Open Full Page

Published OnlineFirst August 27, 2014; DOI: 10.1158/1078-0432.CCR-14-0706

Clinical Cancer Biology of Human Tumors Research

Identiﬁcation of Differentially Expressed Long Noncoding RNAs in Bladder Cancer

Stefan Peter1, Edyta Borkowska1, Ross M Drayton1, Callum P. Rakhit1, Aidan Noon1,2, Wei Chen3, and James WF Catto1

Abstract Purpose: Loss of epigenetic gene regulation through altered long noncoding RNA (lncRNA) expression seems important in human cancer. LncRNAs have diagnostic and therapeutic potential, and offer insights into the biology disease, but little is known of their expression in urothelial cancer. Here, we identify differentially expressed lncRNAs with potential regulatory functions in urothelial cancer. Experimental Design: The expression of 17,112 lncRNAs and 22,074 mRNAs was determined using microarrays in 83 normal and malignant urothelial (discovery) samples and selected RNAs with qPCR in 138 samples for validation. Significantly differentially expressed RNAs were identified and stratified according to tumor phenotype. siRNA knockdown, functional assays, and whole-genome transcriptomic profiling were used to identify potential roles of selected lncRNAs. Results: We observed upregulation of many lncRNAs in urothelial cancer that was distinct to corresponding, more balanced changes for mRNAs. In general, lncRNA expression reflected disease phenotype. We identified 32 lncRNAs with potential roles in disease progression. Focusing upon a promising candidate, we implicate upregulation of AB074278 in apoptosis avoidance and the maintenance of a proproliferative state in cancer through a potential interaction with EMP1, a tumor suppressor and a negative regulator of cell proliferation. Conclusions: We report differential expression profiles for numerous lncRNA in urothelial cancer. We identify phenotype-specific expression and a potential mechanistic target to explain this observation. Further studies are required to validate lncRNAs as prognostic biomarkers in this disease. Clin Cancer Res; 1–11. 2014 AACR.

Introduction resection (4). In contrast, high-grade BCs are aggressive Bladder cancer (BC) is the fourth commonest male tumors that may present before or after the onset of malignancy and one of the most expensive human can- muscle invasion (5). These poorly differentiated urothe- cers to manage (1, 2). The majority of tumors are urothelial cancers have widespread chromosomal instability, lial carcinoma in histologic type. Clinicopathological multiple mutations, and are best characterized by defi- data suggest that urothelial cancers are best stratified into ciency of p53-mediated pathways (3). High-grade tumors two distinct phenotypes characterized by low- and high- share many molecular alterations regardless of stage and grade cellular differentiation. Low-grade urothelial can- respond poorly to chemotherapy (6). cers frequently have mutations of FGFR3 and STAG2, Although many reports detail genetic events in urothe- partial deletion of chromosome 9 (3), and rarely progress lial cancer, alterations of epigenetic gene regulation are to muscle invasion or metastases following endoscopic also important in this disease (7). In general, epigenetic alterations reflect urothelial cancer disease phenotypes and associated genetic events (8, 9). For example, Wolff 1Academic Urology Unit and Unit of Molecular Oncology, University of and colleagues reported that high-grade invasive urothe- Sheffield, Sheffield, United Kingdom. 2Department of Urology, University of Toronto, Toronto, Canada. 3Max Delbruck€ Center for Molecular Medicine, lial cancers had widespread aberrant hypermethylation, Berlin, Germany. whereas low-grade noninvasive tumors had regional Note: Supplementary data for this article are available at Clinical Cancer hypomethylation (10). In general, epigenetic events Research Online (http://clincancerres.aacrjournals.org/). reflect chromosomal changes with the disease and can S. Peter and E. Borkowska share first authorship. be used as predictor of disease progression (11). With regard to miRNA, low-grade urothelial cancers are char- Corresponding Author: James Catto, Department of Oncology, G Floor, The Medical School, University of Sheffield, Beech Hill Road, Sheffield S10 acterized by loss of expression of many species, including 2RX, United Kingdom. Phone: 44-114-226-1229; Fax: 44-114-271-2268; miRs99a/100 that target FGFR3 (12, 13). In contrast, E-mail: j.catto@sheffield.ac.uk high-grade tumors have upregulation of many miRs, doi: 10.1158/1078-0432.CCR-14-0706 including miR21 that targets the p53 pathway (14) and 2014 American Association for Cancer Research. miR129 (15).

www.aacrjournals.org OF1

Peter et al.

cle invasive (NMI), high-grade NMI, and invasive urothelial Translational Relevance cancer. All tissues were fresh frozen in liquid nitrogen. Bladder cancer is a common disease whose biology is Histologic conﬁrmation was obtained before use. We also poorly understood. Here, we show that many long analyzed the human telomerase reverse transcriptase noncoding RNAs have altered expression in the disease (hTERT) immortalized normal human urothelial (NHU) and may play key roles in proliferation and disease cells maintained in keratinocyte serum-free medium con- progression. We identify one long noncoding RNA that taining bovine pituitary extract, EGF (Invitrogen), and appears a strong candidate as a prognostic biomarker cholera toxin. The hTERT NHU cells were obtained as a and show that it affects mRNA expression, cell prolifer- gift directly from Prof. M.A. Knowles, University of Leeds ation, and death, and may have potential as a therapeutic (Leeds, United Kingdom), and tested by this laboratory target. using comparative genomic hybridization and mutational analysis of CDKN2A and TP53 as described (25). The cells were not authenticated in Shefﬁeld before use.

Epigenetic gene regulation may occur directly or indirec- Long noncoding RNA expression profiling m tly through noncoding RNA (ncRNA) species. These are From each sample, 10 10 m thick sections were classified according to size, function, and gene location microdissected to extract normal and malignant urothelial > (16). Most reports of ncRNA in cancer have focused upon cells ( 90% content). Total RNA was extracted using the miRNAs, which modulate gene expression through target- mirVana kit (Ambion) according to manufacturer’s proto- ing complementary seed sequences within mRNA 30 UTRs col (26), and measured using a 2100 Bioanalyzer (Agilent). (17). The role of miRNAs in disease varies with cell type, The expression of long ncRNAs and protein coding mRNAs biologic stress, species conservation, and processing (9, 18). was determined using microarrays (NCode Human Non- To date, few data have reported the role of long noncoding coding RNA Microarrays, Invitrogen). Each sample was RNA (lncRNA) in cancer and little is known of their func- prepared according to Agilent’s one-color microarray pro- tion (16, 19). Exceptions include lncRNAs clearly implicat- tocol. Briefly, 200 ng RNA samples were mixed with spike ed in carcinogenesis (such as MALAT-1 or HOTAIR, in control RNA (Agilent), labeled with cyanine 3-CTP (Low reviewed in Spizzo and colleagues; ref. 20). LncRNAs are Input Quick Amp Labeling Kit, one-color, Agilent), and annotated according to gene location (intergenic, exonic, hybridized to the microarray (Gene Expression Hybridiza- intronic, anti-sense etc.) and size (21). The GENCODE tion Kit, Agilent) according to the manufacturer’s protocol. consortium recently annotated 9,277 long ncRNA genes, After washing (Gene Expression Wash Buffer 1 and 2, corresponding to 14,880 transcripts (22). In contrast with Agilent), the microarray slides were scanned using the protein coding genes, the majority of reported lncRNAs had Agilent Microarray Scanner platform (High Resolution few gene exons, were located in the nucleus and chromatin, Microarray Scanner C) and raw probe fluorescence and many were not conserved between species. As multiple extracted. The microarray data are deposited with Gene reports suggest that lncRNAs are important in human cancer Expression Omnibus (GSE55433). The NCode microarray (16, 19, 23) and preliminary data suggest this is true for contains duplicate probes to 17,112 ncRNAs and 22,074 urothelial cancer (24), we hypothesized that they play a role mRNAs. The lncRNAs were identified by various strategies in urothelial carcinogenesis. To test this hypothesis, we without annotation (27). To annotate the array, we con- profiled the expression of 9,351 lncRNAs in normal and verted all probes to hg 19.0 loci (using LiftOver, UCSC malignant urothelial samples. We identified global changes genome browser), matched to target gene, structure and that reflect the disease phenotype and differ to patterns of the Gencode v7.0 annotation (22), and the nearest CpG mRNA alteration. We performed explorative analysis of island/differentially methylated region (28). For microar- selected species and identified those with potential prog- ray validation, individual lncRNA expression was mea- nostic roles. sured in triplicate using qRT-PCR. Total RNA was tran- scribed using random hexamer primers (Applied Biosys- Materials and Methods tems) and the High Capacity Reverse Transcription Kit Patient samples and cell lines (Applied Biosystems) before diluting 10-fold in nucle- We analyzed two patient cohorts in this work (Table 1). ase-free water. cDNA was mixed with SyberGreen Master- For RNA microarray profiling, we examined urothelial Mix (Applied Biosystems) and PCR primers, and analyzed samples from 57 patients with urothelial cancer and 26 using the ABI 7900HT real-time PCR system. cDNA expres- DC disease free controls (urothelium from radical prostatecto- sion was calculated using t values normalized by sub- TEGT HSP90AB1 my cases). For validation, we examined a second cohort of n traction of the mean of and expression ¼ DDCt ¼ 138 samples from patients with urothelial cancer. Tumors (as reference genes) and fold change (FC 2 ) cal- were obtained following trans-urethral resection of the first culated (29). cancer within a patient, na€ve to Bacillus CalmetteGuerin, chemotherapy, or other pretreatments, and classified using Protein coding potential score of ncRNAs the 2004 WHO/ISUP criteria. Tumors were selected to We investigated the protein coding potential of selected reflect the urothelial cancer spectrum: low-grade non-mus- lncRNAs using geneID (30) and CPC against UniProt

OF2 Clin Cancer Res; 2014 Clinical Cancer Research

Integrated Epigenetic Regulation in Bladder Cancer

Table 1. Details of patients and tissue samples analyzed in this study

Microarray cohort Validation cohort UCC UCC

Low-grade High-grade Normal Low-grade High-grade Normal NMI NMI Invasive urothelium NMI NMI Invasive urothelium Total 24 13 19 26 39 31 42 26 Gender Male 13 11 15 6 23 25 29 26 Female 6 2 4 1 16 6 13 Age Mean 76.3 76.4 72.1 65.5 85 79 74 69.5 Range 48–82 54–84 46–89 59–82 54–100 65–95 41—92 61–88 Stage pTa 24 2 0 39 12 0 pTis 0 2 0 0 3 0 pT1 0 9 0 0 16 0 pT2-4 0 0 19 0 0 42 Progression Yes 2 2 11 5 12 18 No 22 11 8 34 19 24 Follow-up (mo) Mean 31.8 31.6 14.1 67 54.4 17.5 Range 13–98 15–107 1–108 4–159 1–129 1–144

Reference Clusters (31). For both, negative scores suggest caspase-3 activity in 3 105 cells, 24 hours after transfection low protein coding potential. using the CaspGLOW Fluorescein Active Caspase-3 Staining Kit (MBL). Briefly, we resuspended the cells in 300 mL PBS, siRNA knockdown of lncRNA expression added 1 mL FITC-DEVD-FMK, and incubated them for 1 We selected representative lncRNAs for further analysis hour at 37C. After pelleting the cells and washing them in and modulated expression using custom siRNA. All experi- washing buffer, they were resuspended in 300 mL wash ments were performed in triplicate using hTERT immortal- buffer and analyzed on a FACSCalibur flow cytometry ized NHU cells at 70% confluence transfected with siRNAs analyzer (Becton Dickinson). For proliferation, cells were (synthesized with LifeTechnologies BLOCK-iT RNAi design- seeded into 96-well plates following siRNA-mediated gene er) specific to lncRNAs and controls (scrambled RNA knockdown. The culture medium was replaced with 50 mL sequence). Cells were seeded into 12-well dishes and incu- MTT-PBS solution (3 mg MTT/1 mL PBS) 24, 48, and 72 bated for 3 hours before transfection with 80 nmol/L RNAi hours after knockdown, respectively. Cells were incubated using 2 mL Lipofectamine RNAiMAX (Invitrogen) per well. at 37C for 3 hours before 200 mL MTT dissolvent (DMSO) Transfection efficiency was determined 72 hours later by was added to each well. MTT absorbance was measured at qRT-PCR (Applied Biosystems). 570 nm using a microplate reader.

Growth analysis of siRNA modulated cells Genetic consequences of lncRNA modulation The growth characteristics of siRNA-transfected cells were mRNA expression was determined in transfected cells analyzed for cell-cycle regulation (propidium iodide flow using HG-U133 Plus 2.0 microarrays (Affymetrix, Cal.; cytometry), apoptosis (caspase-3 activation), and prolifer- ref. 32). Briefly, RNA was prepared with the Affymetrix ation (MTT assay). For cell-cycle analysis, cells were har- protocol and annealed to an oligo-d(T) primer with a T7 vested after 24, 48, and 72 hours of knockdown, centrifuged polymerase-binding site. cDNA was generated using super- (3 minutes at 1,400 g) and washed in PBS. Cells (1 106) script II and Escherichia coli DNA ligase and polymerase I, were resuspended in 1 mL PBS, mixed with 3 mL ice cold before the reaction was completed with T4 DNA polymerase absolute ethanol, and fixed over night at 4C before washing and EDTA. Amplified cDNA was cleaned, biotin-labeled, twice in PBS. Cells were then mixed with 5 mL RNase and fragmented, before hybridizing to the microarray for 16 solution (2 mg/mL) and 300 mL propidium iodide solution hours (45C in a rotating oven at 60 rpm). After washing (50 mg/mL), and incubated over night at 4C. The next day, and staining, the arrays were scanned (GC3000 scanner) cells were analyzed on a FACSCalibur flow cytometry ana- and data processed using Gene Chip Operating System lyzer (Becton Dickinson). For apoptosis, we determined software. mRNA expression was determined using

www.aacrjournals.org Clin Cancer Res; 2014 OF3

Peter et al.

Microarray Analysis Suite 5 (Affymetrix) and defined as v7 catalogue [ref. (22); n ¼ 3,885, Supplementary Table S2], expressed (perfect match probeset intensity greater than of which 355 were lncRNAs differentially expressed in mismatch intensity) or absent (mismatch probeset intensity urothelial cancer. Although the majority were long inter- greater or equal to perfect match intensity). ANOVA analysis genic ncRNA (lincRNA, n ¼ 225), we identified 130 was then performed using Partek Genomic Suite 6.5 b and lncRNAs located within protein coding genes (termed differentially expressed transcripts were defined as 2 rel- genic). These included RNAs coded in the sense [intronic ative fold change using a cutoff of P < 0.05. qRT-PCR was (n ¼ 20) and overlapping (n ¼ 3)] and antisense [exonic (n used to confirm the expression of individual mRNAs of ¼ 50), intronic (n ¼ 51), and overlapping (n ¼ 6)] direction interested in the second patient cohort. with respect to the mRNA gene (Supplementary Fig. S1). The proportion of aberrantly expressed lncRNAs in urothe- Statistical analysis lial cancer, when compared with normal urothelium, did Raw intensity values for each microarray probe were used not vary with gene location (range from 15% for antisense to calculate lncRNA expression. The NCode microarray exonic to 22% for overlapping antisense lncRNAs; Fig. 1B). includes multiple probes for each target RNA. We excluded RNAs whose targeting probes were non-concordant Expression of lncRNAs with respect to tumor (defined as signal ratio less than 0.5 or more than 2.0) in phenotype all the arrays. Normalization was achieved according to the We compared RNA expression across urothelial cancer 1-color default normalization of Agilent’s Gene Spring phenotypes (low-grade NMI, high-grade NMI, and inva- software (version 7), by dividing each raw intensity value sive) with normal urothelium. Unsupervised hierarchical by the median of the chip and the median expression of that clustering revealed three different expression profiles fitting RNA in all samples. Changes in lncRNA expression and the disease-free, the low-grade, and the high-grade/invasive statistical significance were calculated and illustrated using phenotypes (Fig. 1C). For all tumor phenotypes, we Volcano plots. Significant differences in expression between observed more up than downregulated lncRNAs, although malignant versus normal urothelium, or between urothelial the extent varied between groups (91% in invasive tumors, cancer phenotypes were defined using the Significance 61% in low-grade, and 53% in high-grade; Fig. 2A and Analysis of Microarray software (33) as a t test P value of Supplementary Fig. S2a). Accordingly, the magnitude of <0.05, a false discovery rate of <0.01, and an expression FC differential lncRNAs expression varied between the invasive of 2. Hierarchical clustering was performed using Cluster cohort (average fold change 2.57), and the low-grade and 3.0 and visualized in Tree view (Eisen Lab 9). Correlation high-grade cohort (average fold change 1.75 and 1.73, between variables was assessed using Pearson correlation respectively, Supplementary Fig. S2b). There was less dif- coefficient within SPSS (version 14.0, SPSS, Inc.). Area ference in the magnitude of change in expression for proportional to Venn diagrams were produced using Bio- mRNAs than lncRNAs (average fold change of 2.11 in Venn (34). lncRNA expression and tumor outcome were invasive, 1.59 in low-grade, and 1.45 in high-grade, t test investigated using the log-rank test and plotted by the P < 0.01). The majority of aberrantly expressed lncRNAs and Kaplan–Meier method within SPSS. Tumor progression was mRNAs belonged to the invasive subset [1,800/2,034 defined as the presence of pathologic, radiologic, or clinical lncRNAs (88%) and 1,050/1,410 mRNAs (74%)], followed evidence of an increase in tumor stage and measured from by the low-grade cohort [n ¼ 560 lncRNAs (28%) and n ¼ the time of surgery to the time of proven event. 757 mRNAs (54%)] and the high-grade cohort [n ¼ 416 lncRNAs (20%) and n ¼ 661 mRNAs (47%), (Fig. 2B)]. We Results defined a specific signature for each tumor phenotype by lncRNA expression in bladder cancer identifying RNAs significantly differentially expressed We investigated the expression of 17,112 lncRNAs and between each tumor type. Of these, 75% (1,356/1,800) of 22,074 mRNAs. The data are accessible through Gene lncRNAs in the invasive phenotype, compared with 19% Expression Omnibus Series accession number GSE55433. [109/560] of the low-grade cohort and only 3% [15/416] of We filtered to 9,351 lncRNAs and 7,922 mRNAs, for which the high-grade cohort were phenotype specific. Similarly for the microarray probe signals were concordant (Fig. 1A and mRNAs, 45% in the invasive cohort, 17% and 4% of the Supplementary Table S1). Comparison between urothelial low-grade and high-grade cohort, respectively, were phe- cancer and normal urothelium revealed 2,075 differentially notype specific. We identified n ¼ 188 (9%) lncRNAs and n expressed lncRNAs. In general, there was an increased ¼ 374 (26%) mRNAs that were altered in urothelial cancer expression of long ncRNAs in urothelial cancer [1,788 regardless of tumor phenotype. (86%) were upregulated (FC > 2) and 287 (14%) were downregulated (FC < 0.5)] when compared with normal LncRNA expression and tumor progression urothelium. Fewer protein coding mRNAs (n ¼ 1,410) were To identify lncRNAs with roles in urothelial carcino- differentially expressed between the malignant and normal genesis, we searched for those related to disease progres- tissues, and their distribution was more balanced [836 sion. We performed univariate log-rank analysis and (59%) upregulated and 574 (41%) downregulated in identified 32/2,075 (1.5%) lncRNAs associated with urothelial cancer] than for lncRNAs (t test P < 0.001). We tumor progression (Bonferonni adjusted log-rank P value filtered the microarray transcripts to those in the Gencode < 0.05; Table 2, Fig. 3A). More of these tumor

OF4 Clin Cancer Res; 2014 Clinical Cancer Research

Integrated Epigenetic Regulation in Bladder Cancer

A IncRNA (n = 9,351) C Normal 72% 24% Low-grade NMI High-grade NMI Invasive UCC 1% 1% 2% Fold change <0.25 mRNA (n = 7,922) <0.5 76% 15% No change >2.0 >4.0 1% 3% 5%

B 25% 20% 15% 10% IncRNAs 5% 0%

Divergent Convergent –3.00 Intronic sense –2.00 –1.00 ExonicIntronic antisense antisense 0.00 Overlapping sense Genic 1.00 Overlapping antisense 2.00 Intergenic sameIntergenic strand 3.00

Figure 1. A, percentage of significantly differentially expressed RNAs in urothelial cancer when compared with normal urothelium. Analyzed lncRNAs were more frequently aberrantly expressed than mRNAs, and less balanced (most were upregulated). B, significantly aberrant expressed lncRNAs in urothelial cancer when annotated by the Gencode v7 catalogue and mapped to genomic structure (expressed as a percentage of the total number of probes present in the database for that genomic locus). C, LncRNA expression in normal and malignant urothelium. Centroid linkage hierarchical clustering was performed after selecting lncRNAs with >80% expression frequency and median centering data using correlation similarity in Cluster 3.0. progression-related lncRNAs were phenotype specific (n ic), (iii) had low predicted protein-coding scores (thus ¼ 16) than shared between all urothelial cancer (n ¼ 6). likely to be a ncRNA), (iv) worse outcomes with high Eight were annotated within GENCODE. LncRNAs asso- expression (thus a potential oncogenic role), (iv) also ciated with tumor progression were balanced between up upregulated in urothelial cancer, and (v) appeared of (n ¼ 17) and downregulation (n ¼ 15). Protein coding particular interest as it was intronic (sense direction) to potential was calculated using two algorithms (Table 2; a protein coding host gene (sense to TANC2; as were most refs. 24, 25). The resultant scores were closely correlated validated ncRNAs in GENCODE) also upregulated in (Pearson r ¼ 0.81) and suggested only one member was urothelial cancer (thus potentially regulated by the likely to be a misclassified mRNA (CR611332). lncRNA; ref. 35). siRNA Transfection in hTERT-NHU cells reduced expression by 80% after 24 hours and signifi- A role for AB074278 expression in cell proliferation and cantly increased cell apoptosis/death (Fig. 3B). Knock- apoptosis down also significantly reduced proliferation (Fig. 3C) For functional validation, we investigated an aberrantly when compared with scrambled RNA controls. For com- expressed lncRNA using siRNA in normal urothelial parison, we also knocked down two aberrantly expressed (hTERT-NHU) cells. We selected AB074278, as it was lncRNAs not associated with tumor progression (namely, (i) associated with disease progression, (ii) upregulated G36639 and U50531). Proliferation was reduced for both inallurothelialcancerphenotypes (we were interested in these lncRNAs, but significantly less dramatically than for lncRNAs generic to urothelial cancer not subtype specif- AB074278 (Supplementary Fig. S3).

www.aacrjournals.org Clin Cancer Res; 2014 OF5

Peter et al.

A Higher in normal Higher in low Higher in normal Higher in high Higher in normal Higher in 12 urothelium grade tumor urothelium grade tumor urothelium invasive tumor 10 Significant Figure 2. A, LncRNA expression in 8 Nonsignificant

P low-, high-grade, and invasive 10 6 bladder cancer compared with

–log normal urothelium from non- 4 urothelial cancer cases. Changes 2 in expression (fold-change values plotted as log 2) and statistical 0 signiﬁcance (SAM P values plotted –6 –4 –2 0 2 4 6 –4–6 –2 0 2 4 6 –4–6 –2 0 2 4 6 as log 10) reveal both IncRNA expression (log fold change) 2 upregulation and downregulation Upregulated Upregulated mRNA of numerous lncRNAs. For all three B Inv IncRNA Inv Low-grade NMI phenotypes, there is a general High-grade NMI upregulation in lncRNA 447 Invasive UCC 1,336 expression. B, differentially Inv expressed lncRNAs according to Inv 20 LG 78 19 25 each urothelial cancer phenotype. 134 90 57 The majority of the altered lncRNAs 76 84 19 LG 100 was phenotype speciﬁc, with most 76 19 36 30 7 9 50 54 112 284 transcripts found in the invasive HG HG phenotype. 33 8 HG LG 120 LG 56 HG 15 Downregulation Downregulation

mRNA interactions for AB074278: epithelial membrane 4D; P < 0.01). No significant changes were seen with respect protein 1 to phenotype for TANC2, PTGS2, and CKS2. To explore interactions for AB074278, we compared changes in mRNA expression between siRNA knockdown and control NHU cells using HG-U133 Plus 2.0 microarrays Discussion in triplicate. We identified 471 mRNAs that were signifi- Technological improvements have revealed the impor- cantly differentially expressed (359 upregulated and 112 tance of ncRNA in cellular function (20). Although the down regulated; Fig. 4A). As the cellular phenotype of cells recent GENCODE annotated catalogue details an abun- with AB074278-siRNA is increased apoptosis and reduced dance and the distribution of many lncRNAs, little is known proliferation, we selected the 87 of 471 mRNAs with roles in about their function. Here, we have performed the first large proliferation, cell death, and apoptosis (gene functions comprehensive screen of lncRNAs in urothelial cancer to identified using DAVID Bioinformatics Resources, Vsn. identify those likely to play roles in the biology of this 6.7; ref. 36; bold font in Supplementary Table S3). We cancer. Putative lncRNAs were identified using an algorithm identified 3 of 87 mRNAs (EMP1, CKS2, and PTGS2) scoring characteristics of protein-coding genes, including correlated to AB074278 expression in the discovery NCode open reading frame length, synonymous/nonsynonymous microarray dataset (Pearson r < 0.35 or >0.2, P < 0.001; base substitution rates, and similarity to known proteins Supplementary Table S4). We measured the expression of (27). We identified differential expression of many these three mRNAs and TANC2 (the host protein-coding lncRNAs, without selection for genetic location, and found gene for AB074278) in the validation cohort of 138 urothe- that the overall pattern of altered expression (mostly upre- lial samples (Supplementary Fig. S4) to look for associa- gulation) was more extensive to that seen for protein coding tions with disease phenotype. We identified that low expres- mRNAs. Recently Wang and colleagues reported balanced sion of epithelial membrane protein 1 (EMP1) in urothelial changes in lncRNA expression in 12 urothelial cancer and cancer was associated with increased risk of progression and preliniary data to suggest lncRNAs have malignant roles in BC-specific mortality (Fig. 4B, 20% vs. 38% progression rate mTOR and p53 signaling, and other cancer pathways (24). for tumors with high and low expression, respectively, log- Although our observations do not suggest such a balanced rank P < 0.03) and EMP1 was down regulated in urothelial alteration in expression, data in breast and neurologic cancer (Fig. 4C; ANOVA P ¼ 0.005), in contrast with the cancers (23) and in human primary keratinocytes (35) changes seen for AB07428 [worse outcomes with high support our findings. Given a lack of knowledge about expression (log-rank P < 0.05) and upregulation in urothe- lncRNAs and that our panel was designed computationally, lial cancer (ANOVA P ¼ 0.02)]. To support a direct regu- we focused upon the minority (17%) of lncRNAs catalo- lation, we also saw increased EMP1 expression (3.8 fold) gued within GENCODE (22). This percentage compares following AB074278 knockdown in NHU-TERT cells (Fig. with the 12% overlap for lincRNAs reported by Kapranov

OF6 Clin Cancer Res; 2014 Clinical Cancer Research

Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2014 American Association for Cancer Research. www.aacrjournals.org Table 2. Detailed information on all lncRNA associated with tumor progression Downloaded from Expression Progression Genebank ID Protein coding score UCC expressiona (n)b rateb Log lncRNA Size [nt] Exons GeneID scorec CPC scorec Genecode subcategory Phenotype FC High Low High Low rank (P) AK122774 2,196 1 2.42 1.04 Exonic antisense/divergent All 0.2 27 29 7% 45% 0.001 Published OnlineFirstAugust27,2014;DOI:10.1158/1078-0432.CCR-14-0706 AK055767 2,282 1 0.26 1.04 Intergenic opposite strand Inv 2.2 28 28 46% 7% 0.001

clincancerres.aacrjournals.org uc004dpc 2,001 1 0.28 1.14 Intergenic same strand Inv 0.5 28 28 11% 43% 0.004 AF009302 206 1 2.64 1.38 Intronic sense Inv 2.2 27 29 44% 10% 0.005 AK026718 2,193 1 2.52 1.30 Intergenic same strand UCC only 0.4 27 29 11% 41% 0.007 AY174161 239 1 4.22 1.08 Intergenic same strand Inv 0.5 27 29 11% 41% 0.009 CR611332 5,304 1 61.16 2.45 Exonic sense Inv 0.5 27 29 11% 41% 0.010 AK096965 2,407 1 1.11 0.79 Intronic sense UCC only 0.4 28 28 11% 43% 0.012 uc002ibf 7,329 4 1.03 0.94 Exonic antisense LG, HG 0.5 28 28 11% 43% 0.014 AK022361 1,774 1 1.15 1.25 Intronic sense UCC only 0.5 27 29 11% 41% 0.014 AB074278 1,714 1 2.46 1.37 Intronic sense All 3.2 28 28 39% 14% 0.014 AF043897 3,949 1 1.22 0.64 Exonic sense Inv 0.4 27 29 11% 41% 0.015 Cancer Research. AY387664 1,041 1 2.60 1.13 Intronic antisense Inv 2.3 27 29 41% 14% 0.020

on September 24, 2021. © 2014American Association for uc004cem 2,906 1 17.86 0.24 Exonic antisense LG 0.5 28 28 14% 39% 0.021 uc003uxn 4,533 6 17.52 0.69 Exonic antisense LG, Inv 2.2 27 29 41% 14% 0.022 AK130230 1,538 1 1.58 1.07 Exonic antisense All 3.1 27 29 41% 14% 0.022 AY387665 995 3 1.28 0.64 Intronic antisense Inv 2.2 28 28 39% 14% 0.025 G36954 457 1 1.82 1.10 Intergenic same strand Inv 2.3 27 29 41% 14% 0.025 AK127730 5,015 1 2.00 0.75 Intergenic opposite strand All 4.0 28 28 39% 14% 0.026 uc003eiy 253 1 1.67 1.18 Exonic antisense Inv 2.1 27 29 41% 14% 0.027 AK128855 4,453 1 0.88 1.02 Intronic sense LG, Inv 0.4 27 29 15% 38% 0.028 Cancer Bladder in Regulation Epigenetic Integrated uc002nsx 2,282 1 0.00 0.94 Intergenic same strand Inv 2.1 28 28 39% 14% 0.030 BC015007 1,523 1 2.99 0.88 Exonic antisense LG, Inv 0.4 27 29 15% 38% 0.033 uc001qjw 7,375 2 0.57 1.12 Intergenic same strand Inv 2.2 28 28 39% 14% 0.037 AK129609 1,830 1 1.13 1.25 Intronic sense LG, HG 2.1 27 29 41% 14% 0.038 BC033908 3,868 3 0.33 1.14 Intergenic same strand Inv 2.3 27 29 41% 14% 0.039 AF075063 557 1 1.87 1.35 Intergenic same strand All 0.2 27 29 15% 38% 0.041 AF086032 303 1 2.50 1.27 Exonic sense HG, Inv 0.4 27 29 15% 38% 0.042 AK124944 1,867 1 5.89 0.71 Exonic antisense Inv 2.1 28 28 39% 14% 0.043

lnCne e;2014 Res; Cancer Clin AF086164 561 1 2.63 1.39 Intergenic same strand LG, Inv 0.4 28 28 14% 39% 0.045 BX648713 3,726 1 7.90 0.90 Intronic sense All 2.3 27 29 41% 14% 0.046 AK022801 1,880 1 2.51 1.22 Intronic sense Inv 2.1 27 29 37% 17% 0.048

aUrothelial cancer expression [phenotype: in which phenotype was abnormal expression seen (LG, low-grade NMI; HG, high-grade NMI; Inv, invasive urothelial cancer). FC, fold change in urothelial cancer vs. normal urothelium]. bExpression (n): the number of cases with high or low expression and progression: the progression rates in these tumors, respectively. cProtein coding score calculated using GeneID and CPC (coding potential calculator). Negative values suggest a low protein coding risk. OF7 Published OnlineFirst August 27, 2014; DOI: 10.1158/1078-0432.CCR-14-0706

Peter et al.

A 100

Figure 3. Tumor progression with 50 respect to lncRNA expression. A, P = 0.026 P = 0.022 P = 0.014 tumor progression was analyzed with respect to low (dashed line) AK127730 AK130230 AB074278 0 and high (solid line) lncRNA expression (dichotomized around 100 median). Examples of aberrant up (AK127730, AK130230, and AB074278) and down (AF075063, 50 BC015007, and AK122774)

Progression-free survival Progression-free P = 0.001 P = 0.041 P = 0.033 lncRNA regulation in urothelial cancer (plotted by the Kaplan– AK122774 0 AF075063 BC015007 Meier method, log-rank P values 0408012004080120 0 40 80 120 shown). B, siRNA-mediated gene Follow-up (mo) knockdown of the lncRNA AB074278 led to 84% reduction of BCits expression in hTERT NHU cells 3.0 3.5 and caused cell death/apoptosis. Scr RNA Scr RNA Cell death was determined by 2.5 siRNA AB074278 siRNA AB074278 propidium iodide ﬂow cytometry 2.0 2.5 and apoptosis by caspase-3 activation and ﬂow cytometry. C, 1.5 cell proliferation, measured by MTT 1.0 1.5 assay, was also reduced following

Percentage knockdown of AB074278. 0.5 Fluorescence 0.0 0.5 AB074278 Apoptotic Dead cells 24 h 48 h 72 h expression cells

and colleagues (37), and suggests that much work is needed The most altered transcripts were found in invasive to fully map the lncRNA transcriptome. To date, various urothelial cancer. This was especially true for upregulated profiling strategies have been used to identify differential lncRNAs (64%), making the invasive phenotype distin- lncRNA expression in cancer [e.g., analysis of candidate guishable from the low- and high-grade tumors. This obser- ncRNA (38, 39), microarrays (40, 41), and transcriptome vation is in contrast with miRNA profiles within urothelial sequencing (23, 42)]. These reports identify that lncRNA cancer, which often share differences between high-grade expression is often tissue or disease specific (23). Our and invasive tumors (12), and suggests an exciting role to findings support these data and also identify that the aid pathologic disease staging of high-grade tumors. majority of lncRNAs have tumor phenotype-specific expres- Although our data require validation, we noted fewer pro- sion (71% of all aberrant expressed lncRNAS were pheno- gression events in our high-grade NMI tumors than typical type specific vs. 9% were common to all urothelial cancer). (usually around 25%; ref. 5), suggesting that a chance This reflects (but varies in extent) changes in mRNA expres- enrichment for more indolent disease may have affected sion (44% phenotype-specific expression vs. 26% common- our comparisons. We focused upon lncRNAs that play ly expressed), suggesting coordinated anatomical dysregu- potential roles in urothelial carcinogenesis, through select- lation (e.g., through regional chromosomal instability or ing those associated with tumor progression. Most of those regional epigenetic silencing) or direct interaction (43). To identified were phenotype specific, reflecting global trends explore this, we mapped adjacent RNAs in our dataset. We within urothelial cancer, and have not been reported pre- identified many (n ¼ 3,048) neighboring pairs where the viously in cancer. We used loss-of-function studies for expression was correlated (Pearson correlation r > 0.5 and AB074278 in NHU cells to explore carcinogenic roles. This P < 0.05). This relationship was strongest for genes within cell line has more intact cellular processes (e.g., epigenetic 3,000 base pairs (Supplementary Fig. S5a). There were 803 regulation) and fewer genetic events than most malignant lncRNA/mRNA pairs (26%), where both members were cell lines, and so is better to model the subtle impact of significantly differentially expressed in urothelial cancer. epigenetic changes on gene expression (changes from genet- Expression was directly correlated for the vast majority of ic events, such as chromosomal loss/amplification, may these pairs (82%; Supplementary Fig. S5b), suggesting dominate subtle modifications arising from epigenetic common transcriptional control or the epigenetic media- alterations). AB074278 was chosen because it was signifi- tion of mRNA expression by nearby lncRNAs (22, 35), cantly upregulated in all tumor phenotypes (its overexpres- antisense RNAs (44), or promoter-associated lncRNAs (45). sion in urothelial cancer samples could be confirmed by

OF8 Clin Cancer Res; 2014 Clinical Cancer Research

Integrated Epigenetic Regulation in Bladder Cancer

AB 100 Figure 4. Changes in gene High High expression in hTERT-NHU cells after knockdown of lncRNA AB074278. A, the altered Low Scr RNA siRNA AB074278 50 Low expression of 17,601 mRNAs was recorded using HG-U133 Plus 2.0 microarrays. Complete linkage EMP1 P = 0.02 EMP1 P = 0.03 hierarchical clustering was 0 performed after selecting mRNAs with >80% expression frequency (n 100 ¼ 536) and mean centering data Low Low using correlation similarity in Cluster 3.0. Arrow identiﬁes EMP1. UC Specific survival

B, tumor progression and urothelial survival Progression-free 50 cancer-speciﬁc mortality with High High respect to EMP1 and AB074278 expression (dichotomized around AB074278 P = 0.05 AB074278 P = 0.04 the median), was analyzed and 0 plotted by the Kaplan–Meier 0 50 100 150 0 50 100 150 method, log-rank P values shown. Follow-up (months) C, expression of EMP1 and AB074278 in the validation cohort CD 5.0 of urothelial samples [box plots 15 P = 0.02 P = 0.005 show median and 95% CI DCt values (normalized to TERT 4.0 expression), ANOVA P values 10 shown]. D, expression of lncRNA 3.0 AB074278 after siRNA-mediated 5 knockdown and transfection with 2.0 EMP1 DCt control scrambled siRNA (Scr RNA) change Fold

AB074278 DCt 0 and corresponding changes in 1.0 expression of mRNA EMP1 after –3.00 –2.00 –1.00 knockdown of AB074278 as 0.00 –5 0.0 measured by qRT-PCR. 1.00 NU 2.00 NU INV 3.00 INV EMP1 LG NMI LG NMI Scr RNA HG NMI HG NMI AB074278

qPCR, data not shown), strongly associated with tumor NHU-TERT cells. EMP1 is a putative tumor suppressor gene progression, and has a low protein coding potential score. whose decreased expression is associated with advanced Furthermore, lncRNA AB074278 is intronic to TANC2, clinical stage and metastasis in oral squamous cell carcino- which was also upregulated in urothelial cancer, suggesting ma (48) and reported to be directly involved in the inhi- the potential for direct regulation. Although the expression bition of proliferation (47). To explore the link between of lncRNA AB074278 and TANC2 was not correlated AB074278 and EMP1, we compared the sequence of both in urothelial cancer cases, TANC2 expression did drop RNAs and EMP1-associated genes (identified through Inge- significantly following AB0724278 knockdown (Supple- nuity Pathway Analysis, Ingenuity Inc. and EpiTect ChIP mentary Fig. S6), allowing for the speculation that lncRNA qPCR Primers by Transcription Factor search algorithm by AB074278 may regulate (by enhancing) the expression of SABiosciences) using BLAT (UCSC, GRCh37/hg19) to TANC2 in cis (35). Little is known about the function of investigate potential direct interactions. We found a 21nt TANC2, although it is believed to play a role in embryonic stretch of identical sequence for AB074278 and SND1 development (46). Although these data are promising, we (Staphylococcal nuclease domain-containing protein 1; were keen to select mRNAs whose expression was abnormal Supplementary Table S5). SND1 is a transcription cofactor in urothelial cancer and correlated to AB074278. Using two that regulates EMP1 through its interaction with EMP1’s microarray screens, we identified a dynamic correlation transcription factor STAT5A (Supplementary Fig. S7 and between AB074278 and EMP1, and selected this gene due Supplementary Materials S1–S3). These data suggest but do to its potential involvement in disease progression and its not confirm the exact relationship between AB074278 and role in proliferation consistent with our observations. Our EMP1, which now requires further investigation. expression data of EMP1 matched its functional description In summary, we have identified many lncRNAs signifi- in the literature (overexpression of EMP1 was found to cantly altered in urothelial cancer and associated with inhibit the proliferation of EC9706 cells; ref. 47) and could disease progression and tumor subtypes. We specifically explain the observed decrease in proliferation in transfected implicate AB074278 in apoptosis avoidance and cell

www.aacrjournals.org Clin Cancer Res; 2014 OF9

Peter et al.

proliferation, potentially through regulating the expression Acknowledgments of EMP1. The authors thank Ian Sudbery, University of Oxford and Magnus Rattray, Manchester University for help with microarray annotation, and Dr. Paul Heath University of Sheffield for help with microarray studies. Disclosure of Potential Conflicts of Interest No potential conflicts of interest were disclosed. Grant Support Authors' Contributions J.W.F. Catto was supported by a GSK Clinician Scientist fellowship and Conception and design: S. Peter, J.W.F. Catto project/program grants from Yorkshire Cancer Research (Grant number Development of methodology: S. Peter, R.M. Drayton, C.P. Rakhit, S305PA), Astellas Educational Foundation, and the European Union A.P. Noon, J.W.F. Catto (European Community’s Seventh Framework Programme; grants FP7/ Acquisition of data (provided animals, acquired and managed patients, 2007-2013 and HEALTH-F2-2007-201438). E. Borkowska was supported provided facilities, etc.): S. Peter, E. Borkowska, C.P. Rakhit, A.P. Noon, by the Ministry of Science and Higher Education, Poland (grant 617/ W. Chen MOB/2011/0). Analysis and interpretation of data (e.g., statistical analysis, biosta- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked tistics, computational analysis): S. Peter, E. Borkowska, R.M. Drayton, advertisement C.P. Rakhit, J.W.F. Catto in accordance with 18 U.S.C. Section 1734 solely to indicate Writing, review, and/or revision of the manuscript: S. Peter, J.W.F. Catto this fact. Administrative, technical, or material support (i.e., reporting or orga- nizing data, constructing databases): S. Peter Received March 24, 2014; revised August 7, 2014; accepted August 10, Study supervision: A.P. Noon, J.W.F. Catto 2014; published OnlineFirst August 27, 2014.

References 1. Chavan S, Bray F, Lortet-Teulent J, Goodman MM, Jemal A. Interna- 15. Dyrskjot L, Ostenfeld MS, Bramsen JB, Silahtaroglu AN, Lamy P, tional variations in bladder cancer incidence and mortality. Eur Urol Ramanathan R, et al. Genomic profiling of microRNAs in bladder 2014;66:59–73. cancer: miR-129 is associated with poor outcome and promotes cell 2. Svatek RS, Hollenbeck BK, Holmang S, Lee R, Kim S, Stenzl A, et al. death in vitro. Cancer Res 2009;69:4851–60. The economics of bladder cancer: Costs and considerations of caring 16. Martens-Uzunova ES, Bottcher R, Croce CM, Jenster G, Visakorpi T, for this disease. Eur Urol 2014 Jan 21. [Epub ahead of print]. Calin GA. Long noncoding RNA in prostate, bladder, and kidney 3. Cazier JB, Rao SR, McLean CM, Walker AL, Wright BJ, Jaeger EE, et al. cancer. Eur Urol 2014;65:1140–51. Whole-genome sequencing of bladder cancers reveals somatic 17. Catto JW, Alcaraz A, Bjartell AS, De Vere White R, Evans CP, Fussel S, CDKN1A mutations and clinicopathological associations with muta- et al. MicroRNA in prostate, bladder, and kidney cancer: a systematic tion burden. Nat Commun 2014;5:3756. review. Eur Urol 2011;59:671–81. 4. Linton KD, Rosario DJ, Thomas F, Rubin N, Goepel J, Abbod MF, 18. Michlewski G, Guil S, Semple CA, Caceres JF. Posttranscriptional et al. Disease specific mortality in patients with low risk bladder regulation of miRNAs harboring conserved terminal loops. Mol Cell cancer and the impact of cystoscopic surveillance. J Urol 2013; 2008;32:383–93. 189:828–83. 19. Ponting CP, Oliver PL, Reik W. Evolution and functions of long 5. Thomas F, Noon AP, Rubin N, Goepel J, Catto JW. Comparative noncoding RNAs. Cell 2009;136:629–41. outcomes of primary, recurrent and progressive high-risk non-muscle 20. Spizzo R, Almeida MI, Colombatti A, Calin GA. Long non-coding RNAs invasive bladder cancer. Eur Urol 2013;63:145–54. and cancer: a new frontier of translational research? Oncogene 6. Drayton RM, Dudziec E, Peter S, Bertz S, Hartmann A, Bryant HE, et al. 2012;31:4577–87. Reduced expression of microRNA-27a modulates cisplatin resistance 21. Mercer TR, Dinger ME, Mattick JS. Long non-coding RNAs: insights in bladder cancer by targeting the cystine/glutamate exchanger into functions. Nat Rev Genet 2009;10:155–9. SLC7A11. Clin Can Res 2014;20:1990–2000. 22. Derrien T, Johnson R, Bussotti G, Tanzer A, Djebali S, Tilgner H, et al. 7. Su SF, de Castro Abreu AL, Chihara Y, Tsai Y, Andreu-Vieyra C, The GENCODE v7 catalog of human long noncoding RNAs: analysis of Daneshmand S, et al. A panel of three markers hyper- and hypomethy- their gene structure, evolution, and expression. Genome Res 2012;22: lated in urine sediments accurately predicts bladder cancer recur- 1775–89. rence. Clin Cancer Res 2014;20:1978–89. 23. Gibb EA, Vucic EA, Enfield KS, Stewart GL, Lonergan KM, Kennett JY, 8. Dudziec E, Goepel JR, Catto JW. Global epigenetic profiling in bladder et al. Human cancer long non-coding RNA transcriptomes. PLoS ONE cancer. Epigenomics 2011;3:35–45. 2011;6:e25915. 9. Dudziec E, Gogol-Doring A, Cookson V, Chen W, Catto J. Integrated 24. Wang L, Fu D, Qiu Y, Xing X, Xu F, Han C, et al. Genome-wide screening epigenome profiling of repressive histone modifications, DNA meth- and identification of long noncoding RNAs and their interaction with ylation and gene expression in normal and malignant urothelial cells. protein coding RNAs in bladder urothelial cell carcinoma. Cancer Lett PLoS ONE 2012;7:e32750. 2014;349:77–86. 10. Wolff EM, Chihara Y, Pan F, Weisenberger DJ, Siegmund KD, Sugano 25. Chapman EJ, Hurst CD, Pitt E, Chambers P, Aveyard JS, Knowles MA. K, et al. Unique DNA methylation patterns distinguish noninvasive and Expression of hTERT immortalises normal human urothelial cells invasive urothelial cancers and establish an epigenetic field defect in without inactivation of the p16/Rb pathway. Oncogene 2006;25: premalignant tissue. Cancer Res 2010;70:8169–78. 5037–45. 11. Yates DR, Rehman I, Abbod MF, Meuth M, Cross SS, Linkens DA, et al. 26. Miah S, Dudziec E, Drayton RM, Zlotta AR, Morgan SL, Rosario DJ, Promoter hypermethylation identifies progression risk in bladder can- et al. An evaluation of urinary microRNA reveals a high sensitivity for cer. Clin Cancer Res 2007;13:2046–53. bladder cancer. Br J Cancer 2012;107:123–8. 12. Catto JW, Miah S, Owen HC, Bryant H, Dudziec E, Larre S, et al. 27. Mercer TR, Dinger ME, Sunkin SM, Mehler MF, Mattick JS. Specific Distinct microRNA alterations characterize high and low grade bladder expression of long noncoding RNAs in the mouse brain. Proc Natl Acad cancer. Cancer Res 2009;69:8472–81. Sci U S A 2008;105:716–21. 13. Cancer Genome Atlas Research N. Comprehensive molecular char- 28. Dudziec E, Miah S, Choudhury HMZ, Owen HC, Blizard C, Glover M, acterization of urothelial bladder carcinoma. Nature 2014;507: et al. Hypermethylation of CpG islands and shores around specific 315–22. MicroRNAs and mirtrons is associated with the phenotype and pres- 14. Papatsoris AG, Chrisofos M, Skolarikos A, Varkarakis I, Lekas A, Dellis ence of bladder cancer. Clin Cancer Res 2011;17:1287–96. A, et al. Upper urinary tract transitional cell carcinoma. A 10-year 29. Andersen CL, Jensen JL, Orntoft TF. Normalization of real-time experience. Tumori 2008;94:75–8. quantitative reverse transcription-PCR data: a model-based

OF10 Clin Cancer Res; 2014 Clinical Cancer Research

Integrated Epigenetic Regulation in Bladder Cancer

variance estimation approach to identify genes suited for normal- 39. Han Y, Liu Y, Gui Y, Cai Z. Long intergenic non-coding RNA TUG1 is ization, applied to bladder and colon cancer data sets. Cancer Res overexpressed in urothelial carcinoma of the bladder. J Surg Oncol 2004; 2013;107:555–9. 64:5245–50. 40. Perez DS, Hoage TR, Pritchett JR, Ducharme-Smith AL, Halling ML, 30. Blanco E, Parra G, Guigo R. Using geneid to identify genes. Curr Protoc Ganapathiraju SC, et al. Long, abundantly expressed non-coding Bioinformatics 2007;Chapter 4:Unit 4 3. transcripts are altered in cancer. Hum Mol Genet 2008;17:642–55. 31. Kong L, Zhang Y, Ye ZQ, Liu XQ, Zhao SQ, Wei L, et al. CPC: assess the 41. Silva JM, Perez DS, Pritchett JR, Halling ML, Tang H, Smith DI. protein-coding potential of transcripts using sequence features and Identification of long stress-induced non-coding transcripts that have support vector machine. Nucleic Acids Res 2007;35:W345–9. altered expression in cancer. Genomics 2010;95:355–62. 32. Ferraiuolo L, Heath PR, Holden H, Kasher P, Kirby J, Shaw PJ. 42. Cabili MN, Trapnell C, Goff L, Koziol M, Tazon-Vega B, Regev A, et al. Microarray analysis of the cellular pathways involved in the adaptation Integrative annotation of human large intergenic noncoding RNAs to and progression of motor neuron injury in the SOD1 G93A mouse reveals global properties and specific subclasses. Genes Dev model of familial ALS. J Neurosci 2007;27:9201–19. 2011;25:1915–27. 33. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays 43. Guttman M, Amit I, Garber M, French C, Lin MF, Feldser D, et al. applied to the ionizing radiation response. Proc Natl Acad Sci U S A Chromatin signature reveals over a thousand highly conserved large 2001;98:5116–21. non-coding RNAs in mammals. Nature 2009;458:223–7. 34. Hulsen T, de Vlieg J, Alkema W. BioVenn - a web application for the 44. Yu W, Gius D, Onyango P, Muldoon-Jacobs K, Karp J, Feinberg AP, comparison and visualization of biological lists using area-proportional et al. Epigenetic silencing of tumor suppressor gene p15 by its Venn diagrams. BMC Genomics 2008;9:488. antisense RNA. Nature 2008;451:202–6. 35. Orom UA, Derrien T, Beringer M, Gumireddy K, Gardini A, Bussotti G, 45. Kurokawa R. Promoter-associated long noncoding RNAs repress et al. Long noncoding RNAs with enhancer-like function in human transcription through a RNA binding protein TLS. Adv Exp Med Biol cells. Cell 2010;143:46–58. 2011;722:196–208. 36. Huang da W, Sherman BT, Tan Q, Collins JR, Alvord WG, Roayaei J, 46. Han S, Nam J, Li Y, Kim S, Cho SH, Cho YS, et al. Regulation of et al. The DAVID Gene Functional Classification Tool: a novel biological dendritic spines, spatial memory, and embryonic development by the module-centric algorithm to functionally analyze large gene lists. TANC family of PSD-95-interacting proteins. J Neurosci 2010;30: Genome Biol 2007;8:R183. 15102–12. 37. Kapranov P, St Laurent G, Raz T, Ozsolak F, Reynolds CP, Sor- 47. Wang HT, Kong JP, Ding F, Wang XQ, Wang MR, Liu LX, et al. ensen PH, et al. The majority of total nuclear-encoded non-ribo- Analysis of gene expression profile induced by EMP-1 in esoph- somal RNA in a human cell is 'dark matter' un-annotated RNA. BMC ageal cancer cells using cDNA Microarray. World J Gastroenterol Biol 2010;8:149. 2003;9:392–8. 38. Luo M, Li Z, Wang W, Zeng Y, Liu Z, Qiu J. Upregulated H19 contributes 48. Zhang J, Cao W, Xu Q, Chen WT. The expression of EMP1 is down- to bladder cancer cell proliferation by regulating ID2 expression. FEBS regulated in oral squamous cell carcinoma and possibly associated J 2013;280:1709–16. with tumor metastasis. J Clin Pathol 2011;64:25–9.

www.aacrjournals.org Clin Cancer Res; 2014 OF11

Identification of Differentially Expressed Long Noncoding RNAs in Bladder Cancer

Stefan Peter, Edyta Borkowska, Ross M Drayton, et al.

Clin Cancer Res Published OnlineFirst August 27, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-14-0706

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2014/08/27/1078-0432.CCR-14-0706.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://clincancerres.aacrjournals.org/content/early/2014/09/17/1078-0432.CCR-14-0706. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.