Frequent Overexpression of HMGA2 in Human Atypical Teratoid/Rhabdoid Tumor and Its Correlation with Let-7A3/Let-7B Mirna

Published OnlineFirst January 14, 2014; DOI: 10.1158/1078-0432.CCR-13-1452

Clinical Cancer Human Cancer Biology Research

Frequent Overexpression of HMGA2 in Human Atypical Teratoid/Rhabdoid Tumor and Its Correlation with let-7a3/let-7b miRNA

Keqiang Zhang1, Hanlin Gao2, Xiwei Wu3, Jinhui Wang2, Wendi Zhou1, Guihua Sun1, Jinghan Wang1, Yafan Wang5, Bing Mu3, Charles Kim1, Peiguo Chu4, Donald M. Ho6, David K. Ann1, Tai-Tong Wong7, and Yun Yen1,8

Abstract Purpose: Atypical teratoid/rhabdoid tumors (AT/RT) are highly aggressive pediatric malignancies characterized by biallelic inactivation of the SMARCB1 tumor suppressor gene. We searched for novel genomic aberrations by investigating the copy number and expression alterations of let-7a3/let-7b micro- RNA (miRNA) and correlated these with expression of high-mobility group AT-hook 2 (HMGA2) oncoprotein, a target of let-7 miRNA family, in 18 AT/RT samples to elucidate potential roles of HMGA2 in the pathogenesis of AT/RT. Experimental Design: Genomic aberrations, let-7a3/let-7b miRNA and HMGA2 expression in AT/RT tissues were identified using quantitative PCR, reverse transcription PCR (RT-PCR), and immunohistochemistry. The impact of let-7b miRNA on HMGA2 expression and the malignant potential of human rhabdoid tumor cell G401 (SMARCB1 / ) were investigated by antisense inhibition and ectopic overexpression studies. Results: The copy number of let-7a3/let-7b miRNA was substantially decreased in 4 of 11 AT/RT samples. A significantly inverse correlation between let-7a3/let-7b miRNA expression and HMGA2 mRNA expression was observed in AT/RT tissues (R ¼0.34; P < 0.05). Immunohistochemistry analysis demonstrated that HMGA2 was highly overexpressed in 83.3% (15 of 18) of AT/RT tissues. Restoration of let-7 miRNA or knockdown of HMGA2 expression significantly suppressed proliferation and colony formation, and almost abolished the invasive potential of G401 cells. Conclusion: Reduction of let-7a3/let-7b miRNA may be one of mechanisms leading to overexpression of HMGA2 in AT/RT tissues. HMGA2 oncoprotein plays critical roles in the pathogenesis of AT/RT development; and reconstitution of let-7 miRNA or knockdown of HMGA2 oncoprotein may provide a novel therapeutic strategy for the treatment of patients with AT/RT. Clin Cancer Res; 20(5); 1179–89. 2014 AACR.

Introduction (CNS) that preferentially manifests in children less than 3 Atypical teratoid/rhabdoid tumor (AT/RT) is a highly years of age (1–3). Genetic alterations of the SMARCB1 malignant neoplasm (grade 4) of the central nervous system (hSNF5/INI1) gene locus on chromosome 22q11.2 resulting in the loss of tumor suppressor expression can be demonstrated in more than 90% of AT/RT cases (2–4). Authors' Afﬁliations: 1Department of Molecular Pharmacology; 2Solexa Loss of SMARCB1 function is conventionally considered Core Lab; 3Division of Information Sciences, Department of Molecular Medicine; 4Department of Pathology; 5Translational Research Laboratory, as the primary cause responsible for the development of AT/ Beckman Research Institute, City of Hope National Medical Center, Duarte, RT (2–4). These genetic alterations include homozygous 6 California; Department of Pathology and Laboratory Medicine, Taipei deletions, heterozygous deletions, copy number neutral Veterans General Hospital; 7Pediatric Neurosurgery, Department of Sur- gery, Cheng Hsin General Hospital; and 8Taipei Medical University, Taipei, LOH, as well as mutations affecting all nine exons of Taiwan SMARCB1 (2, 3, 5). Germline inactivation of SMARCB1 is Note: Supplementary data for this article are available at Clinical Cancer viewed as the "ﬁrst hit," which predisposes children to the Research Online (http://clincancerres.aacrjournals.org/). malignancies (6). The SMARCB1 protein is a component of Corresponding Authors: Yun Yen, City of Hope, 1500 E Duarte Road, the switch/sucrose non-fermentable (SWI/SNF) chromatin Duarte, CA 91010. Phone: 626-256-4673, ext. 65707; Fax: 626-471-7204; remodeling complex that functions as a tumor suppressor. E-mail: [email protected]; and Tai-Tong Wong, Pediatric Neurosurgery, Department of Surgery, Cheng Hsin General Hospital, Taipei 11220, This complex positively regulates transcription of a partic- Taiwan. Phone: 886-2-28757587; Fax: 886-2-28757587; E-mail: ular set of eukaryotic genes, including c-Myc, involved in [email protected] differentiation and apoptosis (7). doi: 10.1158/1078-0432.CCR-13-1452 AT/RT has a high mortality rate. A previous survey 2014 American Association for Cancer Research. showed that the 2-year disease-free survival for children

www.aacrjournals.org 1179

Zhang et al.

transformation, HMGA2 interacts with many different tran- Translational Relevance scription factors such as E2F1 to influence the expression of Atypical teratoid/rhabdoid tumors (AT/RT) are highly numerous genes (20). Recently, we also found that HMGA2 aggressive pediatric malignancies characterized by bial- may promote tumorigenesis through modulation of lelic inactivation of the SMARCB1 tumor suppressor. human telomerase reverse transcriptase (hTERT) transcrip- Additional pathogenic mechanisms underlying AT/RT tional activity (21). Multiple studies have identified a neg- development for targeted therapeutics are poorly under- ative correlation between reduced expression of the let-7 stood. Here, we report that besides inactivation of miRNA family and overexpression of Ras/HMGA2 in var- SMARCB1 tumor suppressor the frequent overexpres- ious human cancer tissues (22–24). Because HMGA2 plays sion of high-mobility group AT-hook 2 (HMGA2) onco- an oncogenic role in neoplastic transformation, it has been protein in AT/RT tissues, which may be partially caused generally considered as a potential target for anticancer by loss of let-7a3 and let-7b tumor suppressor miRNA, therapeutics (19, 25). also plays a critical role in the pathogenesis of AT/RT In the present study, we initially investigated the whole development. HMGA2 oncoprotein plays essential roles exome of an AT/RT tumor sample from an adult patient. in tumorigenesis and is generally regarded as a potential Using whole-exome sequencing by coupling Agilent whole therapeutic target for various human cancers. Therefore, exome to the Illumina DNA sequencing platform, we identification of the dysregulation of let-7 miRNA and identified a copy number decrease in the genomic locus overexpression of HMGA2 oncoprotein in AT/RT tissues harboring let-7a3/let-7b miRNA. We further identified over- may be beneficial in leading to novel targeted therapeu- expression of HMGA2 oncoprotein in the AT/RT tissue. tics against AT/RT. Deletion of the genomic locus harboring the let-7 miRNA family has been described as a frequent genetic alteration in various human cancers (26). For example, deletion of let- 7a3/let-7b miRNA has been described in 44.1% of ovarian younger than 3 years was 11%; the overall survival rate was cancer specimens (27). Considering the critical roles of let-7 17% (1, 8). Recently, several clinical studies demonstrated and HMGA2 oncoprotein in human cancers, we further that an intensive multimodality regimen usually used in characterized the status of let-7a3/let-7b miRNA and corre- other CNS neoplasms significantly improved the prognosis lated this to HMGA2 oncoprotein expression in 18 AT/RT of patients with AT/RT (9–12). For example, Chi and tissues. We have for the first time identified reduction of let- colleagues reported that the 2-year progression-free and 7a3/let-7b miRNA and overexpression of HMGA2 oncopro- overall survival rates for newly diagnosed patients with tein in AT/RT tissues, and our results suggest that reconsti- CNS AT/RT treated with the aggressive multimodality tution of let-7 expression may provide a novel therapeutic approach were about 53% and 70%, respectively (11). strategy for the treatment of patients with AT/RT. Despite favorable outcomes of aggressive treatment, most patients with AT/RT still suffered rapid disease recurrence Materials and Methods and death owing to progression (11). Identification of additional genetic alterations will not only shed light on Tumor specimens and DNA and RNA preparation the pathogenetic mechanisms underlying AT/RT develop- Four CNS AT/RT samples were from patients at the City of ment, but also potentially lead to novel therapeutic Hope National Medical Center (Duarte, CA), and 14 CNS approaches. Unfortunately, genome-wide analysis thus far AT/RT samples were from Taipei Veterans General Hospital, has not uncovered consistent genetic alterations, and loss of Taipei, Taiwan. Sixteen of the samples were from children SMARCB1 remains essentially the sole recurrent event younger than 6 years of age, only two were from adults. All (3, 13, 14). Therefore, efforts are still needed to develop clinically diagnosed AT/RT were further confirmed by neg- novel therapeutic approaches for AT/RT treatment. ative SMARCB1 immunostaining. All tumor specimens The role of let-7 tumor suppressor microRNA (miRNA) in were formalin-fixed, paraffin-embedded (FFPE) tissues. cancer was first demonstrated by Johnson and colleagues, DNA was isolated from FFPE samples using the QIAamp who found that the let-7 miRNA family negatively regulated FFPE DNA Mini Kit (Qiagen) according to the user manual. let-60/RAS in C. elegans by binding to multiple let-7 com- Saliva of 1 adult patient was collected into an Oragene DNA plementary sites in the 30 untranslated region (30-UTR; tube (DNA Genotek) for genomic DNA extraction to pro- ref. 15). It has been revealed that let-7 functions as a tumor vide a normal control for exome sequencing. suppressor through negatively regulating various proto- oncogenes, such as Ras and high-mobility mobility group Illumina Genome Analyzer (Solexa) library AT-hook 2 (HMGA2) oncoprotein in a variety of human preparation, exome capture sequencing, alignment, tumors (16, 17). Consistent with this, let-7 miRNA is rarely various calling, and annotation detectable in embryonic stages but increases in differenti- Three micrograms of genomic DNA was sheared by son- ated and mature tissues (18). In human cancers, signifi- ication using a Bioruptor (Diagenode). The resultant 150- to cantly reduced expression of the let-7 miRNA family is 250-bp fragmented DNA was end-repaired and ligated to associated with poor survival of patients (19). As an impor- Illumina adaptor oligonucleotides. Ligation products were tant regulator of cell growth, differentiation, apoptosis, and purified and successfully ligated fragments were amplified

1180 Clin Cancer Res; 20(5) March 1, 2014 Clinical Cancer Research

Let-7 miRNA and HMGA2 in Atypical Teratoid/Rhabdoid Tumor

with a 12-cycle PCR. The enriched PCR products were subject Mutation analysis to the exome capture procedure using the SureSelect Human Oligonucleotide primers for exons 1–9 were designed All Exon v4 Target Enrichment Kit (solution magnetic bead from the intron/exon boundary sequence of the SMARCB1 capture) according to manufacturer’s protocols (Agilent gene (GenBank accession Number. NG_009303). Primer Technologies, Inc.). The captured products were then used sequences are available upon request. The DNA samples for cluster generation and sequencing by synthesis using the extracted from tumors and adjacent normal sections pro- Illumina Genome Analyzer IIx (Illumina). Image analysis ducing a PCR product for individual exons were analyzed by and base calling were performed using Illumina’s default direct sequencing. The sequences of individual exons were pipeline. The sequences were aligned to the human genome compared with the reference sequences in Hg18. reference sequence (NCBI36) using Bowtie 0.12.7 (28). All subsequent analyses were done using customized R scripts. FISH assay The average coverage of each coding exon defined in the Dual-color FISH was performed using the FISH-mapped University of California Santa Cruz (UCSC) consensus cod- confirmed the bacterial artificial chromosome (BAC) probes; ing DNA sequence (CCDS) database was calculated based RP11-124F9 (chromosome 22q11.23) for SMARCB1 and on the alignment result. The log2 coverage ratio of each exon RP11-462A13 (chromosome 12q14.3) for HMGA2 labeled was calculated for tumor versus normal tissue and median in spectrum orange (Abbott Laboratories); and CEP12, centered. Log ratios were smoothed by DNAcopy 25 using CEP22 for chromosome 12, 22 centromere labeled in spec- default values, and copy number variation (CNV) was trum green (Abbott Laboratories) were selected for FISH detected by DNAcopy’s circular binary segmentation algo- assays. FISH assays were done on unstained FFPE slides from rithm (29). Coverage difference between tumor and saliva AT/RT tissues according to the method we previously for each exon was represented as the log2 ratio between their reported (31). Two red and two green signals were consid- coverage with an offset of 1 (30). Only regions containing ered normal in a diploid genome. Due to truncation and more than two consecutive exons and with log2 ratio of overlapping cells, at least 20% of 200 interphase cells must more than or less than 0.7 were considered significant. The show an abnormal pattern to be considered abnormal. individual single-nucleotide polymorphisms (SNP) were obtained through subtraction of the single nucleotide poly- Immunohistochemistry and Western blot morphism database (dbSNP) and 1000 genomes first. SNPs The antibodies against GAPDH and SMARCB1 were present in saliva were further filtered out from SNPs in the obtained from Sigma-Aldrich and Cell Signaling Technology, tumor genome, and the remaining SNPs were assessed as Inc., respectively. The anti-HMGA2 antibody was kindly mutations of AT/RT tissue. provided by Dr. H. Helen Lin at City of Hope Medical Center. Immunohistochemical studies of SMARCB1 were done on Quantitative real-time PCR and quantitative reverse FFPE tissues using the avidin–biotin–peroxidase method. transcription PCR After deparaffinization and blocking, the antigen–antibody Total RNAs from cells were extracted using TRizol total reaction was carried out overnight at 4C. The LSAB2 Kit RNA isolation reagent (Invitrogen) according to the man- (Dako) was applied to detect the signal of the SMARCB1 and ufacturer’s protocol. To extract total RNA/miRNA from HMGA2 antigen–antibody reaction (31). Western blot was FFPE tissues, the QIAamp FFPE miRNA Mini Kit (Qiagen) performed routinely using ECL enhancer reagents. was used, following the manufacturer’s instructions. cDNA was synthesized from 1.0 mg of total RNA using the Super- Cell culture, proliferation assay, and soft agar assay script III first-strand cDNA synthesis Kit (Invitrogen). cDNA The human rhabdoid tumor of the kidney G401 cell line for small nucleolar and miRNA let-7a3 (Tm-377) and let-7b (ATCC CRL-1441) purchased from American Type Culture (Tm-378) was synthesized from 100 ng of total RNA using Collection (ATCC) was cultured in McCoy’s 5a Medium gene-specific primers according to the TaqMan miRNA (ATCC) supplemented with 10% fetal bovine serum (FBS), Reverse Transcription Kit protocol (Applied Biosystems). 100 U/mL penicillin, and 100 mg/mL streptomycin. The cell Quantitative real-time PCR (qPCR) was carried out in the line was newly DNA fingerprinted to confirm identity as ABI Prism 7900HT Sequence Detection System (Applied previously described (32). Cells were seeded in 96-well plates Biosystems) using SYBR Green PCR Master Mix to measure in 10 replicates at densities of 5 103 cells per well. Cell copy number of HMGA2, SMARCB1, let-7a3 and let-7b, and proliferation was monitored at 24, 48, and 72 hours using the using TaqMan Real-Time PCR Master Mix (Applied Biosys- colorimetric MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphe- tems) to quantitatively measure mRNA and miRNA expres- nyltetrazolium bromide) assay (CellTiter 96 Aqueous Assay; sion. b-globin and PEG6B genes were used as internal control Promega). For colony formation assays, 1 104 cells were for genomic analysis, whereas glyceraldehyde-3-phosphate mixed in 0.35% top agar diluted in 2 McCoy’s 5a Medium dehydrogenase (GAPDH) and small nuclear RNA RNU-48 containing 20% FBS before plating onto 0.8% bottom agar in were used as internal controls for mRNA and miRNA 6-multiwell plates. After 15 to 20 days, cells were stained (4–6 expression, respectively. Each qPCR experiment was per- hours, 37C) with 0.2 mg/mL of MTT (Sigma-Aldrich). formed three times. Data were presented as the relative Colonies > 0.2 mm in diameter were counted. The number quantity of targets, normalized with respect to internal of colonies was determined by analyzing 5 fields per well control, or relative to a calibrator sample. from 3 wells. All cultures were performed in triplicate.

www.aacrjournals.org Clin Cancer Res; 20(5) March 1, 2014 1181

Zhang et al.

miRNA/siRNA transient transfection and stable short number of two medium-sized DNA regions on chromo- hairpin RNA transfection some 22 was dramatically decreased (log2 ratio of less than Individual miRNA, siRNA, or control small RNA 0.7; Fig. 1A). The first region located on chromosome 22q11 (smRNA, random 22-mer RNA purchased from Invitrogen) contains the featured SMARCB1 gene, and the second and anti-miR let-7 miRNA inhibitor (a chemically modi- region includes let-7a3 and let-7b tumor suppressor miRNA. fied, single-stranded oligonucleotide) was used at a con- The detailed CNVs of chromosome 22 and locations of centration of 60 pmol/well for 1 105/well cells in a 6-well SMARCB1 and let-7a3 and let-7b are shown in Fig. 1B. We plate with Lipofectamine RNAiMAX (Invitrogen). After selected several genes with different locations for qPCR transfection, cells were harvested and analyzed at the indi- validation using genomic DNA of adjacent nontumor tissue cated times. Human let-7 mirVana miRNA (Cat. # 4464066) in the section as reference material to get more accurate data. and HMGA2 siRNA (Cat. # 4392420) were purchased from As shown in Fig. 1B, only the copy number of the genes, Life Technologies, and their efficacy of HMGA2 knockdown which are located at the first region (SMARCB1 and IGLL1) was tested by Western blot analysis. or at the second region (let-7a3, let-7b, LIF, and PPARP), and Human let-7 miRNA (LentimiRa-GFP-hsa-let-7 Vector, not those of genes outside the two regions (PDE6B located mh10004) and HMGA2 short hairpin RNA (shRNA) in on chromosome 4, CP1BB, MYH9, GGT1, and NAGA on piLenti-siRNA-GFP (I009810b) were purchased from chromosome 22) were dramatically decreased. Results of Applied Biological Materials Inc.. Lentiviral production in gene copy number analysis indicated that the AT/RT tissue human embryonic kidney (HEK) 293T cells was as we harbored a homozygous deletion of region 1, and a het- previously described (21). For stable infection, 8 104 cells erozygous deletion of region 2. were plated in each well of 6-well plates. After overnight Tumor-specific mutations may provide essential transla- incubation, 50 mL of concentrated lentiviral particles were tional genetic information for personalized cancer therapy. added with 12 mg/mL polybrene to each well. Forty-eight We did not find mutations or alterations in copy number of hours later, fresh medium containing 2 mg/mL puromycin the common, broadly mutated genes such as p53, PIK3CA, was added to each well. Fresh medium containingpuromycin K-Ras, and B-Raf in the AT/RT tissue. Similarly, a recent was replaced every 3 to 4 days. Single colonies were obtained study showed the absence of mutations in canonical path- after 2 weeks of puromycin selection. Substantial knockdown ways critical for development and progression of adult of HMGA2 protein was tested by Western blot analysis cancers (33). By Sanger DNA sequencing, novel nonsynon- compared with the control lentivirus transfected G401 cells. ymous point mutations including cancer-related genes such as DBF4, PMS1, RASSF4, Rap2b, and Tie2 were identified Matrigel invasion assay (Supplementary Fig. S3). Test cells (5 104) in serum-free medium were plated in the Matrigel-coated inserts of a 24-well pate (8-mm pores; Inactivation of SMARCB1 tumor suppressor in AT/RT BD) and medium with 10% FBS was added into the bottom tissues in accordance with the manufacturer’s protocol. After 24 Loss of SMARCB1 protein is generally accepted as the hours, cells that migrated through and attached to the diagnostic characteristic of AT/RT, and may be the result of bottom membrane were fixed and stained with the Diff- varied genetic alterations in AT/RT (3). We further investi- Quick kit (Fisher Scientific). Invading cells in three different gated genetic alterations of SMARCB1 gene by FISH and fields were counted under a light microscope of 100 exon sequencing in 18 AT/RT tissues. FISH analysis showed magnification. Three replicate experiments were performed. that 7 AT/RT tissues harbored homozygous deletion of SMARCB1, and 5 AT/RT tissues contained heterozygous Statistical analysis deletion of SMARCB1, whereas large deletions of regions Data were collected using an MS Excel spreadsheet. of chromosome 22, including loss of the entire chromo- Group comparisons for continuous data were done with some 22 in one case was discovered in three cases. Infor- the Student t test for independent means or two-way mation about patients’ age and SMARCB1 FISH and immu- ANOVA. Statistical significance was set at P < 0.05. nohistochemistry (IHC) is shown in Supplementary Table S1. Genetic alterations leading to loss of SMARCB1 protein in another three AT/RT cases were not identified by FISH or Results exome sequencing. Identification of genetic abnormalities of an AT/RT Four typical AT/RT tissues (labeled as AT/RT-1, -2, -3, -4) genomic DNA by whole-exome sequencing are shown in Fig. 2. The first AT/RT tissue (AT/RT-1) was We initially adopted whole-exome sequencing to inves- from a 19-year-old female, and other three were from tigate genetic alterations in an AT/RT tissue of a 19-year-old children less than 10 years old. Hematoxylin/eosin staining female. Due to the limited amount of genomic DNA from revealed that tumor cells were morphologically highly adjacent nontumor partial sections, genomic DNA from heterogeneous (Fig. 2A). Double-color interphase FISH saliva was used as reference. Data analysis demonstrated showed that AT/RT-1 harbored an intratumor homoge- that the sequencing quality for genomic DNA from saliva neous homozygous loss of SMARCB1 signal (0 red signal and FFPE AT/RT tissue was highly comparable (Supplemen- of SMARCB1 with 2 green signals of chromosome 22 tary Figs. S1 and S2). CNV analysis revealed that copy centromere; Fig. 2B), AT/RT-2 harbored a heterozygous loss

1182 Clin Cancer Res; 20(5) March 1, 2014 Clinical Cancer Research

Let-7 miRNA and HMGA2 in Atypical Teratoid/Rhabdoid Tumor

A Chromosome 22 6 ) 2

4 IGLL1 SMARCB1 LIF PPARa

2 Let-7a3/7b –2 0 Ratio to saliva (1/log Ratio to saliva –4 –6 0 1,000 2,000 3,000 4,000 B Normal Tumor 1.5 Unchanged-1 Deletion-1 Unchanged-2 Deletion-2

0.5 (ratio to normal)(ratio 0

LIF Relative gene copy number gene copy Relative MYH9 GGT1 NAGA Let-7b GP1BB Let-7a3 PPARa IGLL1-E3 IGLL1-E5 PDE6B (Chr4) SMARCB1-E2SMARCB1-E6 HMGA2 (Chr12)

Figure 1. Identification of deletion of SMARCB1 and let-7a3/let-7b miRNA in an adult AT/RT by whole-exome DNA sequencing. A, relative CNV plot of chromosome 22. SMARCB1, let-7a3, and let-7b were located at the first and second medium-sized deletions, respectively; arrows indicate their location. B, qPCR analysis to validate the loss of SMARCB1, miRNA let-7a3,andlet-7b in the genome of tumor tissue. The copy number of each selected gene was first normalized to b-globin (on chromosome 11) and results presented as the ratio to the adjacent normal tissue copy number. SMARCB1-E2 and SMARCB1-E6 represent PCR amplicons for exon-2 and exon-6 of SMARCB1, IGLL1-E3, and IGLL1-E5 for exon 3 and 5 of IGLL1 (IGLL1; immunogloblin l-like polypeptide 1) gene. Let-7a3, let-7b, LIF (leukemia inhibitory factor), and PPARa (peroxisome proliferation activated receptor a) were selected as representatives of the second deletion. The copy number of HMGA2 of chromosome 12 and PDE6B (phosphodiesterase 6B) of chromosome 4 (unchanged-1) and genes outside the two deletions of chromosome 22 (unchanged-2): GP1BB (platelet glycoprotein 1b, b polypeptide), MYH9 (nonmuscle myosin heavy chain 9), GGT1 (g glutamyltransferase 1), and NAGA (N-acetyl-a-D-galactosaminidase) were unchanged in the AT/RT tissue. of SMARCB1 (one red with two green; Fig. 2B), whereas AT/ expression, we first examined HMGA2 oncoprotein expres- RT-3 harbored a deletion of large DNA region containing sion in the 18 AT/RT tissues. Based on the percentage of SMARCB1 gene on chromosome 22 or monoallelic chro- HMGA2-stained tumor cells, the levels of HMGA2 onco- mosome 22 (one red with one green; Fig. 2B). However, AT/ protein in AT/RT tissues were scored on a scale of 0 to 3 (0 RT-4 displayed normal diploid SMARCB1 (Fig. 2B). For AT/ for <5% tumor cells; 1 for 5%–25% tumor cells; 2 for 25%– RT-2 tissue with heterozygous deletion of SMARCB1, exon 50% tumor cells; 3 for >50% tumor cells with positive sequencing identified a nonsense mutation of c.197G>A, p. HMGA2 nuclear staining). Figure 3A shows four levels of S66 in exon 2 (Fig. 2C), resulting in a truncated nonfunc- HMGA2 oncoprotein expression in AT/RT tissues that were tional protein. For AT/RT-3, exon sequencing found an scored as 0 to 3, respectively. Quantitative analysis showed insertion of c.531_532insG, p.H177 in exon 5 (Fig. 2C). that HMGA2 oncoprotein expression in 16.7% (3/18), However, no mutations were found in the AT/RT-4, and its 16.7% (3/18), 16.7% (3/18), and 50% (9/18) of examined genetic alterations are still unknown. It is also worthwhile AT/RT tissues was scored as 0, 1, 2, and 3, respectively (Fig. pointing out that the technologies we used in the study 3B). The expression score of HMGA2 oncoprotein in each would miss single/multiple exon deletions and copy num- AT/RT tissue is also presented in Supplementary Table S1. ber neutral LOH. Loss of nuclear immunostaining of The results indicate that overexpression of HMGA2 onco- SMARCB1 protein was found in all 4 AT/RT tissues, and protein may be a frequent event in AT/RT tissues. the endothelial cells (AT/RT-1) and infiltrating lymphoid To further investigate copy number alteration of let-7a3/ cells (AT/RT-2) were positively stained for the SMARCB1 let-7b miRNA; and whether overexpression of HMGA2 was protein (Fig. 2D). related to gene amplification, we measured the gene copy number of let-7a3/let-7b and HMGA2 using of qPCR anal- Frequent overexpression of HMGA2 oncoprotein and ysis on genomic DNA extracted from 11 AT/RT tissues that its correlation with let-7a3/let-7b in AT/RT tissues had paired nontumor partial sections. The qPCR analysis The miRNA let-7 family is reported to negatively regulate showed that the copy number of let-7a3/let-7b decreased by HMGA2 expression in a variety of human cancers (16, 17). at least 50% in 4 of 11 AT/RT tissues examined (Fig. 3C), To reveal the correlation of let-7a3/let-7b with HMGA2 whereas the average HMGA2 gene copy number did not

www.aacrjournals.org Clin Cancer Res; 20(5) March 1, 2014 1183

Zhang et al.

A BCD

AT/RT-1 F-19y Deleted

Figure 2. Inactivation of SMARCB1 tumor suppressor in four AT/RT-2 representative AT/RT tissues. A, M-9y representative hematoxylin/eosin staining of AT/RT tissues; magniﬁcation, 200. B, double- color interphase FISH in which SMARCB1 signal is red and chromosome 22 centromere is green signal. C, genomic AT/RT-3 alterations of SMARCB1 identiﬁed M-1.5y by exon sequencing. D, SMARCB1 immunochemistry of tissue sections to those shown in A and B.

AT/RT-4 F-3y No alteration

change (Fig. 3C). Consistent with this, FISH analysis further not affect HMGA2 expression at both the mRNA and protein confirmed that HMGA2 was not amplified in any of the AT/ levels in human fibroblast 293-T cells (Fig. 4A and B). RT tissues. IHC showed overexpression of HMGA2 onco- Moreover, exogenous expression of SMARCB1 in human protein (Fig. 3D, left); however, FISH analysis displayed kidney rhabdoid tumor G401 cells harboring a homozy- normal diploid HMGA2 (Fig. 3D, right) in two represen- gous SMARCB1 deletion did not suppress either HMGA2 tative AT/RT tissues. By quantitative reverse transcription mRNA or protein levels (Fig. 4C and D). These results PCR (qRT-PCR) analyses, HMGA2 mRNA was significantly indicate that the inactivation of SMARCB1 may not directly higher, whereas let-7a3/let-7b was significantly lower, in AT/ induce the overexpression of HMGA2 oncoprotein in the RT tissues than in adjacent normal tissues (Fig. 3E). Com- AT/RT tissues, and instead may be two independent path- pared with let-7a3, the abundance of let-7b was much higher ologic events in AT/RT development. in AT/RT tissues. A significant inverse correlation between the expression of let-7b and HMGA2 in AT/RT tissues (N ¼ Overexpression of let-7b or knockdown of HMGA2 11) was demonstrated using the Pearson correlation coef- significantly suppressed proliferation, and colony ficient analysis (R ¼0.34; P < 0.05; Fig. 3F). In all, the formation of rhabdoid tumor cell line G401 above data indicated that HMGA2 overexpression in AT/RT To assess the biologic significance of let-7/HMGA2 in AT/ tissues may be partially associated with the reduction in let- RT development, we examined the impact of overexpression 7a3/let-7b miRNA. of let-7b or knockdown of HMGA2 oncoprotein on cellular proliferation and colony formation of rhabdoid tumor Overexpression of HMGA2 oncoprotein is not G401 cells. Both let-7b miRNA and HMGA2-specific siRNA associated with inactivation of SMARCB1 tumor efficiently decreased HMGA2 mRNA (Fig. 5A) and HMGA2 suppressor protein (Fig. 5B) in G401 cells at 72 hours after transfection. The SMARCB1 complex regulates expression of many Compared with control smRNA, let-7b miRNA and HMGA2 genes at the transcriptional level (7). We further questioned siRNA significantly suppressed proliferation of G401 cells. whether overexpression of HMGA2 was associated with the Suppression of cell proliferation was slightly stronger with loss of SMARCB1, and whether SMARCB1 regulates the let-7b miRNA than with HMGA2 siRNA, but not significant- expression of HMGA2. siRNA knockdown of SMARCB1 did ly different, indicating let-7 miRNA may function largely

1184 Clin Cancer Res; 20(5) March 1, 2014 Clinical Cancer Research

Let-7 miRNA and HMGA2 in Atypical Teratoid/Rhabdoid Tumor

A B

Figure 3. Overexpression of HMGA2 oncoprotein and its correlation with let-7a3/let7b 60 miRNA in AT/RT tissues. A, representative photomicrographs 50 of HMGA2 oncoprotein levels in 01 AT/RT tissues, showing scoring 40 from 0 to 3; 0 for <5% tumor cells; 1 for 5% to 25% tumor cells; 2 for 30 25% to 50% tumor cells; 3 for 20 >50% tumor cells with positive HMGA2 nuclear staining. B, 10 Percentage of AT/RT with of AT/RT Percentage histogram of the distribution of scored HMGA2 protein (%) HMGA2 oncoprotein levels in 0 0 123 the AT/RT tissues scored as 0 to 3. 23 C, let-7a3/let-7b and HMGA2 relative gene copy number by qPCR analysis in 11 AT/RT CDHMGA2 IHC HMGA2 FISH genomic DNA samples. Gene copy numbers in tumor samples 2.0 normalized to genomic DNA from adjacent normal sections. D, photomicrographs of IHC and FISH 1.5 of HMGA2 in two representative AT/RT tissues. For FISH, red signal 1.0 is the HMGA2 probe, green is the chromosome 12 centromere probe. E, qRT-PCR analysis of let- 0.5 7a3/let-7b miRNA and HMGA2 mRNA expression in AT/RT

Relative gene copy number Relative gene copy 0.0 tissues. The level of let-7a3/let-7b miRNA was normalized to RNU48 let-7b small nucleolar RNA, and let-7a3 HMGA2 presented as the ratio of the level of EF RNU48. The level of HMGA2 mRNA was normalized to GAPDH 18 mRNA, and is presented as the Normal Tumor GAPDH 16 ratio of 1:100 of the level of 30 mRNA (, P < 0.05; P < 0.01 14 , R = –0.34 compared with that of normal 25 12 P section). F, The Pearson < 0.05 correlation coefﬁcient analysis plot 20 10 of the expression of let-7b and 15 8 HMGA2 in AT/RT tissues (N ¼ 11) ** * 6 demonstrated a signiﬁcant inverse 10 R ¼ P 4 correlation ( 0.34; < 0.05). HMGA2 mRNA level

miRNA/mRNA level 5 2 ** 0 0 let-7a3 let-7b HMGA2 0 20 40 let-7b miRNA level through repression of HMGA2 expression (Fig. 5C). Con- Fig. S4B), supporting the notion that HMGA2 oncoprotein sistent with this, we also observed that let-7 inhibitor plays a critical role in the anchorage-independent growth of significantly promoted proliferation of G401 cells, accom- G401 cells. panied by upregulation of HMGA2 protein (Supplementary Fig. S4A). Importantly, compared with G401 cells trans- Overexpression of let-7b or knockdown of HMGA2 duced with control lentivirus, the number and size of significantly decreased invasive potential of rhabdoid colonies formed by G401 cells transduced with lentivirus tumor cell line G401 expressing let-7 miRNA or HMGA2 shRNA was dramatically We also examined the impact of let-7 and HMGA2 on the lower (Fig. 5D). Conversely, overexpression of HMGA2 invasive potential of G401 cells. G401 rhabdoid tumor cells enhanced colony formation by G401 cells (Supplementary were first transfected with control smRNA, let-7b miRNA, or

www.aacrjournals.org Clin Cancer Res; 20(5) March 1, 2014 1185

Zhang et al.

number decrease in the genomic locus harboring let-7a3/let- 0 ABControl siRNA 7b miRNA. Because the 3 -UTR of HMGA2 has seven con- HEK-293 SMARCB1-siRNA served sites complementary to the let-7 miRNA family (16, Control SMARCB1 1.2 17), we then extended our studies to examine the status and 1 HMGA2 correlation of let-7a3/let-7b with HMGA2 in 2 adult and 16 0.8 pediatric AT/RT tissues, and report herein the identification 0.6 SMARCB1 of HMGA2 oncoprotein overexpression in both adult and 0.4 pediatric AT/RT tissues. 0.2 ** GAPDH The oncogenic properties of HMGA2 are known to be Relative mRNA level Relative 0 SMARCB1 HMGA2 involved in tumor cell differentiation, transformation (16, CD 17), aggressive tumor growth (24, 34), early metastasis (35, Control plasmid 36), and DNA damage response (37). HMGA2 has also been SMARCB1 plasmid G401 3 found to participate in the epithelial-to-mesenchymal tran- # 2.5 Control SMARCB1 sition, an essential process important to both normal devel- 2 HMGA2 opment and tumor progression (38, 39). Most recently, a 1.5 study of genetically engineered mouse models of human 1 SMARCB1 lung adenocarcinoma showed that tumors are initiated 0.5 synchronously by deletion of p53; however, only a subset Relative mRNA level Relative GAPDH 0 with overexpressed HMGA2 due to inactivation of NK2- SMARCB1 HMGA2 related homeobox transcription factor NKX2-1 became malignant and metastatic (40). These authors also observed Figure 4. Lack of association between overexpression of HMGA2 and that the expression of HMGA2 was dominant in the mod- inactivation of SMARCB1. A, HMGA2 mRNA expression assessed by erately and poorly differentiated human adenocarcinoma SMARCB1 (qRT-PCR) in G401 cells transfected with or control siRNA. groups, supporting our notion that HMGA2 participates in As expected SMARCB1 mRNA was inhibited by SMARCB1 siRNA (, P < 0.01, compared with control). B, Western blot of protein levels in dedifferentiation and disease progression of AT/RT. Because human fibroblast 293-T cells transfected with either control or SMARCB1 siRNA. Reconstitution of SMARCB1 did not suppress HMGA2 expression at both the mRNA (C; qRT-PCR; #, 1,000) and protein (D) levels (Western blot) in G401 cells transfected with control and SMARCB1 AB expression plasmids. Control smRNA Let-7 miRNA 1.2 HMGA2 siRNA Control Let-7 HMGA2 1 HMGA2 siRNA 24 hours before subjecting equal numbers HMGA2 0.8 of each modified cells to Matrigel invasion assay. Figure 0.6 GAPDH 6A shows representative photomicrographs of cells that 0.4 invaded the Matrigel membrane. Quantitative analysis of 0.2 invaded cells revealed that overexpression of let-7b or 0 knockdown of HMGA2 almost abolished the invasive HMGA2 mRNA level Relative 24 h 48 h 72 h potential of G401 cells (Fig. 6B), implying that HMGA2 CD25 Control smRNA also plays an indispensable role in the invasion of rhab- 2 Let-7 miRNA 20 doid tumor cells. 1.6 HMGA2 siRNA 15 1.2 * 10 ** ** Discussion 0.8 Colony number Colony 5 Due to the highly aggressive nature of the tumor and high 0.4 0 Cell proliferation index Cell proliferation rate of dissemination at diagnosis, intensive multimodality 0 Control let-7 HMGA2 treatment including surgery, radiation, and chemotherapy 24 h 48 h 72 h smRNA miRNA siRNA has been used to treat this disease, and proven to result in

more favorable outcomes (9–12). However, most children Figure 5. Impact of let-7 miRNA overexpression and HMGA2 die of progressive disseminated disease, with or without downregulation on cellular proliferation and colony formation in G401 local recurrence, despite aggressive therapy, and the general cells. Let-7b miRNA and HMGA2 siRNA significantly decreased HMGA2 outcome for patients with AT/RT is still dismal (8, 12). expression at both the mRNA (A) and protein (B) levels in G401 cells, Mean survival with surgical intervention alone is 3 months HMGA2 mRNA was measured by qRT-PCR assay at 24, 48, and 72 hours after transfection; HMGA2 protein was extracted at 72 hours after and with adjuvant chemotherapy and radiotherapy is about transfection. C, cell proliferation assay of G401 cells transfected with 8 months (8, 11). Therefore, more effective therapies are equal amounts of either control smRNA, or let-7 miRNA, or HMGA2 desperately needed. siRNA. Shown are the cell proliferation index for each measured by MTS In the present study, we originally aimed at interrogating assay at 24, 48, and 72 hours after transfection ( , P < 0.05, compared with control smRNA). D, colony formation assay of G401 cells in soft agar. the genomic abnormalities in an adult AT/RT tissue, due to Let-7b miRNA overexpression or HMGA2 downregulation significantly its extreme rarity, using state-of-the-art whole-exome DNA decreased colony formation (, P < 0.01, colony number of HMGA2 sequencing technology. We unexpectedly identified a copy shRNA/let-7 miRNA compared with that of control).

1186 Clin Cancer Res; 20(5) March 1, 2014 Clinical Cancer Research

Let-7 miRNA and HMGA2 in Atypical Teratoid/Rhabdoid Tumor

Figure 6. Impact of let-7 miRNA overexpression and HMGA2 downregulation on cellular invasive potential of G401 cells. G401 cells were transfected with either control smRNA, let-7b miRNA, or HMGA2 Control smRNA let-7 miRNA HMGA2 siRNA siRNA 24 hours before subjecting the same number of each modiﬁed B 60 cells to Matrigel invasion assay for 24 hours. A, representative 50 photomicrographs. B, quantitative analysis of invaded cells presented 40 as mean SD of 3 replicates per cell type (, P < 0.01, compared with control smRNA). 30

20 **

Invasive cell number Invasive ** 10

0 Control let-7 miRNA HMGA2 smRNA siRNA

brain invasion and widespread dissemination along cer- tially lead to novel therapeutic strategies. However, several ebrospinal fluid pathways is a hallmark of CNS malig- studies of genome-wide analysis thus far have yielded no nant rhabdoid tumors (41), our results also suggest consistent genetic alterations besides inactivation of HMGA2 a potential therapeutic target to suppress inva- SMARCB1 (3, 13, 14). Another recent study also demon- sion of AT/RT. strated that besides a single activating mutation in N-RAS in Furthermore, we identified that HMGA2 mRNA was a case of AT/RT, no other mutations in canonical pathways inversely correlated with let-7 miRNA in AT/RT tissues. critical for development and progression of adult cancers Therefore, downregulation of let-7 miRNA may partially were identified in a total of 25 pediatric patients with AT/RT, be ascribed to the decrease in let-7a3/let-7b miRNA copy further implicating that distinct mechanisms might drive number in at least a portion of AT/RT samples. HMGA2 is these highly malignant pediatric tumors (14, 33). Our derepressed upon inhibition of let-7 in cells with high levels whole-exome sequencing analysis also failed to identify of the miRNA (22–24). miRNA deregulation, including mutations in the usual cancer-related genes, such as p53, miRNA gene deletion has been reported in several tumor PIK3CA, K-Ras, and B-Raf in the adult patient with AT/RT. types including ovarian cancer, breast cancer, and melano- However, mutations of DNA repair pathway genes such as ma (26, 42). In epithelial ovarian cancer, deletion of miRNA DBF4 and PMS1 were found in this AT/RT tissue, and these let-7a3/let-7b was found in 44.1% of cases, suggesting the genes may deserve further study to determine whether they locus may be a fragile site (26, 42). Several studies point to are involved in AT/RT development. Although in silico the importance of the HMGA2/let-7 miRNA axis in tumor analysis was not performed on these mutations identified biology and the promise HMGA2 has as a therapeutic target in the AT/RT tissue, it is highly desirable to use the method (18, 19, 25). Consistent with these results, we showed that to more systematically and accurately analyze and interpret overexpression of let-7 miRNA or knockdown of HMGA2 these genetic alterations, after more data are collected in expression significantly suppressed the in vitro growth of future. It is also important to note that genetic alterations in G401 rhabdoid tumor cells. Of note, let-7 miRNAs have adult AT/RTs are likely to be different than in pediatric many other targets besides HMGA2, and overexpression of tumors, because somatic alterations can accumulate over let-7 miRNA could potentially have effects on other targets time in late-stage adult AT/RTs. that also contribute to the antiproliferative effects of let-7 In summary, we have uncovered a high frequency of miRNA. upregulation of HMGA2 oncoprotein in AT/RT tissues, Identification of additional genetic alterations will fur- which may be partially contributed to by deletion of let- ther our understanding of the biology of AT/RT, also poten- 7a3 and let-7b miRNA. A significant inverse correlation

www.aacrjournals.org Clin Cancer Res; 20(5) March 1, 2014 1187

Zhang et al.

between HMGA2 mRNA and let-7b miRNA was observed. Administrative, technical, or material support (i.e., reporting or orga- nizing data, constructing databases): K. Zhang, J. Wang, Y. Wang, D.K. Moreover, we have demonstrated that overexpression of let- Ann, Y. Yen 7 miRNA or knockdown of HMGA2 can significantly reduce Study supervision: K. Zhang, Y. Yen rhabdoid tumor cell proliferation, colony formation, and invasion. Our results suggest that reconstitution of let-7 Acknowledgments The authors thank Loera Sofia of the Department of Pathology for expression or knockdown of HMGA2 may provide a novel SMARCB1 and HMGA2 IHC staining, Victoria Bedell of the Cytogenetic therapeutic strategy for the treatment of patients with AT/ Core Laboratory, City of Hope for assistance in SMARCB1 and HMGA2 FISH RT. assays; Dr. Her Helen Lin of the Department of Molecular pharmacology, City of Hope for her generous gift of anti-HMGA2 antibody; Dr. Lynne Smith for critical reading and editing of the article; and Mengen Chao of the fl Pediatric Brain Tumor Laboratory of Taipei Veterans General Hospital for Disclosure of Potential Con icts of Interest assistance in sample preparation. No potential conflicts of interest were disclosed. Grant Support Authors' Contributions This study was partially supported by the Aim for the Top University Plan Conception and design: K. Zhang, H. Gao, W. Zhou, D.K. Ann, Y. Yen of Ministry of Education, Cancer Excellence Center Plan of Taipei Veteran Development of methodology: K. Zhang, H. Gao, J. Wang, C. Kim, Y. Yen General Hospital, and DOH102-TD-C-111-007 of the Department of Acquisition of data (provided animals, acquired and managed patients, Health, Taiwan. provided facilities, etc.): K. Zhang, H. Gao, J. Wang, W. Zhou, P. Chu, D.M. The costs of publication of this article were defrayed in part by the Ho, T.-T. Wong, Y. Yen payment of page charges. This article must therefore be hereby marked Analysis and interpretation of data (e.g., statistical analysis, biosta- advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate tistics, computational analysis): K. Zhang, H. Gao, X. Wu, G. Sun, B. Mu, this fact. C. Kim, P. Chu, D.K. Ann, Y. Yen Writing, review, and/or revision of the manuscript: K. Zhang, H. Gao, X. Received May 26, 2013; revised December 5, 2013; accepted December 20, Wu, W. Zhou, G. Sun, C. Kim, D.K. Ann, Y. Yen 2013; published OnlineFirst January 14, 2014.

References 1. Biggs PJ, Garen PD, Powers JM, Garvin AJ. Malignant rhabdoid tumor 12. Bruggers CS, Bleyl SB, Pysher T, Barnette P, Afify Z, Walker M, et al. of the central nervous system. Hum Pathol 1987;18:332–7. Clinicopathologic comparison of familial versus sporadic atypical 2. Biegel JA, Tan L, Zhang F, Wainwright L, Russo P, Rorke LB. Altera- teratoid/rhabdoid tumors (AT/RT) of the central nervous system. tions of the hSNF5/INI1 gene in central nervous system atypical Pediatr Blood Cancer 2011;56:1026–31. teratoid/rhabdoid tumors and renal and extrarenal rhabdoid tumors. 13. Hasselblatt M, Isken S, Linge A, Eikmeier K, Jeibmann A, Oyen F, et al. Clin Cancer Res 2002;8:3461–7. High-resolution genomic analysis suggests the absence of recurrent 3. Jackson EM, Sievert AJ, Gai X, Hakonarson H, Judkins AR, Tooke L, genomic alterations other than SMARCB1 aberrations in atypical et al. Genomic analysis using high-density single nucleotide polymor- teratoid/rhabdoid tumors. Genes Chromosomes Cancer 2013; phism-based oligonucleotide arrays and multiplex ligation-dependent 52:185–90. probe amplification provides a comprehensive analysis of INI1/ 14. Lee RS, Stewart C, Carter SL, Ambrogio L, Cibulskis K, Sougnez C, SMARCB1 in malignant rhabdoid tumors. Clin Cancer Res 2009;15: et al. A remarkably simple genome underlies highly malignant pediatric 1923–30. rhabdoid cancers. J Clin Invest 2012;122:2983–8. 4. Biegel JA, Kalpana G, Knudsen ES, Packer RJ, Roberts CW, Thiele 15. Johnson SM, Grosshans H, Shingara J, Byrom M, Jarvis R, Cheng A, CJ,etal.TheroleofINI1andtheSWI/SNFcomplexinthedevel- et al. RAS is regulated by the let-7 microRNA family. Cell 2005;120: opment of rhabdoid tumors: meeting summary from the workshop 635–47. on childhood atypical teratoid/rhabdoid tumors. Cancer Res 16. Lee YS, Dutta A. The tumor suppressor microRNA let-7 represses the 2002;62:323–8. HMGA2 oncogene. Genes Dev 2007;21:1025–30. 5. Eaton KW, Tooke LS, Wainwright LM, Judkins AR, Biegel JA. Spec- 17. Mayr C, Hemann MT, Bartel DP. Disrupting the pairing between let-7 trum of SMARCB1/INI1 mutations in familial and sporadic rhabdoid and Hmga2 enhances oncogenic transformation. Science 2007;315: tumors. Pediatr Blood Cancer 2011;56:7–15. 1576–9. 6. Biegel JA, Zhou JY, Rorke LB, Stenstrom C, Wainwright LM, Fogelgren 18. Yu F, Yao H, Zhu P, Zhang X, Pan Q, Gong C, et al. let-7 regulates self B. Germ-line and acquired mutations of INI1 in atypical teratoid and renewal and tumorigenicity of breast cancer cells. Cell 2007;131: rhabdoid tumors. Cancer Res 1999;59:74–9. 1109–23. 7. Carlson M, Laurent BC. The SNF/SWI family of global transcriptional 19. Takamizawa J, Konishi H, Yanagisawa K, Tomida S, Osada H, Endoh activators. Curr Opin Cell Biol 1994;6:396–402. H, et al. Reduced expression of the let-7 microRNAs in human lung 8. Packer RJ, Biegel JA, Blaney S, Finlay J, Geyer JR, Heideman R, cancers in association with shortened postoperative survival. Cancer et al. Atypical teratoid/rhabdoid tumor of the central nervous Res 2004;64:3753–6. system: report on workshop. J Pediatr Hematol Oncol 2002;24: 20. Fedele M, Visone R, De Martino I, Troncone G, Palmieri D, Battista S, 337–42. et al. HMGA2 induces pituitary tumorigenesis by enhancing E2F1 9. Olson TA, Bayar E, Kosnik E, Hamoudi AB, Klopfenstein KJ, Pieters activity. Cancer Cell 2006;9:459–71. RS, et al. Successful treatment of disseminated central nervous 21. Li AY, Lin HH, Kuo CY, Shih HM, Wang CC, Yen Y, et al. High-mobility system malignant rhabdoid tumor. J Pediatr Hematol Oncol 1995; group A2 protein modulates hTERT transcription to promote tumor- 17:71–5. igenesis. Mol Cell Biol 2011;31:2605–17. 10. Zimmerman MA, Goumnerova LC, Proctor M, Scott RM, Marcus K, 22. Motoyama K, Inoue H, Nakamura Y, Uetake H, Sugihara K, Mori M. Pomeroy SL, et al. Continuous remission of newly diagnosed and Clinical significance of high mobility group A2 in human gastric cancer relapsed central nervous system atypical teratoid/rhabdoid tumor. J and its relationship to let-7 microRNA family. Clin Cancer Res Neurooncol 2005;72:77–84. 2008;14:2334–40. 11. Chi SN, Zimmerman MA, Yao X, Cohen KJ, Burger P, Biegel JA, 23. Rahman MM, Qian ZR, Wang EL, Sultana R, Kudo E, Nakasono M, et al. et al. Intensive multimodality treatment for children with newly Frequent overexpression of HMGA1 and 2 in gastroenteropancreatic diagnosed CNS atypical teratoid rhabdoid tumor. J Clin Oncol neuroendocrine tumours and its relationship to let-7 downregulation. 2009;27:385–9. Br J Cancer 2009;100:501–10.

1188 Clin Cancer Res; 20(5) March 1, 2014 Clinical Cancer Research

Let-7 miRNA and HMGA2 in Atypical Teratoid/Rhabdoid Tumor

24. Qian ZR, Asa SL, Siomi H, Siomi MC, Yoshimoto K, Yamada S, et al. 34. Wang X, Liu X, Li AY, Chen L, Lai L, Lin HH, et al. Overexpression of Overexpression of HMGA2 relates to reduction of the let-7 and its HMGA2 promotes metastasis and impacts survival of colorectal can- relationship to clinicopathological features in pituitary adenomas. Mod cers. Clin Cancer Res 2011;17:2570–80. Pathol 2009;22:431–41. 35. Hristov AC, Cope L, Reyes MD, Singh M, Iacobuzio-Donahue C, Maitra 25. Malek A, Bakhidze E, Noske A, Sers C, Aigner A, Schafer R, et al. A, et al. HMGA2 protein expression correlates with lymph node HMGA2 gene is a promising target for ovarian cancer silencing ther- metastasis and increased tumor grade in pancreatic ductal adenocar- apy. Int J Cancer 2008;123:348–56. cinoma. Mod Pathol 2009;22:43–9. 26. Zhang L, Huang J, Yang N, Greshock J, Megraw MS, Giannakakis A, 36. Chiappetta G, Ferraro A, Vuttariello E, Monaco M, Galdiero F, De et al. microRNAs exhibit high frequency genomic alterations in human Simone V, et al. HMGA2 mRNA expression correlates with the malig- cancer. Proc Natl Acad Sci U S A 2006;103:9136–41. nant phenotype in human thyroid neoplasias. Eur J Cancer 2008; 27. Wang Y, Hu X, Greshock J, Shen L, Yang X, Shao Z, et al. Genomic 44:1015–21. DNA copy-number alterations of the let-7 family in human cancers. 37. Li AY, Boo LM, Wang SY, Lin HH, Wang CC, Yen Y, et al. Suppression PLoS ONE 2012;7:e44399. of nonhomologous end joining repair by overexpression of HMGA2. 28. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory- Cancer Res 2009;69:5699–706. efficient alignment of short DNA sequences to the human genome. 38. Thuault S, Tan EJ, Peinado H, Cano A, Heldin CH, Moustakas A. Genome Biol 2009;10:R25. HMGA2 and Smads co-regulate SNAIL1 expression during induction 29. Venkatraman ES, Olshen AB. A faster circular binary segmentation of epithelial-to-mesenchymal transition. J Biol Chem 2008;283: algorithm for the analysis of array CGH data. Bioinformatics 2007; 33437–46. 23:657–63. 39. Watanabe S, Ueda Y, Akaboshi S, Hino Y, Sekita Y, Nakao M. HMGA2 30. Taylor BS, Barretina J, Socci ND, Decarolis P, Ladanyi M, Meyerson M, maintains oncogenic RAS-induced epithelial-mesenchymal transition et al. Functional copy-number alterations in cancer. PLoS ONE 2008;3: in human pancreatic cancer cells. Am J Pathol 2009;174:854–68. e3179. 40. Winslow MM, Dayton TL, Verhaak RG, Kim-Kiselak C, Snyder EL, 31. Zhang K, Chu K, Wu X, Gao H, Wang J, Yuan YC, et al. Amplification of Feldser DM, et al. Suppression of lung adenocarcinoma progression FRS2 and activation of FGFR/FRS2 signaling pathway in high-grade by Nkx2-1. Nature 2011;473:101–4. liposarcoma. Cancer Res 2013;73:1298–307. 41. Muller M, Hubbard SL, Fukuyama K, Dirks P, Matsuzawa K, Rutka JT. 32. Lorenzi PL, Reinhold WC, Varma S, Hutchinson AA, Pommier Y, Characterization of a pineal region malignant rhabdoid tumor. Towards Chanock SJ, et al. DNA fingerprinting of the NCI-60 cell line panel. understanding brain tumor cell invasion. Pediatr Neurosurg 1995; Mol Cancer Ther 2009;8:713–24. 22:204–9. 33. Kieran MW, Roberts CW, Chi SN, Ligon KL, Rich BE, Macconaill LE, 42. Yang N, Kaur S, Volinia S, Greshock J, Lassus H, Hasegawa K, et al. et al. Absence of oncogenic canonical pathway mutations in MicroRNA microarray identifies Let-7i as a novel biomarker and ther- aggressive pediatric rhabdoid tumors. Pediatr Blood Cancer 2012; apeutic target in human epithelial ovarian cancer. Cancer Res 59:1155–7. 2008;68:10307–14.

www.aacrjournals.org Clin Cancer Res; 20(5) March 1, 2014 1189

Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2014 American Association for Cancer Research. Correction Clinical Cancer Correction: Frequent Overexpression of Research HMGA2 in Human Atypical Teratoid/ Rhabdoid Tumor and Its Correlation with let-7a3/let-7b miRNA

In the article (Clin Cancer Res 2014;20:1179–89), which was published in the March 1, 2014, issue of Clinical Cancer Research (1), the first author alerted us to an error in one figure, specifically, the reuse of a GAPDH gel blot in panels B and D of Fig. 4. The author acknowledged the reuse was the result of administrative error and provided the correct image for the GAPDH gel blot in panel D. A corrected version of Fig. 4 is below. The results and conclusions put forth in this article remain unchanged. The authors regret these errors.

Control siRNA A SMARCB1-siRNA B HEK-293 1.2 Control SMARCB1 1 0.8 HMGA2 0.6 0.4 SMARCB1 0.2 0

Relative mRNA level Relative mRNA GAPDH SMARCB1 HMGA2

C Control plasmid D SMARCB1 Plasmid 3 G401 Control SMARCB1 2.5 2 HMGA2 1.5 1 SMARCB1 0.5 GAPDH Relative mRNA level Relative mRNA 0 SMARCB1 HMGA2

Figure 4.

Reference 1. Zhang K, Gao H, Wu X, Wang J, Zhou W, Sun G, et al. Frequent overexpression of HMGA2 in human atypical teratoid/rhabdoid tumor and its correlation with let-7a3/let-7b miRNA. Clin Cancer Res 2014;20:1179–89.

Published online June 15, 2017. doi: 10.1158/1078-0432.CCR-16-2999 Ó2017 American Association for Cancer Research.

www.aacrjournals.org 3225 Published OnlineFirst January 14, 2014; DOI: 10.1158/1078-0432.CCR-13-1452

Frequent Overexpression of HMGA2 in Human Atypical Teratoid/Rhabdoid Tumor and Its Correlation with let-7a3/let-7b miRNA

Keqiang Zhang, Hanlin Gao, Xiwei Wu, et al.

Clin Cancer Res 2014;20:1179-1189. Published OnlineFirst January 14, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-13-1452

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2014/01/14/1078-0432.CCR-13-1452.DC1

Cited articles This article cites 42 articles, 17 of which you can access for free at: http://clincancerres.aacrjournals.org/content/20/5/1179.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/20/5/1179.full#related-urls

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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/20/5/1179. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.