Neuro-Oncology - University of Toronto

A novel C19MC amplified cell line links Lin28/let-7 to mTOR signalingin embryonal tumor with multilayered rosettes

Tara Spence†, Christian Perotti†, Patrick Sin-Chan, Daniel Picard, Wei Wu, Anjali Singh, Colleen Anderson,Michael D. Blough, J. Gregory Cairncross, Lucie Lafay-Cousin, Douglas Strother, Cynthia Hawkins, Aru Narendran,Annie Huang, and Jennifer A. Chan

Division of Hematology-Oncology, Arthur and Sonia Labatt Brain Tumour Research Centre, Department of Pediatrics, The Hospital for SickChildren, University of Toronto, Toronto, Ontario, Canada (T.S., P.S.-C., D.P., A.H.); Department of Pathology and Laboratory Medicine,University of Calgary, Calgary, Alberta, Canada (C.P., W.W., C.A., J.A.C.); Departments of Oncology and Pediatrics, Alberta Children’s Hospital,Alberta, Canada (A.S., L.L.-C., D.S., A.N.); Department of Clinical Neurosciences, University of Calgary, Calgary, Alberta, Canada (M.D.B., J.G.C.);Department of Pediatric Laboratory Medicine, The Hospital for Sick Children, Toronto, Canada (C.H.)

Corresponding authors: Jennifer A. Chan, MD, Department of Pathology & Laboratory Medicine, University of Calgary, HRIC 2A25, 3330 Hospital Drive NW,Calgary, Alberta, Canada T2N 4N1 ([email protected]); Annie Huang, MD, PhD, Division of Hematology-Oncology, Department of Pediatrics,The Hospital for Sick Children, TMDT, 11-401P, 101 College St., Toronto, Ontario, Canada M5G 1L7 ([email protected]).†Co-first authors.

Background. Embryonal tumor with multilayered rosettes (ETMR) is an aggressive central nervous system primitive neuroectodermaltumor (CNS-PNET) variant. ETMRs have distinctive histology, amplification of the chromosome 19 microRNA cluster (C19MC) atchr19q13.41-42, expression of the RNA binding protein Lin28, and dismal prognosis. Functional and therapeutic studies of ETMR havebeen limited by a lack of model systems.

Methods. We have established a first cell line, BT183, from a case of ETMR and characterized its molecular and cellular features. LIN28knockdown was performed in BT183 to examine the potential role of Lin28 in regulating signaling pathway gene expression in ETMR. Cellline findings were corroborated with immunohistochemical studies in ETMR tissues. A drug screen of 73 compounds was performed toidentify potential therapeutic targets.

Results. The BT183 line maintains C19MC amplification, expresses C19MC-encoded microRNAs, and is tumor initiating. ETMRs, includingBT183, have high LIN28 expression and low let-7 miRNA expression, and show evidence of mTOR pathway activation. LIN28 knockdownincreases let-7 expression and decreases expression of IGF/PI3K/mTOR pathway components. Pharmacologic inhibition of the mTORpathway reduces BT183 cell viability.

Conclusions. BT183 retains key genetic and histologic features of ETMR. In ETMR, Lin28 is not only a diagnostic marker but also a regulatorof genes involved in growth and metabolism. Our findings indicate that inhibitors of the IGF/PI3K/mTOR pathway may be promising noveltherapies for these fatal embryonal tumors. As the first patient-derived cell line of these rare tumors, BT183 is an important, uniquereagent for investigating ETMR biology and therapeutics.

Keywords: ETMR, ETANTR, C19MC, Lin28, mTOR.

Primitive neuroectodermal tumors of the central nervous system(CNS-PNETs) are a class of malignant childhood brain tumors featur-ing poorly differentiated cells of neural origin. Among these, tumorsthat fall under the histologic category of embryonal tumor withmultilayered rosettes (ETMR) (first described as “embryonal tumorwith abundant neuropil and true rosettes” or ETANTR) occur pri-marily in infants and young children and are particularly aggressive,with typical survival of ,1 year after diagnosis despite intensive

therapy.1,2 Recently, ETMRs have been found to harbor recurrentfocal amplifications of chr19q13.41-42, encoding the large polycis-tronic microRNA cluster C19MC, a genetically defining lesion that islikely critical to the biology of the disease.3,4

Amplification of C19MC is associated with marked overexpres-sion of miRNAs in the cluster that can drive proliferation, promotesurvival, and increase tumorigenicity of cells.3 The oncogenic prop-erties of the overexpressed miRNAs reflect some of the known

Received 1 August 2013; accepted 31 August 2013# The Author(s) 2013. Published by Oxford University Press on behalf of the Society for Neuro-Oncology. All rights reserved.For permissions, please e-mail: [email protected].

Neuro-OncologyNeuro-Oncology 16(1), 62–71, 2014doi:10.1093/neuonc/not162Advance Access date 4 December 2013

62

at University of T

oronto Library on O

ctober 22, 2014http://neuro-oncology.oxfordjournals.org/

Dow

nloaded from

http://neuro-oncology.oxfordjournals.org/

functions of C19MC. In normal development, the C19MC locus isposited to control general “stemness” programs, as many C19MCmiRNAs show expression restricted to undifferentiated tissues orgerminal tissues, and their expression decreases with humanembryonic stem cell differentiation.5 – 7 Further supporting the as-sociation of C19MC with stemness, the pluripotency-associatedtranscription factors Oct4 and Nanog may regulate C19MC miRNAsin human embryonic stem cells.5

Interestingly, C19MC-amplified CNS-PNETs also show strikingupregulation of other distinctive primitive markers including theRNA-binding proteins Lin28A and LIN28B,8 a finding that is diag-nostically useful as a marker of ETMRs.9 Beyond its diagnosticutility, however, Lin28 overexpression may be functionally import-ant as well. Through binding and repression of let-7 microRNAs,Lin28 has been found to regulate lineage, stemness, cellular meta-bolism,and the cell cycle in normal and disease states.10 – 12 Lin28also directly binds other mRNAs to increase their translation, withthe cumulative effect that Lin28 may coordinately control bothproliferative growth and metabolism.13 – 19

Despite our increasing knowledge of C19MC and Lin28 func-tions, the biology of ETMR has remained poorly understood. Fur-thermore, the ability to translate the molecular findings topotential therapeutic targets for these tumors has been hamperedby a lack of model systems for the disease. Here, we have estab-lished a patient-derived ETMR cell line that maintains the charac-teristic genetic and histologic features of the disease includingC19MC-amplification and Lin28 expression. We found that Lin28expression was inversely correlated with expression of let-7 micro-RNAs and positively associated with mTOR pathway activation.Lin28 knockdown also increased let-7 expression and decreasedexpression of mTOR signaling pathway components, supportingthe notion that Lin28 dysregulation is functionally important inETMR. This novel cell line will be an important tool for further inves-tigating ETMR biology and for identifying future therapies for thisand other C19MC-overexpressing cancers.

Materials and Methods

Tumor and Nucleic Acid SamplesFormalin-fixed paraffin embedded (FFPE) tumor specimens and freshfrozen tumor specimens were collected with consent per protocolsapproved by the Hospital for Sick Children Research Ethics Board inToronto and the Conjoint Health Research Ethics Board of the Universityof Calgary. Clinical and genetic information on these tumors was reportedpreviously.3 Only tumors classified histologically as CNS-PNET accordingto 2007 WHO CNS tumor classification criteria20 were included. DNA andRNA were extracted from frozen tumor material using Qiagen DNeasy(Qiagen) and Trizol (Invitrogen), respectively, per manufacturer’s protocol.In this paper, we use a genetic definition of ETMR to refer to histologicCNS-PNETs (including those that have a variety of described histologic pat-terns includingbut not restricted to CNS-PNETNOS, ETANTR,medulloepithe-lioma, and ependymoblastoma) that carry C19MC amplification.

Cell CultureFresh tumor tissue was dissociated bygentle manual trituration followed bypassage through a 40 mm mesh filter. Cells were grown in low-adhesiontissue culture flasks (Sarstedt) in defined serum-free media at 378C, 5%CO2. Initial culture media consisted of human neural stem cell proliferationmedia (Stem Cell Technologies) supplemented with heparin, epidermal

growth factor (EGF; 20 ng/mL; Peprotech), fibroblast growth factor (FGF;20 ng/mL; Peprotech), and sonic hedgehog (recombinant human SHH,0.1 mg/mL; R&D Systems)×2 weeks. The cells were subsequently main-tained in the same media without SHH. Cells were fed every 3–4 days.After 1 week in initial culture, small aggregates were apparent. At 2weeks, well-formed tumor neurospheres were present. When tumor-spheres were 100–200 mm in diameter, cells were split using Accumax(Millipore) and manual trituration, and replated at a density of20 000 cells/mL. After �10 consecutive passages, the cell line was desig-nated BT-183-AB (BT183). For differentiation assays, cells were grown for7 days in the media as above, supplemented with 1% fetal bovine serumand without additional EGF or FGF.

Orthotopic Xenografts

Tumorsphere cultures were dissociated into single cells and resuspended insterile phosphate-buffered saline. One hundred thousand viable cells in avolume of 3 ml were implanted into the right striatum of 6–8 week oldCB-17 NOD-SCID mice (Jackson Labs) using the following stereotactic coor-dinates: -1.0 AP, 2.0 ML, 3.0 DV. Mice were euthanized and their brains har-vested at 8 weeks for histologic examination and isolation of nucleic acids.All animal procedures were approved by the Health Sciences Animal CareCommittee at the University of Calgary.

Histology and Immunohistochemistry

For FFPE sections, tissue was fixed in 10% neutral buffered formalin, andprocessed for paraffin embedding by routine methods. For frozen sections,cells grown in tumorspheres or brain tissue with orthotopic xenografts werefixed in 4% paraformaldehyde in phosphate-buffered saline, cryopreservedin 30% sucrose, and embedded in OCT. 5 mm OCT cryosections; 4 mm FFPEsections were collected on Superfrost Plus slides (VWR). Hematoxylin andeosin staining was performed by routine methods. Immunohistochemistrywas performed on FFPE tumor slides. Sections were treated withheat-induced epitope retrieval and blocked for endogenous peroxidaseand biotin. The following primaryantibodies were used: Lin28 (Cell SignalingTechnology), Synaptophysin (Novocastra), b-III-Tubulin (Sigma), Nestin(Thermo Scientific), glial fibrillary acidic protein (GFAP, Dako), Olig2 (Milli-pore), pAKT(S473) (Cell Signaling Technology), p4EBP1(T37/46) (Cell Signal-ing Technology), pS6 (S240/244)(Cell Signaling Technology), and pS6 (S235/236)(Cell Signaling Technology). Antibody reactivity was visualized with theVectaStain ABC detection kit (Vector Laboratories) or Envision+DAB system(Dako). Staining was assessed visually on the basis of intensity and distribu-tionof stains.Tissuemicroarrayspecimenswereassessedbasedonaveragestaining score of at least 2 tissue cores. TS scored all immunohistochemistrystains while masked to cancer status. Scores were reviewed by JC and CEH.Immunofluorescence staining was performed on paraformaldehyde-fixedfrozen sections with antibodies as above and detected with Alexa-488conjugated secondary antibodies (Invitrogen).

Fluorescence In Situ Hybridization

Fluorescence in situ hybridization was performed on individual 5 mm FFPEsections of tumors or tissue microarrays using prelabeled BAC probes(http://www.tcag.ca/cytogenomics) mapping to the chr19q13.42 miRNAcluster (RP11-381E3: 162, 225 bp) and a control chr19p13.11 locus(RP11- 451E20: 165,783 bp) as described previously.3

Copy Number AnalysesTo assess DNA copy number profiles of the C19MC-amplified primarytumor used in this study, we used Illumina Human Omni 2.5 Quadsingle-nucleotide polymorphism (SNP) array for ultra-high resolution copynumber analyses (interrogating 2.5 million SNPs) (Illumina) to generate

Spence et al.: Lin28 regulates mTOR pathway in C19MC-amplified ETMR

Neuro-Oncology 63

at University of T

oronto Library on O


Dow

nloaded from

http://www.tcag.ca/cytogenomics




DNA copy number profiles and HapMap samples from Illumina as controls.Copy number profiles were visualized using DNA Chip Analyzer (dChip). In-ferred copy number was calculated using median smoothing and asliding window of 15 SNP probes. For comparison of copy number altera-tions between primary tumor tissue, BT183 cells, and xenograft tumortissue, array CGH was performed using 0.5 mg of each DNA sample andhuman reference DNA (Promega). The genomic DNA from each samplewas labeled with Cy5 and co-hybridized with gender matched Cy3-labeledreference DNA on SurePrint G3 Human Catalog CGH 8×60 K arrays(Agilent). Arrays were scanned using the Agilent DNA Microarray Scanner,and quantified using Feature Extraction v10.1.1 (Agilent). Normalized CGHsignal (log-ratio) was uploaded into Partek Genomics Suite version 6.12(Partek Incorporated) for visualization and analysis.

MicroRNA AnalysesmiRNA expression profiles were examined using miRNA Nanostringversion1.1 (NanoString Technologies) assessing 654 miRNAs in RNA from12 primary CNS-PNETs in addition to the parent tumor tissue, establishedcell line, and xenograft used in this study. Samples were processed permanufacturers’ instructions, and miRNA expression data were analyzedby hierarchical clustering analyses using Pearson correlation with averagelinkage. Taqman quantitative RT-PCR analysis was also performed toconfirm overexpression of a specific subset of C19MC miRNAs, downstreammiR371-373, and representative let-7 miRNAs using TaqMan MicroRNAReverse Transcription Kit (Applied Biosystems); miRNA expression levels(Ct) determined relative to noncoding RNU6B are represented.

Gene Expression ProfilingMultiple unsupervised hierarchical clustering analyses (genesets of 200–1 000 genes) were performed on human HT-12v4 expression array (Illu-mina) data from 53 primary CNS-PNETs including the cell line and xenograftused in this study. Gene expression data from the primary CNS-PNETs werereported previously.3 Cluster patterns were determined using 300 geneswith the highest coefficient of variation to establish grouping, usingPearson correlation with average linkage. Gene expression profiles werevisualized using Partek Genomics Suite version 6.5 (Partek Incorporated).RNA hybridizations were performed at the Centre of Applied Genomics Facil-ity at the Hospital for Sick Children. Expression levels of the following indi-vidual genes were also assayed by Taqman quantitative RT-PCR geneexpression assays (Applied Biosystems): LIN28A, LIN28B, RPS6, RHEB,IGFBP2. Expression levels of individual transcripts were determined relativeto glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

Lin28 Knockdown

Neurospheres were disaggregated to single-cell suspension using Accumaxand plated in human neural stem cell proliferation media with EGF and FGFat a density of 50 000 cells/mL in 6-well plates. A mix of siRNAs targetingLin28 (OnTarget Plus SMARTpool L-028584-01; Thermo Scientific) was tran-siently transfected into cells using Lipofectamine 2000 (Invitrogen) permanufacturer’s protocol at final siRNA concentration of 40 nM in a totaltransfection volume of 2 mL per well. Cells were harvested at 7 days post-transfection for RNA extraction and analysis by quantitative RT-PCR.

In Vitro Chemosensitivity Testing

Cells were cultured in 96-well plates (5×103 cells/well) and incubated with1 mM concentration of the drugs listed in Supplemental Table 1 and asreported previously.21 All targeted therapeutic agents used in the screeninganalysis were synthesized, checked for purity, and provided by Chemietek(Minneapolis, MN). These agents were dissolved in dimethyl sulfoxide(DMSO) to a final concentration of 10 mM, stored frozen at -208C, and

diluted appropriately in culture medium at the time of study. Each treat-ment was performed in triplicate. Survival was measured after 4 days inculture and compared with vehicle control using Alamar blue viabilityassay (Invitrogen) detected in a microplate reader per manufacturer’sprotocol.

Western BlottingWestern blot analysis and chemiluminescence detection were performedwith standard methods for detection using primary antibodies againstb-Actin (Santa Cruz Biotechnology), phospho-4EBP1(Thr37/46) (Cell Signal-ing Technology), pS6(S235/236) (Cell Signaling Technology), and secondaryanti-mouse and anti-rabbit HRPconjugated antibodies (Santa CruzBiotech-nology).

Statistical Analyses

For analyses of cell viability and changes in protein expression on Westernblots, results were from at least 3 experiments and were analyzed by ttest. To assess deregulation of transcripts from gene expression profiling, in-cluding LIN28 mRNA and let-7 miRNAs, we assessed gene/miRNA enrich-ment with a supervised t test adjusted for multiple hypotheses testingwith the false-discovery-rate method using R (version R ×64 2.15.2).A P value of ,.05 was regarded as significant for all analyses.

Results

Establishment of Cell Cultures from an ETMR

We used tumor tissue from a 2-year-old male with a knownC19MC-amplified posterior fossa ETMR (Figs. 1 and 2) to prepareprimary cell cultures and then propagated the cells to develop astable cell line referred to as BT183. These cells formed nonadher-ent neurospheres within �10 days of initial culture when grown inin serum-free defined media (Fig. 1A).

Brain tumor stem-like cells derived from some medulloblas-toma and glioblastoma tumors expressed stem cell surfacemarkers such as the cell surface markers CD133, CD15, or both,and the intermediate filament nestin.22 – 25 We found that BT183tumorspheres were negative for CD15 and had only rare cells posi-tive for CD133 but showed strong expression of the neural stem/progenitor marker nestin26 in the majority of cells (Fig. 1A). Withregard to lineage markers, BT183 tumorspheres contained onlyscattered cells expressing the neuronal marker b-III-tubulin, theoligodendroglial lineage marker Olig2, or the astrocytic markerGFAP. Individual BT183 tumorspheres could be dissociated tosingle cells and then replated in media to generate secondaryspheres. Furthermore, when grown in differentiation media,BT183 cells attach to the tissue culture plate and extended pro-cesses (Fig. 1B). These morphologically differentiated cells pre-dominantly expressed the neuronal b-III-tubulin, with rare cellsexpressing Olig2 or GFAP (Fig. 1 and data not shown). Together,these findings support that BT183 cells exhibit stem cell-like char-acteristics but preferentially differentiate along neuronal lineages.

BT183 Cells are Tumor-initiating in Vivo

One of the features of brain tumor stem-like cells is the ability to ini-tiate tumors. To test whether BT183 could generate tumors in vivo,we implanted the cells into the brains of NOD-SCID mice after every5 passages in vitro. The cells maintained tumorigenicity in miceover 6 subsequent cohorts of mice injected, after which routine


64

at University of T

oronto Library on O


Dow

nloaded from

http://neuro-oncology.oxfordjournals.org/lookup/suppl/doi:10.1093/neuonc/not162/-/DC1


in vivo propagation was discontinued. The BT183 xenograft tumorsbore striking histologic resemblance to the original patient tumor,with sheets of poorly differentiated cells punctuated by well-developed multilayered rosettes. (Fig. 2).

BT183 Cells Harbor Amplification of chr19q13.41

The genetic hallmark of ETMR is high-level amplification ofchr19q13.41.3,4,27,28 To determine the genetic features of ournewly established cell line, we analyzed the parent tumor, the

cell line, and xenograft tumors using several methods. High-resolution SNP array confirmed the chr19q13.41 amplification inthe parent tumor tissue with amplicon boundaries fromrs4803149 to SNP19-59570640, inclusive (Fig. 3A). Fluorescencein situ hybridization showed chr19q13.41 amplification in boththe parent tumor, and xenograft (Fig. 3B). Furthermore, arrayCGH showed relative similarity between the parent tumor, cellline, and xenograft with respect to genome-wide copy numberalterations. These included focal amplification of 19q13.41, asexpected, but additionally revealed the presence of several large

Fig. 1. BT183 cells show stem cell-like features in vitro. (A) BT183 tumorspheres under phase contrast microscopy, and with immunofluorescence stainingfor Nestin, CD15, CD133, ß-III-Tubulin, Olig2, and GFAP whengrownin neural stemcell proliferation media. (B) BT183 cells after differentiation in mediawith1% serum shown under phase contrast and immunostained for ß-III-tubulin (600×magnification).

Fig. 2. Histopathologic features of BT183 cells in vivo. Features of the patient’s primary tumor (SM2211) and orthotopic xenografts of BT183 in NOD-SCIDmice. Immunohistochemical analyses from primary (SM2211) and xenograft (BT183) tumors: Lin28, Nestin, Synaptophysin, GFAP, and Olig2immunostains are shown in relation to a hematoxylin and eosin stain (100×magnification).


Neuro-Oncology 65

at University of T

oronto Library on O


Dow

nloaded from


Fig. 3. Genetic characterization of the BT183 cell line. (A) Inferred copy number plots generated using dChip pair-matched analysis reveals thechr19q13.41 amplicon in primary tumor and plotted relative to chromosome ideograms. (B) Fluorescence in situ hybridization confirms high-levelamplification of chr19q13.41 in primary tumor and BT183 xenograft tumor (C) Segmentation identifies chr19q13.41 amplicon, detected with AgilentCGH array (8×60k) in SM2211 primary tumor, and BT183 cell line and xenograft. DNA copy number profiles of normal brain, SM2211 primary tumor,and BT183 cell line and xenograft represented as dot plots of genomic region copy number change and plotted across chromosomes with DNA gainsand losses indicated by red and blue, respectively. Y-axis is log2 scale, one copy gain¼ 0.6 of log2 value. X-axis is the chromosome number.


66

at University of T

oronto Library on O


Dow

nloaded from


segmental and whole arm or whole chromosome gains or losses atchromosomes 2(+), 7(+), 6q(2), and regions of 1 and 19 (Fig. 3C).Collectively, these data indicate that BT183 recapitulates the mo-lecular and morphologic features of the parent tumor and repre-sents an important first model for EMTR and related C19MC-amplified or Lin28-overexpressing embryonal brain tumors.

BT183 Cells Show RNA and Protein Expression Characteristicof C19MC-amplified Tumors

C19MC-amplified tumors have distinctive mRNA transcriptionalprofiles and patterns of protein expression.3 When assessedrelative to the gene expression profiles of a previously reportedpanel of 53 CNS-PNETs,3 including 14 others with C19MC amplifica-tion and ETMR-compatible histology, unsupervised hierarchicalclustering revealed that BT183, its parent tumor, and the xenograft

all clustered together and that their transcriptional profiles clus-tered clearly with the ETMR group and included high expressionof LIN28 transcripts (Fig. 4A). At the protein level, BT183 xenografttumors showed immunohistochemical similarity to its parentETMR, with high expression of nestin and Lin28, positivity for Ini1,patchy positivity for synaptophysin, and low expression of GFAPand Olig2 (Fig. 2), a pattern that is characteristically seen inC19MC-amplified tumors.3

We previously observed that the pathognomonic C19MC ampli-fication characterizing ETMR is typically accompanied by high ex-pression of miRNAs in the cluster.3 To test whether BT183 notonly retained the DNA level changes but also the associatedmiRNA profiles, we examined miRNA expression using NanostringmiRNA profiling on the parent tumor, cell line, and xenograft.This global analysis showed that miRNA expression was similarbetween the three (Fig. 4B). Expression of specific mature miRNAs

Fig. 4. BT183 displays a gene expression signature characteristic of C19MC-amplified tumors. (A) Multiple unsupervised hierarchical analyses identifiesthat gene expression profiles of SM2211 primary tumor, and BT183 cell line and xenograft cluster together with C19MC-amplified CNS-PNETs. (B)Nanostring and (C) miRNA-specific qRT-PCR confirm high expression of C19MC miRNAs in primary tumor, and BT183 cell line and xenograft. miRNAs inthe neighboring unamplified miR 371-373 cluster are not highly expressed.


Neuro-Oncology 67

at University of T

oronto Library on O


Dow

nloaded from


encoded in the region was also evaluated by miRNA-specificTaqman quantitative RT-PCR. This confirmed high-level expressionof C19MC-encoded miRNAs, whereas the miRNAs in the adjacentnonamplified miRNA 371-373 cluster were not highly expressed(Fig. 4C).

Lin28 Regulates Expression of let-7 miRNAsin C19MC-amplified Tumors

Lin28 proteins are known to regulate stability and/or translation ofmany different mRNAs13 – 16,17 as well as the processing of non-coding RNAs. With respect to the latter, a key target of Lin28 isthe let-7 family of microRNAs.10 – 12 Multiple lines of evidence indi-cate that each negatively regulates the expression of the other andthat together lin28/let-7 acts as a genetic switch wherein expres-sion of either Lin28 or let-7 promotes a stem-like or differentiatedstate, respectively (reviewed in 29). However the role of Lin28 inETMR is currently unknown. To explore whether a Lin28/let-7switch is active in ETMR, we examined the expression of microRNAtranscripts relative to expression of LIN28 in our panel of CNS-PNETs. We found that high LIN28A and LIN28B expression were in-versely correlated with expression of all let-7 miRNA family

members (Fig. 5A and B). Tumors that harbor C19MC amplification(including BT183 cells, xenograft, and original patient tumor) haveuniformly high LIN28 expression and low let-7 expression, whereasthe two other major subsets of CNS-PNET show low expression ofLIN28 and higher expression of let-7 miRNAs (P , .05 for all com-parisons). Furthermore, using a pool of siRNAs, knockdown ofLIN28 in BT183 cells resulted in significantly increased expressionof representative let-7 family members let-7a (4.34-334% increasechange; P¼ .028) and let-7e (15.25-1425% increase change; P¼.019) compared with treatment with nontargeting siRNAs (Fig. 5Cand C). Together, these findings are consistent with the presenceof a functional Lin28/let-7 axis in ETMR.

BT183 Cells and C19MC-amplified Tumors Show Evidenceof Lin28-mediated Insulin/PI3K/mTOR SignalingDysregulation

Aside from controlling proliferation and differentiation, the Lin28/let-7 axis may also control growth and metabolism throughregulation of multiple components of the insulin-PI3K-mTORpathway.12,30 To determine whether human ETMRs have activemTOR pathway signaling, we assessed 5 FFPE patient ETMR

Fig. 5. LIN28 regulates expression of let-7 miRNAs in C19MC-amplified tumors. (A) Heat map of LIN28 and let-7 family gene expression in primary tumor,and BT183 cell line and xenograft using a supervised t test adjusted for multiple testing (false discovery rate ≤0.05); magnitude (fold change) andsignificance (P value) of genes in C19MC-amplified group in contrast to Groups 2 and 3 CNS-PNETs are shown. LIN28 is inversely correlated with let-7family miRNAs. (B) Taqman miRNA qRT-PCR on primary CNS-PNETs reveals decreased expression of let-7a, -7b and -7 g in LIN28-positive Group 1tumors, as compared with Groups 2 and 3 tumors. (C) Knockdown of LIN28 increases expression of let-7a and let-7e miRNAs in BT183 cells comparedwith treatment with nontargeting control siRNA (*P , .05).


68

at University of T

oronto Library on O


Dow

nloaded from


samples for expression of phosphorylated-S6 (pS6 S235/236 andS240/244) and phosphorylated 4EBP1 (p4EBP1 Thr37/46) byimmunohistochemistry as readouts of pathway activation. Wefound that all 5 ETMR samples stained positively for bothphosphorylated-4EBP1 and phosphorylated-S6 (Fig. 6A and B)consistent with mTOR activation. To then determine whetherLin28 might regulate the insulin/PI3K/mTOR pathway in C19MC-amplified tumors through modulating expression of mTORpathway components, we performed siRNA knockdown of Lin28in BT183 cells (Fig. 6C) and monitored expression of severalpathway genes. In our knockdown experiments, LIN28A andLIN28B transcripts were reduced by 60%–70% compared withtransfection with nontargeting siRNA control (P , .05). DecreasingLIN28 resulted in both an upregulation of let-7e and let-7a, asdescribed above (Fig. 5C) and a concomitant decrease in ex-pression of IGFBP2, RHEB, and RPS6 (Fig. 6C; P , .05 for each),demonstrating that the Lin28/let-7 axis regulates growth andmetabolism signaling components in ETMR.

To determine whether BT183 cells show evidence of a depend-ence on insulin/PI3K/mTOR signaling, we then treated cells with apanel of 73 pharmacologic inhibitors spanning several classes oftargets. BT183 cells showed sensitivity to IGF1R, PI3K, and mTOR in-hibition in contrast to little or no sensitivity to a variety of other re-ceptor tyrosine kinase and tyrosine kinase inhibitors (Fig. 7A,Supplemental Table 2). In the panel of drugs screened, the onlyother pharmacologic agents that showed similar responses at1 mM concentration were more generalized cell cycle or structuraldirected therapies (eg, topoisomerase inhibitors, tubulin stabilizers,Bcl-2 inhibitors, nucleic acid synthesis inhibitors) that are not tar-geted at specific signaling pathways. Overall, the data support arole for the Lin28/let-7 axis and signaling through insulin-PI3K-mTOR as an important factor in these tumors’ biology. The datafurther highlight the utility of this line for in vitro screening, andsuggest that other types of agents (eg, HDAC inhibitors amongothers) may also have therapeutic potential in these aggressivetumors.

DiscussionETMR (now the proposed name for tumors previously calledETANTR) is a recently discerned type of CNS-PNET that is clinico-pathologically distinct from other subgroups of CNS-PNETs due toits very young demographic, strikingly poor survival, characteristichistopathology featuring multilayered rosettes, and hallmarkamplification of chromosome 19q13.41 that encodes the largemicroRNA cluster C19MC. We have established and characterizedthe first known ETMR cell line, BT183. We showed that BT183harbors the hallmark C19MC amplification, displays transcriptand protein expression profiles characteristic of ETMR, and iscapable of initiating tumor growth when orthotopically xeno-grafted into mice. Although a small number of CNS-PNET celllines have been generated and used by others, their C19MCstatus is either unknown, not reported, or known to be unampli-fied.3,31 – 33 Furthermore, their histologies and resemblance to spe-cific variants are unclear. As the only cell line derived from agenetically well-documented ETMR, BT183 represents an import-ant tool for further investigation into ETMR and biology and willenable future efforts to find effective therapies for this disease.

Characterization of ETMR using BT183 supports is a role for theRNA-binding protein Lin28. High Lin28a and Lin28b expression hasbeen observed previously in C19MC-amplified PNETs by our groupand others,3,9 and this expression has been found to be a usefuldiagnostic tool for distinguishing these ETMR from atypical teratoidrhabdoid tumor, medulloblastoma, anaplastic ependymoma, glio-blastoma, and non-ETMR PNETs.9 However whether Lin28 proteinsare simply markers of the primitive cell state in these tumors orserve a more active role in ETMR biology is unknown. In our cellline and panel of tumors, we found that C19MC amplificationwas not only associated with high LIN28 expression, but alsothat LIN28 levels were inversely correlated with let-7 microRNA ex-pression. Functionally, Lin28 appears to regulate let-7 levels, asknockdown of LIN28 results in increased let-7 levels. Amongother roles, Lin28 proteins can repress let-7 microRNAs,34 – 37 and

Fig. 6. C19MC-group PNETs show evidence of mTOR activation. (A) Representative images of ETMRs immunostained with Lin28, p4EBP1 (Thr37/46), andpS6 (Ser240/244). The mTOR targets, 4EBP1 and S6, are activated in primary and BT183 xenograft tumors. (B) Immunoreactivity of C19MC-amplifiedprimary tumors for LIN28, phospho-4EBP1, and phospho-S6. (C) LIN28 knockdown reduces expression of S6, RHEB, and IGFBP2. (*P , .05, **P , .01).


Neuro-Oncology 69

at University of T

oronto Library on O


Dow

nloaded from



together lin28/let-7 form a conserved genetic switch controllinglineage, stemness, cellular metabolism, and the cell cycle innormal and disease states.10,11,12,30 Our results support that thisswitch is operational in ETMR.

As emerging evidence indicates that Lin28/let-7 controls insulin-PI3K-mTOR signaling, we also investigated whether ETMRs showevidence of aberrant activation of mTOR. We found expression ofboth phosphorylated 4EBP1 and phosphorylated S6, supportingthat mTOR is indeed activated in these tumors. This was alsoevident in BT183 cells and xenografts. Furthermore, when wetreated BT183 cells with a panel of inhibitors, the cells showed sen-sitivity to agents targeting IGF1R, PI3K, and mTOR but not to thosetargeting other receptor tyrosine kinases or intracellular signalingkinases. Beyond corroborating the potential activity of lin28/let-7in regulating metabolism in these cells, these finding have thera-peutic implications as well. Therapies targeting the insulin/mTORpathway, such as rapamycin and metformin, have already beendeveloped and are in clinical use for other indications. It is possiblethat these drugs, likely in combination therapy, will have clinicalvalue as therapy for ETMR.

Finally, our small drug screen highlights the utility of our cells forin vitro studies and the identification of potential other classes ofdrugs that might be investigated for ETMR therapy as well. Our invitro findings suggest a rationale for differentiating therapies (asexemplified by HDAC inhibitors, valproic acid, or retinoids) in com-bination with topoisomerase inhibitors (such as etoposide). Someof these drugs are already in clinical trials for pediatric brain tumors(eg, ClinicalTrials.gov Identifiers NCT01294670, NCT00867178,NCT01076530, NCT00994500). Although individual trials are un-likely to have sufficient numbers of ETMR patients from which todraw conclusions for this particular patient subset, the results ofthese studies will be followed with interest. As is the case with

other rare pediatric malignancies, true progress in understandingtumor biology and achieving higher cure rates will require inter-national collaboration.

Overall, our data show that the unique cell line BT183 representsETMR and will enable further studies aimed at understanding thefunction of C19MC microRNAs, translational regulation by Lin28,and potential biologic dependencies that may be exploited fortherapeutic benefit.

Supplementary MaterialSupplementary material is available online at Neuro-Oncology(http://neuro-oncology.oxfordjournals.org/).

FundingThis work was supported by the Clark H. Smith Neurologic and PediatricTumor Bank at the University of Calgary, the Canadian Institute of HealthResearch (AH; grant no. 102684), Brainchild (AH), Kids Cancer Care Founda-tion of Alberta (JAC), Alberta Innovates Health Solutions (JAC), Alexander’sQuest (JAC), Brain Tumour Foundation of Canada (AN).

Conflict of interest statement. None declared.

References1. Eberhart CG, Brat DJ, Cohen KJ, Burger PC. Pediatric neuroblastic brain

tumors containing abundant neuropil and true rosettes. Pediatr DevPathol. 2000;3(4):346–352.

Fig. 7. Inhibition of mTOR pathway reduces survival of BT183 cell line. (A) Survival assay of BT183 cells following pharmacological inhibition of multiplesignaling pathways; see Supplemental Table 1 for drug names. Treatment with the mTOR inhibitors Pp242 and rapamycin reveals reduced viability ofBT183 cells as detected by Alamar blue assay. (B) Western blot of phospho-S6 (Ser235/236) and phospho-4EBP1 (Thr37/46) in BT183 cells treated withvehicle or rapamycin. Actin is loading control.


70

at University of T

oronto Library on O


Dow

nloaded from








2. Gessi M, Giangaspero F, Lauriola L, et al. Embryonal tumors withabundant neuropil and true rosettes: a distinctive CNS primitiveneuroectodermal tumor. Am J Surg Pathol. 2009;33(2):211–217.

3. Li M, Lee KF, Lu Y, et al. Frequent amplification of a chr19q13.41microRNA polycistron in aggressive primitive neuroectodermal braintumors. Cancer Cell. 2009;16(6):533–546.

4. Korshunov A, Remke M, Gessi M, et al. Focal genomic amplification at19q13.42 comprises a powerful diagnostic marker for embryonaltumors with ependymoblastic rosettes. Acta Neuropathol. 2010;120(2):253–260.

5. Bar M, Wyman SK, Fritz BR, et al. MicroRNA discovery and profiling inhuman embryonic stem cells by deep sequencing of small RNAlibraries. Stem Cells. 2008;26(10):2496–2505.

6. Laurent LC, Chen J, Ulitsky I, et al. Comprehensive microRNA profilingreveals a unique human embryonic stem cell signature dominatedby a single seed sequence. Stem Cells. 2008;26(6):1506–1516.

7. Ren J, Jin P, Wang E, Marincola FM, Stroncek DF. MicroRNA and geneexpression patterns in the differentiation of human embryonic stemcells. J Transl Med. 2009;7:20.

8. Picard D, Miller S, Hawkins CE, et al. Markers of survival and metastaticpotential in childhood CNS primitive neuro-ectodermal brain tumours:an integrative genomic analysis. Lancet Oncol. 2012;13(8):838–848.

9. Korshunov A, Ryzhova M, Jones DT, et al. LIN28A immunoreactivity is apotent diagnostic marker of embryonal tumor with multilayeredrosettes (ETMR). Acta Neuropathol. 2012;124(6):875–881.

10. Viswanathan SR, Powers JT, Einhorn W, et al. Lin28 promotestransformation and is associated with advanced humanmalignancies. Nat Genet. 2009;41(7):843–848.

11. West JA, Viswanathan SR, Yabuuchi A, et al. A role for Lin28 inprimordial germ-cell development and germ-cell malignancy.Nature. 2009;460(7257):909–913.

12. Zhu H, Shyh-Chang N, Segre AV, et al. The Lin28/let-7 axis regulatesglucose metabolism. Cell. 2011;147(1):81–94.

13. Xu B, Huang Y. Histone H2a mRNA interacts with Lin28 and contains aLin28-dependent posttranscriptional regulatory element. NucleicAcids Res. 2009;37(13):4256–4263.

14. Balzer E, Heine C, Jiang Q, Lee VM, Moss EG. LIN28 alters cell fatesuccession and acts independently of the let-7 microRNA duringneurogliogenesis in vitro. Development. 2010;137(6):891–900.

15. Wilbert ML, Huelga SC, Kapeli K, et al. LIN28 binds messenger RNAs atGGAGA motifs and regulates splicing factor abundance. Mol Cell. 2012;48(2):195–206.

16. Peng S, Chen LL, Lei XX, et al. Genome-wide studies reveal that Lin28enhances the translation of genes important for growth and survivalof human embryonic stem cells. Stem Cells. 2011;29(3):496–504.

17. Hafner M, Max KE, Bandaru P, et al. Identification of mRNAs bound andregulated by human LIN28 proteins and molecular requirements forRNA recognition. Rna. 2013;19(5):613–626.

18. Polesskaya A, Cuvellier S, Naguibneva I, Duquet A, Moss EG,Harel-Bellan A. Lin-28 binds IGF-2 mRNA and participates in skeletalmyogenesis by increasing translation efficiency. Genes Dev. 2007;21(9):1125–1138.

19. Li N, Zhong X, Lin X, et al. Lin-28 homologue A (LIN28A) promotes cellcycle progression via regulation of cyclin-dependent kinase 2 (CDK2),cyclin D1 (CCND1), and cell division cycle 25 homolog A (CDC25A)expression in cancer. J Biol Chem. 2012;287(21):17386–17397.

20. McLendon RE, Judkins AR, Eberhart CG, Fuller GN, Sarkar C, Ng HK.Central nervous system primitive neuroectodermal tumours. In:Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, eds. WHOClassification of Tumours of the Central Nervous System. 4th edn.Lyon: International Agency for Research on Cancer (IARC); 2007:141–146.

21. Singh A, Lun X, Jayanthan A, et al. Profiling pathway-specific noveltherapeutics in preclinical assessment for central nervous systematypical teratoid rhabdoid tumors (CNS ATRT): Favorable activity oftargeting EGFR- ErbB2 signaling with lapatinib. Mol Oncol. 2013;7(3):497–512.

22. Hemmati HD, Nakano I, Lazareff JA, et al. Cancerous stem cells canarise from pediatric brain tumors. Proc Natl Acad Sci USA. 2003;100(25):15178–15183.

23. Singh SK, Hawkins C, Clarke ID, et al. Identification of human braintumour initiating cells. Nature. 2004;432(7015):396–401.

24. Singh SK, Clarke ID, Terasaki M, et al. Identification of a cancer stem cellin human brain tumors. Cancer Res. 2003;63(18):5821–5828.

25. Read TA, Fogarty MP, Markant SL, et al. Identification of CD15 as amarker for tumor-propagating cells in a mouse model ofmedulloblastoma. Cancer Cell. 2009;15(2):135–147.

26. Lendahl U, Zimmerman LB, McKay RD. CNS stem cells express a newclass of intermediate filament protein. Cell. 1990;60(4):585–595.

27. Pfister S, Remke M, Castoldi M, et al. Novel genomic amplificationtargeting the microRNA cluster at 19q13.42 in a pediatric embryonaltumor with abundant neuropil and true rosettes. Acta Neuropathol.2009;117(4):457–464.

28. Nobusawa S, Yokoo H, Hirato J, et al. Analysis of chromosome 19q13.42amplification in embryonal brain tumors with ependymoblasticmultilayered rosettes. Brain Pathol. 2012;22(5):689–697.

29. Thornton JE, Gregory RI. How does Lin28 let-7 control developmentand disease? Trends Cell Biol. 2012;22(9):474–482.

30. Frost RJ, Olson EN. Control of glucose homeostasis and insulinsensitivity by the Let-7 family of microRNAs. Proc Natl Acad Sci USA.2011;108(52):21075–21080.

31. Dallas PB, Terry PA, Kees UR. Genomic deletions in cell lines derived fromprimitive neuroectodermal tumors of the central nervous system.Cancer Genet Cytogenet. 2005;159(2):105–113.

32. Inda MM, Mercapide J, Munoz J, et al. PTEN and DMBT1 homozygousdeletion and expression in medulloblastomas and supratentorialprimitive neuroectodermal tumors. Oncol Rep. 2004;12(6):1341–1347.

33. Fults D, Pedone CA, Morse HG, Rose JW, McKay RD. Establishment andcharacterization of a human primitive neuroectodermal tumor cellline from the cerebral hemisphere. J Neuropathol Exp Neurol. 1992;51(3):272–280.

34. Viswanathan SR, Daley GQ, Gregory RI. Selective blockade of microRNAprocessing by Lin28. Science. 2008;320(5872):97–100.

35. Heo I, Joo C, Cho J, Ha M, Han J, Kim VN. Lin28 mediates the terminaluridylation of let-7 precursor MicroRNA. Mol Cell. 2008;32(2):276–284.

36. Newman MA, Mani V, Hammond SM. Deep sequencing of microRNAprecursors reveals extensive 3′ end modification. Rna. 2011;17(10):1795–1803.

37. Rybak A, Fuchs H, Smirnova L, et al. A feedback loop comprising lin-28and let-7 controls pre-let-7 maturation during neural stem-cellcommitment. Nat Cell Biol. 2008;10(8):987–993.


Neuro-Oncology 71

at University of T

oronto Library on O


Dow

nloaded from


Documents

Neuro-Oncology - University of Toronto