Abstract
Recurrent gene fusions involving ETS family genes are a distinguishing feature of human prostate cancers, with TMPRSS2–ERG fusions representing the most common subtype. The TMPRSS2–ERG fusion transcript and its splice variants are well characterized in prostate cancers; however, not much is known about the levels and regulation of wild-type ERG. By employing an integrative approach, we show that the TMPRSS2–ERG gene fusion product binds to the ERG locus and drives the overexpression of wild-type ERG in prostate cancers. Knockdown of TMPRSS2–ERG in VCaP cells resulted in the downregulation of wild-type ERG transcription, whereas stable overexpression of TMPRSS2–ERG in the gene fusion-negative PC3 cells was associated with the upregulation of wild-type ERG transcript. Further, androgen signaling-mediated upregulation of TMPRSS2–ERG resulted in the concomitant upregulation of wild-type ERG transcription in VCaP cells. The loss of wild-type ERG expression was associated with a decrease in the invasive potential of VCaP cells. Importantly, 38% of clinically localized prostate cancers and 27% of metastatic prostate cancers harboring the TMPRSS2–ERG gene fusions exhibited overexpression of wild-type ERG. Taken together, these results provide novel insights into the regulation of ERG in human prostate cancers. Cancer Res; 71(16); 5387–92. ©2011 AACR.
Introduction
Recurrent gene fusions involving ETS family genes are observed in a majority of human prostate cancers and have potential implications for diagnosis, prognosis, and therapy (1–3). These gene fusions typically result in the juxtaposition of the 5′-untranslated region of androgen-regulated genes to the oncogenic ETS family genes, resulting in their overexpression. The TMPRSS2–ERG fusions represent the most common subtype among ETS fusions, with a prevalence of approximately 50% in clinically localized prostate cancers (1, 4). Additional ETS family genes, which together account for approximately 10% of prostate cancer gene fusions, include ETV1 (5), ETV4 (6), and ETV5 (7). TMPRSS2 is an androgen-regulated gene located 3 Mb upstream of ERG in human chromosome 21q22.2, the upstream regulatory elements and promoter of which drive the overexpression of ERG upon the formation of gene fusion. Androgen signaling has been shown to induce the proximity of TMPRSS2 and ERG locus in androgen responsive cells, and in combination with agents causing DNA double-strand breaks induces TMPRSS2–ERG gene fusions (8–10). In prostate cancer specimens, TMPRSS2–ERG gene fusions can be formed with the deletion of the intervening sequences (referred to as Edel) or without the deletion (11, 12). Castration-resistant metastatic prostate cancer sites harboring TMPRSS2–ERG are associated with Edel, suggesting that Edel is more aggressive (13). To date, multiple TMPRSS2–ERG isoforms and splice variants with variable biological activities have been described (14–17). For example, inclusion of a 72-bp exon has been reported to significantly enhance proliferation of prostate-derived cells (17). However, not much is known about the levels and regulation of wild-type ERG in prostate cancers. By employing an integrative approach, we show that the TMPRSS2–ERG gene fusion product binds to the ERG locus and drives the overexpression of wild-type ERG in human prostate cancers and suggest functional and clinical implications of this feed-forward regulation of ERG expression.
Materials and Methods
Samples
Prostate tissues were from the radical prostatectomy series and the Rapid Autopsy Program, which are both part of the University of Michigan Prostate Cancer Specialized Program of Research Excellence (SPORE) Tissue Core. Samples were collected with informed consent and prior institutional review board approval. Total RNA was isolated with Trizol (Invitrogen) according to the manufacturer's instructions.
Cell culture and cloning
VCaP cells were cultured in Dulbecco's modified Eagle's medium high glucose medium containing 10% FBS; LNCaP cells, PC3 cells stably overexpressing TMPRSS2–ERG (PC3-TMPRSS2–ERG), and PC3 cells stably overexpressing LacZ (PC3-LacZ) were cultured in RPMI 1640 medium containing 10% FBS in a 5% CO2 humidified incubator. Parental VCaP, LNCaP, and PC3 cells were obtained from American Type Culture Collection and were passaged for fewer than 6 months. Lentiviral LACZ and TMPRSS2:ERG-3×FLAG overexpression vectors were created by cloning into the pLL_ires_GFP backbone available from the University of Michigan vector core. Briefly, the coding sequence was amplified by PCR, digested with Xho1 and Xba1, and ligated into the pLL vector. The TMPRSS2–ERG variant used is the most prevalent gene fusion product [TMPRSS2 exon 1 fused to ERG exon 2 (NM_182918.3)]. Lentivirus was created at the University of Michigan virus core and used to transduce PC3 cells in the presence of 4μg/mL polybrene (Sigma). PC3-LACZ and PC3-TMPRSS2–ERG cells were then sorted for green fluorescent protein (GFP) expression (top 25%) and used for subsequent assays. Genetic identity of the cells was confirmed by genotyping. GFP expression was monitored every 3 days, and cells were kept in culture for no more than 4 weeks. The adenoviral constructs for ERG (exons 2–10) and LacZ are described elsewhere (4).
Chromatin immunoprecipitation and chromatin immunoprecipitation-seq analysis
Published chromatin immunoprecipitation (ChIP)-seq sequence data were downloaded from the following GEO datasets: VCaP H3K36me3 (GSM353624), LNCaP H3K36me3 (GSM353627), VCaP ERG (GSM353647), LNCaP ERG (GSM353648), and VCaP Pan-H3 (GSM353622; ref. 18). LNCaP Pan-H3 ChIP, VCaP Input DNA, and LNCaP Input DNA sequence data were prepared into libraries and sequenced by using the Genome Analyzer (Illumina) following manufacturer's protocols. The detailed ChIP procedure is described elsewhere (18). Reads were mapped to the human genome (hg19) using the BWA v0.5.8 (http://bio-bwa.sourceforge.net) short read aligner (19). Only reads with a single best alignment were considered for further analysis. Enriched DNA–protein binding sites were determined by using the MACS (version 1.4.0beta) peak-calling software using a threshold P value of 1e−5 and default settings for all other parameters (20). Input DNA was used as a control for the ERG peak-calling analyses, and the Pan-H3 antibody ChIP-seq was used as a control for the H3K36me3 analyses. The ChIP quantitative real-time PCR primers for amplifying wild-type ERG (NM_182918.3) promoter are: forward: GTGCTTGCAGCCCGTGTGAA and reverse: TGTCCAGCCCAAAGAAACAGGATA. The antibodies for ChIP assays are rabbit IgG (Santa Cruz Biotechnology, #sc-2027), rabbit α-FLAG (Sigma, #F7425), rabbit α-ERG (Epitomics, #2805-1).
Results and Discussion
We first studied the ERG locus in the androgen-sensitive VCaP and LNCaP cell lines by using our previously published ChIP-seq compendium (18). Specifically, we analyzed the landscape of trimethylated lysine 36 of histone H3 (H3K36me3) enrichment and ERG binding in these cell lines. The genes actively transcribed by RNA polymerase II (Pol II) are marked by H3K36me3 enrichment along the length of the transcribed region (21). As VCaP cells endogenously harbor the TMPRSS2–ERG gene fusion which produces a chimeric transcript comprising of TMPRSS2 (exon 1) and ERG (exons 2–10), the ERG locus is enriched for the H3K36me3 mark in this cell line (Fig. 1A). By contrast, the LNCaP cells do not harbor the TMPRSS2–ERG gene fusion, lack ERG expression, and do not exhibit H3K36me3 enrichment at the ERG locus. Interestingly, the H3K36me3 epigenetic mark spanned the entire ERG locus (exons 1–10) in VCaP cells, even though gene fusion drives the overexpression of ERG exon 2 and beyond, indicating that the entire ERG locus is transcriptionally active in these cells. High expression levels of ERG exon1 was observed in VCaP cells, suggesting that in addition to the TMPRSS2–ERG chimeric transcript, these cells also express wild-type ERG (Supplementary Fig. S1). The results of 5′ rapid amplication of cDNA ends experiments indicated that Exon 1 of ERG represents bona fide wild-type ERG, and its expression is not driven by fusion to any other gene (Supplementary Fig. S2). In addition, we observed strong binding of the ERG protein to the ERG locus in VCaP cells, but not in the LNCaP cells (Fig. 1B). The majority of the ERG enrichment peaks clustered near exon 1 of wild-type ERG. Taken together, these observations suggest that in addition to the TMPRSS2–ERG transcript, VCaP cells have high levels of wild-type ERG transcript, the expression of which is presumably regulated by ERG protein.
ChIP-seq coverage data showing the H3K36me3 chromatin mark and the ERG transcription factor binding at the ERG locus in VCaP and LNCaP cells. An arrow denotes the direction of ERG transcription; vertical bars represent exons. In each coverage plot, the y-axis represents the number of overlapping reads at each position, normalized to the reads per kilobase per million metric. Significant enrichment (P < 10−5) is denoted by blue boxes below the raw data. A, the transcriptionally active ERG locus in VCaP cells is marked by significant enrichment of the H3K36me3 mark. The TMPRSS2–ERG gene fusion results in upregulation of ERG exon 2 (denoted by the red arrowhead) and beyond in VCaP cells. However, the H3K36me3 domain extends upstream of ERG exon 2, suggesting that the entire ERG locus is transcriptionally active (and not just the exons driven by gene fusion). By contrast, LNCaP cells lack H3K36me3 enrichment. B, the ERG protein binds to multiple locations within the ERG locus in VCaP cells but not LNCaP cells.
ChIP-seq coverage data showing the H3K36me3 chromatin mark and the ERG transcription factor binding at the ERG locus in VCaP and LNCaP cells. An arrow denotes the direction of ERG transcription; vertical bars represent exons. In each coverage plot, the y-axis represents the number of overlapping reads at each position, normalized to the reads per kilobase per million metric. Significant enrichment (P < 10−5) is denoted by blue boxes below the raw data. A, the transcriptionally active ERG locus in VCaP cells is marked by significant enrichment of the H3K36me3 mark. The TMPRSS2–ERG gene fusion results in upregulation of ERG exon 2 (denoted by the red arrowhead) and beyond in VCaP cells. However, the H3K36me3 domain extends upstream of ERG exon 2, suggesting that the entire ERG locus is transcriptionally active (and not just the exons driven by gene fusion). By contrast, LNCaP cells lack H3K36me3 enrichment. B, the ERG protein binds to multiple locations within the ERG locus in VCaP cells but not LNCaP cells.
We next conducted gene silencing experiments to determine the role of TMPRSS2–ERG and wild-type ERG transcripts in VCaP cells. We designed siRNAs that specifically target (i) TMPRSS2–ERG transcript, (ii) wild-type ERG transcript, or (iii) both TMPRSS2–ERG and wild-type ERG transcripts. Specific knockdown of TMPRSS2–ERG transcript resulted in a reduction in levels of wild-type ERG transcript in addition to total ERG transcription as represented by ERG-ALL primers (Fig. 2A). However, specific knockdown of wild-type ERG transcript resulted in reduction of total ERG transcription without affecting the levels of TMPRSS2–ERG transcript. These results indicate that ERG protein can upregulate the levels of wild-type ERG transcription, but not the levels of TMPRSS2–ERG transcript. Further support is obtained by the observation that PC3 cells stably overexpressing TMPRSS2–ERG show higher transcript levels of wild-type ERG as compared with LacZ controls (Fig. 2B). This upregulation is associated with the binding of FLAG-tagged ERG protein to the promoter of wild-type ERG (Fig. 2B). Similar results were obtained upon ectopic overexpression of the TMPRSS2–ERG gene fusion product in LNCaP cells (Supplementary Fig. S3).
TMPRSS2–ERG-mediated feed-forward regulation of ERG. A, schematic representation of the wild-type ERG (ERG-WT, accession: NM_182918.3) and TMPRSS2–ERG transcript (top). The locations of primers for gene expression analysis (solid line) and siRNA sequences (dotted line) are shown. The TMPRSS2–ERG siRNA and ERG-WT siRNA specifically target the TMPRSS2–ERG and wild-type ERG transcripts, respectively. The ERG-ALL siRNA targets both these transcripts. The TMPRSS2–ERG and ERG-WT primers specifically detect TMPRSS2–ERG and wild-type ERG transcripts, respectively; the ERG-ALL primers detect both these transcripts. Targeted knockdown of TMPRSS2–ERG transcript is associated with the downregulation of wild-type ERG transcript. However, targeted knockdown of wild-type ERG does not lead to the downregulation of the TMPRSS2–ERG transcript. *, P < 0.01 by Student's t-test. B, lentivirus-mediated stable overexpression of FLAG-tagged TMPRSS2–ERG fusion product in PC3 cells results in the upregulation of wild-type ERG transcript (top). ChIP assay with α-FLAG antibody shows binding of ERG protein to the promoter of wild-type ERG locus (bottom). C, stimulation with 1 nmol/L R1881, a synthetic androgen agonist, for 24 hours resulted in the upregulation of both the TMPRSS2–ERG and wild-type ERG transcripts. D, knockdown of TMPRSS2–ERG or wild-type ERG transcripts individually results in reduced ERG protein levels. Maximal reduction of ERG protein levels is observed with the ERG-ALL siRNA that targets both the TMPRSS2–ERG and wild-type ERG transcripts. Starvation or androgen deprivation also results in a significant reduction in ERG protein levels.
TMPRSS2–ERG-mediated feed-forward regulation of ERG. A, schematic representation of the wild-type ERG (ERG-WT, accession: NM_182918.3) and TMPRSS2–ERG transcript (top). The locations of primers for gene expression analysis (solid line) and siRNA sequences (dotted line) are shown. The TMPRSS2–ERG siRNA and ERG-WT siRNA specifically target the TMPRSS2–ERG and wild-type ERG transcripts, respectively. The ERG-ALL siRNA targets both these transcripts. The TMPRSS2–ERG and ERG-WT primers specifically detect TMPRSS2–ERG and wild-type ERG transcripts, respectively; the ERG-ALL primers detect both these transcripts. Targeted knockdown of TMPRSS2–ERG transcript is associated with the downregulation of wild-type ERG transcript. However, targeted knockdown of wild-type ERG does not lead to the downregulation of the TMPRSS2–ERG transcript. *, P < 0.01 by Student's t-test. B, lentivirus-mediated stable overexpression of FLAG-tagged TMPRSS2–ERG fusion product in PC3 cells results in the upregulation of wild-type ERG transcript (top). ChIP assay with α-FLAG antibody shows binding of ERG protein to the promoter of wild-type ERG locus (bottom). C, stimulation with 1 nmol/L R1881, a synthetic androgen agonist, for 24 hours resulted in the upregulation of both the TMPRSS2–ERG and wild-type ERG transcripts. D, knockdown of TMPRSS2–ERG or wild-type ERG transcripts individually results in reduced ERG protein levels. Maximal reduction of ERG protein levels is observed with the ERG-ALL siRNA that targets both the TMPRSS2–ERG and wild-type ERG transcripts. Starvation or androgen deprivation also results in a significant reduction in ERG protein levels.
We next determined if androgen signaling could affect the levels of TMPRSS2–ERG and wild-type ERG in VCaP cells. Administration of synthetic androgen (R1881) resulted in robust upregulation of both TMPRSS2–ERG and wild-type ERG transcription (Fig. 2C). Given the observation that ERG was not androgen responsive in a TMPRSS2–ERG negative setting (e.g., LNCaP cells; ref. 2), we hypothesized that upregulation of wild-type ERG by androgen is mediated by TMPRSS2–ERG. To study the effects of siRNA against different ERG transcripts and androgen depletion (starvation) in regulating ERG protein levels, immunoblot analysis was conducted (Fig. 2D). Although siRNA against TMPRSS2–ERG or wild-type ERG reduced the levels of ERG protein, maximal reduction of ERG protein was obtained with ERG-ALL siRNA that targets all the ERG transcripts and also starvation or androgen deprivation. The proteins encoded by wild-type ERG and TMPRSS2–ERG transcripts are indistinguishable in terms of molecular weight. These results suggest that (i) both TMPRSS2–ERG and wild-type ERG transcripts contribute toward total ERG protein levels in VCaP cells, and (ii) the protein product encoded by TMPRSS2–ERG upregulates the levels of wild-type ERG.
To determine the functional consequence of the presence of wild-type ERG and TMPRSS2–ERG transcripts in VCaP cells, invasion assays were conducted. Knockdown of TMPRSS2–ERG and wild-type ERG individually or in a combined fashion with the ERG-ALL siRNA resulted in a significant loss of cell invasion (Fig. 3A). Although TMPRSS2–ERG siRNA and wild-type ERG siRNA individually do not result in maximal loss of ERG protein unlike ERG-ALL siRNA (Fig. 2D), the results of invasion assay indicate that VCaP cells are ultrasensitive to ERG levels. Given that wild-type ERG has functional relevance in cell-based assays, we next determined if TMPRSS2–ERG gene fusion upregulates wild-type ERG expression in clinical samples of prostate cancer. The expression levels of TMPRSS2–ERG, wild-type ERG, and total ERG were measured in a panel of benign prostates (n = 13), clinically localized prostate cancers (n = 67), and metastatic prostate cancers (n = 29; Fig. 3B). None of the benign prostate tissues expressed either TMPRSS2–ERG or wild-type ERG transcripts. In clinically localized prostate cancers, 37.5% (12/32) of TMPRSS2–ERG positive cases and none (0/35) of the TMPRSS2–ERG negative cases exhibited high levels of wild-type ERG (Fig. 3C, and Supplementary Table S1). Similarly, 27.3% (3/11) of TMPRSS2–ERG positive metastatic prostate cancers expressed high levels of wild-type ERG, whereas none (0/18) of the TMPRSS2–ERG negative metastatic prostate cancers showed high wild-type ERG expression. By analyzing published prostate cancer array CGH datasets (22, 23), we could rule out ERG locus amplification as a possible cause of increased wild-type ERG expression (Supplementary Fig. S4). Taken together, these results suggest that TMPRSS2–ERG-mediated upregulation of wild-type ERG transcription occurs in a subset of gene fusion positive prostate cancers.
Functional studies and clinical relevance of TMPRSS2–ERG-mediated feed-forward regulation of ERG. A, a reconstituted basement membrane invasion chamber assay (Boyden chamber assay) was used to assess the invasive potential of VCaP cells treated with siRNA's against ERG isoforms. Representative fields of invaded and stained cells are shown (left). Invasion was quantified using colorimetry (absorbance at 560 nm, right). *, P < 0.01 by Student's t-test (right). B, matrix representation of the transcript status of ERG isoforms in benign prostate, clinically localized prostate cancer (PCA), and metastatic prostate cancer (MET) as measured by quantitative reverse transcription PCR. Each column represents 1 case; the top, middle, and bottom rows represent the status of wild-type ERG transcript (ERG-WT), TMPRSS2–ERG fusion transcript, and all ERG transcripts (ERG-ALL), respectively. A colored cell indicates that the tissue is positive for the corresponding transcript. C, scatter plot representing the relative expression levels of wild-type ERG in TMPRSS2–ERG negative and positive clinically localized prostate cancers. Vertical axes represent the relative expression levels; the red bars represent the mean, dashed line represents the threshold used to classify samples as either wild-type ERG positive or negative. For clarity, points have been horizontally displaced within each sample class. D, proposed model for TMPRSS2–ERG-mediated feed-forward regulation of ERG. Androgen signaling can increase the levels of TMPRSS2–ERG transcript, whereas antiandrogens have the opposite effect. The presence of TMPRSS2–ERG transcript results in the production of ERG protein, which then binds to the promoter of wild-type ERG to drive its transcription resulting in the production of wild-type ERG protein, which is functionally identical to the TMPRSS2–ERG gene product. This creates a feed-forward loop that upregulates wild-type ERG transcription.
Functional studies and clinical relevance of TMPRSS2–ERG-mediated feed-forward regulation of ERG. A, a reconstituted basement membrane invasion chamber assay (Boyden chamber assay) was used to assess the invasive potential of VCaP cells treated with siRNA's against ERG isoforms. Representative fields of invaded and stained cells are shown (left). Invasion was quantified using colorimetry (absorbance at 560 nm, right). *, P < 0.01 by Student's t-test (right). B, matrix representation of the transcript status of ERG isoforms in benign prostate, clinically localized prostate cancer (PCA), and metastatic prostate cancer (MET) as measured by quantitative reverse transcription PCR. Each column represents 1 case; the top, middle, and bottom rows represent the status of wild-type ERG transcript (ERG-WT), TMPRSS2–ERG fusion transcript, and all ERG transcripts (ERG-ALL), respectively. A colored cell indicates that the tissue is positive for the corresponding transcript. C, scatter plot representing the relative expression levels of wild-type ERG in TMPRSS2–ERG negative and positive clinically localized prostate cancers. Vertical axes represent the relative expression levels; the red bars represent the mean, dashed line represents the threshold used to classify samples as either wild-type ERG positive or negative. For clarity, points have been horizontally displaced within each sample class. D, proposed model for TMPRSS2–ERG-mediated feed-forward regulation of ERG. Androgen signaling can increase the levels of TMPRSS2–ERG transcript, whereas antiandrogens have the opposite effect. The presence of TMPRSS2–ERG transcript results in the production of ERG protein, which then binds to the promoter of wild-type ERG to drive its transcription resulting in the production of wild-type ERG protein, which is functionally identical to the TMPRSS2–ERG gene product. This creates a feed-forward loop that upregulates wild-type ERG transcription.
TMPRSS2–ERG, the first recurrent gene fusion to be identified in human prostate cancers, is also the most prevalent. Although it is apparent that the androgen-responsive TMPRSS2 promoter drives the overexpression of ERG, we identified a novel TMPRSS2–ERG-mediated feed-forward loop that upregulates wild-type ERG, thereby, increasing the steady-state levels of ERG protein (Fig. 3D). We have further subclassified TMPRSS2–ERG-positive prostate cancers in to 2 groups based on wild-type ERG expression. In the future, it will be interesting to determine if the 2 groups have distinguishing clinical features that are prognostically or therapeutically relevant. Why TMPRSS2–ERG-mediated upregulation of ERG gets selected in human prostate cancers is not very clear, however, some possibilities do exist. Because ERG is a key driver of cancer progression, any change in cellular circuitry that increases the steady-state ERG protein levels may be positively selected. This may also result in oncogene addiction, and it will be interesting to test if prostate cancers with TMPRSS2–ERG and wild-type ERG represent a subtype that is ultrasensitive to ERG inhibition, as this appears to be the case with VCaP cells. Thus, by employing an integrative approach, we have identified a novel mechanism of regulation of ERG transcription that holds functional significance and clinical relevance.
Disclosure of Potential Conflicts of Interest
A.M. Chinnaiyan serves on the advisory board of Gen-Probe, and is a coinventor on a patent filed by the University of Michigan covering the diagnostic and therapeutic field of use for ETS fusions in prostate cancer. The diagnostic field of use has been licensed to Gen-Probe, Inc. Gen-Probe did not play a role in the design and conduct of this study, in the collection, analysis, or interpretation of the data, or in the preparation, review, or approval of the article. The other authors disclosed no potential conflicts of interest.
Acknowledgments
We thank Ken Pienta for establishing the UM Rapid Autopsy Series, J. Yu for assistance with ChIP-seq experiments, S. Patel and M. Tran for technical assistance, and C. Maher for useful discussions.
Grant Support
This work was supported in part by the National Institutes of Health Prostate SPORE P50CA69568, Early Detection Research Network U01 CA111275, R01CA132874, and the Prostate Cancer Foundation. A.M. Chinnaiyan is supported by a Burroughs Welcome Foundation Award in Clinical Translational Research, a Doris Duke Charitable Foundation Distinguished Clinical Investigator Award, and an American Cancer Society Research Professor Award. R.S. Mani is supported by the Stewart Rahr–PCF Young Investigator Award from the Prostate Cancer Foundation.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.