c-myc drives breast cancer metastasis to the brain, but … · centroid linkage clustering. for...

12
Oncogenes and Tumor Suppressors c-MYC Drives Breast Cancer Metastasis to the Brain, but Promotes Synthetic Lethality with TRAIL Ho Yeon Lee 1 , Junghwa Cha 2 , Seon Kyu Kim 3 , Jun Hyung Park 1 , Ki Hoon Song 4 , Pilnam Kim 2 , and Mi-Young Kim 1,5 Abstract Brain metastasis in breast cancer is particularly deadly, but effective treatments remain out of reach due to insuf- cient information about the mechanisms underlying brain metastasis and the potential vulnerabilities of brain-metastatic breast cancer cells. Here, human breast cancer cells and their brain-metastatic derivatives (BrMs) were used to investigate synthetic lethal interactions in BrMs. First, it was demonstrated that c-MYC activity is increased in BrMs and is required for their brain-metastatic ability in a mouse xenograft model. Specically, c-MYC enhanced brain metastasis by facilitating the following processes within the brain microenvironment: (i) invasive growth of BrMs, (ii) macrophage inltration, and (iii) GAP junction formation between BrMs and astrocytes by upregulating connexin 43 (GJA1/Cx43). Furthermore, RNA- sequencing (RNA-seq) analysis uncovered a set of c-MYCregulated genes whose expression is associated with higher risk for brain metastasis in breast cancer patients. Paradox- ically, however, increased c-MYC activity in BrMs rendered them more susceptible to TRAIL (TNF-related apoptosis- inducing ligand)induced apoptosis. In summary, these data not only reveal the brain metastasis-promoting role of c-MYC and a subsequent synthetic lethality with TRAIL, but also delineate the underlying mechanism. This suggests TRAIL-based approaches as potential therapeutic options for brain-metastatic breast cancer. Implications: This study discovers a paradoxical role of c-MYC in promoting metastasis to the brain and in rendering brain- metastatic cells more susceptible to TRAIL, which suggests the existence of an Achilles' heel, thus providing a new therapeutic opportunity for breast cancer patients. Introduction To successfully form distant metastases, migrating cancer cells must adapt to new microenvironments within secondary organs. This process of adaptation typically involves changes in gene expression and in the signal transduction proles of cancer cells that will eventually give rise to the distinct features of organ-tropic metastatic cells (1). Although several genes that contribute to organ-specic metastasis have been identied, little is known about whether these genes can cause synthetic lethality in organ-specic metastatic cancer cells when combined with anti- cancer agents. Synthetic lethality in cancer drug discovery refers to the selective killing of cancer cells harboring a specic mutation by compromising the function of a secondary gene (2). This approach is widely used as an alternative to directly targeting the oncogene, which is often challenging. For example, inhibition of PLK1, STK33, or TBK1 can induce death in several types of cancer cells containing KRAS mutations (35). Synthetic lethal approaches are also being used to discover therapeutic agents that cause lethality in cancer cells containing molecular lesions in tumor suppressor genes. The best-known example is the synthetic lethality of cancers lacking breast cancer 1 or 2 (BRCA1 or 2) induced by treatment with an inhibitor of poly (ADP-ribose) polymerase1 (PARP1; refs. 6, 7). Although the studies mentioned above provide insight into the synthetic lethality of cancers with specic genetic lesions, syn- thetic lethal interactions in organ-tropic metastatic cancer cells are largely unexplored. Breast cancer metastasizes to several organs including lungs, brain, bones, and liver, and a majority of breast cancer mortality is attributed to metastasis (8, 9). Especially, brain metastasis is generally associated with shorter survival (10) and there are no effective treatments for this disease. Thus, in the present study, we investigated synthetic lethal interactions in brain-metastatic breast cancer. We found that c-MYC function is essential for breast cancer metastasis to the brain, but increased c-MYC activity makes brain-metastatic breast cancer cells (BrM-BCC) highly sensitive to 1 Department of Biological Sciences, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea. 2 Department of Bio and Brain Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea. 3 Personalized Genomic Medicine Research Center, Korea Research Institute of Bioscience & Biotechnology (KRIBB), Daejeon, Republic of Korea. 4 Ajou University, Suwon, Republic of Korea. 5 KAIST Institute for the BioCentury, Cancer Metastasis Control Center, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea. Note: Supplementary data for this article are available at Molecular Cancer Research Online (http://mcr.aacrjournals.org/). J. Cha and S.K. Kim contributed equally to this article. Corresponding Author: Mi-Young Kim, Korea Advanced Institute of Science and Technology, 291 Gwahakro, Daejeon 305-701, Republic of Korea. Phone: 82-42- 350-2615; Fax: 82-42-350-2610; E-mail: [email protected] doi: 10.1158/1541-7786.MCR-18-0630 Ó2018 American Association for Cancer Research. Molecular Cancer Research Mol Cancer Res; 17(2) February 2019 544 on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Upload: others

Post on 20-Mar-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

Oncogenes and Tumor Suppressors

c-MYC Drives Breast Cancer Metastasisto the Brain, but Promotes SyntheticLethality with TRAILHo Yeon Lee1, Junghwa Cha2, Seon Kyu Kim3, Jun Hyung Park1, Ki Hoon Song4,Pilnam Kim2, and Mi-Young Kim1,5

Abstract

Brain metastasis in breast cancer is particularly deadly,but effective treatments remain out of reach due to insuf-ficient information about the mechanisms underlyingbrain metastasis and the potential vulnerabilities ofbrain-metastatic breast cancer cells. Here, human breastcancer cells and their brain-metastatic derivatives (BrMs)were used to investigate synthetic lethal interactions inBrMs. First, it was demonstrated that c-MYC activity isincreased in BrMs and is required for their brain-metastaticability in a mouse xenograft model. Specifically, c-MYCenhanced brain metastasis by facilitating the followingprocesses within the brain microenvironment: (i) invasivegrowth of BrMs, (ii) macrophage infiltration, and (iii)GAP junction formation between BrMs and astrocytes byupregulating connexin 43 (GJA1/Cx43). Furthermore, RNA-sequencing (RNA-seq) analysis uncovered a set of c-MYC–

regulated genes whose expression is associated with higherrisk for brain metastasis in breast cancer patients. Paradox-ically, however, increased c-MYC activity in BrMs renderedthem more susceptible to TRAIL (TNF-related apoptosis-inducing ligand)–induced apoptosis. In summary, thesedata not only reveal the brain metastasis-promoting roleof c-MYC and a subsequent synthetic lethality with TRAIL,but also delineate the underlying mechanism. This suggestsTRAIL-based approaches as potential therapeutic options forbrain-metastatic breast cancer.

Implications: This study discovers a paradoxical role of c-MYCin promoting metastasis to the brain and in rendering brain-metastatic cells more susceptible to TRAIL, which suggests theexistence of an Achilles' heel, thus providing a new therapeuticopportunity for breast cancer patients.

IntroductionTo successfully form distant metastases, migrating cancer cells

must adapt to new microenvironments within secondary organs.This process of adaptation typically involves changes in geneexpression and in the signal transduction profiles of cancer cellsthat will eventually give rise to the distinct features of organ-tropicmetastatic cells (1). Although several genes that contribute toorgan-specific metastasis have been identified, little is knownabout whether these genes can cause synthetic lethality in

organ-specific metastatic cancer cells when combined with anti-cancer agents.

Synthetic lethality in cancer drug discovery refers to theselective killing of cancer cells harboring a specific mutationby compromising the function of a secondary gene (2). Thisapproach is widely used as an alternative to directly targeting theoncogene, which is often challenging. For example, inhibitionof PLK1, STK33, or TBK1 can induce death in several types ofcancer cells containing KRAS mutations (3–5). Synthetic lethalapproaches are also being used to discover therapeutic agents thatcause lethality in cancer cells containing molecular lesions intumor suppressor genes. The best-known example is the syntheticlethality of cancers lacking breast cancer 1 or 2 (BRCA1 or 2)induced by treatment with an inhibitor of poly (ADP-ribose)polymerase1 (PARP1; refs. 6, 7).

Although the studiesmentioned above provide insight into thesynthetic lethality of cancers with specific genetic lesions, syn-thetic lethal interactions in organ-tropicmetastatic cancer cells arelargely unexplored.

Breast cancer metastasizes to several organs including lungs,brain, bones, and liver, and amajority of breast cancermortality isattributed to metastasis (8, 9). Especially, brain metastasis isgenerally associated with shorter survival (10) and there are noeffective treatments for this disease. Thus, in the present study, weinvestigated synthetic lethal interactions in brain-metastaticbreast cancer.We found that c-MYC function is essential for breastcancermetastasis to the brain, but increased c-MYC activitymakesbrain-metastatic breast cancer cells (BrM-BCC) highly sensitive to

1Department of Biological Sciences, Korea Advanced Institute of Science andTechnology (KAIST), Daejeon, Republic of Korea. 2Department of Bio and BrainEngineering, Korea Advanced Institute of Science and Technology (KAIST),Daejeon, Republic of Korea. 3Personalized Genomic Medicine Research Center,Korea Research Institute of Bioscience & Biotechnology (KRIBB), Daejeon,Republic of Korea. 4Ajou University, Suwon, Republic of Korea. 5KAIST Institutefor the BioCentury, Cancer Metastasis Control Center, Korea Advanced Instituteof Science and Technology (KAIST), Daejeon, Republic of Korea.

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

J. Cha and S.K. Kim contributed equally to this article.

CorrespondingAuthor:Mi-YoungKim, KoreaAdvanced Institute of Science andTechnology, 291 Gwahakro, Daejeon 305-701, Republic of Korea. Phone: 82-42-350-2615; Fax: 82-42-350-2610; E-mail: [email protected]

doi: 10.1158/1541-7786.MCR-18-0630

�2018 American Association for Cancer Research.

MolecularCancerResearch

Mol Cancer Res; 17(2) February 2019544

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 2: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

TRAIL-induced apoptosis. Thus, our study provides evidence ofa synthetic lethal interaction between c-MYC–induced brain-metastatic potential and TRAIL.

Materials and MethodsExpression constructs

c-MYCoverexpression construct: pCDH-puro-c-MYC(AddgeneNo. 46970; ref. 11). c-MYC shRNA construct: pRetrosuper-puro-shMYC (Addgene No. 15662; ref. 12). MISSION shRNA (TRCN0000039640, Sigma-Aldrich) GJA1 overexpression construct:pLVX-IRES-hygro-GJA1 (13).

Cell viability assaysMDA-MB-231, MCF7, and their metastatic derivatives were

treated with TRAIL of for 24 and 72 hours, respectively. Cellviability was measured using the CellTiter-Glo system accordingto the manufacturer's protocol (Promega).

Animal studiesAll animal works were done in accordance with a protocol

approved by the KAIST Institutional Animal Care and UsageCommittee. For subcutaneous injections, 4 � 105 of 231-BrMcells were mixed with growth factor–reduced matrigel (3:1), fol-lowedby injection into the lowerflanksof 6- to8-week-oldBALB/cNude mice (Orient Bio). Tumor volumes were calculatedusing (width)2 � (length)/2. For brain metastasis assays, 1 � 105

of 231-BrMor 2� 105 of 231-Par cells were intracardiacally injectedand extracted brains were imaged with IVIS system. MCF7-HER2cells (1 � 106) were intracardiacally injected after pretreating micewith 10mgofb-Estradiol (Sigma-Aldrich) for 3days.b-Estradiolwasadministrated everyday of imaging until the completion of theexperiments. Brain metastases were assessed based on the weightloss as well as the presence of behavioral abnormalities. For TRAILtreatment assays, mice were treated with 500 mg of APO2/TRAIL byintraperitoneal bolus injection every day for 14 days, starting 8 daysafter intracardiac injection of 231-BrM.

Cell culture and cell line generationMDA-MB-231 and MCF7-HER2 cell lines were cultured in

Dulbecco's modified Eagles medium (DMEM) and RPMI1640with 10% FBS, respectively. Human astrocytes were purchasedfrom ScienCell and cultured in complete Astrocyte Media(ScienCell). c-MYC knockdown and rescued 231-BrM cell lineswere generated by using retro- and lentiviral system, respectively.c-MYC–overexpressing 231-Par and MCF7-Par and Cx43-rescuedcell lines were generated by a lentiviral system. Cell lines wererecently authenticated by DNA fingerprinting analysis, andMycoplasma was periodically tested by e-myco plus detection kit(Intron).

Glutamine consumption assayConditioned media (CM) were collected 24 hours after cell

plating and glutamine concentration were measured with Gluta-mine/Glutamate-Glo Assay kit (Promega) according to the man-ufacturer's protocol. Glutamine consumption rate was calculatedby [glutamine] control media � [glutamine] CM and normalized bycell numbers.

Bax oligomerization assaysCells were homogenized by using a 23-gauge syringe and

centrifuged at 2,500 RPM for 10 minutes at 4�C. Supernatantwas collected and recentrifuged at 8,000 RPM for 10 minutes at

4�C. Pellet was resuspended with DPBS and treated with BMH(final 1 mmol/L) for 30 minutes in ice. Samples were quenchedwith 25 mmol/L DTT for 10 minutes at room temperature andsubjected to Western blotting analysis.

Immunostaining and microscopic analysisBrains were isolated and embedded in OCT (Leica). Ten-mm-

thick sectionswere subjected to immune-fluorescent stainingwithCD45 and F4/80 antibodies and microscopic analysis was per-formed using Zeiss Imager M1 microscope at�20 magnification.

Microwell-based three-dimensional (3D) tumor sphereformation assay

The agarose was dissolved by boiling inDMEMand stamped tofabricate microwell to generate nonadherent surfaces. Single cellswere seeded onto microwell and cultured for 7 days. Microscopicanalysis was performed with phase-contrast microscope (Leica).The area and perimeter of each sphere was measured by usingImageJ software. The circularity of spheres was calculated by 4p�(area)/(perimeter)2.

3D invasion assaysTumor spheres were embedded in hyaluronic acid (HA)-

collagen1 hydrogels. After 72 hours, spheres were fixed with4% paraformaldehyde (Sigma-Aldrich) and permeabilized withTriton X-100 (Sigma-Aldrich). Spheres were then stained withDAPI and Phalloidin (Sigma-Aldrich). Microscopic analysis wasperformedwith phase-contrastmicroscope (Leica) and a confocalmicroscope (Nikon). The invading area for initial and final timepoint was measured using ImageJ software. The invasion wasquantified by calculating the ratio of invading area changes frominitial tofinal time point, comparingwith the initial invading area(A ¼ (Afinal � Ainitial) � 100/Ainitial).

Macrophage (RAW 264.7) recruitment assayTumor spheres and RAW 264.7 cells were labeled with

CellTracker Green CMFDA and Red CMTPX (Invitrogen), respec-tively. Tumor spheres were embedded with HA-collagen type Ihydrogels. RAW 264.7 cells mixed with the same matrix wereadded onto tumor sphere–containing HA-collagen matrix. RAWcell infiltration was observed for 24 hours using confocal micro-scope (Nikon). The infiltrated Raw 264.7 cells were quantified bycounting the infiltrated cells area inside the boundary ofmicrowell.

Dye transfer assaySuspended cancer cells were labeled with 3 mmol/L of Calcein

Red-Orange, AMdye (Invitrogen) for 30minutes at 37�C. Labeledcancer cellsweremixedwith unlabeled human astrocytes at a ratioof 4:1 and incubated for 5 hours at 37�C. Dye transferred astro-cytes were analyzed using FACS LSRFortessa.

qRT-PCRFour nanograms of cDNA was subjected to qRT-PCR by using

anCFX-96Real-TimePCRSystem (Bio-Rad).HPRTwas used as anendogenous control. Primer sequences are provided in Supple-mentary Materials and Methods.

Genome-wide RNA-seq analysis and MYC-BrMGS derivationRNAs isolated from 231 cell lines were subjected to strand-

specific whole transcriptome analysis by Illumina Next-seq 500using 75 nt single-end methodology (SE75). The RNA sequence

Achilles' Heel of Brain-Metastatic Breast Cancer

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 545

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 3: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

readswere aligned to thehumanGRCh38using the STAR software(ver. 2.5). For quantification of mRNA, the counts per millionfragmentsmappedof each samplewere calculated. EdgeRpackagethat uses a negative binomial model was used to detect differen-tially expressed genes from RNA-seq count data. Genes with thefold difference of >1.5.and P value <0.001 between shcntr andshMYC as well as shMYC-Vec and rescue groups were selected. Agene set composed of these genes (total 125) and c-MYC wasdefined as MYC-BrMGS. The data are available in the NCBI Gene-Expression Omnibus public database under accession numberGSE106312.

Clinical sample analysisEMC 192 (GSE12276) and EMC 286 (GSE2034) data sets were

used to analyze association between c-MYC-BrMGS and brainmetastasis-free survival (BrMFS). To divide breast cancer patientsinto subgroups based on expression levels of c-MYC-BrMGS, ahierarchical clustering algorithm was applied that used the cen-tered correlation coefficient as the measure of similarity andcentroid linkage clustering. For cluster analysis, gene-expressiondata were normalized by the quantile method, log2-transformed,andmedian-centered across genes and samples. TheKaplan–Meiermethod was used to calculate the time to metastasis, and differ-ences between the times were assessed using log-rank statistics.

FACS analysis for apoptosisCells were treated with 10 ng/mL of TRAIL for 2 hours and

stained with Annexin V- Alexa488 (Invitrogen) and 7-AAD solu-tion (BD), followed by flow cytometry analysis.

Statistical analysisStatistical analysis was performed by an unpaired Student t test,

two-way ANOVA, Mann–Whiney tests, or log-rank test as indi-cated in each figure legend. P values of <0.05 were considered asstatistically significant. Results are reported as mean � SEM.

Detailed information is available in Supplementary Materialsand Methods.

ResultsElevated c-MYC activity in highly BrM-BCCs is required for theirbrain-metastatic ability

The oncogenic function of c-MYC is well established, and itsrole in cancer metastasis has been documented in some cancertypes (14–17). In the case of breast cancer metastasis, both lungmetastasis-promoting and -suppressive functions of c-MYC havebeen reported (18–20).However, whether or not c-MYCactivity isaltered in BrM-BCCs and this is required for breast cancer metas-tasis to the brain remains mostly unexplored.

To investigate this, we first examined whether highly brain-metastatic BCCs have differential expression levels of c-MYCcompared with poorly brain-metastatic BCCs. We used MDA-MB-231 and MCF7-HER2 (Erb-B2 Receptor Tyrosine Kinase 2)human breast cancer cell lines as model systems because of theavailability of the low metastatic parental population (231-Parand MCF7-Par) and highly brain-metastatic sublines (231-BrMand MCF7-BrM3; refs. 21, 22). c-MYC protein levels wereincreased in 231-BrM cells compared with 231-Par cells withno increase in its transcript levels (Fig. 1A). Increased c-MYC in

Figure 1.

Elevated c-MYC activity in BrM-BCCs is requiredfor their brain-metastatic ability. A, Relativec-MYC protein (left) and mRNA levels (right) in231-Par and -BrM cells by immunoblottinganalysis and qRT-PCR, respectively. P values,two-tailed unpaired Student t test.B, Subcutaneous tumor growth rates of control(shcntr) and c-MYC knockdown (shMYC)231-BrM cells. Tumor volumes were measured atthe indicated days. C and D, Relative brain-metastatic abilities of c-MYC knockdown231-BrM (C) and c-MYC–overexpressing 231-Par(D), compared with their corresponding controlcell lines. Top, c-MYC protein levels in theindicated cells prior to the injection. Bottom,cells were injected into the left ventricle ofimmunodeficient mice. At the indicated days,luminescent signals from extracted brains weremeasured by using bioluminescence imaging(BLI) system. Normalized BLI signals inextracted brains and representative BLI imagesof brains are shown. P values, one-tailedMann–Whitney test. All data representmean � SEM. �, P < 0.05.

Lee et al.

Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research546

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 4: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

231-BrM #2 cells, another independent brain-metastatic variantof MDA-231, was also observed but from the transcript level(Supplementary Fig. S1A). In the MCF7-HER2 system, MCF7-Parand BrM3 showed similar total c-MYC protein levels, but MCF7-BrM3 exhibited an increased pS62/pT58-MYC ratio comparedwithMCF7-Par (Supplementary Fig. S1B). It has been shown thatactivity of c-MYC can be regulated by relative levels betweenphospho-S62-MYC (pS62-MYC) and phospho-T58-MYC (pT58),in which a high pS62/pT58-MYC ratio indicates the activatedc-MYC pathway (23). Thus, our data suggest that the c-MYCpathway is more active in MCF7-BrM3 compared with the Parcells. Further supporting this, MCF7-BrM3 showed increasedglutamine consumption rates, which is consistent with previousdata that overexpression of c-MYCpromotes glutamine consump-tion in MCF7 (refs. 24, 25; Supplementary Fig. S1C).

On the basis of our data indicating the activated c-MYCpathway in BrM-BCCs, we tested whether c-MYC is requiredfor breast cancer metastasis to the brain. c-MYC knockdown in231-BrM had no effect in their general growth rate as subcu-taneous tumors (Fig. 1B). However, c-MYC knockdown led to asignificant decrease in the formation of brain metastases (Fig.1C; Supplementary Fig. S1D). Consistent with this, overexpres-sion of c-MYC in 231-Par cells increased their ability to formbrain metastases (Fig. 1D). Similar results were observed withc-MYC–overexpressing MCF7-Par (Supplementary Fig. S1E).Collectively, our data suggest that c-MYC is a key player inbreast cancer metastasis to the brain.

Increased c-MYC function in BrM-BCCs is essential for invasiveoutgrowth in the brain microenvironment

Next, we investigated the mechanism by which increasedc-MYC activity enhances the ability of BrM-BCCs to form brainmetastases. Brain-metastatic cancer cells exhibit expansive growthwhen growing as spheres, and this phenotype is correlated wellwith their invasiveness in a 3D matrix (26, 27). To determinewhether c-MYC contributes to this process, we used a nonadher-ent microwell platform, each well of which is designed to containa single tumor sphere (Fig. 2A, schematic). In this assay, whereascontrol 231-BrM cells formed spheres with disorganizedmorphologies (i.e., expansive or grape-like structures; ref. 27),c-MYC–depleted 231-BrM cells produced much more compactand circular spheres, which was reversed upon c-MYC rescue (Fig.2A and B; Supplementary Fig. S2A). Consistent with this, over-expression of c-MYC in MCF7-Par cells led to formation ofdisorganized spheres as demonstrated by decreased circularity(Supplementary Fig. S2B).

On the basis of the aforementioned association betweenexpansive growth and invasiveness, we next investigated therole of c-MYC in the invasive growth of BrM-BCCs. We used apublished spheroid-based invasion assay in which tumorspheres are embedded in a 3D matrix composed of collagen(type 1) and HA, one of the major components in brainextracellular matrix (28, 29). 231-BrM cells exhibited moreinvasive behavior in this HA-collagen matrix compared withthe 231-Par (Supplementary Fig. S2C), but invasive growth of

Figure 2.

Increased c-MYC function in BrM-BCCs isessential for invasive outgrowth in the brainmicroenvironment. A (top), A simplifiedschematic of the microwell system used inthis study. Single cells were seeded ontononadherent microwell, and bright fieldimages were taken at day 7. A (bottom),Representative images of spheres formedbythe indicated 231-BrM cells. B, Thequantification of sphere circularity from A.Lower circularity indicates less compactand disorganized spheres. The circularity ofspheres was calculated by 4p � (area)/(perimeter)2. n ¼ 3. C and D, Comparison ofinvasive growth of the indicated 231-BrMcells. Tumor spheres were embedded withinthematrix composed of collagen (type1) andHA and stained with Phalloidin. Images weretaken by using a confocal microscope (C).The invasion was quantified by calculatingthe changes of invaded areas from 0 hours(A0) to 72 hours (A72; D). Invaded area(A) ¼ (A72 – A0) � 100/A0. n ¼ 3. P values,two-tailed unpaired Student t test. All datarepresent mean � SEM. � , P < 0.05;�� , P < 0.01; ���� , P < 0.0001.

Achilles' Heel of Brain-Metastatic Breast Cancer

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 547

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 5: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

231-BrM was significantly reduced upon knockdown of c-MYCand restored upon c-MYC rescue (Fig. 2C and D). Together,these data suggest c-MYC promotes the invasive growth ofBrM-BCCs in the brain microenvironment.

c-MYC creates a favorable brain microenvironment bypromoting macrophage infiltration

In addition to changes in their intrinsic properties such asinvasive growth, BrM-BCCs can also modify their own microen-vironment by modulating the infiltration of inflammatory cells(e.g., macrophages) to improve their survival and growth in thebrain (10, 30). To determine whether c-MYC also plays a role inmodulation of the brain microenvironment, we looked forchanges in the stromal composition of brain lesions arising fromc-MYC–depleted 231-BrM cells. Similar to our observation in vitro(refer to Fig. 2A), lesions formed by control 231-BrM cells exhib-ited more disorganized and expanded shape whereas thoseformed by c-MYC–depleted BrM cells showed more compactstructure (Fig. 3A, green). More importantly, immuno-stainedbrain sections revealed that lesions formed by control 231-BrMcells contained large numbers of leukocytes (CD45þ), includingmacrophages (F4/80þ; Fig. 3A, red). In contrast, brain lesionsformed by c-MYC–depleted 231-BrM cells only contained veryfew CD45þ and F4/80þ cells (Fig. 3A, red).

To further support c-MYC's role in macrophage recruitment inthe brain, we used a 3D macrophage recruitment assay (Fig. 3B,experimental scheme). Briefly, macrophages (RAW 264.7) wereadded to a HA–collagen matrix in which tumor spheres wereembedded, and their migration toward tumor spheres was mon-itored. Consistent with our in vivo staining data, 231-BrM cellscaused rapid infiltration of RAW 264.7 toward tumor spheres inthe HA–collagen matrix. Importantly, c-MYC knockdown dra-matically reduced thismacrophage infiltration, but it was restoredupon c-MYC rescue (Fig. 3B). Similarly, overexpression of c-MYCinMCF7-Par increased RAW cell recruitment (Supplementary Fig.S3).Ourdata suggest that elevated c-MYC inBrM-BCCs induce theinfiltration of macrophages within brain metastases.

c-MYC promotes GAP junction formation between BrM-BCCsand astrocytes via upregulation of Cx43

To better understand how c-MYC promotes brain metastasis inbreast cancer, we compared the transcriptomes of 231-BrM cellsexpressing shcntr, shMYC, shMYC-Vec, and c-MYC rescue con-structs by using genome-wide RNA-sequencing. Through thisanalysis, we identified 20 and 105 genes whose expression wereupregulated anddownregulated by c-MYC, respectively (P<0.001and fold change > 1.5; Supplementary Table S1).

To identify the genes that mediate c-MYC's role in promotingbrain metastasis, we compared our gene set with a previouslyreported list of genes that are differentially expressed in 231-BrMand CN34-BrM (BrM subline derived from CN34 breast cancercell line) compared with their corresponding Par lines (21). Thisanalysis identified 4 genes (1 up and 3 down) that were differ-entially expressed in 231-BrM cells compared with 231-Par andregulated by c-MYC (Fig. 4A). c-MYC–dependent expression ofthese four genes was verified by qRT-PCR (Fig. 4B; SupplementaryFig. S4A and S4B; Supplementary Table S1). From this analysis, wefound that major gap junction protein GJA1 (Cx43, connexin43)was upregulated by c-MYC in BrM-BCCs. Cx43-mediated gapjunction formation between BCCs and astrocytes has recentlybeen suggested to support BCC growth in the brain (13). How-

ever, the upstream regulator of Cx43 in BrM-BCCs has not beenidentified. Thus, we hypothesized that c-MYCmay promote brainmetastasis by stimulating gap junction formation betweenBrM-BCCs and astrocytes and that Cx43 functions as a down-stream mediator of c-MYC in this process.

To test this hypothesis, we performed dye (calcein) transferassays, which allowed us tomonitor gap junction formation (13).Although c-MYC knockdown significantly reduced dye transferfrom231-BrM to astrocytes, c-MYC rescue restored thedye transferactivity (Fig. 4C and D). This indicates that c-MYC function isrequired for the gap junction formation between BrM-BCCs andastrocytes. Furthermore, the lost dye transfer ability of c-MYC–depleted 231-BrM cells was fully recovered upon the ectopicexpression of Cx43 (Fig. 4E and F; Supplementary Fig. S4C).Collectively, these data suggest that c-MYCpromotes gap junctionformation through upregulating CX43 expression.

Activated c-MYC pathway predicts brain metastasis in breastcancer patients

On the basis of our discovery of a role for c-MYC in promotingbrain metastasis, we next asked whether c-MYC is clinicallyassociated with the risk of brain metastasis in breast cancerpatients. To this, we used publicly available breast cancer micro-array data sets (EMC192 andEMC286) that included informationabout BrMFS (21, 31). In both EMC192 and EMC286, c-MYCtranscript levels showed no association with BrMFS (Fig. 5A).

We suspected that c-MYC transcript levels may not represent asufficient reflection of active c-MYC signaling because our exper-imental data suggest that the c-MYC pathway in BrM-BCCs canalso be activated by an increased c-MYC protein level and bychanges in phosphorylation status of the protein (refer to Fig 1A;Supplementary Fig. S1B).

To address this, we performed a similar analysis as above butusing a gene set that is composed of 125 c-MYC–regulated genes,identified in our RNA-seq analysis (Supplementary Table S1), andc-MYC. We named this gene set the "c-MYC brain metastasis geneset (MYC-BrMGS)." Compared with c-MYC expression alone,MYC-BrMGS showed a significant association with the risk ofbrain metastasis in EMC192 and EMC286 data sets (Fig. 5B). Incontrast to increased risks for brain metastasis in MYC-BrMGS–positive patients, the association between MYC-BrMGS and lungmetastasis-free was observed only in EMC 192 (SupplementaryFig. S5A). In case of bonemetastasis-free survival (BoMFS), MYC-BrMGS–positive group showed lower risks for the bonemetastasisin EMC 286, which is opposite of what we found with brainmetastasis (Supplementary Fig. S5B).

Thus, clinical data analyses in combination with our experi-mental data strongly suggest that increased c-MYC activity isrequired for the brain metastasis in breast cancer.

Increased c-MYC activity in BrM-BCCs sensitizes them toTRAIL-induced apoptosis

Changes inmolecular features of cancer cells can alter their drugresponsiveness. Because we found that the c-MYC pathway ishighly activated in BrM-BCCs, we wondered whether this causeschanges in their sensitivity to anticancer drugs. To test this, we firstcompared the responsiveness of 231-Par and BrM to the mostcommonly used chemotherapeutic agents for breast cancer (doc-etaxel, doxorubicin, and methotrexate) and observed no differ-ences between 231-Par and BrM (Supplementary Fig. S6A).

Lee et al.

Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research548

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 6: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

However, 231-BrM cells, but not 231-LM cells (a highly lung-metastatic MDA-MB-231 derivative), exhibited dramaticallyincreased susceptibility to TRAIL-induced cell death comparedwith 231-Par cells (Fig. 6A). FACS analysis confirmed a greaterincrease in apoptotic cells in 231-BrM upon TRAIL treatment(Supplementary Fig. S6B), which is in accordance with previousreports that c-MYC can activate TRAIL-induced apoptosis invarious cell types (2, 25, 32, 33). Increased TRAIL sensitivity wasalso observed with 231-BrM #2 as well as with MCF7-BrM3compared with the corresponding Par cells (Supplementary Fig.S6C and S6D). These data together suggest the synthetic lethalinteraction between BrM-BCCs and TRAIL.

TRAIL-induced apoptosis is initiated by the binding of TRAIL todeath receptors 4 and 5 (DR 4 and 5). This triggers activation ofcaspase-8 and the downstreameffector caspases (caspase-3, 6, and7) through cell type–specific mechanisms (34). In type I cells,direct activation of caspase-3/6/7 by caspase-8, a.k.a. extrinsic

apoptotic pathway, is sufficient to execute apoptosis (35). On theother hand, apoptosis in type II cells requires activation of themitochondrial (intrinsic) pathway. In this pathway, caspase-8–mediated cleavage of BID (BH3 interacting Domain Death Ago-nist) induces BAK/BAXoligomerization and subsequent release ofcytochrome C from mitochondria. This is followed by activationof caspase-9 and caspases 3/6/7 (36).

Upon TRAIL treatment, 231-BrM exhibited faster cleavage ofcaspase-8 (CASP-8), indicating activation of an extrinsic pathway(Fig. 6B). 231-BrM also showed increased BID cleavage (Fig. 6B),faster oligomerization of BAX and BAK (Fig. 6C; SupplementaryFig. S6E), as well as accelerated cleavage of caspase-9 and caspase-3 (Fig. 6B). This suggests that increased TRAIL sensitivity in 231-BrM also involves activation of the intrinsic pathway.

On the basis of this finding, we next examined (i) whetherc-MYC is responsible for the higher TRAIL sensitivity in BrM-BCCsand (2) if so, which apoptotic pathway is regulated by c-MYC.

Figure 3.

c-MYC creates a favorable brainmicroenvironment by promotingmacrophage infiltration. A, Mousebrain sections from metastasis assaysshown in Fig. 1C were subject toimmunofluorescence staining with theindicated antibodies. GFP-positivecells represent tumor cells.Representative microscopic imagesare shown. B, Relative recruitment ofmacrophages (RAW 264.7 cells) bythe indicated 231-BrM cells in theHA–collagen matrix. Tumor spheres(green) were embedded in the matrix,followed by addition of RAW 264.7cells (red) mixed with the samematrix(experimental scheme, top right).Imageswere taken by using a confocalmicroscope. Representative top-viewimages (left) and the normalized RAWcell recruitment (bottom right) areshown. P values, two-tailed unpairedStudent t test. n¼ 3. All data representmean � SEM. � , P < 0.05; ��, P < 0.01.

Achilles' Heel of Brain-Metastatic Breast Cancer

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 549

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 7: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

c-MYC–depleted 231-BrM cells exhibited reduced TRAIL sensi-tivity, and restoring c-MYC expression resensitized them toTRAIL-induced apoptosis (Fig. 6D). Furthermore, overexpressionof c-MYC in 231-Par cells increased their TRAIL sensitivity(Supplementary Fig. S6F). In 231-BrM cells, however, c-MYCoverexpression had no effect on TRAIL sensitivity (Fig. 6D). Thisis probably because endogenous levels of c-MYC are sufficient toinduce maximum TRAIL sensitivity. Consistent with results in231-BrM cells, c-MYC knockdown in MCF7-BrM3 cells reducedtheir TRAIL sensitivity (Supplementary Fig. S6G). Taken together,this suggests the essential role of c-MYC in increased susceptibilityof BrM-BCCs to TRAIL.

Next, we investigated molecular mechanisms underlyingc-MYC–mediated increase in TRAIL sensitivity of BrM-BCCs. Bymonitoring activation of apoptotic pathways, we found that c-MYCknockdown in 231-BrM attenuated cleavage of caspase-8, BID,caspase-9, and caspase-3, which was restored upon c-MYC rescue(Fig. 6E). These indicate that c-MYC activates both extrinsic andintrinsic pathways. Further supporting this finding, DR5 andcaspase-8 inhibitor FLIP was increased and decreased in 231-BrMcomparedwith the231-Par, respectively (Supplementary Fig. S6H),and DR5 increase was reversed by c-MYC knockdown in 231-BrM(Fig. 6F). Finally, accelerated BAX and BAK oligomerizationobserved in 231-BrM was impaired upon c-MYC knockdown and

Figure 4.

c-MYC promotes GAP junctionformation by upregulating Cx43.A, A Venn diagram showing the125 c-MYC–regulated genes in 231-BrMfrom RNA-seq analysis in the presentstudy (left circle) and genes that werepreviously shown to be differentiallyexpressed in 231 and CN34-BrMcompared with their correspondingPar cell lines (right circle). The namesof overlapping genes are indicated.The 125 c-MYC–regulated genes wereselected based on the fold change > 1.5and P < 0.001. B, Relative transcriptlevels of Cx43 in the indicated 231-BrMcells. n ¼ 3. C (top), A schematicdiagram of the dye transfer assay.231-BrM cells were labeledwith calceindye and incubated with unlabeledastrocytes. Calcein-positiveastrocytes were analyzed by flowcytometry. C (bottom), Quantificationof dye transfer from the indicated231-BrM cells to astrocytes. Values arenormalized to shcntr-expressing231-BrM cells. n¼ 3. D, Representativehistograms from experiments in C.The histograms represent redfluorescent signals from astrocytesafter incubation with the indicated231-BrM cell lines. FL2 denotes emittedfluorescence from calcein. The lightgray box represents backgroundsignal from astrocytes and 231-BrMmixture at 0 hours. E and F,Quantification of dye transfer from theindicated 231-BrM cells to astrocytes(E) and representative histograms (F).n¼ 3. All data represent mean� SEM.P values, two-tailed unpaired Studentt test. � , P < 0.05; �� , P < 0.01;���, P < 0.001.

Lee et al.

Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research550

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 8: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

restored by c-MYC rescue (Fig. 6G; Supplementary Fig. S6I).This explains attenuated activation of caspase-9 in 231-BrM uponc-MYC knockdown which we showed in Fig. 6E. Therefore, thesedata together suggest that increased c-MYC in BrM-BCCs sensitizesthem to TRAIL-induced apoptosis via activating both extrinsic andintrinsic pathways.

Finally, as our data indicate that TRAIL may be an effectivedeath inducer of BrM-BCCs, we tested this possibility in vivo. Tothis, 231-BrM cells were introduced into the arterial circulation ofimmune-deficient mice, andmice were treated with either vehicle(water) or TRAIL. Brain metastases were monitored by usingbioluminescence imaging. TRAIL treatment reduced the growthof brain metastases (Fig. 6H), suggesting TRAIL is effective in thetreatment of brain-metastatic breast cancer.

DiscussionEvidence is accumulating that BCC subpopulations acquire

brain-metastatic potential via the activation of distinct molec-ular pathways (21, 37). Despite significant advances in ourunderstanding of the molecular mechanisms that govern breast

cancer metastasis to the brain, currently no effective therapeuticagents are available. This necessitates new approaches to targetthis disease.

In this study, we demonstrate that BrM-BCCs exhibit increasedc-MYC activity, which promotes brain metastasis, and this sub-sequently leads to increased susceptibility to TRAIL-induced apo-ptosis. To the best of our knowledge, this study is the first to reporta link between c-MYC, brain-metastatic potential, and TRAILsensitivity, thus providing evidence of a novel synthetic interac-tion in BrM-BCCs.

We found that c-MYC confers brain-metastatic potential onBCCs without affecting their general tumor-forming capability.This suggests c-MYC may function as a brain virulence gene—agene that promotes BCC colonization within the brain—ratherthan providing BCCs with an advantage to expand within theprimary tumor.

How, then, does c-MYC enable BCCs to formmetastatic lesionwithin the brain? The brain contains a specialized extracellularmatrix (ECM)and cell types including astrocytes (38). To colonizethe brain, incoming BCCs must be able to cope with this uniquebrainmicroenvironment. Our data indicate c-MYC contributes to

Figure 5.

The c-MYC–regulated gene setpredicts brain metastasis in breastcancer patients. A, Kaplan–Meier plotsshowing probability of BrMFS inEMC192 and EMC286 based on theexpression level of c-MYC transcript.c-MYC–high and –low groups werestratified based on the medianexpression value of c-MYC transcript.M indicates the number of patientswith brain metastasis in each group.Total number of patients at risk at eachfollow-up time is also shown (bottom).B, Similar analysis as in A but withusing c-MYC–brain metastasis geneset (MYC-BrMGS). BrMGS-positiveand -negative groups were stratifiedbased on a hierarchical clusteringalgorithm and probability of BrMFSwas analyzed. P values, log-rank test.

Achilles' Heel of Brain-Metastatic Breast Cancer

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 551

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 9: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

the ability of BCCs to colonize the brain not only by promotingtheir invasive growth within the brain ECM, but also by modu-lating and exploiting stromal components such as macrophagesand astrocytes. In this regard, c-MYC in BrM-BCCs enhances theirability to induce macrophage recruitment. This is consistent withprevious reports that c-MYC promotes macrophage infiltrationinto pancreatic cancers (39, 40). Our results along with thesepublished studies strongly suggest c-MYC modulates the tumor

microenvironment, eventually leading to cancer cell survival andgrowth. We suspect this is accomplished via c-MYC–mediatedchanges in cytokines and chemokines. Future studies will providemore insight into the molecular mechanisms by which c-MYC–expressing BCCs induce macrophage influx.

We also discovered a novel role for c-MYC in the formation ofgap junctions between BCCs and astrocytes, which is essential forthe growth and chemoresistance of brain metastases (13). We

Figure 6.

Increased c-MYC activity in BrM-BCCs sensitizes them toTRAIL-induced apoptosis. A, Comparison of cell viabilitybetween 231-Par, BrM, and LM cells upon TRAILtreatment. Cell viability was assessed by the CellTiter-Glo system 24 hours after treatment. n ¼ 3. P values,two-way ANOVA, 231-Par vs. BrM. B, The indicated cellswere treated with 10 ng/mL of TRAIL for 0, 4, or 6 hoursand cleavage of the indicated proteinswasmonitored byimmunoblotting analysis. C, BAX oligomerization in231-Par and -BrM cells upon TRAIL treatment (10 ng/mL,6 hours). Mitochondria was isolated and treated withcross-linking agent BMH for 30 minutes, followed byWestern blotting analysis. 1�, 2x, 3�, and 4� indicatemono-, di-, tri-, and tetramers, respectively. D, TRAILsensitivity (left) and relative c-MYC protein levels (right)of the indicated 231-BrM. Cell viability was assessed24 hours after treatment. P values, two-way ANOVA(cntr vs. shMYC and shMYC vs. c-MYC rescue). n ¼ 3.E, Similar experiments as in Bwith the indicated 231-BrMcells. F, Relative DR5 and FLIP protein levels of theindicated 231-BrM cells. G, Similar experiments asin C with the indicated 231-BrM cells. H, Comparison ofbrain metastasis in mice treated with water or TRAIL.231-BrM cells were intracardiacally introduced into theimmunocompromised mice. After 8 days, mice weretreatedwithwater orAPO2/TRAIL (500mgper injection)every day for 2 weeks. Brains were extracted at day 28after injection and metastatic colonization wasquantified by BLI (top). Representative brain images onthe bottom. P values, one-tailed Mann–Whitney test. Alldata represent mean� SEM. � , P < 0.05; ���� , P < 0.0001.

Lee et al.

Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research552

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 10: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

further showed that this c-MYC–mediated increase in gap junc-tion formation is accomplished via an upregulation of Cx43. Aprevious study reported that Cx43 is upregulated in RAS-expres-sing NIH3T3 cells due to increased occupancy of the Hsp90 andc-MYC complex at the Cx43 promoter (41), which is consistentwith our results.

Our clinical analyses indicate that MYC-BrMGS can be a prog-nostic marker for brain metastasis in breast cancer patients. Weexpect that this is due to the regulation of c-MYC at the posttran-scriptional level (42–44), as we observed in our model system.Consistent with this, one study found elevated levels of c-MYCprotein in brain metastases compared with the matched primarybreast tumors (45).

Elevated c-MYC function in BrM-BCCs increases susceptibilityof these cells to TRAIL-induced apoptosis. Our data demonstratethat the underlying mechanisms of elevated TRAIL sensitivityinclude c-MYC–regulated expression of DR5 as well as activationof the BAX and BAK oligomerization. Consistent with this result,previous studies reported that c-MYC exerts its proapoptoticfunction by upregulating DR5 in mammary epithelial cells (2),by activating BAX in rat fibroblasts and mammary epithelial cells(25, 46) and by promoting BAK oligomerization in MCF7 cells(25). Therefore, c-MYC induces TRAIL sensitivity in BrM-BCCs viaa mechanism that is similar to the one used by mammaryepithelial cells.

If c-MYC promotes brain metastasis but enhances TRAIL sen-sitivity as our data suggest, how do c-MYC–expressing BCCssurvive within the brain in the first place? TRAIL can be secretedby the BCCs themselves and by various organs. Interestingly,TRAIL is downregulated in 231-BrM compared with Par cells(21), suggesting BrM-BCCs that overexpress c-MYC may protectthemselves from apoptosis by suppressing the expression ofTRAIL. In addition, the brain microenvironment produces lessTRAIL than bone and lungs (31). This may help BCCs over-expressing c-MYC survive in the brain microenvironment.

The important clinical implication of our study is that TRAIL-mediated apoptosis may be used to target brain metastases.Despite the efficacy of TRAIL-based therapies in preclinical mod-els, the clinical benefits of these therapies are not yet proven. Thisis typically blamed on TRAIL's short half-life and rapid clearancefrom the body (47). Although recent studies have suggestedsolutions for increasing the efficacy of TRAIL-based therapies(48, 49), insufficient delivery of TRAIL to the brain due to the

blood–brain barrier has remained as a major impediment fortreatment of brainmetastases (50). Several innovative approachesfor resolving this problem have recently proven effective inpreclinical models of glioblastoma and brainmetastases of breastcancer. These include stem cell–mediated delivery of TRAIL andadministration of a TRAIL-inducing compound (50). Thus, ourfindings together with these studies suggest breast cancer patientswith brain metastases may benefit from TRAIL treatment.

Collectively, our results indicate that c-MYC plays an essentialrole in the successful formation of brain metastases and that, as abystander effect, these cells become more susceptible to TRAIL-induced apoptosis.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: H.Y. Lee, J.H. Park, K.H. Song, M.-Y. KimDevelopment of methodology: H.Y. Lee, J.H. ParkAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): H.Y. Lee, J. Cha, S.K. Kim, J.H. Park, K.H. SongAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): H.Y. Lee, J. Cha, S.K. Kim, P. KimWriting, review, and/or revision of the manuscript: H.Y. Lee, J. Cha, P. Kim,M.-Y. KimAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): H.Y. LeeStudy supervision: M.-Y. Kim

AcknowledgmentsThis research was supported by the KIB Cancer Metastasis Control Center

(N1018001), KIAST Grand Challenge Grant (N11180008), the NationalResearch Foundation of Korea (NRF;NRF-2016M3A9B4915818), KoreaHealthTechnology R&D Project through the Korea Health Industry DevelopmentInstitute (KHIDI, HI14C1324), and by KRIBB Research Initiative Program.

We thank Dr. Joan Massagu�e for MDA-MB-231 and its metastatic derivativecell lines andCx43 constructs; Drs. Park,Wang, andEilers for c-MYC and shMYCconstructs, respectively.MCF7-HER2-Par andBrM3 cellswere kindly gifted fromDr. Steeg. Genentech/Amgen provided APO2/TRAIL used in in vivo study.

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received June 19, 2018; revisedAugust 6, 2018; accepted September 19, 2018;published first September 28, 2018.

References1. Jacob LS, Vanharanta S, Obenauf AC, Pirun M, Viale A, Socci ND, et al.

Metastatic competence can emerge with selection of preexisting oncogenicalleles without a need of new mutations. Cancer Res 2015;75:3713–9.

2. Wang Y, Engels IH, Knee DA, Nasoff M, Deveraux QL, Quon KC. Syntheticlethal targeting of MYC by activation of the DR5 death receptor pathway.Cancer Cell 2004;5:501–12.

3. Scholl C, Frohling S, Dunn IF, Schinzel AC, Barbie DA, Kim SY, et al.Synthetic lethal interaction between oncogenic KRAS dependency andSTK33 suppression in human cancer cells. Cell 2009;137:821–34.

4. Smolen GA, Zhang J, Zubrowski MJ, Edelman EJ, Luo B, Yu M, et al. Agenome-wide RNAi screen identifiesmultiple RSK-dependent regulators ofcell migration. Genes Develop 2010;24:2654–65.

5. Barbie DA, Tamayo P, Boehm JS, Kim SY, Moody SE, Dunn IF, et al.Systematic RNA interference reveals that oncogenic KRAS-driven cancersrequire TBK1. Nature 2009;462:108–12.

6. Chan SL, Mok T. PARP inhibition in BRCA-mutated breast and ovariancancers. Lancet 2010;376:211–3.

7. Rios J, Puhalla S. PARP inhibitors in breast cancer: BRCA and beyond.Oncol (Williston Park) 2011;25:1014–25.

8. Chiang AC, Massague J. Molecular basis of metastasis. N Engl J Med2008;359:2814–23.

9. Vanharanta S, Massague J. Origins of metastatic traits. Cancer Cell2013;24:410–21.

10. Zhang C, Yu D. Microenvironment determinants of brain metastasis. CellBiosci 2011;1:8.

11. Cheng Z, Gong Y, Ma Y, Lu K, Lu X, Pierce LA, et al. Inhibition of BETbromodomain targets genetically diverse glioblastoma. Clin Cancer Res2013;19:1748–59.

12. Popov N, Wanzel M, Madiredjo M, Zhang D, Beijersbergen R, Bernards R,et al. The ubiquitin-specific protease USP28 is required for MYC stability.Nat Cell Biol 2007;9:765–74.

13. Chen Q, Boire A, Jin X, Valiente M, Er EE, Lopez-Soto A, et al. Carcinoma-astrocyte gap junctions promote brain metastasis by cGAMP transfer.Nature 2016;533:493–8.

Achilles' Heel of Brain-Metastatic Breast Cancer

www.aacrjournals.org Mol Cancer Res; 17(2) February 2019 553

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 11: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

14. Wolfer A, Ramaswamy S. MYC and metastasis. Cancer Res 2011;71:2034–7.

15. ChoH,Herzka T, ZhengW,Qi J,Wilkinson JE, Bradner JE, et al. RapidCaP, anovel GEM model for metastatic prostate cancer analysis and therapy,reveals myc as a driver of Pten-mutant metastasis. Cancer Discov2014;4:318–33.

16. Rapp UR, Korn C, Ceteci F, Karreman C, Luetkenhaus K, Serafin V, et al.MYC is ametastasis gene for non-small-cell lung cancer. PLoSOne 2009;4:e6029.

17. Lawson DA, Bhakta NR, Kessenbrock K, Prummel KD, Yu Y, Takai K, et al.Single-cell analysis reveals a stem-cell program in humanmetastatic breastcancer cells. Nature 2015;526:131–5.

18. Liu H, Radisky DC, Yang D, Xu R, Radisky ES, Bissell MJ, et al. MYCsuppresses cancer metastasis by direct transcriptional silencing of alphavand beta3 integrin subunits. Nat Cell Biol 2012;14:567–74.

19. Wolfer A,Wittner BS, IrimiaD, FlavinRJ, LupienM,GunawardaneRN, et al.MYC regulation of a "poor-prognosis"metastatic cancer cell state. Proc NatAcad Sci USA 2010;107:3698–703.

20. CappellenD, Schlange T, BauerM,Maurer F,HynesNE.Novel c-MYC targetgenes mediate differential effects on cell proliferation and migration.EMBO Rep 2007;8:70–6.

21. Bos PD, Zhang XH, Nadal C, Shu W, Gomis RR, Nguyen DX, et al.Genes that mediate breast cancer metastasis to the brain. Nature 2009;459:1005–9.

22. Gril B, Palmieri D, Qian Y, Smart D, Ileva L, Liewehr DJ, et al. Pazopanibreveals a role for tumor cell B-Raf in the prevention ofHER2þ breast cancerbrain metastasis. Clin Cancer Res 2011;17:142–53.

23. Myant K, Qiao X, Halonen T, Come C, Laine A, JanghorbanM, et al. Serine62-phosphorylated MYC associates with nuclear lamins and its regulationby CIP2A is essential for regenerative proliferation. Cell Rep 2015;12:1019–31.

24. WiseDR,DeBerardinis RJ,MancusoA, SayedN, ZhangXY, PfeifferHK, et al.Myc regulates a transcriptional program that stimulates mitochondrialglutaminolysis and leads to glutamine addiction. Proc Nat Acad Sci USA2008;105:18782–7.

25. Nieminen AI, Eskelinen VM, Haikala HM, Tervonen TA, Yan Y, Partanen JI,et al. Myc-induced AMPK-phospho p53 pathway activates Bak to sensitizemitochondrial apoptosis. Proc Nat Acad Sci USA 2013;110:E1839–48.

26. Kenny PA, Lee GY,Myers CA, Neve RM, Semeiks JR, Spellman PT, et al. Themorphologies of breast cancer cell lines in three-dimensional assayscorrelate with their profiles of gene expression. Mol Oncol 2007;1:84–96.

27. Cheung WK, Zhao M, Liu Z, Stevens LE, Cao PD, Fang JE, et al. Control ofalveolar differentiation by the lineage transcription factors GATA6 andHOPX inhibits lung adenocarcinoma metastasis. Cancer Cell 2013;23:725–38.

28. Park JB, Kwak HJ, Lee SH. Role of hyaluronan in glioma invasion. CellAdhesion Migr 2008;2:202–7.

29. Jin SG, Jeong YI, Jung S, Ryu HH, Jin YH, Kim IY. The effect of hyaluronicAcid on the invasiveness ofmalignant glioma cells: comparison of invasionpotential at hyaluronic Acidhydrogel andmatrigel. J KoreanNeurosurg Soc2009;46:472–8.

30. Charles NA, Holland EC, Gilbertson R, Glass R, Kettenmann H. The braintumor microenvironment. Glia 2011;59:1169–80.

31. Zhang XH, Wang Q, Gerald W, Hudis CA, Norton L, Smid M, et al. Latentbone metastasis in breast cancer tied to Src-dependent survival signals.Cancer Cell 2009;16:67–78.

32. Ricci MS, Kim SH, Ogi K, Plastaras JP, Ling J, Wang W, et al. Reduction ofTRAIL-induced Mcl-1 and cIAP2 by c-Myc or sorafenib sensitizes resis-tant human cancer cells to TRAIL-induced death. Cancer Cell 2007;12:66–80.

33. Ricci MS, Jin Z, Dews M, Yu D, Thomas-Tikhonenko A, Dicker DT, et al.Direct repression of FLIP expression by c-myc is a major determinant ofTRAIL sensitivity. Mol Cell Biol 2004;24:8541–55.

34. Gonzalvez F, Ashkenazi A. New insights into apoptosis signaling byApo2L/TRAIL. Oncogene 2010;29:4752–65.

35. Fulda S, Meyer E, Debatin KM. Inhibition of TRAIL-induced apoptosis byBcl-2 overexpression. Oncogene 2002;21:2283–94.

36. Scaffidi C, Fulda S, Srinivasan A, Friesen C, Li F, Tomaselli KJ, et al. TwoCD95 (APO-1/Fas) signaling pathways. EMBO J 1998;17:1675–87.

37. Nguyen DX, Chiang AC, Zhang XH, Kim JY, Kris MG, Ladanyi M, et al.WNT/TCF signaling through LEF1 and HOXB9 mediates lung adenocar-cinoma metastasis. Cell 2009;138:51–62.

38. Lau LW, Cua R, KeoughMB, Haylock-Jacobs S, Yong VW. Pathophysiologyof the brain extracellular matrix: a new target for remyelination. Nat RevNeurosci 2013;14:722–9.

39. Sodir NM, Swigart LB, Karnezis AN, Hanahan D, Evan GI, Soucek L.Endogenous Myc maintains the tumor microenvironment. Gen Develop2011;25:907–16.

40. Soucek L, Lawlor ER, Soto D, Shchors K, Swigart LB, Evan GI. Mast cells arerequired for angiogenesis and macroscopic expansion of Myc-inducedpancreatic islet tumors. Nat Med 2007;13:1211–8.

41. Carystinos GD, Kandouz M, Alaoui-Jamali MA, Batist G. Unexpectedinduction of the human connexin 43 promoter by the ras signalingpathway is mediated by a novel putative promoter sequence. MolPharmacol 2003;63:821–31.

42. Sears RC. The life cycle of C-myc: from synthesis to degradation. Cell Cycle2004;3:1133–7.

43. Yeh E, Cunningham M, Arnold H, Chasse D, Monteith T, Ivaldi G, et al. Asignalling pathway controlling c-Myc degradation that impacts oncogenictransformation of human cells. Nat Cell Biol 2004;6:308–18.

44. Junttila MR, Westermarck J. Mechanisms of MYC stabilization in humanmalignancies. Cell Cycle 2008;7:592–6.

45. Singhi AD, Cimino-Mathews A, Jenkins RB, Lan F, Fink SR, Nassar H, et al.MYC gene amplification is often acquired in lethal distant breast cancermetastases of unamplified primary tumors. Mod Pathol 2012;25:378–87.

46. Annis MG, Soucie EL, Dlugosz PJ, Cruz-Aguado JA, Penn LZ, Leber B, et al.Bax forms multispanning monomers that oligomerize to permeabilizemembranes during apoptosis. EMBO J 2005;24:2096–103.

47. StuckeyDW, Shah K. TRAIL on trial: preclinical advances in cancer therapy.Trends Mol Med 2013;19:685–94.

48. Graves JD, Kordich JJ, Huang TH, Piasecki J, Bush TL, Sullivan T, et al.Apo2L/TRAIL and the death receptor 5 agonist antibody AMG 655 coop-erate to promote receptor clustering and antitumor activity. Cancer Cell2014;26:177–89.

49. Holland PM, Miller R, Jones J, Douangpanya H, Piasecki J, Roudier M,et al. Combined therapy with the RANKL inhibitor RANK-Fc andrhApo2L/TRAIL/dulanermin reduces bone lesions and skeletal tumorburden in a model of breast cancer skeletal metastasis. Cancer Biol Ther2010;9:539–50.

50. Bagci-Onder T, Du W, Figueiredo JL, Martinez-Quintanilla J, Shah K.Targeting breast to brain metastatic tumours with death receptor ligandexpressing therapeutic stem cells. Brain 2015;138:1710–21.

Mol Cancer Res; 17(2) February 2019 Molecular Cancer Research554

Lee et al.

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630

Page 12: c-MYC Drives Breast Cancer Metastasis to the Brain, but … · centroid linkage clustering. For cluster analysis, gene-expression data were normalized by the quantile method, log

2019;17:544-554. Published OnlineFirst September 28, 2018.Mol Cancer Res   Ho Yeon Lee, Junghwa Cha, Seon Kyu Kim, et al.   Synthetic Lethality with TRAILc-MYC Drives Breast Cancer Metastasis to the Brain, but Promotes

  Updated version

  10.1158/1541-7786.MCR-18-0630doi:

Access the most recent version of this article at:

  Material

Supplementary

  http://mcr.aacrjournals.org/content/suppl/2018/09/28/1541-7786.MCR-18-0630.DC1

Access the most recent supplemental material at:

   

   

  Cited articles

  http://mcr.aacrjournals.org/content/17/2/544.full#ref-list-1

This article cites 50 articles, 14 of which you can access for free at:

  Citing articles

  http://mcr.aacrjournals.org/content/17/2/544.full#related-urls

This article has been cited by 2 HighWire-hosted articles. Access the articles at:

   

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

  Subscriptions

Reprints and

  [email protected]

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

  Permissions

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

.http://mcr.aacrjournals.org/content/17/2/544To request permission to re-use all or part of this article, use this link

on March 26, 2020. © 2019 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst September 28, 2018; DOI: 10.1158/1541-7786.MCR-18-0630