Full Text

Turn on search term navigation

ARTICLE

Received 5 Dec 2013 | Accepted 26 Sep 2014 | Published 21 Nov 2014

Dimple Chakravarty1,2, Andrea Sboner1,2,3, Sujit S. Nair4, Eugenia Giannopoulou5,6, Ruohan Li7, Sven Hennig8, Juan Miguel Mosquera1,2, Jonathan Pauwels1, Kyung Park1, Myriam Kossai1,2, Theresa Y. MacDonald1, Jacqueline Fontugne1,2, Nicholas Erho9, Ismael A. Vergara9, Mercedeh Ghadessi9, Elai Davicioni9,Robert B. Jenkins10, Nallasivam Palanisamy11,12, Zhengming Chen13, Shinichi Nakagawa14, Tetsuro Hirose15, Neil H. Bander16, Himisha Beltran1,2, Archa H. Fox7, Olivier Elemento2,3 & Mark A. Rubin1,2

ERa-regulated intergenic long non-coding RNAs (lncRNAs), we identied nuclear enriched abundant transcript 1 (NEAT1) as the most signicantly overexpressed lncRNA in prostate cancer. Analysis of two large clinical cohorts also revealed that NEAT1 expression is associated with prostate cancer progression. Prostate cancer cells expressing high levels of NEAT1 were recalcitrant to androgen or AR antagonists. Finally, we provide evidence that NEAT1 drives oncogenic growth by altering the epigenetic landscape of target gene promoters to favour transcription.

1 Department of Pathology and Laboratory Medicine, Weill Medical College of Cornell University, 413 East 69th Street, Room 1402, New York, New York 10021, USA. 2 Institute for Precision Medicine, Weill Medical College of Cornell University and New York Presbyterian Hospital, New York, New York 10021, USA. 3 Institute for Computational Biomedicine, Weill Cornell Medical College of Cornell University, New York, New York 10021, USA. 4 Department of Biochemistry and Molecular Medicine, School of Medicine and Health Sciences, George Washington University, Washington DC 20037, USA. 5 Biological Sciences Department, New York City College of Technology, City University of New York, Brooklyn, New York 11201, USA. 6 Arthritis and Tissue Degeneration Program and the David Z. Rosensweig Genomics Research Center, Hospital for Special Surgery, New York, New York 10021, USA. 7 School of Biomedical, Biomolecular and Chemical Sciences, University of Western Australia, Crawley, Western Australia 6009, Australia. 8 Chemical Genomics Centre, 44227 Dortmund, Germany. 9 Research and Development, GenomeDx Biosciences, Vancouver, British Columbia, Canada V6B 1B8. 10 Department of Pathology and Laboratory Medicine, Mayo Clinic, Rochester, Minnesota 55905, USA. 11 Michigan Center for Translational Pathology, University of Michigan, Ann Arbor, Michigan 48105, USA. 12 Henry Ford Health System, Medical Group Urology - Prostate Cancer Research, 1 Ford Place, Room 2D26, Detroit, Michigan 48202, USA. 13 Division of Biostatistics and Epidemiology, Department of Public Health, Weill Cornell Medical College, New York, New York 10021, USA. 14 RNA Biology Laboratory, RIKEN Advanced Research Institute, Hirosawa 2-1, Wako 351-0198, Japan. 15 Institute for Genetic Medicine, Hokkaido University, Kita-15 Nishi-7, Kita-ku, Sapporo 060-0815, Japan. 16 Department of Urology, Weill Cornell Medical College of Cornell University, New York, New York 10021, USA. Correspondence and requests for materials should be addressed to M.A.R. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

DOI: 10.1038/ncomms6383 OPEN

The oestrogen receptor alpha-regulated lncRNA NEAT1 is a critical modulator of prostate cancer

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383

Steroid receptors are key transducers of hormone signalling and control a wide spectrum of tissue-specic functions that are critical for the physiological homeostasis of reproductive

organs. Aberrant or deregulated expressions of steroid nuclear receptors are often associated with cancer progression and have been a major target for therapeutic intervention. The androgen receptor (AR) plays a central role in the progression of prostate cancer1. Androgen ablation is highly effective in treating metastatic prostate cancer, although resistance inevitably develops leading to castrate-resistant prostate cancer (CRPC). Most cases of CRPC remain dependent on AR signalling, which has led to the clinical development and recent approval of potent AR-targeted therapies for CRPC (that is, abiraterone and enzalutamide)2,3. However, similar to rst-generation anti-androgen therapies, patients develop resistance to these second-generation hormonal therapies. How CRPC tumours bypass AR signalling is emerging as a signicant area of investigation. Many view co-targeting therapies as an important next step to managing the inevitable emergence of resistance to single-agent treatments, but critical to co-targeting is the identication of other biological pathways that drive disease progression and the development of strategies that can target judgmental pathways.

In CRPC, cross-talk between oestrogen- and androgen-signalling pathways may present an opportunity for clinical intervention. Oestrogen receptor (ER) signalling through ERa increases with prostate cancer progression46 and can drive important oncogenic events, including TMPRSS2-ERG expression7. Although ERa signalling has been extensively studied in breast cancer810, our understanding of the potential impact of this nuclear receptor on prostate physiology is less clear. Nevertheless, the connection is a particularly intriguing concept given that most cases of prostate cancer arise in the sixth decade of life, a time when testosterone levels are decreasing and oestrogens are increasing in men. Mouse models suggest that antagonism of ERa may diminish prostate carcinogenesis4.

We posit that ERa is an important alternate signalling pathway for the transcriptional regulation of prostate cancer, allowing refractory disease to bypass androgen/AR signalling. Herein, we provide experimental evidence to support this hypothesis and demonstrate a functional specialization and distinct genomic role of this nuclear receptor in prostate cancer, with signicant implications for prognosis and management. We show that ERa is recruited to both coding and non-coding regions of the prostate genome and orchestrates expression of non-coding regulatory RNAs.

We identied nuclear enriched abundant transcript 1 (NEAT1) long non-coding RNA (lncRNA) as a potential target of ERa and as an important mediator for maintenance of prostate cancer. NEAT1 functions as a transcriptional regulator and contributes to a cancer-favourable transcriptome, thereby promoting tumor-igenesis in experimental animal models. Our analysis of the transcriptional role of NEAT1 identied functions beyond its previously characterized role in maintaining the integrity of subnuclear organelles called paraspeckles5. We demonstrate that NEAT1 is recruited to the chromatin of well-characterized prostate cancer genes and contributes to an epigenetic on state. Analysis of two large clinical cohorts nominated NEAT1 as a novel biomarker of disease progression. Given its signicance within the ERa signalling pathway, we propose that targeting

NEAT1 might represent a novel and important therapeutic strategy for the treatment of prostate cancer.

ResultsERa in transcriptional regulation of prostate cancer. To elucidate the role of ERa in prostate cancer, we analysed ERa protein and transcript levels in a panel of prostate cancer cell lines

(n 5) and in a cohort of matched benign prostate tissue (n 14)

and prostate adenocarcinoma (PCa) (n 14), respectively. We

observed that ERa was signicantly upregulated (P 0.03) in

prostate tumours compared with benign tissues (Fig. 1a). To determine the clinical relevance of ERa in prostate cancer, we performed immunohistochemistry using a tissue microarray composed of tissue cores from 64 samples of benign prostate tissue, 16 high-grade prostate intraepithelial neoplasia, 292 PCa, and 42 neuroendocrine prostate cancer (NEPC). Representative photomicrographs are depicted in Supplementary Fig. 1a. Although benign prostate had only low expression levels of ERa,

ERa was detected in adenocarcinoma and the adjacent high-grade prostate intraepithelial neoplasia through focal nuclear and cytoplasmic staining (Supplementary Fig. 1a). ERa is over-expressed in a signicant number of prostate cancer cohorts. It was also found to be overexpressed in prostate cancers with high Gleason score (GS) compared with those with low GS as well as in those with tumour recurrence when analysed via the Oncomine6 database (Fig. 1b)722. Analysis of subcellular distribution in prostate cancer cell lines revealed signicant nuclear distribution of ERa in all cell lines tested (Supplementary Fig. 1b). ERa protein levels were similar in both AR-positive LnCaP and VCaP cells (Fig. 1a, inset). We used parental VCaP and the ERa-positive prostate cancer cell line NCI-H660 as model cell lines to further explore and delineate the specic contribution of ERa to prostate cancer. A ligand-dependent modulation of invasive potential was observed in VCaP cells on oestrogen (E2) treatment (Fig. 1c). These results suggest that a functionally relevant, ligand-dependent ERa signalling pathway is active in prostate cancer cell lines.

To further understand the impact of ERa, we generated VCaP cells that overexpress ERa (VCaP ERa). Stable expression of ERa was conrmed by western blotting (Supplementary Fig. 1c). VCaP ERa exhibited signicantly higher invasive potential than

VCaP parental cells or the vector control cells (Fig. 1c). Intriguingly, the noted effects of ERa overexpression were independent of AR status, as experimental silencing of AR in VCaP ERa cells (Supplementary Fig. 1c) did not compromise the increased invasive potential of E2-treated VCaP ERa cells (Fig. 1c). These data suggest that prostate cancer cells can use alternate nuclear receptor signalling (for example, ERa signalling)

to propagate, and understanding these mechanisms will help discern the complete spectrum of key regulators of prostate cancer progression.

Studies have established ERas dominant role in transcriptional regulation of target genes in breast cancer23,24. Likewise, high nuclear levels of ERa in prostate cancer cells (Supplementary

Fig. 1b) and their direct association with chromatin implicate ERa in the transcriptional regulation of this cancer, as well. We used ERa chromatin immunoprecipitation coupled with high-throughput sequencing (ChIP-seq) in VCaP cells, with and without E2 treatment, and also in VCaP ERa and NCI-H660 cells with E2 treatment to investigate the underlying mechanisms by which ERa might drive a transcriptional programme in prostate cancer. The majority of ERa-binding sites were cell specic (Supplementary Fig. 1d). Analysis of the ChIP-seq data for ERa in NCI-H660 and VCaP ERa cells revealed that 64.9% of ERa binding occurred within intergenic regions of the prostate genome. This fraction is higher than the expected fraction if peaks were randomly distributed across the genome (P 3e 05)

(Supplementary Fig. 1e).

Using publicly available data sets25, we found that 28% of the intergenic ERa-binding sites in the prostate cancer genome (from

VCaP ERa and NCI-H660 cell lines) overlapped with the active histone marks trimethylated lysine 4 of histone H3 (H3K4me3) and trimethylated lysine 36 of histone H3 (H3K36me3)

2 NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383 ARTICLE

a b

3.5

MCF7

RWPE1

LnCaP

VCaP

NCI-H660

3 P=0.018 P=0.06

Arredouani

P=0.013

2.5

P=0.0260

Fold change

log2 (P-value)

1.5

-Actin

0.5

Grasso

Liu

Luo

Singh Lapointe

Tomlins

Glinsky

Vanaja

Liu Taylor

LaTulippe

Wallace

Best

TCGA Lapointe

Barwick

Benign prostate (n=14)

PCa (n=14)

Varambally

Cancer vs normal

Advanced Gleasons

Outcome

c d

** ** **

0.90

% Of input

0.80

0.70

% Invasion

0.60

0.50

0.40

0.30

0.20

0.10

0.00

IgG ER IgG ER IgG ER IgG ER IgG ER

P1 P2 P3 P4 P5

E2 +E2

1.60

VCaP

Ctrl-si

AR-si

Ctl-vector

ER+Ctrl-si

ER+AR-si

VCaP

Ctrl-si

ER+Ctrl-si

ER+AR-si

AR-si

Ctl-vector

1.40

1.20

% Of input

1.00

0.80

0.60

+E2

0.40

0.20

0.00

IgG ER IgG ER IgG ER IgG ER IgG ER IgG ER

P6 P7 P8 P9 P10 P11

P=0.00002

Log 2 (RPKM+1)Log2 (RPKM+1)Log 2 (RPKM +1)

RefSeq,

GENCODE,

ncRNA.org,

Mattick et al.

Known

lncRNAs

73 Prostate

samples

(PE-RNA-SEQ)

2,000

1,000

NEAT1

Chr 11q13.1

NEAT1

Fold change

P=0.004

P<0.01

P <0.01

(12,143)

Benign

PCa

Identification of differentially expressed lncRNAs

P=0.00007

P=0.004

NR_024490

FR349599

8 76 54 3

3.5

2.5

Chr 15q 21.2

NR_024490

0 Benign prostate (n=14)

PCa(n =14)

LnCaP

VCaP

Histone

marks

H3K4 - H3K36

H3K9 - H3K27

binding

sites

(3,819)

VCaP ER

NCI-H660

Benign

PCa

PCa (n=14)

Identification of differentially expressed ER regulated lncRNAs (140)

Chr 22q 11.21

FR349599

Benign

PCa

0 Benign prostate (n=14)

PCa (n=14)

Figure 1 | ERa plays a distinct role in prostate cancer. (a) ERa is upregulated in prostate cancer compared with matched benign controls. Waterfall plots depict the qRT-PCR expression levels of ERa mRNA in an independent cohort of benign (n 14) and PCa (n 14). (inset A) The expression of ERa in

different prostate cancer cell lines was determined by western blotting and compared with MCF7, a breast cancer cell line. (b) Analysis of ERa expression in Oncomine public data sets of normal versus prostate cancer and advanced disease. (c) Invasion of VCaP and VCaP ERa cells analysed 48 h post treatment with vehicle control or E2 (10 nM) in the presence of control or AR-siRNA. Results are expressed as the means.d. of three independent experiments. Students t-test was performed for comparisons (% Invasion) between E2 and E2 conditions for ERa, ERa-Ctrl siRNA and AR-siRNA, and

**Po0.01 were considered statistically signicant. (d) Recruitment of endogenous ERa to target gene chromatin was analysed in VCaP cells with or without E2 treatment. Results are expressed as the percentage of input of two independent experiments. Error bars represent the range of data. (e) Computational pipeline for identication of ERa-regulated lncRNAs upregulated in prostate cancer: a schematic overview of the methodology employed to identify ERa-regulated lncRNAs that are differentially expressed between benign versus prostate cancer and prostate cancer versus NEPC. (f) Box plots show expression levels of the top three ERa-regulated lncRNAs from 26 benign and 40 PCa cases, with ideogram depicting their chromosomal position. Waterfall plots depict the qRT-PCR expression levels on an independent cohort of benign (n 14) and PCa (n 14) of the three nominated lncRNAs: NEAT1, NR_024490

and FR349599.

(Po1e 7). On the other hand, 20.7% of those sites overlapped

with histone marks typical of inactive chromatin, such as trimethylated lysine 9 of histone H3 or trimethylated lysine 27 of histone H3. To prioritize experimental validation of ERa

targets, we ranked the peaks according to the average P-value determined by the peak-calling algorithm ChIPSeeqer26 and selected the highest ranking peaks for further analysis. We analysed recruitment of endogenous ERa to the top 11 binding

NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383

ERa-regulated NEAT1 lncRNA is upregulated in prostate cancer. In the Oncomine database, we observed signicant overexpression of NEAT1 lncRNA in several prostate cancer data sets (normal versus cancer) and aggressive prostate cancer (Fig. 2a)7,1022,2831. We rst conrmed that amplication of chromosome 11q (where NEAT1 resides) was not seen across 109 adenocarcinoma cases32, eliminating chromosome 11q13.1 amplication as an explanation for high NEAT1 expression (Supplementary Fig. 2a)33,34. The expression of NEAT1 in two radical prostatectomy cohorts with long-term clinical follow-up from the Mayo Clinic35,36 was measured using Affymetrix HuEx microarrays (see Methods). Supplementary Table 1 contains the patient characteristics of the data sets. NEAT1s expression ranked in the 99th percentile of all genes on the microarray (Fig. 2b). We determined levels of NEAT1 by RNA in situ hybridization (ISH) in a tissue microarray that included 16 benign prostate tissues, 21 PCa, 12 PCa with neuroendocrine differentiation and 7 NEPC cases. NEAT1 was found to be highly expressed in prostate cancer compared with that in benign tissue (Supplementary Fig. 2b).

We observed that in a panel of prostate cancer cell lines, ERa overexpression and E2 treatment upregulated NEAT1 transcript levels in a time-dependent manner (Fig. 2c). In DU145, an ERG-negative cell line37, E2/ERa signalling was intact (Fig. 2c), supporting an ERG-independent phenomenon. Following ERa overexpression, we also recorded an increase in expression of the long isoform NEAT1_2 (Supplementary Fig. 2c). This was not surprising as both isoforms of NEAT1 are driven by the same promoter38. The preferential upregulation and increase in the NEAT1 long form alone is not well understood and is not further addressed in this study. Interestingly, knockdown of ERb did not alter NEAT1 levels, suggesting that NEAT1 regulation is specic for ERa (Supplementary Fig. 2d).

NEAT1 was originally identied localized to subnuclear organelles called paraspeckles that are free of chromatin and function as repositories of edited RNA and a number of nuclear RNA-binding proteins5. Loss of NEAT1 dramatically reduces the formation of paraspeckles. Treatment of VCaP cells with E2 resulted in re-distribution of NEAT1 from paraspeckles to an enhanced distribution throughout the nucleus (Supplementary Fig. 2e).

We inspected our ERa ChIP-seq data in VCaP ERa and NCIH660 cells, and identied two ERa-binding sites on the NEAT1 promoter (Fig. 2d). Analysis of chromatin marks using ChIP-seq data sets for histone marks25 revealed the presence of active histone marks H3 Acetyl K9 and H3K4me3 in the promoter region of NEAT1, while H3K36Me3 marks were abundant in the

Figure 2 | ERa regulated NEAT1 lncRNA is upregulated in prostate cancer. (a) NEAT1 is overexpressed in various prostate data sets (Oncomine).(b) Distribution of the median expression of all genes (core transcript clusters) on the Human Exon 1.0 STarray in the pooled Mayo Clinic cohort (n 594).

NEAT1s expression ranks in the 99th percentile of all genes on the array. (c) Expression of NEAT1 with/without ERa overexpression and E2 treatment (10 nM) at different time points in a panel of prostate cancer cell lines. Results are expressed as the means.d. of three independent experiments.

(d) View of NEAT1 genomic location indicates presence of two ERa-binding sites in the promoter region. Read coverage tracks derived from RNA-sequencing data indicates a higher abundance of NEAT1 transcripts in PCa compared with benign tumours in three representative cases. The gure also reports the ChIP-sequencing coverage tracks for ERa (VCaP ERa, VCaP and input DNA as control). The bottom panel shows the binding sites of ERa, AR (GEO Accession GSM353651-tissue AR (ref. 25)), Ace-H3, H3K4me3 and H3K36me3 in VCaP cell line (GEO Accession GSM353629, GSM353620 and

GSM353624 (ref. 25)), respectively. (e) Chromatin immunoprecipitation followed by qPCR to study ERa recruitment to NEAT1 promoter in VCaP cells with/without E2 treatment (10 nM) was performed with primers spanning the binding regions identied by ERa ChIP-seq data. Primers for nonspecic region were used as negative control for ChIP studies. Results are expressed as percentage of input from two independent experiments. Vertical error bars represent the range of data. (f) Luciferase-based promoter reporter assays was used to analyse effect of ERa and/or AR on ERE-Luc promoter in VCaP cells. Cells were transiently transfected with the (ERE)3-SV40-luc reporter plasmid and ERa, or AR-treated with/without E2 or R1881 (1 nM) for 48 h.

Results are expressed as the means.d calculated from three independent experiments. (g) Luciferase-based promoter reporter assays were used to analyse NEAT1 promoter activity following ERa expression / E2 (10 nM) for 24 h. Results are expressed as the meanss.d. calculated from three

independent experiments. Students t-test was performed for comparisons where indicated, and *Po0.05 and **Po0.01 were considered statistically signicant.

sites in parental VCaP cells (Fig. 1d), providing an experimental validation of the ChIP-seq data. A signicantly higher recruitment of ERa was evident at the binding sites compared with control IgG.

Given the enhanced recruitment of ERa to intergenic regions of the prostate genome, we evaluated the likelihood that ERa might inuence transcriptional output and thereby the repertoire of non-coding RNA in the context of prostate cancer. We thus analysed the abundance of non-coding transcripts in RNA-seq data derived from a cohort of 73 prostate tissues, which included 26 benign prostate samples, 40 PCa and 7 NEPC (Supplementary Dataset 1), focusing our analysis on 6,850 intergenic lncRNAs out of 12,143 known lncRNAs (see Supplementary Methods). We identied 1,314 and 1,399 intergenic lncRNAs that are differentially expressed between benign and PCa, and between PCa and NEPC, respectively (false discovery rate o0.01). We identied 140 intergenic lncRNAs putatively regulated by ERa (Fig. 1e and Supplementary Dataset 2). An analysis of AR-binding sites25 identied 98 lncRNAs that have an AR-binding site within the promoter. This supported the view that ERa might signicantly inuence the non-coding transcriptome in prostate cancer. Using the RNA-seq data on VCaP and VCaP ERa cell lines to validate the expression levels of the top differentially expressed ERa-regulated lncRNAs, we selected six potential candidate lncRNAs that had higher expression in VCaP ERa compared with VCaP. We used quantitative real-time PCR (qRTPCR) to validate expression for these six ERa-regulated lncRNAs in VCaP and VCaP ERa-expressing cell lines (Supplementary Fig. 1f). Expression of three of these lncRNAs was further determined in a cohort of 28 matched benign and prostate cancer samples, conrming upregulation of these three nominated lncRNAs in prostate cancer compared with benign prostate (Fig. 1f). Taken together, these analyses indicate that ERa is a transcriptional regulator of the non-coding transcriptome in prostate cancer.

Among the putatively ERa-regulated intergenic lncRNAs, we identied NEAT1 as the most signicantly overexpressed lncRNA in prostate cancer versus benign prostate in our patient cohort (73 samples) (Fig. 1f and Supplementary Dataset 2). The NEAT1 gene is located on chromosome 11q13.1 and produces two RNA isoforms that overlap completely at the 50-end. The shorter isoform (hereafter abbreviated as NEAT1/NEAT1_1) is 3.7 kb in length and more abundant than the longer, 23 kb isoform (NEAT1_2)27. NEAT1 lncRNA is essential for the formation of subnuclear bodies called paraspeckles27, and although both isoforms localize to paraspeckles, their physiological role in prostate cancer remains unknown.

4 NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383 ARTICLE

gene body (Fig. 2d). A recent study revealed that bivalent H3K4Me3 and H3K36Me3 marks are indicators of functional transcriptional loci from the non-coding genome39. ERa recruitment to specic regions of the NEAT1 promoter was independently validated by ERa ChIP in VCaP, VCaP ERa and

NCI-H660 cells (Fig. 2e and Supplementary Fig. 2f,g) using specic primers encompassing ERa-binding sites in the NEAT1 promoter. We found that a functional oestrogen/ERa signalling

pathway was active in VCaP cells, as determined by reporter-based estrogen response element (ERE) luciferase assays in VCaP cells, with ERa and AR overexpression, and E2 or R1881 treatment, respectively, for 48 h (Fig. 2f). To further test whether ERa is required for NEAT1 transcriptional activation, we generated luciferase promoter reporter constructs with both ERa-binding sites upstream of the luciferase-coding region.

Luciferase reporter assays in VCaP cells conrmed that NEAT1

a b

log2 (P-value)

0.30

P-value = 0.0007 P-value = 0.02

0.20

Density

0.10

NEAT1

Lapointe

Grasso

Wallace

Singh

Welsh

Luo

LaTulippe

Vanaja

Holzbeierlein

Tomlins

Arredouani

Luo

Lapointe

Wallace

Glinsky

Tomlins

Vanaja

Grasso

Magee

Varambally

Tomlins

Tamura

Taylor

0.00

10.6

2 4 6 8 10 12 14 log2(Expression)

Advanced Gleasons score vs low Gleasons score

P-value = 0.006

Metastasis vs primary

Cancer vs normal

Tumour 1 Tumour 2 Tumour 3

Benign 3

Benign 1 Benign 2

AR Ace-H3

NEAT1 mRNA

Levels +E2/E2

RefSeq gene

H3K36me3

H3K4me3

6 h

14 h

24 h

48 h

6 h

14 h

24 h

48 h

6 h

14 h

24 h

48 h

6 h

14 h

24 h

48 h

6 h

14 h

24 h

48 h

DU145+ER LnCaP+ER PC3+ER VCaP+ER NCI-H660

VCaP ER

VCaP

VCaP Input

1.20

1.00

0.80

% Of input

0.60

0.40

0.20

0.00

IgG

Region 1 Region2 Region 3 Region 4 NT-Region

VCaP

E2 48h

E2 14h E2 E2 48h

E2 14h

IgG

E2 E2 48h

E2 14h

IgG

E2 E2 48h

E2 14h

f g

0.25

0.2

E2 +E2 R1881 + R1881

Relative luc units

0.15

0.1

Relative luc units

1000

0.05

pGL3-Vector +

0 ERE

ERE+ER

ERE+AR

pGL3-NEAT1(R1+R2) +

E2 +

+ +

ERE+NEAT1

ERE+NEAT1+ER

ERE+NEAT1+AR

NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383

promoter activity was upregulated in an ERa-dependent manner and further enhanced with E2 treatment (Fig. 2g).

ERa and NEAT1 regulate several prostate cancer genes. We next sought to understand the physiological role of NEAT1 and to determine the downstream targets of the ERa-NEAT1 axis in prostate cancer. We were particularly interested in identifying genes signicantly deregulated in prostate cancer and positively correlated with ERa and NEAT1 expression. Transcriptome sequencing of VCaP and VCaP ERa cells and pairwise comparison revealed 588 genes to be upregulated in VCaP ERa cells (log2-fold change 42) (Supplementary Dataset 3 and Fig. 3a). We performed a comparative analysis of this 588 gene signature using Oncomine concept analysis. We focused on data sets from prostate cancer studies that included both prostate tumour and benign prostate tissues. The analysis revealed that the ERa gene signature was signicantly upregulated in a number of prostate cancer data sets, but was downregulated in other non-prostate data sets, indicating that ERa regulates prostate cancer-specic genes (Fig. 3b and Supplementary Dataset 4).

To validate whether ERa targets identied by in silico analysis are dependent on cellular levels of ERa, we experimentally silenced ERa in VCaP cells using an small interfering RNA (siRNA) approach and determined transcript levels of ten target genes using qRTPCR. The target genes selected for validation were those genes that demonstrated the highest log2-fold difference in VCaP and VCaP ERa cells. Results indicated that messenger RNA levels of the target genes selected were dependent on ERa (Fig. 3c), suggesting a distinct contribution of ERa in determining the transcriptional programme.

NEAT1 is a downstream target in the ERa signalling pathway. After determining an ERa signature, we next investigated the potential role of NEAT1. Interestingly, knockout of NEAT1 compromised the expression of ERa target genes, suggesting that

NEAT1 is not only a downstream target but also a mediator of ERa signalling in prostate cancer cells (Fig. 3d). To evaluate this further and to determine whether a functional synergy between ERa and

NEAT1 pathways exists in prostate cancer cells, we performed RNA-seq of vector control and NEAT1-overexpressing VCaP cells to determine a NEAT1 signature. To achieve this, we limited our analysis to genes that were upregulated four-fold in NEAT1-expressing cells (Supplementary Dataset 5). Interestingly, the NEAT1 signature showed a strong correlation with the ERa signature genes (q 1.90E 120). Analysis of the top 1,000 genes of

the NEAT1 signature revealed that this signature is upregulated in prostate cancer data sets when compared with other cancer data sets (Fig. 3e and Supplementary Dataset 4). Furthermore, the NEAT1 signature was also upregulated in all prostate cancer data sets (comparing benign versus PCa; odds ratio 42.0 and

Po1 10 6) (Supplementary Fig. 3a).

We also queried Oncomine prostate data sets to identify genes whose mRNA levels correlate with those of NEAT1 (correlation coefcient 40.5). We compared this gene list with the ERa signature genes from our analysis in Fig. 3a and identied 155 genes in common. These 155 genes were also found to be upregulated in all prostate cancer data sets compared with other cancer data sets (only normal versus cancer data sets were considered; odds ratio 43.0 and Po1 10 6) (Supplementary

Dataset 4 and Supplementary Fig. 3b).

To determine whether the genes identied by in silico analysis are indeed inuenced by NEAT1, we silenced NEAT1 in VCaP cells and determined transcript levels of potential target genes using qRTPCR. We selected the top ten genes that were signicantly correlated to NEAT1 expression across all prostate cancer concepts. As expected, mRNA levels of these selected target genes were indeed dependent on NEAT1, further conrming a denite role of NEAT1 in the transcriptional programme (Fig. 4a). In addition to cell lines, we also determined transcript levels of these ERa-NEAT1 signature-selected genes in a small patient cohort (n 26) of 13 matched benign and PCa,

respectively. We observed that relative mRNA levels of these NEAT1-ERa signature-selected genes revealed signicant upregulation in prostate cancer (Fig. 4b). We computed the log2-fold change of expression levels using the 13 paired tumour/benign samples for NEAT1 and for these selected genes. We then correlated the fold change values and observed a moderate-to-strong correlation between NEAT1 and the associated genes in clinical samples (Fig. 4c). Among these seven genes, prostate-specic membrane antigen (PSMA) and alpha-methylacyl-CoA racemase (AMACR) are well-known diagnostic and, in the case of PSMA, prognostic markers of prostate cancer progression4044. Furthermore, knocking down ERb did not alter expression of key signature genes in LnCaP, PC3, VCaP and NCI-H660 cells (Supplementary Fig. 3c), suggesting a non-redundant regulatory role for ERa.

NEAT1 and chromatin regulation. To study the potential role of NEAT1 in regulation of target genes in vivo, we performed luciferase reporter assays using PSMA-luc as a candidate NEAT1 target. NEAT1 induced robust activation of the PSMA promoter in PC3 cells (Fig. 5a) and VCaP cells (Fig. 5b). These results prompted us to investigate whether NEAT1 is recruited to chromatin of target genes. We used the chromatin isolation by RNA purication (ChIRP) approach45 to pull down endogenous NEAT1 from VCaP cells. Analysis of the ChIRP data revealed that NEAT1 is recruited to the PSMA promoter, but not the downstream exon 1 (Fig. 5c). In addition to PSMA, we also tested NEAT1 recruitment to other target genes described in Figs 3c and 4a, and observed that in addition to PSMA, NEAT1 was also recruited to the promoter region of GJB1 (Supplementary Fig. 4a). This suggests that NEAT1 transcriptionally regulates a

Figure 3 | NEAT1 ERa signature correlates with prostate cancer. (a) Scatter plots for gene expression levels in VCaP ERa compared with VCaP cell lines. (b) Five hundred and eighty-eight genes that are overexpressed in VCaP ERa (log2-fold change 42) were used for Oncomine concept analysis across different cancer data sets (see Methods for detail). (c) qRTPCR analysis of relative mRNA levels of ERa target genes in VCaP cells with knockout of ERa with and without E2 treatment. The target genes selected for validation are the ones that had the highest log2-fold difference in VCaP and VCaP ERa cell lines. Results are expressed as the means.d. calculated from three independent experiments. Students t-test was performed (as indicated) for comparisons between E2 and E2 conditions for Ctrl siRNA and ERa-siRNA transfections, and *Po0.05 and **Po0.01 were considered statistically

signicant. A representative example is shown for ERG target expression. (d) qRTPCR analysis of ERa target genes in VCaP cells with ERa overexpression and NEAT1 knockout with and without E2 treatment. Results are expressed as the means.d. calculated from three independent experiments. Students t-test was performed for comparisons between E2 and E2 conditions for scrambled shRNA and NEAT1 shRNA transfections in VCaP and VCaP ERa

cells, and *Po0.05 and **Po0.01 were considered statistically signicant. A representative example is shown for SPDEF target expression. (e) Network representation of NEAT1 signature, derived from genes overexpressed in VCaP NEAT1 (NEAT1 signature) cells, across different cancer data sets using

Oncomine concept analysis.

6 NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383 ARTICLE

compendium of genes known to be involved in prostate cancer progression. We hypothesized that NEAT1 might contribute to gene transcription by interacting with chromatin-modifying proteins and/or interacting with histones. Several recent studies support the view that lncRNAs recruit chromatin-modifying machinery4649. To test this hypothesis, we analysed the chromatin landscape at the PSMA promoter and observed that

NEAT1_1, and not NEAT1_2, facilitated gene transcription by promoting an active chromatin state (Fig. 5d). Overexpression of NEAT1_1 signicantly increased active chromatin marks at the PSMA promoter (that is, H3K4Me3 and H3AcK9). Of note, ERa was not signicantly recruited to the PSMA promoter when expressed alone. Overexpression of NEAT1_1 resulted in subsequent recruitment of NEAT1_1 and ERa to the PSMA

a b

log2 (RPKM+1)VCaP ER

0 5 10 15

588

log 2(RPKM+1) VCaP

c e

Relative mRNA levels

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.06

0.05

0.04

0.03

0.02

0.01

0.00

+ + + + + + + + + + + + + + + + + + + +

si-Ctrl

ER-si

si-Ctrl

ER-si

si-Ctrl

ER-si

si-Ctrl

ER-si

si-Ctrl

ER-si

si-Ctrl

ER-si

si-Ctrl

ER-si

si-Ctrl

ER-si

si-Ctrl

ER-si

si-Ctrl

ER-si

ERG NPY KLK2 LDHB CRISP3 ACPP CLDN8 KLK4 PSMA TRPM8

100

+E2

Relative mRNA levels

1.5

1.0

0.5

* *

0.0

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

Scra shRNA

NEAT1 shRNA

VCaP VCaP ER VCaP VCaP ER VCaP VCaP ER VCaP VCaP ER VCaP VCaP ER

KLK2 PSMA

ERG SPDEF TMPRSS2

NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383

a c

NEAT1

PSMA

PDLIM3

RTN1

AMACR

+E2

GJB1

HPN

EHF

Relative mRNA levels

AMACR

EHF

HPN

RTN1

PDLIM3

GJB1

0.8

** * * *

PSMA

0.4

RTN1

NEAT1

0.0

+ + + + + + + + + + + + + + + +

shRNA N1 Sc N1 Sc N1 Sc N1 Sc N1 Sc N1 Sc N1 Sc N1

ADRB1 AMACR EHF HPN PDLIM3 PLA2G7 RTN1 SIM2

+ + + + + + + + + + + + + + + + + + + + Sc N1 Sc N1 Sc N1 Sc N1 Sc N1 Sc N1 Sc N1 Sc N1 Sc N1

Sc N1

100

0.5

E2 +E2

Colour key

and histogram

Value

0.5

Relative mRNA levels

20 15 10 5 0 Count

0.6

0.4

0.2

0.0

* *

shRNA

PSMA PRAC ARHGAP6 KLK4 TRPM8 ERG NPY KLK2 GJB1 AR

Relative mRNA levels

tumour/benign

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

HPN

EHF

AMACR

15489 T 15508 T 15608 T 15969 T 16108 T 16588 T

NEAT1

PSMA

GJB1

PDLIM3

RTN1

16728 T

Relative mRNA levels

tumour/benign

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

NEAT1

PSMA

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

GJB1

PDLIM3

RTN1

HPN

EHF

AMACR

16748 T 17368 T 20750 T

NEAT1

PSMA

21090 T

NEAT1

PSMA

21130 T

NEAT1

PSMA

21211 T

Figure 4 | NEAT1 ERa signature is upregulated in prostate cancer. (a) Relative mRNA levels of genes nominated from analysis in Fig. 3b,e, analysed using qRTPCR in parental VCaP cells transfected with scrambled (Sc) and NEAT1 shRNA (N1), respectively, with and without E2 (10 nM) treatment. Results are expressed as the means.d. calculated from three independent experiments. Students t-test was performed for comparisons (relative mRNA levels of target gene expression) between E2 and E2 conditions for scrambled shRNA and NEAT1 shRNA transfections. A representative example is shown for

ADRB1 and PSMA target expression. *Po0.05 and **Po0.01 were considered statistically signicant. (b) Validation of expression of the top target NEAT1 ERa signature genes in a small matched patient cohort of 13 benign and 13 PCa, n 26. Results are expressed as the relative mRNA levels tumor/benign

from two independent experiments. Error bars represent the range of data. (c) Heatmap shows the Spearmans correlation results from b.

8 NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383 ARTICLE

R1881

E2 + R1881

+E2

b c

8 VCaP

0.70

1.8

PC3

E2 +E2

0.60

Relative luciferase units

Empty vector

PSMA

PSMA+NEAT1

PSMA+NEAT1+ER

PSMA+NEAT1+AR

Relative luciferase units

1.6

R1881 + R1881

% ChIP Input

1.4

0.50

Promoter +E2 Promoter E2 Exon +E2 Exon E2

1.2

0.40

0.8

0.30

0.6

0.20

0.4

0.2

0.10

0.00

Empty vector

PSMA

PSMA+NEAT1

0 PSMA+NEAT1+ER

PSMA+NEAT1+AR

Sc-NEAT1

-Biotin

NEAT1-Biotin

Pool1

NEAT1-Biotin

Pool2

Sc-LacZ-

Biotin

Lac-Z-

Biotin

PSMA

1.20

1.00

0.80

E2 +E2

% ChIP input

0.60

0.40

0.20

0.00

Rabbit IgG

H3K4Me3

H3K9Me3

H3K27Me3

H3AcK9

Rabbit IgG

H3K4Me3

H3K9Me3

H3K27Me3

H3AcK9

Rabbit IgG

H3K4Me3

H3K9Me3

H3K27Me3

H3AcK9

Rabbit IgG

H3K4Me3

H3K9Me3

H3K27Me3

H3AcK9

Rabbit IgG

H3K4Me3

H3K9Me3

H3K27Me3

H3AcK9

Rabbit IgG

H3K4Me3

H3K9Me3

H3K27Me3

H3AcK9

Con

NEAT1 ER ER+NEAT1 ER+Neat1 ER+Neat1

NEAT1_1-si1

NEAT1_2-si2

VCaP

Input IP

Sc-NEAT1-As-Biotin

NEAT1-As-Biotin

NEAT1-siRNA/Neat1-As-Biotin

Sc-NR_024490-As-Biotin

NR_024490-As-Biotin

Sc-NEAT1-As-Biotin

NEAT1-As-Biotin

NEAT1-siRNA/Neat1-As-Biotin

Sc-NR_024490-As-Biotin

NR_024490-As-Biotin

Sc-NEAT1-As-Biotin

NEAT1-As-Biotin

NEAT1-siRNA/NEAT1-As-Biotin

Sc-NR_024490-As-Biotin

NR_024490-As-Biotin

Histone H3AcK9

Histone H3K4Me3

Histone-H3 Total

1 2 3 4 5 6 7 8 9 10 11

VCaP

Histone-H3

Figure 5 | NEAT1 is a transcriptional regulator. (a,b) Promoter luciferase reporter assay shows that NEAT1 activates PSMA promoter in PC3 and VCaP cells. Cells were co-transfected with empty vector or PSMA luc and Renilla-luc reporter genes alone or with NEAT1, NEAT1 ERa and NEAT1 AR.

Luciferase activity was measured 48 h post treatment with E2 (10 nM) or R1881 (1 nM). Results are expressed as the means.d. calculated from three independent experiments. Students t-test was performed for comparisons (relative PSMAluciferase activity) between E2 and E2 conditions for

vector control, NEAT1 and NEAT1 ERa transfections in PC3 and VCaP cells. *Po0.05 and **Po0.01 were considered statistically signicant. (c)

Quantitative analysis of NEAT1 ChIRP in VCaP cells with or without E2 treatment (10 nM). Recruitment proles of NEAT1 to PSMA are shown. Results are expressed as the percentage of input calculated from two independent experiments. Error bars represent the range of data. Results were reproducible between representative experiments. **Po0.01 was considered statistically signicant. (d) Analysis of the chromatin landscape at the PSMA promoter performed by ChIP in VCaP cells alone or transected with NEAT1, ERa, NEAT1 ERa, NEAT1 ERa NEAT1_1 siRNA and NEAT1 ERa NEAT1_2 siRNA with and without E2 treatment. qPCR was performed with specic primers for the PSMA promoter. Results are expressed as the percentage of input calculated from two independent experiments. Error bars represent the range of data. Results were reproducible between representative experiments. (e) NEAT1 binds to Histone H3. 20 mer-biotinylated NEAT1 and NR_024490 antisense probes were used to immunoprecipitate NEAT1 and NR_024490 from nuclear lysates of VCaP cells using streptavidin magnetic beads. Immunoprecipitates from Streptavidin-IP were analysed on 15% gel and probed for Histone H3. NEAT1 is shown to also bind with active histone H3 modications, including H3AcK9 and H3K4Me3.

NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383

promoter. These studies indicate that although NEAT1_1 may function as a chaperone for ERa and other chromatin-modifying machinery to target promoters, binding of ERa and/or recruitment to NEAT1_1 targets is not necessary for transcriptional activation.

As our data suggests that NEAT1 overexpression favours a chromatin landscape for active transcription, we investigated whether NEAT1 could directly interact with nucleosomal histones. Nuclear lysates from VCaP cells were used in an immunoprecipitation experiment with streptavidin beads coupled with either scrambled, antisense NEAT1, or antisense NR_024490 (another ERa lncRNA target) oligonucleotides. NEAT1 was found to specically associate with histone H3 (Fig. 5e, left panel, lane 8) and the specicity of this binding is apparent when comparing lanes 7 and 9, which represent Streptavidin-IP using scrambled biotinylated oligos and Streptavidin-IP using anti-sense-NEAT1 oligos and nuclear lysates from NEAT1 siRNA-treated cells, respectively. As an additional negative control, we used scrambled and specic antisense oligos for a different lncRNA, NR_024490, another ERa target. The results indicate that NEAT1 can associate with chromatin via a specic interaction with histone H3. We also determined association of NEAT1 with active histone H3 modications, including H3AcK9 and H3K4Me3 (Fig. 5e, right panel). Similar association patterns were seen for NEAT1 in NCI-H660 cells (Supplementary Fig. 4b).

To complement this nding, we performed RNA immunoprecipitation from VCaP ERa cells using anti-histone H3 and anti

SNRNP70 (positive control) as the immunoprecipitating antibody. qRTPCR showed robust binding of NEAT1 to histone H3 (Supplementary Fig. 4c). The positive control U1 small nuclear RNA showed high enrichment in the immunoprecipitate with SNRNP70. To further conrm the specicity of NEAT1 binding to histone H3, we performed a streptavidinbiotin pull-down assay in VCaP and VCaP ERa cells with and without E2 (Supplementary Fig. 4d). These data suggest that NEAT1 can directly interact with the histone H3 component of chromatin.

NEAT1 promotes prostate tumorigenesis. To better understand the physiological role of NEAT1 in the context of ERa in prostate cancer, we rst determined the levels of NEAT1 in VCaP cells overexpressing ERa (Supplementary Fig. 5a). Further, we generated stable VCaP and VCaP ERa cell lines that overexpress NEAT1 (Supplementary Fig. 5b). We also knocked down NEAT1 in VCaP and VCaP ERa-expressing cells by stably expressing

NEAT1 shRNA targeting two different regions of NEAT1 and non-targeting shRNA (Supplementary Fig. 5c). Although over-expression of NEAT1 signicantly increased proliferation and cell invasion, knockdown of NEAT1 signicantly decreased proliferation and the invasive properties of the cells (Fig. 6a,b).

Soft agar assays were performed in both VCaP and VCaP NEAT1 cells. Colonies were monitored over a period of 21 days. Overexpression of NEAT1 resulted in a signicantly higher number of viable colonies (Fig. 6c). Colony-forming assays performed in NEAT1 clones in VCaP cells with and without E2 demonstrated that E2 treatment in NEAT1-overexpressing cells signicantly increased the number of colonies (Fig. 6d). These in vitro assays establish an oncogenic role for NEAT1.

To further validate the oncogenic role of NEAT1, we extended our studies to an in vivo model system. We performed xenograft studies in NOD-SCID mice. The mice were treated with time-release oestrogen pellets. They were divided into two groups and one group was implanted subcutaneously with VCaP ERa cells expressing control shRNA luciferase reporter, and the other group with VCaP ERa cells expressing NEAT1 shRNA luciferase reporter. The mice from both groups were imaged weekly for luciferase activity and Fig. 6e shows the bioluminescent signals at

day 7 and day 35. The tumour growth was monitored weekly for 45 days and was found to be signicantly lower in the NEAT1 shRNA-expressing group compared with the control group (Fig. 6f). The tumours were excised and weighed, and the NEAT1 shRNA group had signicantly smaller tumours (Supplementary Fig. 6a,b). We conrmed the efcacy of the shRNA in vivo by measuring the NEAT1 and ERa levels in the tumours (Supplementary Fig. 6c).

To further substantiate our hypothesis that NEAT1 plays a role in tumorigenesis, we repeated the experiment in athymic nude mice using VCaP control and VCaP NEAT1-overexpressing cells, as well as NCI-H660 and NCI-H660 NEAT1-overexpressing cells. In both these experiments, a signicantly higher tumour growth was seen in the NEAT1-overexpressing cells (Fig. 6g,h and Supplementary Fig. 6d,e) further conrming its oncogenic potential. qRTPCR analysis conrmed an increased expression of the NEAT1 signature genes in VCaP NEAT1 xenografts compared with control VCaP xenograft tissue (Supplementary Fig. 6f).

NEAT1 is associated with therapeutic resistance. The study presented so far shows that ERa establishes an oncogenic cascade and that NEAT1 functions as a downstream mediator of ERa signalling. The ERa-NEAT1 axis is functional both in AR-positive and -negative cell lines, and drives prostate carcinogenesis. We hypothesized that targeting NEAT1 using mechanisms that can constrain ERa might represent a novel therapeutic strategy in prostate tumours that are resistant to anti-androgen therapy. To test this hypothesis in vitro, we evaluated the effect of anti-oestrogens and anti-androgens on NEAT1 levels in prostate cancer cell lines. As shown in Fig. 7a,b, NEAT1 expression is constrained when cells are treated with the ERa antagonists ICI 182,780 (ICI)

and 4-hydroxy tamoxifen (4OHT) in combination with E2. Intriguingly, treatment of ICI and 4OHT alone for longer periods can enhance NEAT1 expression (Fig. 7a,b). We observed similar results with AR antagonists enzalutamide and bicalutamide (Fig. 7c,d). These results provide compelling evidence to evaluate NEAT1 levels in advanced CRPC cases. RNA-uorescent ISH analysis of benign and advanced prostate tumours, including CRPC and NEPC tumour tissue samples, illustrated signicantly upregulated NEAT1 levels in advanced prostate cancer, with enhanced focal staining throughout the tumour tissue (Fig. 7e). We also screened nine cases of benign prostate, seven PCa and seven CRPC (Supplementary Table 2) for NEAT1 and ERa expression by qRT-PCR (Fig. 7f), and both NEAT1 and ERa levels were signicantly higher in the CRPCs. We determined the correlation between NEAT1 and ERa expression by estimating the Pearsons correlation coefcient R. The results indicate a strong positive correlation: R 0.86 (P-value 1.9e 07). Taken

together, our results present a novel role for the non-coding transcriptome in cancer-favourable adaptations.

NEAT1 is associated with aggressive prostate cancer. Given the importance of NEAT1 in promoting tumorigenesis both in vitro and in vivo, we sought to determine the relationship between NEAT1 levels and prostate cancer clinical outcomes in 594 patients from two radical prostatectomy cohorts with long-term clinical follow-up from the Mayo Clinic35,36. Supplementary Table 1 contains the patient characteristics of men who underwent radical prostatectomy at the Mayo Clinic Comprehensive Cancer Center between 1987 and 2001 for clinically localized prostate cancer.

We assessed the prognostic potential of NEAT1 expression using several statistical measures and correlating it with biochemical recurrence (BCR) and metastasis (MET), prostate

10 NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383 ARTICLE

a b c d

** **

4 E2

+E2 24 h +E2 48 h

E2 +E2

Relative cell proliferation

Vector con

0 NEAT1 Cl-1

NEAT1 Cl-2

si-scrambledNEAT1 shRNA 1

NEAT1 shRNA 2

Vector con

NEAT1 Cl-1

NEAT1 Cl-2

E2 +E2

** **

250

% Invasion

Number of colonies

+E2E2

Number of colonies

***

* **

** **

200

0.6

150

0.4

100

0.2

0.0

Vector con

NEAT1 Cl-1

NEAT1 Cl-2

si-scrambled

NEAT1 shRNA 1

NEAT1 shRNA 2

VCaP con

0 VCaP NEAT1

0 Con

Clone 1

Clone 2

Luminescence

Radiance(p s1 cm2 per sr)

scrambled shRNA NEAT1 shRNA

5.0

4.0

3.0

2.0

1.0

Day 7 Day 35

f g

VCaP NEAT1 + E2 VCaP vector con + E2

Mean bioluminescence

12 NCI-H660 NEAT1 + E2 NCI-H660 vector con + E2

Day 21

photons per second 106

Scrambled shRNA

NEAT1 shRNA *

Mean bioluminescence

photons per second 106

5 Day 7

Day 14

Day 21

Day 28

Day 35

Day 45

0 Day 7

Day 14

Day 21

Day 28

Day 35

Day 45

0 Day 7

Day 14

Day 35

Day 28

Day 45

Figure 6 | NEAT1 is a driver of oncogenic cascade. (a) Cell proliferation assays were performed in VCaP vector control, NEAT1-overexpressing cells and also in si scrambled and NEAT1 knockout cells with or without E2 treatment (10 nM) at 24 and 48 h time points. Results are expressed as the means.d. calculated from three independent experiments. Students t-test was performed for comparisons (relative cell proliferation) between E2 conditions for vector control, NEAT1 Cl-1, and NEAT1 Cl-2 and E2 conditions for si-scrambled, Neat1-shRNA1 and shRNA2 transfections. **Po0.01 was considered statistically signicant. (b) Quantitative bar chart for depicting percentage of cells invaded at the completion of invasion assays performed in VCaP vector control, NEAT1-overexpressing cells and also in si scrambled and NEAT1 knockout cells with or without E2 treatment (10 nM). Results are expressed as the means.d. of three independent experiments. *Po0.05 and **Po0.01, Students t-test. (c) Soft agar assays were performed with VCaP control and

NEAT1-expressing cells. Quantitative bar-plot analysis of stained colonies at 21 days are shown. Results are expressed as the means.d. of three independent experiments. ***Po0.001, Students t-test. (d) Colony-forming assay were performed in VCaP vector control, NEAT1-overexpressing cells with or without E2 treatment (10 nM). The right panel depicts the number of colonies at 21 days. Results are expressed as the means.d. calculated from three independent experiments. *Po0.05 and **Po0.01, Students t-test. (e) VCaP ERa cells expressing con shRNA luciferase (luc) and NEAT1 shRNA luc were injected subcutaneously into the ank of male NOD-SCID mouse. Bioluminescent imaging on Day 7 and Day 35 in the VCaP ERa scrambled shRNA (top panel) and VCaP ERa NEAT1 shRNA (bottom panel) injected mice is shown. (f) Growth curve for the tumours monitored upto 45 days. Results are expressed as the means.d. calculated from three independent experiments. *Po0.05, Students t-test. (g,h) VCaP and NCI-H660 vector control and

NEAT1-overexpressing cells were injected subcutaneously into the ank of male NOD-SCID mouse. Bioluminescence imaging monitored the tumour growth. Growth curve for the tumours monitored upto 45 days is shown; VCaP (g) and NCI-H660 (h).

cancer-specic mortality (PCSM) and GS47. To evaluate endpoints of disease aggressiveness and progression based on NEAT1 expression, KaplanMeier (KM) analysis was performed for the BCR and MET endpoints. The resulting KM curves (Fig. 8a,b) demonstrate that patients with higher NEAT1 expression have signicantly worse outcomes for both BCR (P-value: 0.028) and MET events (P-value: 0.016).

Patient risk discrimination based on the expression prole of NEAT1 was assessed by area under the receiver operating characteristic curve with 95% condence intervals (Supplementary Fig. 7). NEAT1 signicantly segregates patients who exhibited BCR, MET, PCSM and GS47.

To further compare NEAT1s prognostic ability to other clinic-opathologic variables, univariable odds ratios were computed for

NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383

a c

b d

E2 E2+ICI ICI

E2 E2+TAM TAM

Relative mRNA levels

+E2/E2

+E2+Enz/E2

Relative mRNA levels

1.5

1.0

Fold change

0.5

** **

0.5

0.03

0.02

0.01

0.00

0 6 h 14 h 24 h 48 h 24 h 36 h 48 h 60 h 72 h 96 h

VCaP

NEAT1

Bicalutamide

+ + +

+ + + + + +

+ + + + + + + +

NEAT1

LNCaP VCaP

NEAT1

GJB1

TRPM8

R=0.86 (P-value=1.9e07)

120

800

LPCA CRPC Benign

100

600

# Cells

NEAT1

400

200

1 1 0.33

2.4 8.6 34.2

4.2 6.4 108.2

Benign PCa CRPC

0 2 4 6 8

H&E 20X NEAT1 ISH 20X NEAT1 ISH 40X

13 Signals 36 Signals 6+ Signals Benign

LPCa

CRPC

Figure 7 | NEAT1 in therapy resistance. (a) NEAT1 expression in VCaP cells treated with E2 (10 nM) at different time points alone, E2 ICI

(10 nM 10 mM) or ICI (10 mM) alone. Results are expressed as the means.d. of three independent experiments. *Po0.05, **Po0.01, Students t-test.

(b) NEAT1 expression in VCaP cells treated with E2, E2 4OHT (10 nM 10 nM) and 4OHT (10 nM) alone for 48 h. (c) NEAT1 expression in VCaP cells

treated with or without E2 (10 nM) or E2 Enzalutamide (10 nM 10 mM) at different time points. Results are expressed as the means.d. of three

independent experiments. *Po0.05 and **Po0.01, Students t-test. (d) qRTPCR analysis of NEAT1, GJB1 and TRPM8 in LnCaP and VCaP control cells, with bicalutamide treatment (10 mM) alone or in combination with E2 (10 nM) for 48 h. Results are expressed as the meanss.d. of three independent experiments. Results were reproducible between representative experiments. (e) Representative image for RNA ISH of NEAT1 in benign, localized PCa and in advanced disease (top panel). Quantication for the RNA ISH signals shown in the bottom using RNA Spot Studio. (f) Scatter plot showing the correlation between ERa and NEAT1 expression by qRT-PCR in nine cases of benign prostate, seven PCa and seven CRPC. Pearsons correlation coefcient

R 0.86 (P-value 1.9e 07).

a b

0 10 20 30 40 50 60 70

1.0

Probability of BCR-free

survival

Probability of MET-free

survival

0.8

0.6

0.4

0.2

NEAT1 high

NEAT1 low

NEAT1 High

NEAT1 Low

0.0

0 10 20 30 40 50 60 70

Time (Months)

111 62 30 105 70 41

Time (Months)

P-value: 0.028

P-value: 0.016

High

Low

High

Low

85 65 36 131 118 67

Figure 8 | NEAT1 overexpression is associated with aggressive prostate cancer. (a,b) KM curves showing (a) BCR-free survival and (b) MET-free survival for NEAT1 low and high expression groups of samples from the Mayo casecohort data set35 (n 216). The cut points to dene high and

low NEAT1 expression were selected using patients from the Mayo nested casecontrol data set (n 378)58 by maximizing the product of the

sensitivity and specicity for each endpoint. The number of patients at risk for each group is shown beneath the plot.

the BCR, MET, PCSM and GS47 endpoints (Fig. 9). NEAT1 was signicantly prognostic for segregating high-risk from low-risk patients for each of the endpoints (Po0.05). Further multivariate analysis adjusting for adjuvant radiation and hormone treatment, in addition to the other clinicopathological variables assessed, also demonstrates that NEAT1 was signicantly prognostic for BCR,

MET and GS47, supporting NEAT1 as a prognostic biomarker for aggressive prostate cancer independent of common clinical and pathologic variables (Supplementary Fig. 8). Overall, these results show that NEAT1 is signicantly prognostic for several clinically relevant endpoints.

DiscussionThe tissue-specic role of ERa in breast and other gynaecological malignancies is well understood. Interestingly, ERa is expressed in all prostate cancers, including those that lack AR expression, while it is absent in normal prostate epithelium. Studies from our and other laboratories have examined the relevance of ERa in prostate cancer4,5052. ERa-mediated regulation of oncogenic TMPRSS2-ERG fusion and oestrogen regulation of the EBAG9 gene, which conrms aggressive behaviour of prostate cancer, are noted examples that suggest a functional ERa-signalling pathway exists in prostate cancer. From a clinical perspective, the association of a polymorphism in ERa with prostate cancer with a favourable GS or cancers of late onset has also been reported53. These initial observations prompted us to evaluate whether re-expression of ERa and the establishment of an alternate nuclear receptor-signalling axis (that is, ERa versus AR) in prostate cancer cells could represent an adaptive mechanism to evade AR-directed therapies.

Analysis of global ERa recruitment in prostate cancer cells using a ChIP-seq approach revealed that ERa is preferentially recruited to intergenic regions of the prostate genome. Comparison of binding proles with transcriptome sequencing data

12 NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383 ARTICLE

3.0

2.5

Odds ratio

2.0

1.5

1.0

NEAT1_1

NEAT1_2

log(pPSA)

GS>7

SVI

ECE

SMS

LNI

NEAT1_1

NEAT1_2

log(pPSA)

GS>7

SVI

ECE

SMS

LNI

c d

Odds ratio

NEAT1_1

NEAT1_2

log(pPSA)

GS>7

SVI

ECE

SMS

LNI

NEAT1_1

NEAT1_2

log(pPSA)

SVI

ECE

SMS

LNI

P-value < 0.001 P-value < 0.01 P-value < 0.05 P-value <= 1

Figure 9 | NEAT1 is a strong prognosticator of prostate cancer. (ad) Univariable forest plots comparing the expression of NEAT1s short (NEAT1_1) and long isoform (NEAT1_2) to clinicopathologic variables in the pooled Mayo cohort (n 594) (a) BCR, (b) MET, (c) PCSM and (d) GS47.

Pathological tumour stage 3 or greater (pT3 ), lymph node invasion (LNI),

surgical margin status (SMS) positive, seminal vesicle invasion (SVI), extra capsular extension (ECE), preoperative PSA (pPSA), adjuvant hormone therapy and adjuvant radiation therapy are shown.

suggested that ERa drives expression of non-coding transcripts. These results led us to analyse the functional consequences of ERa recruitment to non-coding regions. From a large compendium of ERa-regulated non-coding transcripts, we selected NEAT1 for a detailed biochemical and in vivo evaluation, based on an in silico approach that demonstrated a strong association of NEAT1 with prostate cancer progression. We show that ERa transcriptionally regulates NEAT1. NEAT1 is recruited to the promoter of several key target genes and induces an active chromatin state favourable for transcription. Our studies indicate that ERa does not function as a molecular chaperone to guide

NEAT1 to target chromatin; rather, we suspect that a complex proteome of chromatin-interacting proteins interacts with and guides NEAT1 to promoter targets. Interestingly, both ERa and

NEAT1 signalling were refractory to AR inhibitors and the lack of AR or ERb, thus indicating a functional specialization of the ERa-

NEAT1 axis for prostate cancer progression. Furthermore, introduction of cells overexpressing NEAT1 could clearly induce prostate cancer progression in experimental animal models.

The current study opens up a new arena of alternative mechanisms of tumorigenesis by ERa in prostate cancer. We show that ERa regulates NEAT1 lncRNA with distinct chromatin regulatory functions. Large-scale bioinformatic analysis of SAGE libraries has identied NEAT1 as one of the differentially regulated lncRNAs between some types of cancer and normal tissue54. However, its possible role in promoting tumorigenesis has never been explored. We show here that NEAT1 regulates expression of prostate cancer genes by altering the epigenetic landscape at target gene promoters to favour transcription. A closer examination of NEAT1 revealed a previously uncharacterized role in recognition of modied histones. We have not tested whether NEAT1 is a reader of multiple histone

H3 post-translational modications (acetylation, methylation and so on) and our laboratory is actively pursuing this intriguing question. NEAT1 expression independently was sufcient to activate prostate cancer genes in an AR-independent manner. Further, our results conrmed an oncogenic role for NEAT1 in an experimental animal model of prostate cancer and in cell culture models.

Molecular sieving of the net non-coding transcriptome using comprehensive bioinformatic approaches and wet-lab validation over a decade has indicated that the non-coding transcriptome has a regulatory role beyond the speculated transcriptional noise and a direct inuence on the coding transcriptome and biologic homeostasis. We observed that several lncRNAs such as NEAT1 respond to cellular cues and ligand signalling in a manner reminiscent of the coding transcriptome. Thus far, the literature on NEAT1 has focused on its architectural role in forming subnuclear paraspeckles27. Our results indicate a role for NEAT1 beyond that of paraspeckles. It would be interesting in the future to reconcile how the formation of paraspeckles in the inter-chromosomal space ties in with the role of NEAT1 in activating gene expression at promoters. Our lab continues to pursue some of these unanswered questions to better understand the role of NEAT1.

Our identication of an ERa-NEAT1 axis illustrates a mechanism whereby prostate cancer cells may develop therapeutic resistance through positive selection of an alternate nuclear receptor signalling pathway in the absence of AR or during androgen ablation therapy (Fig. 10). However, we cannot exclude the presence of other NEAT1-interacting chromatin factors. This is the subject of ongoing investigation.

From a clinical perspective, our studies indicate for the rst time that NEAT1 is signicantly prognostic for several clinically relevant endpoints. In prostatectomy specimens from two large cohorts, high NEAT1 expression was associated with a signicant increase in both biochemical and metastatic recurrence rates compared to those with low NEAT1 expression.

In summary, this study provides important insights into a unique mechanism of ERa regulation in prostate cancer and identies NEAT1 as a novel prognostic marker and potential therapeutic target in this disease. Although our studies have identied a previously unexplored function of ERa in regulating lncRNAs, it is also the rst of its kind to demonstrate transcriptional regulation of lncRNAs by an alternative steroid receptor in prostate cancer. We propose that NEAT1 is directly involved in modulation of the phenotype of a leading disease. Combinatorial targeting of NEAT1 and AR may represent a unique therapeutic regimen within a subset of patients with advanced prostate cancer.

Methods

Cell culture and treatments. LnCaP and PC3 cells were grown in RPMI 1640 (Invitrogen) and supplemented with 10% fetal bovine serum (FBS) and 1% penicillinstreptomycin. RWPE1 cells were grown in Keratinocyte Serum Free Medium Kit (Gibco, 17005-042). VCaP and DU145 cells were grown in DMEM (Invitrogen) and supplemented with 10% FBS with 1% penicillinstreptomycin. NCI-H660 cells were grown in RPMI 1640 supplemented with 0.005 mg ml 1 insulin, 0.01 mg ml 1 transferrin, 30 nM sodium selenite, 10 nM hydrocortisone, 10 nM b-estradiol, 5% FBS, 1% penicillinstreptomycin and an extra 2 mM of

L-glutamine (for a nal concentration of 4 mM). For cell treatments in several experiments, we used 10100 nM b-estradiol (Sigma Aldrich, St Louis, MO), 10 mM

Enzalutamide (Astellas), 10 mM bicalutamide (Sigma Aldrich), 110 nM R1881 (PE Biosystems), 10100 nM 4OHT (Sigma Aldrich) and 110 mM ICI (Tocris Bioscience).

Plasmids, siRNAs and transfection. Plasmids, pcDNA 3.1, pcDNA3.1-ERa, pcDNA 3.1 AR, piLenti-GFP, piLenti-NEAT1 siRNA-GFP (set of four, sequences provided in Supplementary Table 3), iLenti-si-scrambled, pLenti-bicistronic-luc-NEAT1 were used. siRNAs for ERa, ERb, AR, NEAT1 and NEAT1_2 were used, and the sequence is provided in Supplementary Table 3. For the mammalian expression vectors, Lipofectamine 3000 (Invitrogen) and Lonza nucleofection were

NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383

ER signaling

Antiandrogens

Non-coding transcriptome

Alternate receptor

(ER) signaling

IncRNA2

NEAT1

ER ER

PCa

NEAT1

ER ER

Normal prostate

AR-mediated oncogenic cascade

PCa PCa

Therapy resistanceoncogenic adaptation Cancer progression

IncRNA1

Future directions/functional relevance

Clinicalsignificance NEAT1/NEAT1

signature

Prognostic marker

Therapeutic target

Clinical management of PCa

NEAT1

Figure 10 | Model for NEAT1 function in prostate cancer. Functional ERa signalling in prostate cancer modulates expression of the lncRNA NEAT1. Prostate epithelial cells positive for NEAT1 have an oncogenic advantage and are refractile to androgen inhibitors or androgen ablation therapy. NEAT1, a histone interacting lncRNA and transcriptional regulator, is recruited to promoters of several prostate cancer-specic genes. NEAT1 can modulate the epigenetic landscape of target promoters and maintains expression of AR-dependent and -independent genes. The selection of alternate nuclear receptor signalling is a novel hallmark of prostate cancer progression.

used for transfection. The stable clones for NEAT1 overexpression as well as the scrambled and NEAT1 shRNA-expressing cells were generated by using the lentiviral vectors and by selection in puromycin.

Identication of ERa-regulated lncRNA. A set of known lncRNAs was generated from various data sources: RefSeq: GENCODE v7, - ncRNA.org and lncRNAdb55 (see Supplementary Material) and those that were at least 200 nt long were selected, resulting in 12,483 lncRNAs. These lncRNAs were characterized according to their potential of being regulated by ERa by using ERa-binding sites information from

ChIP-seq experimental data. Moreover, several histone marks were considered to provide evidence of transcription, including H3K4me3 and H3K36me3 (details in Supplementary Material).

Differential expression analysis. To prioritize the experimental validation of lncRNAs, a pair-wise differential expression analysis was performed on the expression values determined by paired-end transcriptome sequencing of 73 samples (26 benign prostate, 40 PCa and 7 NEPC). A pair-wise Wilcoxon test was performed and all P-values were corrected for multiple hypotheses testing using BenjaminiHochberg56 (details in Supplementary Material).

ERa and NEAT1 signature via Oncomine concept analysis. RNA sequencing was done for VCaP and VCaP ERa-expressing cells, as well as in vector control and NEAT1-overexpressing VCaP cells (detailed in Supplementary Methods). The expression of the genes was computed and those genes with a log2-fold change 42 were selected. Results are reported in Supplementary Datasets 3 and 5. Five hundred and eighty-eight genes were found to be overexpressed in VCaP ERa cells. A custom concept of this gene list was generated in Oncomine (Supplementary Dataset 4). Similarly, genes from the VCaP NEAT1 group with a log2-fold change 42 were selected and a custom concept was built in Oncomine using the top 1,000 genes from NEAT1 signature (Supplementary Dataset 4). The signicantly associated tumour versus normal concepts with odds ratio 42.0 and Po1 10 6

considering tumour versus normal analysis was determined. The resulting concepts and associations are represented through a concept network using Cytoscape version 2.8.2. Each node represents a concept to which the signature is associated at a 43-fold odds ratio for ERa signature and 42-fold odds ratio for NEAT1 signature. Node size reects the concept size, that is, the number of genes in each concept; red and green colours represent correlation with over- or underexpressed genes in the concept, respectively; and edge thickness represents the odds ratio of

the association between concepts, ranging from 1.4 to 29.9 and 1.2 to 637 for ERa and NEAT1 signatures, respectively. The border colour of each node represents the tumour type. The layout of the network is based on the Edge-weighted spring-embedded algorithm.

Luciferase reporter assays. For ERE luciferase assays, VCaP cells were transiently transfected with the (ERE)3-SV40-luc reporter plasmid and/or ERa and/or AR, as well as an internal control construct pRL harbouring the renilla luciferase gene. VCaP cells were also transfected with empty vector or NEAT1 promoter (1 2)

luciferase reporter constructs alone or with ERa, as well as an internal control construct pRL harbouring renilla luciferase gene. To determine the PSMA reporter activity, 293T and PC3 cells were co-transfected with empty vector or PSMA luc and Renilla-luc reporter genes alone or with NEAT1, NEAT1 ERa or

NEAT1 AR.

Twenty-four hours post transfection, the media was changed to 5% charcoal-stripped media and the cells indicated were treated with E2 (10 nM) or R1881 (1 nM) for 14 h. At 48 h, cells were lysed with passive lysis buffer and luciferase activities were measured using the dual luciferase system (E1910, Promega) and normalized with renilla luciferase activity.

RNA ISH for NEAT1. RNA ISH for NEAT1 was performed on ve benign, ve PCa and three CRPC cases using kits and probes designed by Advanced Cell Diagnostics. Briey, the single-colour chromogenic detection assay uses pairs of specially designed oligonucleotide probes that through sequence-specic hybridization, recognize both the specic target NEAT1 RNA sequence and the signal amplication system. Unique target probe oligonucleotides were designed to hybridize in tandem to the target RNA. Cross-hybridization to other sequences is minimized by screening against the entire human RNA sequence database.

The signal amplication system consists of the pre-amplier, amplier and enzyme-conjugated label probe, which assemble into a tree-like complex through sequential hybridization. Signal amplication occurs at target sites bound by probe pairs only. Nonspecic off-target binding by single probes does not result in signal amplication.

All steps of NEAT1 RNA ISH staining of the slides are performed manually, optimized in tissue microarrays. Briey, formalin-xed, parafn-embedded unstained tissue sections (5 mm) were mounted on positively charged microscopic glass slides, deparafnized in xylene and rehydrated through a series of alcohols. The rehydrated sections were treated with 3% hydrogen peroxide at room temperature (RT) for 10 min to block endogenous peroxidase. Sections were then boiled in 1

14 NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383 ARTICLE

citric buffer (10 nmol l 1 Nacitrate, pH 6.0) for 15 min and incubated with protease(2.5 mg ml 1; Sigma Aldrich) at 40 C for 30 min. The slides were hybridized sequentially with target probes (20 nmol l 1) in hybridization buffer A (6 saline

sodium citrate (SSC) buffer (1 SSC is 0.15 mol l 1 NaCl and 0.015 mol l 1 Na

citrate), 25% formamide, 0.2% lithium dodecyl sulfate (LDS) and blocking reagents) at 40 C for 2 h, signal pre-amplier in hybridization buffer B (20% formamide,5 SSC, 0.3% LDS, 10% dextran sulfate and blocking reagents) at 40 C for 30 min,

amplier in hybridization buffer B at 40 C for 30 min and horseradish peroxidaseor alkaline phosphatase-labelled probes in hybridization buffer C (5 SSC, 0.3%

LDS and blocking reagents) at 40 C for 15 min.

Hybridization signals were detected under bright-eld microscope as red colorimetric staining (using Fast Red chromogen, BioCare Biomedical, Concord, CA) followed by counterstaining with haematoxylin. Signals were granular and discrete red signals corresponding to individual lncRNA targets. The signals were scored using the RNA Spot Studio software.

Chromatin immunoprecipitation. All ChIP experiments were carried out using Millipore EZ-Magna ChIP kit (catalogue number 1710086). Briey, 510 106 cells

were cross-linked with 1% formaldehyde for 10 min at rRT. The cross-linking was then quenched with 0.125 M glycine. Chromatin was sonicated in the lysis buffer to 300500 bp and the extraction of ChIP DNA was done as per the kit protocol. Antibodies used include ERa (AC-066-100, diagenode, 5 mg), AR (06680, Millipore, 5 mg), H3K4me3 (ab8580, Abcam, 5 mg), trimethylated lysine 9 of histone H3 (ab8898, Abcam, 5 mg), H3K36me3 (ab9050, Abcam, 5 mg), trimethylated lysine 27 of histone H3 (07449, Millipore, 5 mg) and Ace-H3 (no. 06599, Millipore, 5 mg).

ERa ChIP was also performed in cross-linked VCaP cells with E2 treatment for 0, 14 and 48 h. In VCaP ERa cells, E2 treatment was for 6, 14 and 48 h. The primer sequences are provided in Supplementary Table 4.

Chromatin isolation by RNA purication. ChIP for NEAT1 was done in VCaP control and NEAT1-expressing cells, with and without E2 treatment using the ChiRP protocol45. Briey, biotin TEG antisense oligos were generated using singlemoleculesh.com for NEAT1, Lac Z and scrambled NT NEAT1. The NEAT1 probes were divided into two pools. Cells cross-linked in 1% glutaraldehyde were lysed and sonicated. The biotinylated probes were hybridized followed by RNA and DNA isolation. qPCR was performed on the DNA samples. Probe sequences are described in Supplementary Table 5.

RNA ISH for NEAT1 on cell lines. Cells were grown on a 15 mm, poly-L-lysine-coated glass coverslip. At B70% conuence, cells were serum starved in 8% charcoal-stripped media for 48 h, followed by 48 h treatment with 10 nM E2. At the end of treatment, cells were xed in 4% formaldehyde, dehydrated by an ethanol gradient (50100%) and stored at 20 C. For the hybridization assay, cells were

rehydrated by an ethanol gradient (10050%) into PBS. Between subsequent steps, cells were washed with PBS. The Affymetrix QuantiGene ViewRNA ISH cell assay kit was used for NEAT1 staining. Cells were permeabilized by 5 min incubation at RT in Detergent Solution QC and digested for 10 min at RT by Protease QS (1:4,000 in PBS). Next, the target-specic Probe Set (1:100 in Diluent QF) was allowed to hybridize for 3 h at 401 C. Between subsequent steps, cells were washed by soaking in Wash Buffer. Sequential hybridization steps were conducted for signal amplicationPreAmplier Mix (1:25 in Diluent QF), Amplier Mix (1:25 in Diluent QF) and Label Probe Mix (1:25 in Diluent QF), each incubated 30 min at 401 C. After two 10-min washes in Wash Buffer, nuclei were stained with 40,6-diamidino-2-phenylindole and cover slips were mounted to slides with Prolong Gold Antifade Reagent (Life Technologies) for visualization.

Proliferation assay. Cell proliferation was assessed using the CyQUANT NF cell proliferation assay kit (Life Technology). Cells were seeded in 96-well plates at 3 4 104 cells per well. Cells were incubated in DMEM media with 10% FBS for 24 h.

The cells were then serum starved in 8% charcoal-stripped DMEM medium for 48 h followed by E2 treatment at 10 nM concentration for indicated time points. The media was then aspirated and replaced with the dye-binding solution followed by incubation for 3060 min. The uorescence was then measured in a microplate reader using excitation at 48510 nm and uorescence detection at 53015 nm. The assay was performed in triplicates.

Invasion assay. The CHEMICON cell invasion assay kit (EMD MIlipore) was used for determining the cell invasion. Cells were serum-starved for 48 h and then seeded at a density of 2 105 cells per well in the upper well of the invasion chamber. Five

hundred microlitres of phenol red-free DMEM media supplemented with 8% charcoal-stripped serum and 10 nM E2 was added to the lower chamber. After 48-h incubation, the invaded cells were stained by dipping the inserts in the staining solution for 20 min. The stained cells were then dissolved in 10% acetic acid and transferred to a 96-well plate for colorimetric reading of optical density at 560 nm.

Migration assay. The Cell Biolabs Inc. Radius 96-Well Cell Migration Assay was used to determine cell migration. Cells were serum starved for 48 h, then seeded to a pretreated (incubated 20 min in Radius Gel Pretreatment solution and washed

with Radius Wash Solution) Radius 96-Well Plate at a density of 8 104 cells per

well with or without E2 (10 nM). After 24 h incubation, the Radius Gel Spot was removed via the Radius Gel Removal Solution and pre-migration images were captured. After 24-h incubation, cells were stained with Cell Stain Solution and post-migration images were captured for analysis using the CellProler Cell Image Analysis Software (Broad Institute).

Statistical analysis. The Wilcoxon test was employed with Benjamini Hochberg56 correction for multiple hypotheses for pair-wise comparisons for differential expression analysis. The w2-test was used for comparison of proportions and the Pearsons correlation was used to compare the expression of selected genes. For qRTPCR, we computed the Delta CT value according to the ABI qPCR guidelines as described in Supplementary Methods. To compare qPCR data, a Students t-test was employed. Median rank statistics results are reported for analyses with the Oncomine data sets57.

Analysis of Mayo clinic cohort. Affymetrix HuEx microarrays were used to analyse NEAT1 expression in two post-radical prostatectomy cohorts from the Mayo Clinic. Details on tissue preparation, RNA extraction, amplication, hybridization and clinical characteristics for these cohorts have been described previously35,36. Both cohorts were ltered using the same criteria (patient either exhibiting pre-operative prostate-specic antigen 420 ng ml 1, GS Z8, pT3b or GPSM58 score

Z10) to increase the homogeneity of patient characteristics. The two sets were pooled to improve analytic power, resulting in a data set of 594 patients. The patient characteristics of the pooled data set can be found in Supplementary Table 1.

A representative Probe Selection Region for the genomic span of the short and long NEAT1 isoforms was selected by minimizing the technical variance across the pooled data set. Based on these two Probe Selection Regions, the prognostic performance of NEAT1 short and long isoforms was evaluated using univariable and multivariable odds ratios, and area under the receiver operating characteristics curve for BCR, MET, PCSM and GS47 endpoints. KM curves were used to perform survival analysis on the Mayo casecohort patients only35, as the nested casecontrol cohort58 was not suitable for KM analysis.

References

1. Heinlein, C. A. & Chang, C. Androgen receptor in prostate cancer. Endocr. Rev.

25, 276308 (2004).

2. de Bono, J. S. et al. Abiraterone and increased survival in metastatic prostate cancer. New Engl. J. Med. 364, 19952005 (2011).

3. Scher, H. I. et al. Increased survival with enzalutamide in prostate cancer after chemotherapy. New Engl. J. Med. 367, 11871197 (2012).

4. Ricke, W. A. et al. Prostatic hormonal carcinogenesis is mediated by in situ estrogen production and estrogen receptor alpha signaling. FASEB J. 22, 15121520 (2008).

5. Clemson, C. M. et al. An architectural role for a nuclear noncoding RNA: NEAT1 RNA is essential for the structure of paraspeckles. Mol. Cell 33, 717726 (2009).

6. Rhodes, D. R. et al. Oncomine 3.0: genes, pathways, and networks in a collection of 18,000 cancer gene expression proles. Neoplasia 9, 166180 (2007).

7. Arredouani, M. S. et al. Identication of the transcription factor single-minded homologue 2 as a potential biomarker and immunotherapy target in prostate cancer. Clin. Cancer Res. 15, 57945802 (2009).

8. Barwick, B. G. et al. Prostate cancer genes associated with TMPRSS2-ERG gene fusion and prognostic of biochemical recurrence in multiple cohorts. Br. J. Cancer 102, 570576 (2010).

9. Best, C. J. et al. Molecular alterations in primary prostate cancer after androgen ablation therapy. Clin. Cancer Res. 11, 68236834 (2005).

10. Glinsky, G. V., Glinskii, A. B., Stephenson, A. J., Hoffman, R. M. & Gerald, W.L. Gene expression proling predicts clinical outcome of prostate cancer. J. Clin. Invest. 113, 913923 (2004).11. Grasso, C. S. et al. The mutational landscape of lethal castration-resistant prostate cancer. Nature 487, 239243 (2012).

12. Lapointe, J. et al. Gene expression proling identies clinically relevant subtypes of prostate cancer. Proc. Natl Acad. Sci. USA 101, 811816 (2004).

13. LaTulippe, E. et al. Comprehensive gene expression analysis of prostate cancer reveals distinct transcriptional programs associated with metastatic disease. Cancer Res. 62, 44994506 (2002).

14. Liu, P. et al. Sex-determining region Y box 4 is a transforming oncogene in human prostate cancer cells. Cancer Res. 66, 40114019 (2006).

15. Luo, J. et al. Human prostate cancer and benign prostatic hyperplasia: molecular dissection by gene expression proling. Cancer Res. 61, 46834688 (2001).

16. Singh, D. et al. Gene expression correlates of clinical prostate cancer behavior. Cancer Cell 1, 203209 (2002).

17. Taylor, B. S. et al. Integrative genomic proling of human prostate cancer. Cancer Cell 18, 1122 (2010).

18. Tomlins, S. A. et al. Integrative molecular concept modeling of prostate cancer progression. Nat. Genet. 39, 4151 (2007).

NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 15

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6383

19. Vanaja, D. K., Cheville, J. C., Iturria, S. J. & Young, C. Y. Transcriptional silencing of zinc nger protein 185 identied by expression prolingis associated with prostate cancer progression. Cancer Res. 63, 38773882 (2003).

20. Varambally, S. et al. Integrative genomic and proteomic analysis of prostate cancer reveals signatures of metastatic progression. Cancer Cell 8, 393406 (2005).

21. Wallace, T. A. et al. Tumor immunobiological differences in prostate cancer between African-American and European-American men. Cancer Res. 68, 927936 (2008).

22. Yu, Y. P. et al. Gene expression alterations in prostate cancer predicting tumor aggression and preceding development of malignancy. J. Clin. Oncol. 22, 27902799 (2004).

23. Lin, C. Y. et al. Discovery of estrogen receptor alpha target genes and response elements in breast tumor cells. Genome Biol. 5, R66 (2004).

24. Romano, A. et al. Identication of novel ER-alpha target genes in breast cancer cells: gene- and cell-selective co-regulator recruitment at target promoters determines the response to 17beta-estradiol and tamoxifen. Mol. Cell. Endocrinol. 314, 90100 (2010).

25. Yu, J. et al. An integrated network of androgen receptor, polycomb, and TMPRSS2-ERG gene fusions in prostate cancer progression. Cancer Cell 17, 443454 (2010).

26. Giannopoulou, E. G. & Elemento, O. An integrated ChIP-seq analysis platform with customizable workows. BMC Bioinformatics 12, 277 (2011).

27. Bond, C. S. & Fox, A. H. Paraspeckles: nuclear bodies built on long noncoding RNA. J. Cell Biol. 186, 637644 (2009).

28. Holzbeierlein, J. et al. Gene expression analysis of human prostate carcinoma during hormonal therapy identies androgen-responsive genes and mechanisms of therapy resistance. Am. J. Pathol. 164, 217227 (2004).

29. Magee, J. A. et al. Expression proling reveals hepsin overexpression in prostate cancer. Cancer Res. 61, 56925696 (2001).

30. Tamura, K. et al. Molecular features of hormone-refractory prostate cancer cells by genome-wide gene expression proles. Cancer Res. 67, 51175125 (2007).

31. Welsh, J. B. et al. Analysis of gene expression identies candidate markers and pharmacological targets in prostate cancer. Cancer Res. 61, 59745978 (2001).

32. Barbieri, C. E. et al. Exome sequencing identies recurrent SPOP, FOXA1 and MED12 mutations in prostate cancer. Nat. Genet. 44, 685689 (2012).

33. Cerami, E. et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2, 401404 (2012).

34. Gao, J. et al. Integrative analysis of complex cancer genomics and clinical proles using the cBioPortal. Sci. Signal. 6, pl1 (2013).

35. Erho, N. et al. Discovery and validation of a prostate cancer genomic classier that predicts early metastasis following radical prostatectomy. PLoS ONE 8, e66855 (2013).

36. Karnes, R. J. et al. Validation of a genomic classier that predicts metastasis following radical prostatectomy in an at risk patient population. J. Urol. 190, 20472053 (2013).

37. Rickman, D. S. et al. ERG cooperates with androgen receptor in regulating trefoil factor 3 in prostate cancer disease progression. Neoplasia 12, 10311040 (2010).

38. Nakagawa, S., Naganuma, T., Shioi, G. & Hirose, T. Paraspeckles are subpopulation-specic nuclear bodies that are not essential in mice. J. Cell Biol. 193, 3139 (2011).

39. Guttman, M. et al. Chromatin signature reveals over a thousand highly conserved large non-coding RNAs in mammals. Nature 458, 223227 (2009).

40. Burger, M. J. et al. Expression analysis of delta-catenin and prostate-specic membrane antigen: their potential as diagnostic markers for prostate cancer. Int. J. Cancer 100, 228237 (2002).

41. Gumulec, J. et al. Evaluation of alpha-methylacyl-CoA racemase, metallothionein and prostate specic antigen as prostate cancer prognostic markers. Neoplasma 59, 191201 (2012).

42. Jiang, N., Zhu, S., Chen, J., Niu, Y. & Zhou, L. A-methylacyl-CoA racemase (AMACR) and prostate-cancer risk: a meta-analysis of 4,385 participants. PLoS ONE 8, e74386 (2013).

43. Ross, J. S. et al. Correlation of primary tumor prostate-specic membrane antigen expression with disease recurrence in prostate cancer. Clin. Cancer Res. 9, 63576362 (2003).

44. Xiao, Z. et al. Quantitation of serum prostate-specic membrane antigen by a novel protein biochip immunoassay discriminates benign from malignant prostate disease. Cancer Res. 61, 60296033 (2001).

45. Chu, C., Quinn, J. & Chang, H. Y. Chromatin isolation by RNA purication (ChIRP). J. Vis. Exp. 61, pii: 3912 (2012).

46. Bernstein, E. & Allis, C. D. RNA meets chromatin. Genes Dev. 19, 16351655 (2005).

47. Khalil, A. M. et al. Many human large intergenic noncoding RNAs associate with chromatin-modifying complexes and affect gene expression. Proc. Natl Acad. Sci. USA 106, 1166711672 (2009).

48. Tsai, M. C. et al. Long noncoding RNA as modular scaffold of histone modication complexes. Science 329, 689693 (2010).

49. Wang, K. C. et al. A long noncoding RNA maintains active chromatin to coordinate homeotic gene expression. Nature 472, 120124 (2011).

50. Bonkhoff, H., Fixemer, T., Hunsicker, I. & Remberger, K. Estrogen receptor expression in prostate cancer and premalignant prostatic lesions. Am. J. Pathol. 155, 641647 (1999).

51. Setlur, S. R. et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J. Natl Cancer Inst. 100, 815825 (2008).

52. Singh, P. B., Matanhelia, S. S. & Martin, F. L. A potential paradox in prostate adenocarcinoma progression: oestrogen as the initiating driver. Eur. J. Cancer 44, 928936 (2008).

53. Nicolaiew, N. et al. Association between estrogen and androgen receptor genes and prostate cancer risk. Eur. J. Endocrinol. 160, 101106 (2009).

54. Gibb, E. A. et al. Human cancer long non-coding RNA transcriptomes. PLoS ONE 6, e25915 (2011).

55. Amaral, P. P., Clark, M. B., Gascoigne, D. K., Dinger, M. E. & Mattick, J. S. lncRNAdb: a reference database for long noncoding RNAs. Nucleic Acids Res. 39, D146D151 (2011).

56. Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. R. Stat. Soc. Series B (Methodol.) 57, 289300 (1995).

57. Rhodes, D. R. et al. ONCOMINE: a cancer microarray database and integrated data-mining platform. Neoplasia 6, 16 (2004).

58. Blute, M. L., Bergstralh, E. J., Iocca, A., Scherer, B. & Zincke, H. Use of Gleason score, prostate specic antigen, seminal vesicle and margin status to predict biochemical failure after radical prostatectomy. J. Urol. 165, 119125 (2001).

Acknowledgements

This work was supported by NCI R01 CA152057 (A.S. and M.A.R.), the Early Detection Research Network NCI U01 CA111275 (M.A.R.) and the Prostate Cancer Foundation Young Investigator award (D.C. and M.B.). We thank Francesca Demichelis for identication of somatic copy number variations, Wasay Hussain for the RNA sequencing analysis, Anastas Popratiloff from George Washington University for advice on imaging. We also thank John S. Mattick and Marcel Dinger for their help in accessing and using data from http://www.lncrnadb.org

Web End =www.lncrnadb.org . This investigation received editorial support from grant UL1TR000457 of the Clinical and Translation Science Center at Weill Cornell Medical College.

Author contributions

Conception and design: D.C., S.S.N. and M.A.R. Development of reagents and methodology: D.C., S.S.N., A.S., A.H.F., S.N., T.H., R.L., S.H., N.P. and N.H.B. Acquisition of data: D.C., A.S., K.P., Z.C., N.E., I.A.V., M.G., E.D., R.B.J., T.Y.M, J.F., J.P. and M.K. Analysis and interpretation of data: D.C., A.S., S.S.N., J.M.M., E.G. and M.A.R. Writing, review and/or revision of the manuscript: D.C., S.S.N., A.S., O.E., H.B. and M.A.R. Administrative, technical or material support (that is, reporting or organizing data and constructing databases): D.C., A.S. and M.A.R. Study supervision: M.A.R.

Additional information

Accession codes: Genome-wide data generated in this study have been deposited with The Gene Expression Omnibus (GEO) under accession number GSE43988.

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Chakravarty, D. et al. The oestrogen receptor alpha-regulated lncRNA NEAT1 is a critical modulator of prostate cancer. Nat. Commun. 5:5383doi: 10.1038/ncomms6383 (2014).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

16 NATURE COMMUNICATIONS | 5:5383 | DOI: 10.1038/ncomms6383 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

Word count: 14398

Show less

Abstract

Translate

The androgen receptor (AR) plays a central role in establishing an oncogenic cascade that drives prostate cancer progression. Some prostate cancers escape androgen dependence and are often associated with an aggressive phenotype. The oestrogen receptor alpha (ERα) is expressed in prostate cancers, independent of AR status. However, the role of ERα remains elusive. Using a combination of chromatin immunoprecipitation (ChIP) and RNA-sequencing data, we identified an ERα-specific non-coding transcriptome signature. Among putatively ERα-regulated intergenic long non-coding RNAs (lncRNAs), we identified nuclear enriched abundant transcript 1 (NEAT1) as the most significantly overexpressed lncRNA in prostate cancer. Analysis of two large clinical cohorts also revealed that NEAT1 expression is associated with prostate cancer progression. Prostate cancer cells expressing high levels of NEAT1 were recalcitrant to androgen or AR antagonists. Finally, we provide evidence that NEAT1 drives oncogenic growth by altering the epigenetic landscape of target gene promoters to favour transcription.

Details

Title

The oestrogen receptor alpha-regulated lncRNA NEAT1 is a critical modulator of prostate cancer

Author

Chakravarty, Dimple; Sboner, Andrea; Nair, Sujit S; Giannopoulou, Eugenia; Li, Ruohan; Hennig, Sven; Mosquera, Juan Miguel; Pauwels, Jonathan; Park, Kyung; Kossai, Myriam; Macdonald, Theresa Y; Fontugne, Jacqueline; Erho, Nicholas; Vergara, Ismael A; Ghadessi, Mercedeh; Davicioni, Elai; Jenkins, Robert B; Palanisamy, Nallasivam; Chen, Zhengming; Nakagawa, Shinichi; Hirose, Tetsuro; Bander, Neil H; Beltran, Himisha; Fox, Archa H; Elemento, Olivier; Rubin, Mark A

Pages

5383

Publication year

2014

Publication date

Nov 2014

Publisher

Nature Publishing Group

e-ISSN

20411723

Source type

Scholarly Journal

Language of publication

English

DOI

https://doi.org/10.1038/ncomms6383

ProQuest document ID

1626667722

The oestrogen receptor alpha-regulated lncRNA NEAT1 is a critical modulator of prostate cancer

Jump to:

Full Text

Abstract

Details

Suggested sources