RESEARCH ARTICLE

Modelling the MYC-driven normal-to-tumour switch in breastcancerCorey Lourenco1,2, Manpreet Kalkat1,2, Kathleen E. Houlahan2,3, Jason De Melo1, Joseph Longo1,2,Susan J. Done1,2, Paul C. Boutros2,3 and Linda Z. Penn1,2,*

ABSTRACTThe potent MYC oncoprotein is deregulated in many human cancers,including breast carcinoma, and is associated with aggressivedisease. To understand the mechanisms and vulnerabilities ofMYC-driven breast cancer, we have generated an in vivo modelthat mimics human disease in response to MYC deregulation.MCF10A cells ectopically expressing a common breast cancermutation in the phosphoinositide 3 kinase pathway (PIK3CAH1047R)led to the development of organised acinar structures in mice.Expressing both PIK3CAH1047R and deregulated MYC led to thedevelopment of invasive ductal carcinoma. Therefore, thederegulation of MYC expression in this setting creates a MYC-dependent normal-to-tumour switch that can be measured in vivo.TheseMYC-driven tumours exhibit classic hallmarks of human breastcancer at both the pathological andmolecular level. Moreover, tumourgrowth is dependent upon sustained deregulated MYC expression,further demonstrating addiction to this potent oncogene and regulatorof gene transcription. We therefore provide a MYC-dependent modelof breast cancer, which can be used to assay in vivo tumour signallingpathways, proliferation and transformation from normal breast acini toinvasive breast carcinoma. We anticipate that this novel MYC-driventransformation model will be a useful research tool to betterunderstand the oncogenic function of MYC and for the identificationof therapeutic vulnerabilities.

KEY WORDS: MYC, Driver oncogene, PI3K, Breast cancer, Cancermodel, Microenvironment

INTRODUCTIONMYC deregulation occurs in the majority of human cancers andis therefore an ideal therapeutic target (Kalkat et al., 2017).Deregulation can be broadly defined as any genomic, epigeneticor signalling pathway aberration that results in sustained andoften elevated MYC function. Targeting this potent oncoproteinusing traditional approaches has been difficult as MYC is atranscription factor lacking enzymatic pockets. Alternative

strategies are therefore under intense investigation, includingapproaches that target MYC-DNA interactions with small-protein inhibitors such as Omomyc or ME47 (Lustig et al.,2017; Soucek et al., 2013). In addition, identifying the uniquevulnerabilities that arise in a MYC-driven tumour can revealtargetable synthetic lethal interactions (Horiuchi et al., 2016;Hsu et al., 2015; Sato et al., 2015; Toyoshima et al., 2012; Zhouet al., 2014). Thus, there is a need for a diverse set of MYC-driven models to identify and test new therapeutic strategies in aMYC-transformed setting.

Many approaches have been developed to model MYC in cancer,including cancer cell lines and genetically engineered mousemodels (GEMMs). Rodent models include the widely used Rat-1Acell line, the Eμ-Myc transgenic model of B-cell lymphoma and theMMTV-MYC model of breast cancer (Sabò et al., 2014; Stewartet al., 1984; Stone et al., 1987; Strasser et al., 1990). Additionally,human MYC-dependent models, such as the Burkitt lymphomacell-line model P493-6 with the ability to regulate MYC levels, havebeen used extensively to interrogate transcriptional regulation byMYC (Lin et al., 2012; Pajic et al., 2000; Schuhmacher et al., 1999).Moreover, the MycER chimeric protein system, which can beapplied to cell-line and mouse models of many cancer types, hasbeen used for studies of MYC in the cell cycle, genomic instability,transformation and synthetic lethal interactions (Eilers et al., 1989;Felsher and Bishop, 1999; Kessler et al., 2012; Liu et al., 2012;Pelengaris et al., 1999, 2001, 2004). These models have providedseveral new insights into MYC biology, suggesting that the furtherdevelopment of diverse models will also provide importantcontributions, particularly if they are complementary and obviateassociated limitations with existing models. For instance, GEMMsare time-consuming to manipulate genetically, are resource-intensive and often do not accurately model human disease at thehistological level. Specifically, when modelling human breastcancer, there are numerous differences between mouse and humanmammary gland development and the surrounding stroma that needto be considered (McNally and Stein, 2017). In addition, mostcancer cell line models do not accurately reflect human tumoursin vivo, are derived from already fully transformed tumour tissue,and are not necessarily driven by or dependent on deregulatedMYC. We have therefore set out to design a functionallycomplementary MYC-driven transformation model that meetsseveral criteria: it is fully transformed in response to, anddependent upon, deregulated MYC; it recapitulates human diseaseat both the pathological and molecular level in vivo; and it is easy tomanipulate genetically. These criteria outline a model that bestreflects the complexity of human tumours in vivo, while ensuringthat it is possible to identify MYC dependencies. To this end, weused a non-transformed parental cell (MCF10A) and developed anisogenic panel of cell lines that could meet these standards andmodel the transformation of breast cancer upon the introduction ofReceived 13 November 2018; Accepted 3 June 2019

1Princess Margaret Cancer Centre, University Health Network, 101 College St,Toronto, ON M5G 0A3, Canada. 2Department of Medical Biophysics, University ofToronto, 101 College Street Suite 15-701, Toronto, ON M5G 1L7, Canada. 3OntarioInstitute for Cancer Research, 661 University Ave, Suite 510, Toronto, ONM5G 0A3,Canada.

*Author for correspondence (lpenn@uhnres.utoronto.ca)

C.L., 0000-0001-9075-7661; P.C.B., 0000-0003-0553-7520; L.Z.P., 0000-0001-8133-5459

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2019. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms

deregulated MYC. We evaluate MYC dependency in vivo andcharacterise the histological and mRNA expression changes thatoccur in these MYC-driven tumours.

RESULTS AND DISCUSSIONIn vivo transformation of MCF10A cells is dependent onactive PI3K signalling, deregulatedMYC and the presence ofMYC Box IIWe chose to model breast cancer as MYC is often deregulated andcontributes to the progression of this disease (Singhi et al., 2011;Stratikopoulos et al., 2015). Recent large whole-genomesequencing studies of breast cancer patient tumours revealed thatthe most frequently mutated and amplified oncogenes in breastcancer are PIK3CA and MYC, respectively (Nik-Zainal et al.,2016). PIK3CA is mutated in approximately 30% of breast cancerpatient tumours, whereas MYC deregulation, primarily scoredthrough amplification, occurs in approximately 20% of breastcancers (Chen and Olopade, 2008; Nik-Zainal et al., 2016). Tomodel breast cancer transformation, we sequentially introducedthese commonly mutated oncogenes into MCF10As, an immortal,non-transformed breast cell line (Soule et al., 1990). MCF10Aswere chosen owing to their inherently stable genome, which ensuresthat transformation is a consequence of the introduction of ectopiconcogenes rather than an indirect consequence of genomicinstability and the selection of a transformed clone. In addition,the MCF10As are a relevant model as the few mutations acquiredduring spontaneous immortalisation are prevalent in human breastcancer, including CDKN2A deletion (Kadota et al., 2010). Finally,despite harbouring a focal amplification at chromosome 8q24, totalMYC levels remain highly regulated, decreasing rapidly in responseto antiproliferative conditions; this observation is characteristic ofnormal MYC regulation in non-transformed cells (Kadota et al.,2010; Wasylishen et al., 2013). Thus, the MCF10A cells fulfilled ourcriteria of a parental breast cell system on which to base our model.We first stably introduced the empty vector control or the

activated PIK3CAH1047R allele into MCF10A cells. The latter wasactive in these cells as protein kinase B (AKT) phosphorylation atSer473 was readily detectable (Fig. 1A). To this pair of cell lines, weintroduced empty vector control or ectopic MYC expression, tomodel MYC deregulation through sustained expression aspreviously shown (Wasylishen et al., 2013). This resulted in apanel of four cell lines consisting of 10A.EE (empty vector, emptyvector), 10A.EM (empty vector, MYC), 10A.PE (PIK3CAH1047R,empty vector) and 10A.PM (PIK3CAH1047R, MYC). Undergrowing conditions, we observed no significant change in eithercell size or proliferation between any groups (Fig. S1A), suggestingthat these cellular properties would not contribute to anyphenotypes observed. Additionally, despite minor cellmorphology differences (Fig. S1B), epithelial-to-mesenchymaltransition (EMT) markers in the isogenic panel were not evidentproviding confidence that our cell-line panel had not undergoneEMT transition, and thus represents a model of epithelial breastcancer (Fig. S1C).To assay for transformation, we cultured our isogenic cell-line

panel as breast organoids on Matrigel to model three-dimensionalacini formation in vitro (Fig. 1B) (Debnath et al., 2003). The10A.EE and 10A.EM cells formed round, organised acinarstructures consistent with previous publications (Wasylishen et al.,2013). By contrast, 10A.PE and 10A.PM cells formed large anddisorganised acinar structures, demonstrating a transformedphenotype driven by PIK3CA activation in vitro. MYCderegulation alone was insufficient to transform these acini, as

previously described (Wasylishen et al., 2013), and did notpotentiate the transformation of PI3KCAH1047R-expressing acini(Fig. 1B), suggesting that aberrant PI3K signalling was the maindriver of transformation in this in vitro setting.

Despite no significant proliferative difference between these celllines, activated PI3K signalling was sufficient for transformation inMatrigel. Considering the importance of environmental context onthe phenotypes observed, we progressed to evaluate xenograft growthin vivo. Xenografts recapitulate the tumour microenvironment andoffer an increasingly relevant context to study tumour formation andgrowth. MCF10A cells are non-transformed and are therefore notcapable of forming xenografts in immunocompromised mice. In thismanuscript, we refer to a xenograft as any human cell capable ofengraftment in foreign species (mouse). We therefore assayed theisogenic panel developed in Figure 1A into female NOD-SCID mice(three mice per group) to observe if transformation had been achievedin vivo through expression of PIK3CAH1047R and/or MYC(Fig. S1D). As expected 10A.EE cells, containing no ectopiconcogenes, did not form any palpable masses in vivo, nor did10A.EM cells. This was similar to our observation in organoidcultures showing that ectopic MYC alone was not sufficient fortransformation. The most transformed organoid groups (10A.PE and10A.PM) were able to establish palpable xenograft growths in allmice tested, suggesting that activated PI3K signalling is required forxenograft formation. Interestingly, despite no significant differencebetween the 10A.PE and 10A.PM cells in organoid transformationfrequency, 10A.PM xenografts were much larger (Fig. S1D) andfaster growing (data not shown) than the small, but palpable massesthat arose from 10A.PE cells. To further interrogate the function ofMYC in this assay, we generated cells expressing PIK3CAH1047R andthen introduced empty vector (10A.PE), MYC (10A.PM) or MYCwith a deletion of MYC Box II (10A.PΔMBII) (Fig. S1E). The latterserves as a negative control, as MBII is evolutionarily conserved andrequired for MYC-dependent transformation (Oster et al., 2003;Stone et al., 1987). As observed previously (Fig. S1D), 10A.PE cellsengrafted as small but measurable masses in vivo and 10A.PM cellswere able to form significantly larger, and faster growing massescompared with 10A.PE cells (Fig. 1C). Cells expressing MYCΔMBII

(10A.PΔMBII) formed small palpable masses, similar to 10A.PE,demonstrating that tumour xenograft growth was dependent onMYCand the conserved MBII region. Doubling times were 6 days for10A.PM and 60 days for both 10A.PE and 10A.PΔMBII xenografts.Histological examination showed that 10A.PE and 10A.PΔMBIIxenograft masses were composed of acinar structures embeddedwithin an extracellular matrix, strikingly similar to the histology ofnormal human breast acini (Fig. 1D, Fig. S2A). By contrast, 10A.PMxenografts possessed major phenotypic differences, appearing asdense, fully transformed breast tumours with features similar tohuman disease. Given the role of PI3K signalling in the activation ofprosurvival pathways (Manning and Toker, 2017), PIK3CAH1047R

enables MCF10A cells to establish small xenograft masses, yet doesnot provide the oncogenic activity to form fully transformed tumours.Although active PI3K signalling transforms breast organoids in vitro,both PIK3CAH1047R and deregulated MYC are required fortransformation in vivo.

Deregulated MYC transforms breast acini into invasiveductal carcinoma in vivoIn clinical practice, tumours are evaluated through histology,immunohistochemistry (IHC) and/or molecular profiling tocharacterise tumour subtype and features of aggressive disease.We applied these analyses to our xenografts, in a blinded-manner,

2

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms

with the assistance of practicing breast cancer pathologist Dr SusanDone. 10A.PM and 10A.PΔMBII growths had significantlyincreased Ki67 staining compared with 10A.PE, whereas TUNELstaining was low in all samples (Fig. 2A). Interestingly, although10A.PΔMBII xenografts had an increase in Ki67 signal, the overallxenograft volume was equal to 10A.PE xenograft growths. Thissuggests that MYCΔMBII is able to confer some proliferativeadvantage owing to retention of partial MYC-related activity, but isnot able to drive transformation; this observation is consistent withprevious reports (Oster et al., 2003; Stone et al., 1987). We alsoanalysed clinically relevant markers such as the estrogen andprogesterone receptors (ER and PR), human epidermal growth

factor receptor 2 (HER2) and cytokeratin 5 (CK5) as a marker ofbasal cells (Fig. 2B). The MCF10A cell line has been previouslycharacterised as basal and triple-negative for breast cell receptorsin vitro (Subik et al., 2010). Introduction of PIK3CAH1047R andMYC maintained the basal and triple-negative state in vivo.

We then used clinical markers to distinguish the degree oftransformation within each xenograft with the assistance of DrSusan Done. Malignant disease can be categorised as eithercarcinoma in situ (CIS) or invasive ductal carcinoma (IDC) and isdistinguished by transformed cells contained within or invadingthrough a layer of myoepithelial cells surrounding the acinarstructure, respectively. To assess these conditions, we stained for

Fig. 1. In vivo transformation of MCF10A cells is dependent on active PI3K signalling, deregulated MYC and the conserved MBII region. (A) Schematicoverview of the generation of an isogenic panel of MCF10A cell lines. Immunoblotting was performed against pAKTS473, total AKT, MYC and actin and quantifiedusing ImageJ (n=3). One-way ANOVA with Bonferroni post-test for multiple testing. **P≤0.01, *P≤0.05. (B) The MCF10A isogenic panel cultured in Matrigelfor 12 days. Images were acquired on day 12 and quantified based on an assessment of acinar circularity and size. Individual mean values from three biologicalreplicates are shown; ****P≤0.0001, one-way ANOVA with Bonferroni post-test for multiple testing. Scale bars: 500 μm. (C) Individual tumour volumemeasurements from 10A.PE (n=6), 10A.PM (n=7) and 10A.PΔMBII (n=7) xenografts 49 days after injection. Representative images and take rate(mass formed/mouse injected) is indicated below each image. Scale bar: 0.5 cm. ***P≤0.001, one-way ANOVA with Bonferroni post-test for multiple testing.(D) Histology samples from xenografts in (C) stainedwith H&E. Representative images are shown. Scale bar for 5× images: 50 μm.Scale bar for 10× images: 500 μm.

3

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms

myoepithelial cells using twomarkers, p63 and smoothmuscle actin(SMA). 10A.PE and 10A.PΔMBII acini were surrounded bymyoepithelial cells and were identified with human-specificepidermal growth factor receptor (EGFR) antibody, indicating thatthese cells were of MCF10A origin. In contrast, myoepithelial cellswere not detected in 10A.PM tumours, as evidenced by a lack of p63and SMA co-staining. Our pathology report revealed that a majorityof 10A.PE growths were normal breast acinar structures displaying

hollow lumen and a surrounding myoepithelial layer, with few casesof hyperplasia (Fig. 2B,C, Fig. S2B) (Dontu and Ince, 2015). The10A.PΔMBII growths developed moderately more hyperplasticlesions than 10A.PE xenografts, despite MBII being essential fortumour growth. These data align with our Ki67 proliferation dataand previously published results showing that although MBII isrequired for transformation, MYCΔMBII can retain some MYC-related activity in tissue culture assays (Oster et al., 2003; Stone

Fig. 2. Deregulated MYC transforms breast acini into invasive breast carcinoma in vivo. (A) Tumours harvested in Figure 1C were stained (above)for Ki67 and TUNEL and were quantified (below) for the degree of Ki67 and TUNEL staining. Individual quantifications per tumour are shown; *P≤0.05,***P≤0.001, one-way ANOVA with Bonferroni post-test for multiple testing. Scale bar: 100 μm. (B) Tumours harvested in Figure 1C were stained for ER,PR, HER2, CK5, EGFR, p63 andSMA by the Pathology Research Program Laboratory. (C) A pathology report was produced for each tumour using a combinationof H&E, as shown in Figure 1D, and IHC markers used in this study. Scale bar: 100 μm.

4

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms

et al., 1987). By contrast, 10A.PM tumours, which were highlyproliferative and lacked myoepithelial cells, were scored as high-grade IDC. Therefore, the addition of deregulated MYC in thisin vivo model system transforms cells into IDC in an MBII-dependent manner. Thus, we are able to evaluate MYC-driventumour progression (10A.PM) against a non-transformed control(10A.PE) in vivo, as well as being able to measure tumour volumeand evaluate tumour pathology: a rich data set that, to the best of ourknowledge, uniquely distinguishes this model from previous solidhuman MYC-driven cancer models.

10A.PM tumours have MYC and invasive ductal carcinomaexpression signaturesOur histological analysis demonstrates that the in vivo 10A.PE and10A.PM isogenic xenograft pair represents normal breast acini andIDC, respectively. We took this opportunity to identify an in vivoMYC-dependent transformation expression data set. We re-established 10A.PE and 10A.PM xenografts and isolated RNAfrom three biological replicates for RNA-sequencing (RNA-seq)(Fig. 3A). Contaminating mouse RNA was removed using Xenome(Conway et al., 2012), leaving ∼5000 significantly differentiallyexpressed genes (PPDE ≥0.95, Table S1). A subset of upregulatedand downregulated genes were validated through reverse transcriptionquantitative PCR (RT-qPCR) using human-specific primers and the2(−ddC(T)) method (Fig. S1F) (Livak and Schmittgen, 2001).Following RNA-seq, we performed gene set enrichment analysis(GSEA) (Reimand et al., 2019; Subramanian et al., 2005) on rankedgene list expression data to generate a gene ontology (GO) map ofupregulated and downregulated biological functions in 10A.PMtumours (Fig. 3B, Table S2). Upregulated biological functionsincluded protein translation, chromatin organisation and RNAprocessing. Interestingly, downregulated processes included cell-celladhesion and cell differentiation, reflective of aggressive breastcancer. Additionally, we performed GSEA to compare our observeddifferentially expressed genes with well-established MYC-driven andbreast cancer gene sets (Fig. 3C, Table S2) (Liberzon et al., 2011). TheMYC_UP.V1_UP gene set was significantly enriched in theupregulated portion of our xenograft gene expression list, whereasthe MYC_UP.V1_DN gene list was significantly enriched in thedownregulated portion of our xenograft gene expression list. Thesedata show that the expression of known upregulated anddownregulated MYC target genes is also observed in the 10A.PMmodel. Additionally, we observed significant and positive normalisedenrichment scores (NES) with IDC gene sets (Fig. 3C, Table S2),indicating that these gene sets were identified in the upregulatedportion of our xenograft gene expression list. These data reinforce theidea that ourMYC-driven breast cancer model accurately recapitulatesthe activated and downregulated signalling pathways associated withMYC deregulation and IDC observed in human disease.

Constitutive expression of deregulated MYC is required fortumour growthIntroducing deregulated MYC into MCF10A cells expressingPIK3CAH1047R transforms these cells in vivo; however, it is notclear whether deregulated MYC is necessary to sustain tumourgrowth. To address this, we generated cell lines with doxycycline(DOX)-inducibleMYC to regulate ectopic expression (Fig. 4A, left).Cells were treated with DOX to induce ectopic MYC expression andinjected subcutaneously into NOD-SCID mice, which were givensupplemented DOX-treated drinking water. Once tumours reached200 mm3, mice were randomly assigned to ‘MYC ON’ or ‘MYCOFF’ groups, the latter being removed from DOX-supplemented

drinking water (Fig. 4A, right). Over 12 days of treatment, MYC ONtumours continued to grow and MYC OFF tumours slightly declined(Fig. 4B, left). MYC ON tumours had a doubling time of 7-8 daysduring this 12 day treatment period, whereas MYC OFF tumours, onaverage, were shrinking at a negative doubling rate of approximately 39days. After 12 days of treatment, the first MYC ON tumour reachedhumane end point (Fig. 4B, right). Treatment was continued for31 days and the overall time to humane end point for MYC ONtumours was significantly shorter than for MYC OFF tumours(Fig. 4C, left). In addition, we calculated the percent volume changefor every tumour by comparing the final volume measurements (eitherat humane end point or 31 days of treatment) to the initial tumourvolumes (approximately 200mm3) and observed a significantdifference between MYC ON and MYC OFF groups (Fig. 4C,right). These data demonstrate that this model is both driven by, andaddicted to, deregulated MYC in a reproducible and quantitativemanner. Furthermore, we show that targeting the deregulated fractionof MYC is sufficient to inhibit tumour growth, as endogenous MYCexpression is not affected by DOX treatment (Fig. 4A). Therefore, theobserved phenotypes, signalling pathways and vulnerabilities that arisein vivo are a consequence of the regulatable and ectopic allele of MYCand, importantly, are not due to permanent genomic ormicroenvironmental changes following deregulated MYC-driventransformation.

The isogenic MCF10A panel presented here offers uniqueadvantages and is complementary to current MYC-dependentmodels of transformation (Fig. S1G) (Boxer et al., 2004; Wanget al., 2011; Zhao et al., 2003). Interestingly, the observed MYC-driven phenotype was exclusive to the in vivo xenograft assay. Thisobservation highlights the importance of identifying a relevantcontext when studying MYC-dependent transformation, as ourin vitro studies demonstrated little oncogenic MYC activity towardsproliferation or in vitro acini transformation and was quite unable topredict the striking in vivo response. Moreover, studying cancerbiology using in vivo models most accurately reflects humandisease, and to some extent incorporates the contribution of tumourmicroenvironment and cellular stress. The in vivo context istherefore ideal for studying deregulated MYC and for identifyingMYC-dependent vulnerabilities, as the relevant stress-signallingpathways will be active.

We have demonstrated that this model can recapitulate humanbreast cancer at both the molecular and pathological level, and istransformed by deregulated MYC. We anticipate that this modelwill be a useful tool for identifying the context-dependent functionsand vulnerabilities of MYC-driven cancers, which are more difficultto discover using traditional in vitro or murine models.Additionally,this system is reproducible and scalable: ideal characteristics for theidentification and validation of anti-MYC therapies. Indeed, wehave recently demonstrated the utility of this model to interrogateMYC mutants for their role in transformation (Kalkat et al., 2018).We anticipate that this system will be able to predict the efficacy ofnovel therapeutic drugs targeting MYC and will identify novelvulnerabilities of MYC-dependent breast cancers.

MATERIALS AND METHODSCell cultureMCF10A epithelial cells were a kind gift of Dr Senthil Muthuswamy andwere cultured in MCF10A culture media, as previously described (Debnathet al., 2003). All cell lines were routinely monitored for mycoplasmacontamination and tested negative; in vitro experiments were conductedwithin five passages from selection. Cell culture images were taken with a32× (Ph) objective on an AxioObserver microscope (Zeiss).

5

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms

Lentiviral vectors and cell-line transductioncMYC and PIK3CAH1047R (cloned from Addgene plasmid #12524) alleleswere first cloned into pENTR4 no ccDB (686-1) (Addgene plasmid #17424)and subsequently shuttled into pLenti CMV/TO Puro DEST (670-1)(Addgene plasmid #17293) and pLenti CMV Hygro DEST (w117-1)

(Addgene plasmid #17454), respectively (Campeau et al., 2009).Tetracycline-inducible cells were generated using pLenti CMV TetR Blast(716-1) (Addgene plasmid #17492). Lentiviral particles were generatedusing HEK293Tv (kind gift from Sam Benchimol) and MCF10A cell lineswere transduced at the same time and with an equivalent viral titre.

Fig. 3. 10A.PM tumours have MYC and invasive breast cancer expression signatures. (A) Female NOD-SCID mice were injected with 10A.PE or 10A.PMcells and allowed to form tumours (left). For RNA-seq analysis, 10A.PE growths (n=3) and actively growing10A.PM tumours (n=3) were harvested (right, indicatedin red). ***P≤0.001, unpaired, two-tailed t-test. (B) Changes in gene expression, relative to 10A.PE xenografts, were ranked and tested for enrichment in GOBiological Process gene sets. The 10A.PM upregulated or downregulated genes that were significantly enriched in a GO Biological Process gene set arerepresented as red or blue circles, respectively (all data are provided in Table S2). Edges represent the number of overlapping genes within each gene set andnode size represents the number of genes within a single gene set. Finally, clusters of gene sets and annotations were created using clusterMaker2 andAutoAnnotate Cytoscape applications. Data has been filtered (for visualisation) to show highest/lowest enriched GO processes (Reimand et al., 2019).(C) The gene expression data were analyzed by GSEAwith MYC and IDC signature data sets. Gene sets used are labelled on the top of each enrichment plot (alldata is provided in Table S2). NES, normalised enrichment score (the enrichment score for the gene set after it has been normalised across analysed gene sets).NOM P-value, the statistical significance of the enrichment score. The nominal P-value is not adjusted for gene set size or multiple hypothesis testing and,therefore, is of limited use in comparing gene sets. FDR q-value, false discovery rate (i.e. the estimated probability that the normalised enrichment scorerepresents a false-positive finding).

6

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms

AntibodiesImmunoblotting was performed against pAKTS473 (Cat# 4051, CellSignaling Technology), total AKT (Cat# 4685, Cell SignalingTechnology), MYC (9e10, homemade, 1:1000 dilution), actin (Cat#A2066, Sigma), fibronectin (Cat# Ab32419, Abcam), vimentin (Cat# 5741,Cell Signaling Technology) and tubulin (Cat# CP06, Millipore). Antibodydilutions were used according to manufacturer’s recommendations.

EdU incorporation assayThe MCF10A isogenic panel was grown for 24 h in fully supplementedmedia and allowed to reach 50% confluency. At 50% confluency, cells weretreated with 5 ethynyl-2′-deoxyuridine (EdU) for 4 h, processed accordingto manufacturer’s instructions (Click-iT EdU Alexa Fluor 488 FlowCytometry Assay Kit, Cat# C10420, Life Technologies) and analysed forEdU positivity using the LSR II flow cytometer (BD Biosciences) andFlowJo v10 software.

Three-dimensional Matrigel morphogenesis assayCells were cultured for 12 days on Matrigel (Cat# 354230, Corning) inspecialised MCF10A media for three-dimensional cultures, as previouslydescribed (Debnath et al., 2003). Images were acquired on day 12 using aFLUAR 5×/0.25 NA lens on an AxioObserver microscope (Zeiss). Cell-line groups were blinded and quantified on the basis of a qualitativeassessment of acinar circularity and size. Small round acini wereconsidered normal; large and disorganised acini were consideredtransformed. One-way ANOVA with Bonferroni post-test for multipletesting was used for analysis.

XenograftsA total of 1×106 cells were suspended in 50% Matrigel (Cat# 354262,Corning) to a final volume of 0.2 ml and injected subcutaneously into theflanks of six- to eight-week old female NOD-SCID mice. For tetracycline-repressed cell lines, 100 μg/ml DOX (Sigma) was supplemented into the

Fig. 4. Constitutive expression of deregulated MYC is required for tumour growth. (A) With the addition of 1 μg/ml doxycycline (DOX) in vitro, ectopic MYCis expressed (MYC ON). The arrow indicates ectopic MYC; the asterisk indicates endogenous MYC. A total of 1×106 cells were injected into femaleNOD-SCID mice (n=18) and given DOX-treated water (100 μg/ml) to induce MYC expression. Tumours were allowed to grow until 200 mm3, at which pointthe animals were randomised into MYC ON (n=9) or MYC OFF (n=9) groups. (B) Individual mean values from all measured tumours (nine per group) during theinitial 12 day treatment period are shown (left) and individual tumour volumes from day 12 (final day with all animals present) are also shown (right); **P≤0.01,***P≤0.001, ****P≤0.0001, unpaired, two-tailed t-test. Error bars represent s.d. (C) Left, tumours were measured until the mice reached a humane end pointof 1000 mm3 or until they had received treatment for 1 month. Time to humane end point was plotted as percentage survival; ****P≤0.0001, log-rank test. Middle,percent volume changes for each tumour after the entire course of treatment are reported. ****P≤0.0001, unpaired, two-tailed t-test. Right, representative imagesare shown. Scale bars: 1 cm.

7

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms

drinking water, prepared fresh twice a week. Animals were randomised toreceive DOX or control water when tumours reached 200 mm3. The end pointfor all experiments was determined when tumours exceeded 1000 mm3 or ifotherwise stated. Percent volume changewas calculated for each tumour usingthe following formula: 100 × [(Vend − Vstart)/Vstart], where Vend is the finalvolume of each tumour at end-point andVstart is the initial volumewhenDOX-treatment began. Animal oversight was performed by the Animal ResourcesCentre (ARC) affiliated with the University Health Network (UHN).

EMT analysisRecombinant human transforming growth factor β (TGF-β) was purchasedfrom PeproTech (Rocky Hill, NJ, USA) and reconstituted according to themanufacturer’s instructions. The isogenicMCF10Apanel presented in Fig. 1Awas cultured in vehicle control or 5 ng/ml TGF-β for 72 h before harvest.

ImmunohistochemistryExtracted tumours were fixed in 10% buffered formalin (Sigma) for 24-48 hat room temperature and then stored in 70% ethanol at 4°C. Samples werethen paraffin-embedded, sectioned at 4 μm thickness and stained withhaematoxylin and eosin (H&E). Ki67 (NB110-90592, Novus), TUNEL(homemade by the Pathology Research Program Laboratory), ER (ab80922,Abcam), PR (NCL-PR-312, Leica), HER2 (RM9103, Thermo FisherScientific), CK5 (NCL-LCK5, Leica), EGFR (28-0005, Invitrogen), p63(VP-P960, Vector) and SMA (M0851, Dako) were used to further analysetumour sections. All immunohistochemistry was performed as a service bythe Pathology Research Program Laboratory (University Health Network,Toronto, Canada).

RT-qPCRThe cDNAwas synthesised using SuperScript III as per the manufacturer’sprotocol (Invitrogen). The mRNA expression was analysed by RT-qPCR usingprimers listed in Fig. S1H and normalised to the housekeeping gene TBPusing SYBR Green (Thermo Fisher Scientific).

RNA sequence analysisTumours were extracted and immediately flash-frozen. RNA from xenografttissue was harvested using the RNeasy Plus Universal Kit (Cat# 73404,Qiagen) following the manufacturer’s protocol. RNA was sent to thePrincess Margaret Genomics Centre and sequenced using the IlluminaNextSeq 500 for an average of 40 M reads per xenograft growth. Fastq fileswere filtered using Xenome in order to filter out mouse contamination. Onlyreads mapping exclusively to the mouse reference were filtered out. Theremaining reads were mapped to the human reference genome (hg19) andread counts estimated using STAR. Expressed genes were defined asrequiring at least 10 reads in at least two samples. Read counts were upperquantile normalised. Finally, EBSeq was used to test for differentialexpression between experimental conditions (Leng et al., 2013). Genes weredetermined to be differentially expressed given a false discovery rate (FDR)threshold of 0.05. Raw sequencing files can be found on GEO underaccession number GSE130513.

Gene ontology and gene set enrichment analysisAll available gene expression data between 10A.PE and 10A.PM were firstranked [GSEA Rank=−log(PPEE)×PostFC] and then used for GSEA withGO Biological Process gene sets using GSEA software (https://cytoscape.org/) following the protocol outlined in Reimand et al. (2019). Gene setsincluded the entire ‘GO gene set’ collection and selected MYC and breastcancer gene sets (Table S2). Significantly upregulated or downregulated GOprocesses were then used to generate a cytoscape map. Groups visualisedhave been filtered to NES=1.75-2.163 or −1.716-0. We acknowledge ouruse of the gene set enrichment analysis, GSEA software and MolecularSignature Database (MSigDB) (Subramanian et al., 2005).

AcknowledgementsWe thank the members of the Penn Laboratory, Animal Resources Centre,Pathology Research Program Laboratory and Princess Margaret Genomics Centreteams for their advice, support and expertise.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: C.L., M.K., J.D.M., L.Z.P.; Methodology: C.L., M.K., J.D.M.,S.J.D.; Validation: C.L., K.E.H., S.J.D., P.C.B.; Formal analysis: C.L., K.E.H., P.C.B.;Investigation: C.L., M.K., J.D.M., J.L., S.J.D.; Resources: L.Z.P.; Data curation:K.E.H., P.C.B.; Writing - original draft: C.L.; Writing - review & editing: C.L., M.K.,K.E.H., J.D.M., J.L., S.J.D., P.C.B., L.Z.P.; Visualization: C.L.; Supervision: L.Z.P.;Project administration: L.Z.P.; Funding acquisition: L.Z.P.

FundingThis work was supported with funding from the Tier 1 Canada Research ChairsProgram (L.Z.P.), Canadian Institutes of Health Research (L.Z.P., P.C.B.), CanadianBreast Cancer Foundation doctoral fellowship (M.K.), and Terry Fox ResearchInstitute New Investigator Awards (P.C.B.).

Data availabilityRNA-seq data have been deposited in GEO under accession number GSE130513.

Supplementary informationSupplementary information available online athttp://dmm.biologists.org/lookup/doi/10.1242/dmm.038083.supplemental

ReferencesBoxer, R. B., Jang, J. W., Sintasath, L. and Chodosh, L. A. (2004). Lack of

sustained regression of c-MYC-induced mammary adenocarcinomas followingbrief or prolonged MYC inactivation. Cancer Cell 6, 577-586. doi:10.1016/j.ccr.2004.10.013

Campeau, E., Ruhl, V. E., Rodier, F., Smith, C. L., Rahmberg, B. L., Fuss, J. O.,Campisi, J., Yaswen, P., Cooper, P. K. and Kaufman, P. D. (2009). A versatileviral system for expression and depletion of proteins in mammalian cells. PLoSONE 4, e6529. doi:10.1371/journal.pone.0006529

Chen, Y. and Olopade, O. I. (2008). MYC in breast tumor progression. Expert Rev.Anticancer Ther. 8, 1689-1698. doi:10.1586/14737140.8.10.1689

Conway, T., Wazny, J., Bromage, A., Tymms, M., Sooraj, D., Williams, E. D. andBeresford-Smith, B. (2012). Xenome—a tool for classifying reads from xenograftsamples. Bioinformatics 28, i172-i178. doi:10.1093/bioinformatics/bts236

Debnath, J., Muthuswamy, S. K. and Brugge, J. S. (2003). Morphogenesis andoncogenesis of MCF-10A mammary epithelial acini grown in three-dimensionalbasement membrane cultures. Methods 30, 256-268. doi:10.1016/S1046-2023(03)00032-X

Dontu, G. and Ince, T. A. (2015). Of mice and women: a comparative tissue biologyperspective of breast stem cells and differentiation. J. Mammary Gland Biol.Neoplasia 20, 51-62. doi:10.1007/s10911-015-9341-4

Eilers, M., Picard, D., Yamamoto, K. R. and Bishop, J. M. (1989). Chimaeras ofMyc oncoprotein and steroid receptors cause hormone-dependent transformationof cells. Nature 340, 66-68. doi:10.1038/340066a0

Felsher, D. W. and Bishop, M. J. (1999). Transient excess of MYC activity can elicitgenomic instability and tumorigenesis. Proc. Natl. Acad. Sci. USA 96, 3940-3944.doi:10.1073/pnas.96.7.3940

Horiuchi, D., Camarda, R., Zhou, A. Y., Yau, C., Momcilovic, O., Balakrishnan,S., Corella, A. N., Eyob, H., Kessenbrock, K., Lawson, D. A. et al. (2016). PIM1kinase inhibition as a targeted therapy against triple-negative breast tumors withelevated MYC expression. Nat. Med. 22, 1321-1329. doi:10.1038/nm.4213

Hsu, T. Y. T., Simon, L. M., Neill, N. J., Marcotte, R., Sayad, A., Bland, C. S.,Echeverria, G. V., Sun, T., Kurley, S. J., Tyagi, S. et al. (2015). The spliceosomeis a therapeutic vulnerability in MYC-driven cancer. Nature 525, 384-388. doi:10.1038/nature14985

Kadota, M., Yang, H. H., Gomez, B., Sato, M., Clifford, R. J., Dunn, B. K.,Wakefield, L. M. and Lee, M. P. (2010). Delineating genetic alterations for tumorprogression in theMCF10A series of breast cancer cell lines.PLoSONE 5, e9201.doi:10.1371/journal.pone.0009201

Kalkat, M., Melo, J., Hickman, K., Lourenco, C., Redel, C., Resetca, D., Tamachi,A., Tu, W. B. and Penn, L. Z. (2017). MYC Deregulation in primary humancancers. Genes (Basel) 8, 151. doi:10.3390/genes8060151

Kalkat, M., Resetca, D., Lourenco, C., Chan, P.-K., Wei, Y., Shiah, Y.-J., Vitkin,N., Tong, Y., Sunnerhagen, M., Done, S. J. et al. (2018). MYC proteininteractome profiling reveals functionally distinct regions that cooperate to driveTumorigenesis. Mol. Cell 72, 836-848.e7. doi:10.1016/j.molcel.2018.09.031

Kessler, J. D., Kahle, K. T., Sun, T., Meerbrey, K. L., Schlabach, M. R., Schmitt,E. M., Skinner, S. O., Xu, Q., Li, M. Z., Hartman, Z. C. et al. (2012).A SUMOylation-dependent transcriptional subprogram is required for Myc-drivenTumorigenesis. Science 335, 348-353. doi:10.1126/science.1212728

Leng, N., Dawson, J. A., Thomson, J. A., Ruotti, V., Rissman, A. I., Smits,B. M. G., Haag, J. D., Gould, M. N., Stewart, R. M. and Kendziorski, C. (2013).Gene expression EBSeq : an empirical Bayes hierarchical model for inference

8

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms

in RNA-seq experiments. Bioinformatics 29, 1035-1043. doi:10.1093/bioinformatics/btt087

Liberzon, A., Subramanian, A., Pinchback, R., Thorvaldsdottir, H., Tamayo, P.and Mesirov, J. P. (2011). Molecular signatures database (MSigDB) 3.0.Bioinformatics 27, 1739-1740. doi:10.1093/bioinformatics/btr260

Lin, C. Y., Loven, J., Rahl, P. B., Paranal, R. M., Burge, C. B., Bradner, J. E., Lee,T. I. and Young, R. A. (2012). Transcriptional amplification in tumor cells withelevated c-Myc. Cell 151, 56-67. doi:10.1016/j.cell.2012.08.026

Liu, L., Ulbrich, J., Muller, J., Wustefeld, T., Aeberhard, L., Kress, T. R.,Muthalagu, N., Rycak, L., Rudalska, R., Moll, R. et al. (2012). Deregulated MYCexpression induces dependence upon AMPK-related kinase 5. Nature 483,608-612. doi:10.1038/nature10927

Livak, K. J. and Schmittgen, T. D. (2001). Analysis of relative gene expression datausing real-time quantitative PCR and the 2-ΔΔCT method. Methods 25, 402-408.doi:10.1006/meth.2001.1262

Lustig, L. C., Dingar, D., Tu, W. B., Lourenco, C., Kalkat, M., Inamoto, I.,Ponzielli, R., Chan, W. C. W., Shin, J. A. and Penn, L. Z. (2017). Inhibiting MYCbinding to the E-box DNA motif by ME47 decreases tumour xenograft growth.Oncogene 36, 6830. doi:10.1038/onc.2017.275

Manning, B. D. and Toker, A. (2017). AKT/PKB Signaling: navigating the network.Cell 169, 381-405. doi:10.1016/j.cell.2017.04.001

McNally, S. and Stein, T. (2017). Overview of mammary gland development: acomparison of mouse and human. Methods Mol. Biol. 1501, 1-17. doi:10.1007/978-1-4939-6475-8_1

Nik-Zainal, S., Davies, H., Staaf, J., Ramakrishna, M., Glodzik, D., Zou, X.,Martincorena, I., Alexandrov, L. B., Martin, S., Wedge, D. C. et al. (2016).Landscape of somatic mutations in 560 breast cancer whole-genome sequences.Nature 534, 47-54. doi:10.1038/nature17676

Oster, S. K., Mao, D. Y. L., Kennedy, J. and Penn, L. Z. (2003). Functional analysisof the n-terminal domain of the myc oncoprotein. Oncogene 22, 1998-2010.doi:10.1038/sj.onc.1206228

Pajic, A., Spitkovsky, D., Christoph, B., Kempkes, B., Schuhmacher, M.,Staege, M. S., Brielmeier, M., Ellwart, J., Kohlhuber, F., Bornkamm, G. W.et al. (2000). Cell cycle activation by c-myc in a burkitt lymphoma model cell line.Int. J. Cancer 87, 787-793. doi:10.1002/1097-0215(20000915)87:6<787::AID-IJC4>3.0.CO;2-6

Pelengaris, S., Littlewood, T., Khan, M., Elia, G. and Evan, G. (1999). Reversibleactivation of c-Myc in skin induction of a complex neoplastic phenotype by a singleoncogenic lesion. Mol. Cell 3, 565-577. doi:10.1016/S1097-2765(00)80350-0

Pelengaris, S., Khan, M. and Evan, G. I. (2001). Suppression of Myc-inducedapoptosis in β cells exposes multiple oncogenic properties of Myc and triggerscarcinogenic progression. Cell 109, 321-334. doi:10.1016/S0092-8674(02)00738-9

Pelengaris, S., Abouna, S., Cheung, L., Ifandi, V., Zervou, S. and Khan, M.(2004). Brief inactivation of c-Myc is not sufficient for sustained regression of c-Myc-induced tumours of pancreatic islets and skin epidermis. BMC Biol. 2, 1-14.doi:10.1186/1741-7007-2-26

Reimand, J., Isserlin, R., Voisin, V., Kucera, M., Tannus-Lopes, C.,Rostamianfar, A., Wadi, L., Meyer, M., Wong, J., Xu, C. et al. (2019).Pathway enrichment analysis and visualization of omics data using g:Profiler,GSEA, Cytoscape and EnrichmentMap. Nat. Protoc. 14, 482-517. doi:10.1038/s41596-018-0103-9

Sabo , A., Kress, T. R., Pelizzola, M., de Pretis, S., Gorski, M. M., Tesi, A., Morelli,M. J., Bora, P., Doni, M., et al. (2014). Selective transcriptional regulation by Mycin cellular growth control and lymphomagenesis. Nature 511, 488-492. doi:10.1038/nature13537

Sato, M., Rodriguez-Barrueco, R., Yu, J., Do, C., Silva, J. M. and Gautier, J.(2015). MYC is a critical target of FBXW7. Oncotarget 6, 3292-3305. doi:10.18632/oncotarget.3203

Schuhmacher, M., Staege, M. S., Pajic, A., Polack, A.,Weidle, U. H., Bornkamm,G. W., Eick, D. and Kohlhuber, F. (1999). Control of cell growth by c-Myc in theabsence of cell division. Curr. Biol. 9, 1255-1258. doi:10.1016/S0960-9822(99)80507-7

Singhi, A. D., Cimino-Mathews, A., Jenkins, R. B., Lan, F., Fink, S. R., Nassar,H., Vang, R., Fetting, J. H., Hicks, J., Sukumar, S. et al. (2011). MYC geneamplification is often acquired in lethal distant breast cancer metastases ofunamplified primary tumors. Mod. Pathol. 25, 378-387. doi:10.1038/modpathol.2011.171

Soucek, L., Whitfield, J. R., Sodir, N. M., Masso, D., Serrano, E., Karnezis, A. N.,Swigart, L. B. and Evan, G. I. (2013). Inhibition of Myc family proteins eradicatesKRas-driven lung cancer in mice. Genes Dev. 27, 504-513. doi:10.1101/gad.205542.112

Soule, H. D., Maloney, T. M., Wolman, S. R. and Peterson, W. D. (1990). Isolationand characterization of a spontaneously immortalized human breast epithelial cellline, MCF-10. Cancer Res. 50, 6075-6086.

Stewart, T. A., Pattengale, P. K. and Leder, P. (1984). Spontaneous mammaryadenocarcinomas in transgenic mice that carry and express MTV/myc fusiongenes. Cell 38, 627-637. doi:10.1016/0092-8674(84)90257-5

Stone, J., de Lange, T., Ramsay, G., Jakobovits, E., Bishop, J. M., Varmus, H.and Lee, W. (1987). Definition of regions in human c-myc that are involved intransformation and nuclear localization. Mol. Cell. Biol. 7, 1697-1709. doi:10.1128/MCB.7.5.1697

Strasser, A., Harris, A. W., Bath, M. L. and Cory, S. (1990). Novel primitivelymphoid tumours induced in transgenic mice by cooperation between myc andbcl-2. Nature 348, 331-333. doi:10.1038/348331a0

Stratikopoulos, E. E., Dendy, M., Szabolcs, M., Khaykin, A. J., Lefebvre, C.,Zhou, M.-M. and Parsons, R. (2015). Kinase and BET inhibitors together clampinhibition of PI3K signaling and overcome resistance to therapy. Cancer Cell 27,837-851. doi:10.1016/j.ccell.2015.05.006

Subik, K., Lee, J.-F., Baxter, L., Strzepek, T., Costello, D., Crowley, P., Xing, L.,Hung, M.-C., Bonfiglio, T., Hicks, D. G. et al. (2010). The expression patterns ofER, PR, HER2, CK5/6, EGFR, Ki-67 and AR by immunohistochemical analysis inbreast cancer cell lines. Breast Cancer (Auckl) 4, 35. doi:10.1177/117822341000400004

Subramanian, A., Tamayo, P., Mootha, V. K., Mukherjee, S., Ebert, B. L., Gillette,M. A., Paulovich, A., Pomeroy, S. L., Golub, T. R. and Lander, E. S. (2005).Gene set enrichment analysis: a knowledge-based approach for interpretinggenome-wide expression profiles. Proc. Natl. Acad. Sci. USA 102, 15545-15550.doi:10.1073/pnas.0506580102

Toyoshima, M., Howie, H. L., Imakura, M., Walsh, R. M., Annis, J. E., Chang,A. N., Frazier, J., Chau, B. N., Loboda, A., Linsley, P. S. et al. (2012). Functionalgenomics identifies therapeutic targets for MYC-driven cancer. Proc. Natl. Acad.Sci. USA 109, 9545-9550. doi:10.1073/pnas.1121119109

Wang, X., Cunningham, M., Zhang, X., Tokarz, S., Laraway, B., Troxell, M. andSears, R. C. (2011). Phosphorylation regulates c-Myc’s oncogenic activity in themammary gland. Cancer Res. 71, 925-936. doi:10.1158/0008-5472.CAN-10-1032

Wasylishen, A. R., Chan-Seng-Yue, M., Bros, C., Dingar, D., Tu, W. B., Kalkat,M., Chan, P.-K., Mullen, P. J., Huang, L., Meyer, N. et al. (2013). MYCphosphorylation at novel regulatory regions suppresses transforming activity.Cancer Res. 73, 6504-6515. doi:10.1158/0008-5472.CAN-12-4063

Zhao, J. J., Gjoerup, O. V., Subramanian, R. R., Cheng, Y., Chen, W., Roberts,T. M. and Hahn, W. C. (2003). Human mammary epithelial cell transformationthrough the activation of phosphatidylinositol 3-kinase. Cancer Cell 3, 483-495.doi:10.1016/S1535-6108(03)00088-6

Zhou, Z., Patel, M., Ng, N., Hsieh, M. H., Orth, A. P., Walker, J. R., Batalov, S.,Harris, J. L. and Liu, J. (2014). Identification of synthetic lethality of PRKDC inMYC-dependent human cancers by pooled shRNA screening. BMC Cancer 14,944. doi:10.1186/1471-2407-14-944

9

RESEARCH ARTICLE Disease Models & Mechanisms (2019) 12, dmm038083. doi:10.1242/dmm.038083

Disea

seModels&Mechan

isms