6

1

siBCL6

2

siSTAT3

Rel

ativ

e vi

abili

ty

BCL6

HSP90

siBCL6 - + - + - + - + - + - +

T47D

0

05

1

15

2

25


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

02

04

06

08

1

12


Rel

ativ

e vi

abili

ty

sicontrolsiBCL6

BCL6

Tubulin


STAT3

BCL6

Tubulin


STAT3

SK-BR-3

MCF-7 ZR-75

BT-549


V B S3 V B S3

sicontrolbcl6

sicontrolstat3

sibcl6bcl6

sibcl6stat3



5











6







0

02

04

06

08

1

12

cp 5 cp10

cp15

cp20

bpi5

bpi10

bpi15

bpi20

Rel

ativ

e vi

abili

ty

T-47d

SKBR3

MDA-468

0

1

2

3

4

5

6


Rel

ativ

e m

RN

A e

xpre

ssio


PARP

RI-BPI - +

HSP90

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleRI-BPI

002040608

112

sicontrol sijak2

Rel

ativ

e vi

abili

ty vehicleRI-BPI

vehicle

RI-BPI

002040608

112

0 2 3 4uM TG101348

Rel

ativ

e vi

abili

ty vehicleBPI

002040608

112


Rel

ativ

e vi

abili

ty

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleTG101348

05 41

12941

520 45

62 373



7















8










































9











10





BCL6

STAT3

STAT5

siControl siBCL6

Plasmid

0

05

1

15

2

25

3

35


Rel

ativ

e gr

owth

siControlsiBCL6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl61

0

02

04

06

08

1

12

14

sicontrol sibcl6

Rel

ativ

e ce

ll vi

abili

ty

0

02

04

06

08

1

12

14

16

siCon

trol2

siCon

trol3

siBCL6

1

siBCL6

2

siSTAT3

Rel

ativ

e vi

abili

ty

BCL6

HSP90

siBCL6 - + - + - + - + - + - +

T47D

0

05

1

15

2

25


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

05

1

15

2

25

3


Rel

ativ

e gr

owth

sicontrol

sibcl6

0

02

04

06

08

1

12


Rel

ativ

e vi

abili

ty

sicontrolsiBCL6

BCL6

Tubulin


STAT3

BCL6

Tubulin


STAT3

SK-BR-3

MCF-7 ZR-75

BT-549


V B S3 V B S3

sicontrolbcl6

sicontrolstat3

sibcl6bcl6

sibcl6stat3



5











6







0

02

04

06

08

1

12

cp 5 cp10

cp15

cp20

bpi5

bpi10

bpi15

bpi20

Rel

ativ

e vi

abili

ty

T-47d

SKBR3

MDA-468

0

1

2

3

4

5

6


Rel

ativ

e m

RN

A e

xpre

ssio


PARP

RI-BPI - +

HSP90

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleRI-BPI

002040608

112

sicontrol sijak2

Rel

ativ

e vi

abili

ty vehicleRI-BPI

vehicle

RI-BPI

002040608

112

0 2 3 4uM TG101348

Rel

ativ

e vi

abili

ty vehicleBPI

002040608

112


Rel

ativ

e vi

abili

ty

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleTG101348

05 41

12941

520 45

62 373



7















8










































9











10











6







0

02

04

06

08

1

12

cp 5 cp10

cp15

cp20

bpi5

bpi10

bpi15

bpi20

Rel

ativ

e vi

abili

ty

T-47d

SKBR3

MDA-468

0

1

2

3

4

5

6


Rel

ativ

e m

RN

A e

xpre

ssio


PARP

RI-BPI - +

HSP90

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleRI-BPI

002040608

112

sicontrol sijak2

Rel

ativ

e vi

abili

ty vehicleRI-BPI

vehicle

RI-BPI

002040608

112

0 2 3 4uM TG101348

Rel

ativ

e vi

abili

ty vehicleBPI

002040608

112


Rel

ativ

e vi

abili

ty

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleTG101348

05 41

12941

520 45

62 373



7















8










































9











10







0

02

04

06

08

1

12

cp 5 cp10

cp15

cp20

bpi5

bpi10

bpi15

bpi20

Rel

ativ

e vi

abili

ty

T-47d

SKBR3

MDA-468

0

1

2

3

4

5

6


Rel

ativ

e m

RN

A e

xpre

ssio


PARP

RI-BPI - +

HSP90

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleRI-BPI

002040608

112

sicontrol sijak2

Rel

ativ

e vi

abili

ty vehicleRI-BPI

vehicle

RI-BPI

002040608

112

0 2 3 4uM TG101348

Rel

ativ

e vi

abili

ty vehicleBPI

002040608

112


Rel

ativ

e vi

abili

ty

002040608

112

sicontrol sibcl6

Rel

ativ

e vi

abili

ty

vehicleTG101348

05 41

12941

520 45

62 373



7















8










































9











10















8










































9











10










































9











10











10


the transcriptional modulator bcl6 as a molecular target for breast cancer therapy

Documents