Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351 Cancer Molecular and Cellular Pathobiology Research

CIC-DUX4 Induces Small Round Cell Sarcomas Distinct from Ewing Sarcoma Toyoki Yoshimoto1,2, Miwa Tanaka1, Mizuki Homme1, Yukari Yamazaki1, Yutaka Takazawa3, Cristina R. Antonescu4, and Takuro Nakamura1

Abstract

CIC-DUX4 sarcoma (CDS) or CIC-rearranged sarcoma is a short spindle cells. -expression profiles of CDS and eMC subcategory of small round cell sarcoma resembling the morpho- revealed upregulation of CIC-DUX4 downstream such as logical phenotypes of Ewing sarcoma (ES). However, recent PEA3 family genes, Ccnd2, Crh, and Zic1. IHC analyses for both clinicopathologic and molecular genetic analyses indicate that mouse and human tumors showed that CCND2 and MUC5AC CDS is an independent disease entity from ES. Few ancillary are reliable biomarkers to distinguish CDS from ES. Gene silenc- markers have been used in the differential diagnosis of CDS, and ing of CIC-DUX4 as well as Ccnd2, Ret, and Bcl2 effectively additional CDS-specific biomarkers are needed for more defini- inhibited CDS tumor growth in vitro. The CDK4/6 inhibitor tive classification. Here, we report the generation of an ex vivo palbociclib and the soft tissue sarcoma drug trabectedin also mouse model for CDS by transducing embryonic mesenchymal blocked the growth of mouse CDS. In summary, our mouse cells (eMC) with human CIC-DUX4 cDNA. Recipient mice trans- model provides important biological information about CDS planted with eMC-expressing CIC-DUX4 rapidly developed an and provides a useful platform to explore biomarkers and ther- aggressive, undifferentiated sarcoma composed of small round to apeutic agents for CDS. Cancer Res; 77(11); 1–11. 2017 AACR.

Introduction CIC–DUX4 and EWS–ETS as well as the potential different cell-of- origin of each tumor type. CIC-DUX4 sarcomas (CDS) belong to a highly aggressive IHC detection of ETV4, a transcriptional target of CIC-DUX4, is subgroup of small round cell sarcoma, affecting predominantly a useful marker for the histologic diagnosis of CDS (2–6). This children and young adults (1). Although EWS–ETS-negative small suggests that further molecular characterization of CDS will round cell sarcomas were previously classified as Ewing-like expand our understanding on their biological behavior, as well sarcoma or Ewing sarcoma (ES)–like round cell sarcoma, increas- as provide novel biomarkers and molecular targets for therapy. To ing evidence suggests the distinct biology of CDS with CIC–DUX4 this end we developed an animal model that recapitulates the gene fusions, secondary to either a t(4;19)(q35;q13) or t(10;19) phenotypes of human CDS. CIC–DUX4 gene fusion is the initi- (q26.3;q13) translocation (1, 2). CDS morphologically show ating and causative event in CDS, encoding a chimeric transcrip- small- to medium-sized, round to ovoid cells, packed in solid tion factor consisting of the large part of CIC including its DNA- sheets, lacking any line of differentiation. Thus, the differential binding HMG box, and the transcriptional activation domain diagnosis from ES is often difficult without detecting the CIC- derived from the DUX4 C-terminus (2). Deregulation of CIC related fusions (1, 3). CDS show a poor outcome and overall target genes such as ETV4 is one of the key events in the devel- survival of CDS patients was worse than that of ES patients (3). It is opment and progression of CDS (2), and chimeric transcription therefore important to clarify the biological characteristics respon- factors act as an oncogene only in proper cellular context (7–9). sible for the different behaviors between CDS and ES. The differ- We have previously successfully generated an ES mouse model ences may be caused by distinct signaling pathways regulated by using an ex vivo–based technology (7). In this model, EWS–FLI1 appropriately upregulates its target genes in the sarcoma cells as well as in the embryonic osteochondrogenic progenitors, consid- 1 Division of Carcinogenesis, The Cancer Institute, Japanese Foundation for ered the likely cell-of-origin of ES, exhibiting morphologic and 2 Cancer Research, Tokyo, Japan. Department of Pathology, Toranomon Hos- molecular features of human ES. In addition, an alveolar soft pital, Tokyo Japan. 3Division of Pathology, The Cancer Institute, Japanese Foundation for Cancer Research, Tokyo Japan. 4Department of Pathology, part sarcoma model mouse that recapitulates histologic features Memorial Sloan-Kettering Cancer Center, New York, New York. and high metastatic potency of human counterpart has been generated (10). Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). Taking advantage of this ex vivo gene transduction system, we have developed a similar mouse model for human CDS. Mouse T. Yoshimoto and M. Tanaka contributed equally to this article. embryonic mesenchymal cells transduced with the CIC–DUX4 Corresponding Author: Takuro Nakamura, The Cancer Institute, Japanese develop small blue-round cell tumors, as seen in human CDS, Foundation for Cancer Research, 3-8-31 Ariake, Koto-ku, Tokyo 135-8550, Japan. which showed a highly aggressive growth. Gene microarray anal- Phone: 813-3570-0462; Fax: 813-3570-0463; E-mail: [email protected] yses of CDS revealed distinct gene-expression profiles of CDS doi: 10.1158/0008-5472.CAN-16-3351 from ES, with significant upregulation of the cell-cycle progres- 2017 American Association for Cancer Research. sion pathway, such as Ccnd2 gene in CDS. Furthermore, cyclin D2

www.aacrjournals.org OF1

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

Yoshimoto et al.

and MUC5AC were found as novel biomarkers for diagnosis of In vivo imaging CDS, and a growth inhibitory effect of trabectedin was observed. The mice transplanted CDS tumor cells introduced with the Modeling fusion gene-associated sarcomas thus provides impor- luciferase cDNA were administrated with luciferin. Illumination tant tools for further our understanding in the pathogenesis of from the labeled tumor cells were monitored by the IVIS Lumina these rare types of neoplasms. LT imaging system (PerkinElmer).

Materials and Methods RT-PCR and real-time quantitative RT-PCR Total RNA extraction, reverse transcription and RNA quantifi- Generation and characterization of the CDS model mouse cation were performed according to methods described previously N-terminal HA-tagged CIC–DUX4 was introduced into the (2). Conventional RT-PCR and real-time quantitative RT-PCR pMYs–IRES–GFP vector. Limbs of Balb/c mouse embryo (Clea were performed using a Gene Amp 9700 thermal cycler (Applied Japan) were removed aseptically on 18.5 dpc, and embryonic Biosystems) and a 7500 Fast Real-Time PCR System (Applied mesenchymal cells (eMC) were obtained by dissection using Biosystems), respectively. The sequences of the oligonucleotide two mg/mL collagenase (Wako Pure Chemical) at 37Cfor primers used are shown in Supplementary Table S1. 2 hours. eMC were cultured in growth medium composed of Iscove's Modified Dulbecco's Medium (Invitrogen) supplemen- ted with 15% FBS and subjected immediately to retroviral Microarray analysis infection without further purification. The analysis of cell GeneChip analysis was conducted to determine gene expres- fi surface marker expression suggested that eMC might share sion pro les. The murine HT MG-430 PM array (Affymetrix) was overlapping characteristics with mesenchymal stem cells (Sup- hybridized with aRNA probes generated from eMC 48 hours after plementary Fig. S1). Retroviral stock was added into the medi- transduction with pMYs-CIC-DUX4 or empty vector, CDS and ES um containing eMC with 6 mg/mL of Polybrene (Sigma) and tumor tissues, or a mixture of mouse normal tissues according to then spun at 700 g for 1 hour. The spin infection was repeated methods described previously (11). The expression data were after 24 hours. Transduced eMC were mixed with growth factor- analyzed using GeneSpring ver 12.6 (Agilent Technologies) reduced Matrigel (Becton Dickinson) and 1 106 cells were and gene set enrichment analysis (GSEA) was performed using transplanted into the subcutaneous regions of Balb/c nude GSEA-P 2.0 software (12). The microarray datasets are acces- mice was performed as described previously (7). Transduction sible through the NCBI Gene Expression Omnibus database efficiency of the CIC-DUX4 retrovirus was confirmed by flow (http://www.ncbi.nlm.nih.gov/geo), with an accession number cytometry using FACSCalibur (Beckton Dickinson; Supplemen- GSE90978. tary Fig. S2). All experiments described in this study were performed in strict accordance with standard ethical guidelines Data comparisons and clustering between murine and human and approved by the animal care committee at the Japanese microarray data Foundation for Cancer Research under licenses 10-05-9 and The microarray data from six mCDS to five hCDS samples (1) 0604-3-13. were compared with human sarcoma microarray data sets. Data from the ONCOMINE data base (https://www.oncomine.org/) Human sarcoma specimens were accessed in June 2011. CEL files from E-MEXP-1142, Paraffin blocks from 10 cases of both CDS and ES specimens E-MEXP-353, GSE21122, GSE7529, mCDS and hCDS samples were obtained from the Memorial Sloan-Kettering Cancer Center. were summarized by MAS5 algorithm with Affymetrix Power The study was approved by the Institutional Review Board at Tool software version 1.12.0 (Affymetrix). Each probe set that Memorial Sloan-Kettering Cancer Center (Protocol 02-060). had same Gene Symbols was collapsed into a single probe set Synovial sarcoma (hSS), rhabdomyosarcoma (hRS), and extra- that showed the highest median value in the raw MAS5 dataset. skeletal myxoid chondrosarcoma (hEMCS) samples were Each mouse gene of collapsed datasets was joined into the same obtained from the Japanese Foundation for Cancer Research. gene symbol of human collapsed datasets that resulted in one dataset with 10,166 probe sets. Signal intensities of combined Histopathology and IHC dataset were normalized by quantile algorithm with "prepro- For light microscopic analysis, formaldehyde-fixed, paraffin- cessCore" library package on Bioconductor software (13, 14). embedded tumor tissues were stained with hematoxylin and To obtain the genes that are highly (or lowly) expressed in eosin (H&E) using standard techniques. IHC staining was per- hCDS than other human samples, linear models for microarray formed using the Simple Stain MAX-PO Kit (Nichirei Bioscience), analysis (limma) package of Bioconductor software were the Histofine SAB-PO (R) Kit (Nichirei Bioscience) or the Bench- applied (15). Hierarchical clustering was performed using Mark GX automated slide preparation system (Ventana). The 1,450 probe sets as hCDS-specific genes by MeV software and following primary antibodies were used: anti-MUC5AC (Santa Pearson's centered measurements (16). The criteria of the Cruz Biotechnology), anti-ETV4 (Abcam), anti-CCND2 (Santa specific genes were limma P value of <0.05 and absolute log- Cruz Biotechnology). fold-change (|logFC|) > 1.

Immunoprecipitation and Western blotting RNA interference assays The CIC–DUX4 was immunoprecipitated using total siRNAs against mouse Ccnd2, mouse Ret, human CIC (Qiagen) cell lysates of mouse CDS cells using the anti-DUX4 antibody and mouse Bcl2 (Life Technologies) were transfected into mouse (Santa Cruz Biotechnology). The samples were then subjected to CDS cell lines using Lipofectamine 2000 (Invitrogen). The list of SDS-PAGE and CIC-DUX4 was detected using the polyclonal anti- siRNAs is shown in Supplementary Table S2. Knockdown effi- CIC antibody (2). ciencies of each gene were confirmed by RT-PCR.

OF2 Cancer Res; 77(11) June 1, 2017 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

CIC-DUX4 Sarcoma Mouse Model

Pharmacological experiments survival between groups was compared with the log-rank test. Mouse CDS cells were treated with palbociclib, vandetanib, Continuous distributions were compared with a two-tailed ABT199, lenalidomide, pazopanib, bortezomib (Selleckchem), Student t test. trabectedin and eribulin (Taiho Pharmaceutical) in vitro. The cell proliferation and viability were determined by measuring cell Results numbers with Trypan blue exclusion. For in vivo experiments, Generation and characterization of the CDS ex vivo model 1 106 cells were transplanted subcutaneously into Balb/c nude First, we aimed to establish an ex vivo mouse model expressing mice, and the mice were treated with palbociclib and/or trabec- CIC–DUX4 using an established ex vivo technology (7). We tedin. Palbociclib was orally administrated 150 mg/kg daily for introduced the human CIC-DUX4 cDNA into murine eMC 2 weeks, and trabectedin was intravenously 100 mg/kg/d three derived from embryo limbs on dpc 18.5. The eMC were then times for every 4 days, total 300 mg/kg in a course of experiment. transplanted subcutaneously into nude mice. All the recipient mice developed solid tumor masses at 100% penetrance with a Statistical analysis mean latency of 24 days (Fig. 1A and B). The tumor latency of All data are given as the means SEM. Survival analysis was mouse CDS (mCDS) was significantly shortened than that of performed using the Kaplan–Meier life table method and mouse ES (mES) reported previously (Fig. 1A; ref. 7). Although

Figure 1. The CDS ex vivo model. A, Cumulative incidence of CDS tumors arising in the transplanted mice induced by eMC- expressing CIC-DUX4 (blue line) or containing an empty vector (black line). The survival curve of mouse ES is shown (red line) and survival was compared by the log-rank test; , P < 0.01. B, The CDS tumors could be observed as subcutaneous masses in recipient nude mice (arrows). C–F, Histology of murine CDS. H&E staining for primary sites (C–E) and metastasis in the lung (F). Mouse CDS are composed of predominantly small round cells (C) or short spindle cells (D). Abundant ECM accumulation was indicated as dense fibrous septa (E). Scale bars, 50 mm(C and D) and 200 mm (E and F). Insets show high- magnification images. G, Systemic metastases by injecting CDS cells via tail vein. H, Expression of the CIC-DUX4 chimeric protein was detected by immunoprecipitation with an anti- DUX4 antibody followed by Western blotting by an anti-CIC antibody. Immunoprecipitated samples with rabbit IgG were loaded as negative controls.

www.aacrjournals.org Cancer Res; 77(11) June 1, 2017 OF3

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

Yoshimoto et al.

Figure 2. Gene expression profiles of eMC-expressing CIC–DUX4 and mCDS tumors. A, Venn diagram showing upregulated genes in eMC-expressing CIC–DUX4 (n ¼ 4) or mCDS tumors (n ¼ 6) versus those containing an empty vector (n ¼ 4). B, GSEA of eMC-expressing CIC-DUX4 versus those with an empty vector shows enrichment of the gene set involved in ECM organization (left), and mCDS tumors versus eMC with an empty vector shows the cyclin D1 pathway (right). C, Principal component analysis for gene expression profiles of mCDS and mES. D, Hierarchical clustering with 1,661 differentially upregulated genes in mCDS or mES. E, Real-time quantitative RT-PCR for Ccnd1, Ccnd2, Bcl2, Ret, Muc5ac, and Etv4 in six independent mCDS and mES tumors, eMC with an empty vector, and the CIC-DUX4 retrovirus. F, GSEA shows enrichment of the VEGF signaling pathway and the cyclin D1 signature in mCDS. G, Hierarchical clustering of gene expression profiles of 6 samples of mCDS, 5 cases of hCDS, 20 myxoid liposarcomas (MLS), 21 malignant fibrous histiocytomas (MFH), 16 synovial sarcomas (SS), 11 osteosarcomas (OS), 7 chondrosarcomas (CS), 32 Ewing sarcomas (ES), and 10 neuroblastomas (NB).

OF4 Cancer Res; 77(11) June 1, 2017 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

CIC-DUX4 Sarcoma Mouse Model

transduced cell types were different between CDS and ES models, expression signatures in mCDS were also compared to the mouse CIC-DUX4 expression in osteochondrogenic progenitors also ES (mES). The microarray analysis showed that 4,659 and 353 developed CDS in a similar manner with limb eMC (data not genes were upregulated in CIC–DUX4-expressing eMC vs eMC shown), suggesting that CIC–DUX4 might be acceptable by more with an empty vector; fold change > 2.0, P < 0.05, and in mCDS diverse cell lineages than EWS-FLI1. Characterization of eMCs by tumors vs eMC with an empty vector; fold change > 1.5, P < 0.05, the cell surface marker analysis suggests that they are composed of respectively. Furthermore, 69 genes were shown to be upregulated multilineage stem/progenitors (Supplementary Fig. S1). in both categories (Fig. 2A). In addition, 758 and 89 genes were Histological analysis showed that the majority of mCDS downregulated in CIC–DUX4-expressing eMC and mCDS tumors tumors consisted of small round to short spindle cells arranged in the same criteria, respectively. The lists of genes significantly up- in solid sheet pattern with rather abundant stroma (Fig. 1C and or down-regulated in each category are shown in Supplementary D). Tumors were frequently associated with a prominent extra- Tables S3 and S4. Notably a subset of PEA3 family genes of ETS cellular matrix (ECM; Fig. 1E). This histologic heterogeneity of transcription factors such as Etv1 and Etv4 were significantly CDS was also described in human tumors (1, 3, 4). The tumors upregulated in mCDS tumors and Etv1 and Etv5 in CIC–DUX4- (1 106 cells) were serially transplantable at 100% penetrance expressing eMC (Table 1 and Supplementary Table S3), consistent to both nude mice (16/16) and immune competitive Balb/c with the previous findings that PEA3 family genes are transcrip- mice (9/9), and spontaneous metastasis to the lung was tional targets of CIC-DUX4 (2). Moreover, the upregulated genes observed in 28% (total 7/25, 2/16 in nude mice and 5/9 in of CIC–DUX4-expressing eMC, mCDS tumors or both include Balb/c mice) of secondary transplanted recipients with mean Crh, Vgf, Dlk1, Zic1 and Ccnd2 (Table 1 and Supplementary Table latency of 5.9 weeks (5.8 weeks in nude mice and 6.6 weeks in S3) that were found upregulated in human CDS (hCDS; ref. 1), Balb/cmice;Fig.1F).Whentumorcellswereinjectedviatail indicating that mouse and hCDS share common gene expression vein, systemic metastases in the lung, liver, ovary and thoracic signatures. cavities were observed (Fig. 1G). The CIC–DUX4 protein was Gene set enrichment analysis (GSEA) showed strong correla- detected by immunoprecipitation followed by Western blotting tion between gene expression in CIC–DUX4-expressing eMC and in CDS cells (Fig. 1H). ECM organization, and in mCDS tumors and the cyclin D1 signature (Fig. 2B). The data suggest that induction of ECM Gene expression profiles of mouse CDS production might be initiated by CIC-DUX4 expression, and that To examine the characteristic molecular signatures in CDS, cyclin D-associated cell cycle progression might be one of the comprehensive gene expression profiling was carried out. Com- important genetic hallmarks of CDS. paring the gene expression profiles of the eMC transduced with The expression profile of mCDS was then compared with that of the CIC-DUX4 retrovirus versus the empty vector, and between mES. A principal component analysis showed distinct gene murine CDS (mCDS) and eMC with an empty vector. Gene expression signatures in two tumor types (Fig. 2C), and

Table 1. Selected upregulated genes in CDS and eMCs-expressing CIC-DUX4 Gene name FC in tumora FC in eMCa Gene product Muc5ac 4663.1 b 5AC Crh 3686.0 b Corticotropin releasing hormone Zic1 2680.2 23.5 Zinc finger protein Etv4 2239.8 b ets variant gene Etv1 2023.7 39.1 ets variant gene Col24a1 987.2 b Collagen, type XXIV, alpha 1 Fgfbp3 940.2 b Fibroblast growth factor binding protein Etv5 611.2 42.0 ets variant gene Col9a1 441.7 50.7 Collagen, type IX, alpha 1 Col13a1 230.5 b Collagen, type XIII, alpha 1 Col9a3 192.6 45.8 Collagen, type IX, alpha 3 Fli1 142.5 15.8 Friend leukemia integration 1 Col8a2 140.2 40.3 Collagen, type VIII, alpha 2 Fmod 132.2 51.0 Irf4 113.5 b Interferon regulatory factor 4 Cdkn2a 81.5 b Cyclin-dependent kinase inhibitor 2A Kit 62.4 29.7 Kit oncogene Bcl2 41.4 19.6 B-cell leukemia/lymphoma 2 Col25a1 21.4 16.2 Collagen, type XXV, alpha 1 Ccnd2 11.4 b Cyclin D2 Cdkn1c 7.3 c Cyclin-dependent kinase inhibitor 1C Ccnd1 5.1 c Cyclin D1 Gpc3 5.1 c 3 Ets1 4.0 b ets transcription factor Col8a1 3.6 b Collagen, type VIII, alpha 1 Vcan 3.1 b NOTE: Genes upregulated in human CDS are underlined. aFold change (FC) of gene expression versus eMCs with an empty vector. bFold change >1.5 with P > 0.05. cNot upregulated.

www.aacrjournals.org Cancer Res; 77(11) June 1, 2017 OF5

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

Yoshimoto et al.

Table 2. Selected upregulated genes in CDS vs ES fibromodulin, genes were also manifested in mCDS Gene name FC Gene product compared with mES (Table 2). Furthermore, expression profiles Muc5ac 7022.8 Mucin 5, subtypes A and C, tracheobronchial/gastric of mCDS were compared with eight types of human sarcomas, fi Zic1 6571.5 Zinc nger protein of the cerebellum 1 including hCDS, hES, synovial sarcoma, osteosarcoma, chondro- Crh 5368.0 Corticotropin releasing hormone fi Col9a1 4165.7 Collagen, type IX, alpha 1 sarcoma, myxoid liposarcoma, malignant brous histiocytoma Ccnd2 3799.6 Cyclin D2 and neuroblastoma. Hierarchical clustering using common gene Vgf 3596.9 VGF nerve growth factor inducible sets between mice and humans (1,450 probes selected from Col2a1 2948.0 Collagen, type II, alpha 1 10,166 probe sets) showed that the mCDS was most closely Col24a1 2010.0 Collagen, type XXIV, alpha 1 related to hCDS (Fig. 2G), indicated similar gene expression Hmga2 1941.4 High mobility group AT-hook 2 profiles between mCDS and hCDS. Dlk1 1899.6 Delta-like 1 homolog (Drosophila) Col4a5 1245.3 Collagen, type IV, alpha 5 Fgfbp3 1027.0 Fibroblast growth factor binding protein 3 Cyclin D2 and MUC5AC are novel biomarkers useful for CDS Cdkn1c 880.8 Cyclin-dependent kinase inhibitor 1C (P57) diagnosis Gpc3 857.8 The distinct gene expression profiles between mCDS and mES Vcan 771.1 Versican provide useful information to identify CDS-specific biomarkers Etv4 687.3 ets variant gene for potential use for IHC. CCND2 (cyclin D2) and MUC5AC were Prg4 529.7 Proteoglycan 4 Cldn9 338.3 Claudin 9 selected for further validation by IHC, using ETV4 as a positive Ets1 322.5 E26 avian leukemia oncogene 1, 50 domain control. Consistent nuclear overexpression for CCND2 and ETV4 Ret 321.7 Ret proto-oncogene was noted in mCDS tumor cells, but CCND2 was negative in mES Etv1 305.6 ets variant gene 1 (Fig. 3A). ETV4 was occasionally detected in mES. MUC5AC, a Col13a1 283.0 Collagen, type XIII, alpha 1 mucin component, was positive in mCDS and negative in mES Col9a3 276.9 Collagen, type IX, alpha 3 (Fig. 3A). The frequencies of positive cells for CCND2, MUC5AC, Ncam1 258.8 Neural cell adhesion molecule 1 Fmod 228.7 Fibromodulin or ETV4 were quantitated. The results for CCND2 and MUC5AC ¼ ¼ Fndc1 216.9 Fibronectin type III domain containing 1 were consistent in all mCDS (n 6) and mES tumors (n 9) Fli1 147.9 Friend leukemia integration 1 tested. Etv5 135.2 ets variant gene 5 CCND2 and MUC5AC expression was further examined in Col11a1 117.0 Collagen, type XI, alpha 1 human CIC-rearranged sarcoma (hCIC) and human ES (hES). Irf4 101.3 Interferon regulatory factor 4 Nuclear and cytoplasmic expression with variable proportion Lmna 82.7 lamin A > Gpc4 77.2 Glypican 4 of CCND2 in 70% of tumor cells was evident in all hCIC Gpc6 72.7 Glypican 6 tested (10/10 cases), whereas negative in 7 cases or only limited Bcl2 51.2 B-cell leukemia/lymphoma 2 expression (<20% tumor cells) in 2 cases of 10 ES samples (Fig. Ccnd1 9.4 Cyclin D1 3B). Eight out of 10 hCIC samples showed scattered positive Col5a2 6.7 Collagen, type V, alpha 2 staining of MUC5AC in cytoplasm and/or extracellular matrix. Col5a1 6.6 Collagen, type V, alpha 1 In contrast, no MUC5AC staining was identified in 10 hES Ccnd3 6.1 Cyclin D3 Fbln2 5.3 Fibulin 2 samples (Fig. 3B). Moreover, CCND2 and MUC5AC was rarely Col3a1 5.0 Collagen, type III, alpha 1 positive in hSS, hRS, and hEMCS, which often show small NOTE: Genes upregulated in human CDS are underlined. round cell morphologies. These results indicate that CCND2 and MUC5AC are promising biomarkers useful for diagnosis of CIC-rearranged sarcomas. hierarchical clustering showed 1,661 genes differentially expressed in mCDS or mES (Fig. 2D; Supplementary Table Inhibition of CDS tumor growth by suppression of critical S5). In these 952 and 709 genes were upregulated in mCDS or signals mES, respectively, and Bcl2,Ccnd1,Ccnd2,Crh,Etv1,Etv4,Etv5, Gene-expression data provide further support of the critical role Fli1, Muc5ac, Ret and Zic1 were found as significantly upregu- of CIC-DUX4 and its downstream signals in mCDS growth, which lated genes in mCDS (Table 2). Expression of these genes were can be used as effective targets for therapy. Ccnd2, Ret, and Bcl2 confirmed by quantitative RT-PCR (Fig. 2E; Supplementary Fig. were selected from the upregulated genes in the mCDS based on S3). In addition, expression of these genes was also upregulated expression rates and biological importance in cancer. RNA inter- when CIC-DUX4 was introduced into NIH3T3 cells (Supple- ference-mediated gene knockdown experiments showed signifi- mentary Fig. S4). Our previous study also showed upregulation cant growth inhibition by suppression of these genes as well as of ETV1 and ETV5 by CIC-DUX4 in the human osteosarcoma CIC (Fig. 4A and B). U2OS cell, and CIC-DUX4 expression induced anchorage-inde- pendent growth of NIH3T3 cells (2). In contrast, upregulation Use of the mouse model to test therapy targeted against CDS of Bcl2,Ccnd1,Ccnd2,Muc5acand Ret, and growth promotion We have previously shown that the mouse model for ES could were not observed when CIC-DUX4 was introduced into bone be used as a platform for evaluation of inhibitory drugs targeting marrow-derived MSC (BM-MSC; Supplementary Fig. S4), sarcoma (7). Therefore, the current CDS model was expected to suggesting CIC-DUX4 function is cell context–dependent at provide important information for novel therapeutics. To this end least in part. specific inhibitors for cyclin-dependent kinases 4 and 6 (palbo- GSEA showed significant enrichment of gene sets such as the ciclib), Ret (vandetanib) and Bcl2 (ABT199) were tested for their VEGF signaling pathway and cyclin D1 (Fig. 2F), the latter tumor suppressive activities against mCDS cells (17–19), as indicating that the cell-cycle progression is even more aggressive knockdown of Ccnd2, Ret and Bcl2 genes was effective for growth in mCDS than in mES. Many ECM genes such as collagen, inhibition. These inhibitors showed substantial growth inhibitory

OF6 Cancer Res; 77(11) June 1, 2017 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

CIC-DUX4 Sarcoma Mouse Model

A mCDS mES

10

8

6 mCDS mES CCND2 4 2

Number of cases 0 mES (n = 9) ≥50% ≥20% mCDS (n = 6) ≥5% <5% Positive rate (%)

6 5 Figure 3. 4 3 Immunostaining of CCND2 and MUC5AC 2 MUC5AC in mouse and human CDS and 1 Number of cases 0 mES (n = 9) ES. A, mCD shows positive staining for ≥ 20% ≥ mCDS (n = 6) 1% <1% CCND2, ETV4, and MUC5AC (left). No 0% Positive rate (%) expression for CCND2 and MUC5AC and little expression for ETV4 in mES 3 m (center). Scale bars, 40 m. The 2 frequencies of positive cells for CCND2, MUC5AC or ETV4 were quantitated ETV4 1 Number of cases (right). B, Human CIC-rearranged 0 mES (n = 9) ≥50% ≥ mCDS (n = 6) sarcoma (hCIC) shows CCND2 and 20% ≥ 5% <5% MUC5AC expression (left). No Positive rate (%) expression in hES (middle). Scale bars, 40 mm. The frequencies of positive cells for CCND2 or MUC5AC in hCIC, hES, B hCIC hES rhabdomyosarcoma (hRS), 10 hCDS hES extraskeletal myxoid chondrosarcoma 8 hRMS (hEMCS), and synovial sarcoma (hSS) 6 hESMCS were quantitated (right). The numbers hSS CCND2 4 hSS (n = 10) of cases are indicated in parentheses. 2 hESMCS (n = 8) Number of cases hRMS (n = 10) 0 hES (n = 10) ≥50% ≥20% ≥ hCDS (n = 10) 5% <5% Positive rate (%)

10

8

6

MUC5AC 4 hSS (n = 10) 2 hESMCS (n = 8) Number of cases hRMS (n = 10) 0 hES (n = 10) ≥20% ≥1% < hCDS (n = 10) 1% 0% Positive rate (%) effect with palbociclib as the highest efficacy (Fig. 5A). However, model provides an effective tool to explore and evaluate novel mCDS-specific inhibitory effect of palbociclib was not observed as therapeutic drugs both in vitro and in vivo. the effects were similar with the human clear cell sarcoma cell line, KAS, and the osteosarcoma cell line, U2OS. Anticancer drugs and inhibitors, bortezomib, eribulin, lenalidomide, pazopanib and Discussion trabectedin that are used for soft tissue neoplasms were also tested Human CDS and CIC-rearranged sarcoma is a recently estab- for mCDS cells (20–24). Among these a proteasome inhibitor lished disease entity, formerly considered as Ewing-related sar- bortezomib and an alkylating agent trabectedin showed signifi- coma. The morphologic overlap might be related in part to the cant effects both for mCDS and mES cells (Fig. 5B and data not activation of PEA3 family genes, as some of them are also involved shown). in ES gene fusions (2, 26). However, more recent clinical, genetic Antitumor effect of the trabectedin, palbociclib and bortezo- and histopathologic investigation elucidated the distinct nature of mib were then tested in vivo. Because preliminary experiments CDS from ES (1, 3, 4, 27). Here, we have established the first showed little effect of bortezomib (data not shown), trabectedin mouse model for CDS expressing the human CIC–DUX4 fusion and palbociclib were administrated alone or in combination to gene. The model faithfully recapitulates the highly aggressive the mCDS-bearing mice. Trabectedin showed marked growth nature of mCDS, the distinct morphologic and biologic charac- suppression, whereas palbociclib did not show significant effect teristics similar to hCDS. nor enhancement of trabectedin's inhibition (Fig. 5C). There was The distinct gene expression profiles of these mouse models no significant difference in the degree of necrotic areas and suggest a molecular basis for the phenotypic differences between BrdUrd-labeling index (data not shown). Trabectedin efficacy for CDS and ES. In CDS, the abundant ECM noted correlated with an ES was reported (25). Our results suggest that similar treatment upregulation of various collagen genes, their modifying enzymes might be selected for CDS; however, it is ideal to develop treat- and matrix proteases, consistent with enhanced ECM induction ment targeting CIC–DUX4 and related signaling pathways for (28). Similarly, abundant ECM in CDS was also reported in precision medicine-based treatment. Nonetheless, the current human CDS either as dense fibrous septa or myxoid stroma

www.aacrjournals.org Cancer Res; 77(11) June 1, 2017 OF7

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

Yoshimoto et al.

Figure 4. Growth suppression of tumor cells by gene silencing. A, Inhibition of cell proliferation by knockdown of CIC- DUX4 and upregulated genes in mCDS. Relative growth of tumor cells 48 hours after siRNA treatment was calculated by comparing each cell number with cells treated with control siRNA (NC). Two oligonucleotide sequences were used for each gene and the symbols of siRNA used are indicated. B, Gene knockdown was confirmed by RT-PCR. Knockdown of CIC-DUX4 was achieved using human CIC-specific siRNAs. CIC-DUX4 of human origin was extensively downregulated (bottom left), whereas endogenous mouse Cic expression was not reduced significantly (bottom right).

(29). Interestingly, spontaneous hematogenous metastases, use of novel CDK4/6 inhibitors may improve the effect of which were never found in the ES mouse model, were observed CCND2 targeting. In contrast, trabectedin was effective for in the CDS model, which might be related to enhanced vasculo- mCDS tumor growth both in vitro and in vivo. Trabectedin is genesis in CDS. The ECM in CDS contains mucin components, a Caribbean tunicate-extracted alkylating agent and is used for and MUC5AC is one such component secreted by tracheobron- ovarian cancer and soft part sarcoma including leiomyosar- chial and gastrointestinal epithelium (30), which was expressed coma and liposarcoma (24, 25, 36, 37). Previous study in a small subset of the tumor cells and ECM of CDS, though its reported that trabectedin exhibited growth inhibition of ES as functional significance in sarcoma cells remain to be clarified. a single agent, and improved effect in combination with an The upregulation of cell-cycle progression signature was IGFR inhibitor (25). Therefore, trabectedin is a promising prominent in mCDS, even more than in mES, in which acti- therapeutic agent for CDS despite lack of the synergistic effect vation of CCND1 was reported (31). In mCDS, the top upre- with palbociclib. gulated cyclin D gene was Ccnd2. A previous report showed that CIC is an HMG-box transcription factor homologous to transcription of Ccnd2 is directly regulated by ETV4 (32), drosophila capicua (38). As a downstream of receptor tyrosine suggesting that Ccnd2 upregulation is achieved by multiple kinase signaling, it acts as a transcriptional repressor associating PEA3 family induced by CIC-DUX4 in CDS. Therefore, with TLE corepressors (39). By fusion to DUX4 the transrepres- the CCND2 can be used as an informative diagnostic tool, sional activity of CIC is converted to strong transactivation by though CCND1 expression is also enhanced. Like cyclin D1, replacement with the DUX4 C-terminal transactivation domain cyclin D2 is a regulatory subunit of the holoenzymes function- whose activity was indicated in the wild-type DUX4 function (2, ing with catalytic partners CDK4/6 that phosphorylates and 40). In our model, upregulation of the CIC-DUX4 targets such inactivates RB. Although cyclin D2 is less frequently involved in as Etv4, Crh and Zic1 is consistently manifested, indicating the human cancer, it has an ability to replace cyclin D1 function important role of CIC-DUX4 transactivation in tumorigenesis. (33–35). Inhibition of CDK4/6 is therefore expected to sup- In this context, juxtaposition of DUX4 to the IgG enhancer in press cell proliferation of mCDS. The CDK inhibitor palbociclib human B-cell lymphoblastic leukemia where the transactiva- indeed inhibited mCDS growth in vitro, however, it showed tion domain of DUX4 is deleted is intriguing (41). CIC is also only limited effect for tumor growth in vivo. The cause of weak involved in the majority (70%–80%) of human oligodendro- response of mCDS to palbociclib in vivo remains unclear. The glioma and infantile primitive neuroectodermal neoplasms of

OF8 Cancer Res; 77(11) June 1, 2017 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

CIC-DUX4 Sarcoma Mouse Model

A Palbociclib Vandetanib ABT199 1.0 1.0 1.0 CDS 4 0.8 0.8 0.8 CDS 15 0.6 0.6 0.6 CDS 16 0.4 0.4 0.4 U2OS Viability rate Viability Viability rate Viability 0.2 0.2 rate Viability 0.2 KAS 0 0 0 0 1.25 2.5 5 10 20 0 1.25 2.5 5 10 20 0 1.25 2.5 5 10 20 Concentration (mmol/L) Concentration (mmol/L) Concentration (mmol/L)

CDS 4 IC = 4.08 mmol/L CDS 4 IC = 6.01 mmol/L CDS 4 IC50 = 6.52 mmol/L 50 50 CDS 15 IC CDS 15 IC50 = 5.65 mmol/L CDS 15 IC50 = 2.03 mmol/L 50 = 4.25 mmol/L CDS 16 IC50 = 1.89 mmol/L CDS 16 IC50 = 2.20 mmol/L CDS 16 IC50 = 3.19 mmol/L U2OS IC50 = 9.10 mmol/L U2OS IC50 = 6.69 mmol/L U2OS IC50 = 11.41 mmol/L KAS IC50 = 3.40 mmol/L KAS IC50 = 3.83 mmol/L KAS IC50 = 1.27 mmol/L

B Bortezomib Trabectedin 1.0 1.0

0.8 0.8 CDS 16

0.6 0.6 ES 49 U2OS 0.4 0.4 Viability rate Viability Viability rate Viability KAS 0.2 0.2

0 0 00.625 1.25 2.5 5 10 20 40 00.125 0.25 0.5 1 2 4 8 Concentration (nmol/L) Concentration (nmol/L)

CDS 16 IC = 1.95 nmol/L CDS 16 IC50 = 0.84 nmol/L 50 ES 49 IC ES 49 IC50 = 2.20 nmol/L 50 = 0.87 nmol/L U2OS IC U2OS IC50 = 5.84 nmol/L 50 = 1.88 nmol/L KAS IC KAS IC50 = 6.88 nmol/L 50 = 2.46 nmol/L C 3,000

2,500 ) 3 2,000 Vehicle Trabectedin 1,500 Palbociclib

1,000 Trab.+ palbo. Tumor volume (mm Tumor 500

0 0 4 8 12 Trabectedin Palbociclib Days

Figure 5. Effects of small-molecule inhibitors and anticancer agents on mCDS. A, In vitro growth inhibition of mCDS (mCDS 4, 15, and 16) cell lines by palbociclib, vandetanib, and ABT199. Human clear cell sarcoma (KAS) and human osteosarcoma (U2OS) cell lines were also tested. Cells were treated with each reagentatthe indicated concentration for 48 hours. The experiment was performed in triplicate, and average suppression rates with SEM and IC50 are indicated. B, In vitro growth inhibition of mCDS and ES cell lines by bortezomib and trabectedin as in A. C, Growth inhibitory effects of bortezomib and palbociclib for mCDS in vivo. mCDS 16 cells were transplanted subcutaneously into nude mice, and tumor volume was measured every other day. Mean tumor volumes SEM for 5 mice of each group are plotted. The treatment times of trabectedin and palbociclib are indicated by arrows. , P < 0.01 and , P < 0.001. brain (42–44). In oligodendroglioma, mutations result in loss- tolerate cancer-associated fusion genes, possibly by bypassing of-function of CIC. Collectively these findings indicate aberra- oncogene-induced senescence (9). mCDS of the current model tions of CIC function is one of key molecular events in multiple can be generated rapidly and be maintained in the immune human cancer. competent mouse, which enable to evaluate host responses In summary, we generate a CDS mouse model by CIC-DUX4 against the tumor more precisely than in the xenograft model. expression in murine eMC. The eMC exhibit sufficient plasticity to Introduction of fusion genes into eMC will, therefore, provide

www.aacrjournals.org Cancer Res; 77(11) June 1, 2017 OF9

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

Yoshimoto et al.

accurate and reliable modeling of human sarcoma, useful for Administrative, technical, or material support (i.e., reporting or organizing biomarker and novel therapy discovery. data, constructing databases): Y. Yamazaki, Y. Takazawa, T. Nakamura Study supervision: T. Yoshimoto, M. Tanaka, T. Nakamura

Disclosure of Potential Conflicts of Interest Acknowledgments T. Nakamura reports receiving a commercial research grant from Otsuka The authors would like to thank Dr. Masahiro Yanagisawa for gene silencing Pharmaceutical Co. Ltd. No potential conflicts of interest were disclosed by the experiments and Mr. Nobuyuki Nakamura, Ms. Takako Fukawa, and Mr. Yuki other authors. Fujita for excellent technical assistance of immunostaining.

Authors' Contributions Grant Support Conception and design: T. Yoshimoto, M. Tanaka, T. Nakamura The study was supported by Grants-in-Aid for Scientific Research from Japan Development of methodology: T. Yoshimoto, M. Tanaka, T. Nakamura Society for the Promotion of Science (no. 26250029 to T. Nakamura) and Acquisition of data (provided animals, acquired and managed patients, Cancer Center Support grants (P50 CA140146 and P30-CA008747 to C.R. provided facilities, etc.): T. Yoshimoto, M. Tanaka, M. Homme, Y. Takazawa, Antonescu) from the NIH (Bethesda, MD). T. Nakamura The costs of publication of this article were defrayed in part by the payment of Analysis and interpretation of data (e.g., statistical analysis, biostatistics, page charges. This article must therefore be hereby marked advertisement in computational analysis): T. Yoshimoto, M. Tanaka, Y. Yamazaki, C.R. Anto- accordance with 18 U.S.C. Section 1734 solely to indicate this fact. nescu, T. Nakamura Writing, review, and/or revision of the manuscript: T. Yoshimoto, M. Tanaka, Received December 9, 2016; revised January 4, 2017; accepted March 30, M. Homme, Y. Takazawa, C.R. Antonescu, T. Nakamura 2017; published OnlineFirst April 12, 2017.

References 1. Specht K, Sung YS, Zhang L, Richter GH, Fletcher CD, Antonescu CR. 14. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, et al. Distinct transcriptional signature and immunoprofile of CIC-DUX4 Bioconductor: open software development for computational biology and fusion-positive round cell tumors compared to EWSR1-rearranged Ewing bioinformatics. Genome Biol 2004;5:R80. sarcomas: further evidence toward distinct pathologic entities. Genes 15. Smyth GK. limma: linear models for microarray data. In: Gentleman R, Cancer 2014;53:622–33. Carey V, Huber W, Irizarry R, Dudoit S, editors. Bioinformatics and 2. Kawamura-Saito M, Yamazaki Y, Kaneko K, Kawaguchi N, Kanda H, Mukai Computational Biology Solutions using R and Bioconductor. New York, H, et al. Fusion between CIC and DUX4 up-regulates PEA3 family genes in NY: Springer; 2005. p. 397–420. Ewing-like sarcomas with t(4;19)(q35;q13) translocation. Hum Mol Genet 16. Saeed AI, Sharov V, White J, Li J, Liang W, Bhagabati N, et al. TM4: a free, 2006;15:2125–37. open-source system for microarray data management and analysis. Bio- 3. Yoshida A, Goto K, Kodaira M, Kobayashi E, Kawamoto H, Mori T, et al. Cic- techniques 2003;34:374–8. rearranged sarcomas: a study of 20 cases and comparisons with Ewing 17. Fry DW, Harvey PJ, Keller PR, Elliott WL, Meade M, Tracher E, et al. Specific sarcomas. Am J Surg Pathol 2016;51:313–23. inhibition of cyclin-dependent kinase 4/6 by PD 0332991 and associated 4. Italiano A, Sung YS, Zhang L, Singer S, Maki RG, Coindre JM, et al. High antitumor activity in human tumor xenografts. Mol Cancer Ther prevalence of CIC fusion with double homeobox (DUX4) transcription 2004;3:1427–38. factors in EWSR1-negative undifferentiated small blue round cell sarcomas. 18. Carlomagno F, Vitagliano D, Gulda T, Ciardiello F, Tortora G, Vecchio G, Genes Chromosomes Cancer 2012;51:207–18. et al. ZD6474, an orally available inhibitor of KDR tyrosine kinase activity, 5. Le Guellec S, Velasco V, Perot G, Watson S, Tirode F, Coindre JM. ETV4 is a efficiently blocks oncogenic RET kinases. Cancer Res 2002;62:7284–90. useful marker for the diagnosis of CIC-rearranged undifferentiated round- 19. Souers AJ, Leverson JD, Boghaert ER, Ackler SL, Catron ND, Chen J, et al. cell sarcomas: a study of 127 cases including mimicking lesions. Mod ABT-199, a potent and selective BCL-2 inhibitor, achieves antitumor Pathol 2016;29:1523–1531. activity while sparing platelets. Nat Med 2013;19:202–8. 6. Smith SC, Palanisamy N, Martin E, Almenara J, McHugh JB, Choi EK, et al. 20. Maki RG, Kraft AS, Scheu Y, Yamada J, Wadler S, Antonescu CR, et al. A The utility of ETV1, ETV4 and ETV5 RNA in-situ hybridization in the multicenter Phase II study of bortezomib in recurrent or metastatic sar- diagnosis of CIC-DUX sarcomas. Histopathology 2017;70:657–63. comas. Cancer 2005;103:1431–8. 7. Tanaka M, Yamazaki Y, Kanno Y, Igarashi K, Aisaki K, Kanno J, et al. Ewing's 21. Schoffski P, Ray-Coquard IL, Cioffi A, Bui NB, Bauer S, Hartmann JT, et al. sarcoma precursors are highly enriched in embryonic osteochondrogenic Activity of eribulin mesylate in patients with soft-tissue sarcoma: a phase 2 progenitors. J Clin Invest 2014;124:3061–74. study in four independent histological subtypes. Lancet Oncol 8. Tanaka M, Yamaguchi Y, Yamazaki Y, Kinoshita H, Kuwahara K, Nakao K, 2011;12:1045–52. et al. Somatic chromosomal translocation between Ewsr1 and Fli1 loci 22. Said R, Kakadiaris E, Piha-Paul S, Fu S, Falchook G, Janku F, et al. Phase I leads to dilated cardiomyopathy in a mouse model. Sci Rep 2015;5:7826. clinical trial of lenalidomide in combination with bevacizumab in patients 9. Minas TZ, Surdez D, Javaheri T, Tanaka M, Howarth M, Kang HJ, et al. with advanced cancer. Cancer Chemother Pharmacol 2016;77:1097–102. Combined experience of six independent laboratories attempting to create 23. Sleijfer S, Ray-Coquard I, Papai Z, Le Cesne A, Scurr M, Schoffski P, et al. an Ewing sarcoma mouse model. Oncotarget 2016 May 15. [Epub ahead Pazopanib, a multikinase angiogenesis inhibitor, in patients with relapsed of print]. or refractory advanced soft tissue sarcoma: a phase II study from the 10. Tanaka M, Homme M, Yamazaki Y, Shimizu R, Takazawa Y, Nakamura T. European organization for research and treatment of cancer-soft tissue Modeling alveolar soft part sarcoma unveils novel mechanisms of metas- and bone sarcoma group (EORTC study 62043). J Clin Oncol 2009; tasis. Cancer Res, 2017;77:897–904. 27:3126–32. 11. Yokoyama T, Nakatake M, Kuwata T, Couzinet A, Goitsuka R, Tsutsumi S, 24. Yovine A, Riofrio M, Blay JY, Brain E, Alexandre J, Kahatt C, et al. Phase II et al. MEIS1-mediated transactivation of synaptotagmin-like 1 promotes study of exteinascidin-743 in advanced pretreated soft tissue sarcoma CXCL12/CXCR4 signaling and leukemogenesis. J Clin Invest 2016; patients. J Clin Oncol 2004;22:890–9. 126:1664–78. 25. Amaral AT, Garofalo C, Frapolli R, Manara MC, Mancarella C, Uboldi S, 12. Subramanian A, Kuehn H, Gould J, Tamayo P, Mesirov JP. GSEA-P: a et al. Trabectedin efficacy in Ewing sarcoma is greatly increased by com- desktop application for gene set enrichment analysis. Bioinformatics bination with anti-IGF signaling agents. Clin Cancer Res 2015;21: 2007;23:3251–3. 1373–82. 13. Bolstad BM, Irizarry RA, Astrand M, Speed TP. A comparison of normal- 26. Lawlor ER, Sorensen PH. Twenty years on – What do we really know about ization methods for high density oligonucleotide array data based on Ewing sarcoma and what is the path forward? Crit Rev Oncol 2015;20: variance and bias. Bioinformatics 2003;19:185–93. 155–71.

OF10 Cancer Res; 77(11) June 1, 2017 Cancer Research

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

CIC-DUX4 Sarcoma Mouse Model

27. Antonescu C. Round cell sarcomas beyond Ewing: Emerging entities. ovarian cancer. Focus on trabectedin. Cancer Treatment Rev 2014;40: Histopathology 2014;64:26–37. 366–75. 28. Frantz C, Stewart KM, Weaver VM. The extracellular matrix at a glance. J Cell 37. Demetri GD, von Mehren M, Jones RL, Hensley ML, Schuetze SM, Staddon Sci 2010;123:4195–200. A, et al. Efficacy and safety of trabectedin or dacarbazine for metastatic 29. Gambarotti M, Benini S, Gamberi G, Cocchi S, Palmerini E, Sbaraglia M, liposarcoma and leiomyosarcoma after failure of conventional chemo- et al. CIC-DUX4 fusion-positive round-cell sarcomas of soft tissue and therapy: results of a phase III randomized multicenter clinical trial. J Clin bone: a single-institution morphological and molecular analysis of seven Oncol 2016;34:786–93. cases. Histopathology 2016;69:624–34. 38. Jimenez G, Guichet A, Ephrussi A, Casanova J. Relief of gene repression by 30. Moehle C, Ackermann N, Langmann T, Aslanidis C, Kel A, Kel-Margoulis O, torso RTK signaling: role of capicua in Drosophila terminal and dorso- et al. Aberrant intestinal expression and allelic variants of mucin genes ventral patterning. Genes Dev 2000;14:224–31. associated with inflammatory bowel disease. J Mol Med 2006;84:1055–66. 39. Cinnamon E, Gur-Wahnon D, Helman A, St Johnston D, Jimenez G, 31. Kennedy AL, Vallurupalli M, Chen L, Crompton B, Cowley G, Vazquez F, Paroush Z. Capicua integrates input from two maternal systems in Dro- et al. Functional, chemical genomic, and super-enhancer screening identify sophila terminal patterning. EMBO J 2004;23:4571–82. sensitivity to cyclin D1/CDK4 pathway inhibition in Ewing sarcoma. 40.ChoiSH,GearhartMD,CuiZ,BosnakovskiD,KimM,SchennumN, Oncotarget 2015;6:30178–93. et al. DUX4 recruits p300/CBP through its C-terminus and induces 32. Ladam F, Damour I, Dumont P, Kherouche Z, de Launoit Y, Tulasne D, et al. global H3K27 acetylation changes. Nucl Acids Res 2016;44:5161– Loss of a negative feedback loop involving Pea3 and Cyclin D2 is required 73. for Pea3-induced migration in transformed mammary epithelial cells. Mol 41. Yasuda T, Tsuzuki S, Kawazu M, Hayakawa F, Kojima S, Ueno T, et al. Cancer Res 2013;11:1412–24. Recurrent DUX4 fusions in B-cell lymphoblastic leukemia of adolescents 33. Fu M, Wang C, Li Z, Sakamaki T, Pestell RG. Minireview: Cyclin D1: normal and young adults. Nat Genet 2016;48:569–74. and abnormal functions. Endocrinology 2004;145:5439–47. 42. Bettegowda C, Agrawal N, Jiao Y, Sausen M, Wood LD, Hruban RH, et al. 34. Mullany LK, White P, Hanse EA, Nelsen CJ, Goggin MM, Mullany JE, et al. Mutations in CIC and FUBP1 contribute to human oligodendroglioma. Distinct proliferative and transcriptional effects of the D-type cyclins in vivo. Science 2011;333:1453–5. Cell Cycle 2008;7:2215–24. 43. Yip S, Butterfield YS, Morozova O, Chittaranjan S, Blough MD, An J, et al. 35. Salaverria I, Royo C, Carvajal-Cuenca A, Clot G, Navarro A, Valera A, et al. Concurrent CIC mutations, IDH mutations, and 1p/19q loss distinguish CCND2 rearrangements are the most frequent genetic events in cyclin D1- oligodendrogliomas from other cancers. J Pathol 2012;226:7–16. mantle cell lymphoma. Blood 2013;121:1394–402. 44. Sturm D, Orr BA, Toprak UH, Hovestadt V, Jones DTW, Capper D, et al. 36. Poveda A, Ray-Coquard I, Romero I, Lopez-Guerrero JA, Colombo N. New brain tumor entities emerge from molecular classification of CNS- Emerging treatment strategies in recurrent platinum-sensitive PNETs. Cell 2016;164:1060–72.

www.aacrjournals.org Cancer Res; 77(11) June 1, 2017 OF11

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research. Published OnlineFirst April 12, 2017; DOI: 10.1158/0008-5472.CAN-16-3351

CIC-DUX4 Induces Small Round Cell Sarcomas Distinct from Ewing Sarcoma

Toyoki Yoshimoto, Miwa Tanaka, Mizuki Homme, et al.

Cancer Res Published OnlineFirst April 12, 2017.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-16-3351

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2017/04/12/0008-5472.CAN-16-3351.DC1

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2017/05/22/0008-5472.CAN-16-3351. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 2017 American Association for Cancer Research.