Although common in hematologic and mesenchymal malignancies, recurrent gene fusions have not been well characterized in epithelial carcinomas. Recently, using a novel bioinformatic approach, we identified recurrent gene fusions between TMPRSS2 and the ETS family members ERG or ETV1 in the majority of prostate cancers. Here, we interrogated the expression of all ETS family members in prostate cancer profiling studies and identified marked overexpression of ETV4 in 2 of 98 cases. In one such case, we confirmed the overexpression of ETV4 using quantitative PCR, and by rapid amplification of cDNA ends, quantitative PCR, and fluorescence in situ hybridization, we show that the TMPRSS2 (21q22) and ETV4 (17q21) loci are fused in this case. This result defines a third molecular subtype of prostate cancer and supports the hypothesis that dysregulation of ETS family members through fusions with TMRPSS2 may be an initiating event in prostate cancer development. (Cancer Res 2006; 66(7): 3396-400)

Despite their well-defined role in hematologic and mesenchymal malignancies, recurrent gene fusions have not been well characterized in epithelial carcinomas (13). Recently, in an effort to nominate candidate oncogenes from DNA microarray data, we developed a novel bioinformatic approach termed cancer outlier profile analysis (COPA) to identify genes markedly overexpressed in a subset of cancers. Applying the COPA approach to a compendium of tumor gene expression data, the ETS family transcription factors ERG and ETV1 were identified as outliers across prostate cancer profiling studies. Using a variety of molecular techniques, we characterized fusions of the 5′-untranslated region (5′UTR) of TMPRSS2 (21q22) with ERG (21q22) or ETV1 (7p21) in cases that overexpressed the respective ETS family member (4). TMPRSS2 has been characterized previously as being both androgen responsive and highly expressed in the prostate, presumably through androgen response elements (ARE) in the promoter (57). As a possible mechanism driving the overexpression of ETS family members in cases with the gene fusions, we showed that androgen can induce the overexpression of ERG (presumably through AREs) in a TMPRSS2:ERG-positive cell line (4). Together, these results suggested that dysregulation of ETS family activity through AREs upstream of TMPRSS2 may drive prostate cancer development. Here, we describe a rare third molecular subtype of prostate cancer, characterized by fusion of TMPRSS2 to another ETS family member, ETV4.

ETS family expression in profiling studies. To investigate the expression of ETS family members in prostate cancer, we selected two prostate cancer profiling studies (8)8

8

S.A. Tomlins et al., in preparation.

present in the Oncomine database (9). Genes with an ETS domain were identified by the Interpro filter “Ets” (Interpro ID: IPR000418). Heatmap representations were generated in Oncomine using the “median center per gene” option, and the color contrast was set to accentuate ERG and ETV1 differential expression.

Samples. Prostate cancer tissues (PCA1-5) were from the radical prostatectomy series at the University of Michigan, which is part of the University of Michigan Prostate Cancer Specialized Program of Research Excellence Tissue Core. All samples were collected with informed consent of the patients and prior institutional review board approval. Total RNA was isolated with Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. A commercially available pool of benign prostate tissue total RNA (CPP, Clontech, Mountain View, CA) was also used.

Quantitative PCR. Quantitative PCR was done using SYBR Green dye on an Applied Biosystems 7300 Real-time PCR system (Applied Biosystems, Foster City, CA) as described (4). The amount of each target gene relative to the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) for each sample was reported. The relative amount of the target gene was calibrated to the relative amount from the pool of benign prostate tissue (CPP). All oligonucleotide primers were synthesized by Integrated DNA Technologies (Coralville, IA). GAPDH primers were as described (10). Primers for exons of ETV4 were as follows (listed 5′ to 3′): ETV4_exon2-f, CCGGATGGAGCGGAGGATGA; ETV4_exon2-r, CGGGCGATTTGCTGCTGAAG; ETV4_exon3-f, GCCGCCCCTCGACTCTGAA; ETV4_exon4-r, GAGCCACGTCTCCTGGAAGTGACT; ETV4_exon11-f, CTGGCCGGTTCTTCTGGATGC; ETV4_exon12-r, CGGGCCGGGGAATGGAGT; ETV4_3′UTR-f, CCTGGAGGGTACCGGTTTGTCA; ETV4_3′UTR-r, CCGCCTGCCTCTGGGAACAC. Exons were numbered by alignment of the RefSeq for ETV4 (NM_001986.1) with the May 2004 freeze of the human genome using the University of California Santa Cruz Genome Browser. For quantitative PCR confirmation of TMPRSS2:ETV4 fusion transcripts, TMPRSS2:ETV4a-f (AAATAAGTTTGTAAGAGGAGCCTCAGCATC) and TMPRSS2:ETV4b-f (ATCGTAAAGAGCTTTTCTCCCCGC), which detects both TMPRSS2:ETV4a and TMPRSS2:ETV4b transcripts, were used with ETV4_exon4-r.

RNA ligase–mediated rapid amplification of cDNA ends. RNA ligase–mediated rapid amplification of cDNA ends (RLM-RACE) was done using the GeneRacer RLM-RACE kit (Invitrogen), according to the manufacturer's instructions as described (4). To obtain the 5′ end of ETV4, first-strand cDNA from PCA5 was amplified using the GeneRacer 5′ Primer and ETV4_exon4-r or ETV4_exon7-r (GAAAGGGCTGTAGGGGCGACTGT). Products were cloned and sequenced as described (4). Equivalent 5′ ends of the TMPRSS2:ETV4 transcripts were obtained from both primer pairs.

Fluorescence in situ hybridization. Formalin-fixed, paraffin-embedded tissue sections were used for interphase fluorescence in situ hybridization (FISH). Deparaffinized tissue was treated with 0.2 mol/L HCl for 10 minutes, 2× SSC for 10 minutes at 80°C and digested with Proteinase K (Invitrogen) for 10 minutes. The tissues and BAC probes were codenatured for 5 minutes at 94°C and hybridized overnight at 37°C. Post-hybridization washing was with 2× SSC with 0.1% Tween 20 for 5 minutes, and fluorescent detection was done using anti-digoxigenin conjugated to fluorescein (Roche Applied Science, Indianapolis, IN) and streptavidin conjugated to Alexa Fluor 594 (Invitrogen). Slides were counterstained and mounted in ProLong Gold Antifade Reagent with 4′,6-diamidino-2-phenylindole (Invitrogen). Slides were examined using a Leica DMRA fluorescence microscope (Leica, Deerfield, IL) and imaged with a CCD camera using the CytoVision software system (Applied Imaging, Santa Clara, CA).

All BACs were obtained from the BACPAC Resource Center (Oakland, CA), and probe locations were verified by hybridization to metaphase spreads of normal peripheral lymphocytes. For detection of TMPRSS2:ETV4 fusion, RP11-35C4 (5′ to TMPRSS2) was used with multiple BACs located 3′ to ETV4 (distal to ETV4 to proximal: RP11-266I24, RP11-242D8, and RP11-100E5). For detection of ETV4 rearrangements, RP11-436J4 (5′ to ETV4) was used with the multiple BACs 3′ to ETV4. For each hybridization, areas of cancerous cells were identified by a pathologist, and 100 cells were counted per sample. The reported cell count for TMPRSS2:ETV4 fusions used RP11-242D8, and similar results were obtained with all 3′ ETV4 BACs. To exclude additional rearrangements in PCA5, we did FISH with two probes 3′ to ETV4 (RP11-266I24 and RP11-242D8): ERG split signal probes (RP11-95I21 and RP11-476D17) and TMPRSS2:ETV1 fusion probes (RP11-35C4 and RP11-124L22). BAC DNA was isolated using a QIAFilter Maxi Prep kit (Qiagen, Valencia, CA), and probes were synthesized using digoxigenin- or biotin-nick translation mixes (Roche Applied Science).

Although our initial COPA screen led to the characterization of TMPRSS2 fusions with ERG or ETV1, we hypothesized that prostate cancers negative for these gene fusions may harbor rearrangements involving other ETS family members. By interrogating the expression of all ETS family members monitored in prostate cancer profiling studies from the Oncomine database (http://www.oncomine.org; ref. 9), we identified marked overexpression of the ETS family member ETV4 in a single prostate cancer case from each of two studies: one profiling grossly dissected tissues (ref. 8; Fig. 1A), and the other profiling laser capture microdissected (LCM) tissues8 (Fig. 1B). As these cases did not overexpress ERG or ETV1, and no benign prostate tissues showed overexpression, we hypothesized that fusion with TMPRSS2 may drive the overexpression of ETV4 in these cases. Although ELF3 was also overexpressed in a fraction of prostate cancer cases, in both studies, normal prostate tissue samples also showed marked ELF3 overexpression, suggesting that a gene fusion driving expression in both benign and cancerous tissue is unlikely. Thus, we focused on characterizing the ETV4 overexpressing case (designated here as PCA5) in our LCM cohort.

Figure 1.

Overexpression of ETS family members in prostate cancer. Expression of all monitored ETS family members in profiled benign prostate (blue), prostatic intraepithelial neoplasia (PIN; green), clinically localized prostate cancer (PCa; yellow), and metastatic prostate cancer (Met PCa; orange) from grossly dissected tissue (A; ref. 8) or tissue isolated by laser capture microdissection (B)8 was visualized using Oncomine (http://www.oncomine.org). Columns, samples of the indicated class; rows, indicated gene. Relative underexpression (blue) or overexpression (red; in z-score units, median centered per gene), as indicated by the color scale. Features that did not pass filtering in the original study (gray).

Figure 1.

Overexpression of ETS family members in prostate cancer. Expression of all monitored ETS family members in profiled benign prostate (blue), prostatic intraepithelial neoplasia (PIN; green), clinically localized prostate cancer (PCa; yellow), and metastatic prostate cancer (Met PCa; orange) from grossly dissected tissue (A; ref. 8) or tissue isolated by laser capture microdissection (B)8 was visualized using Oncomine (http://www.oncomine.org). Columns, samples of the indicated class; rows, indicated gene. Relative underexpression (blue) or overexpression (red; in z-score units, median centered per gene), as indicated by the color scale. Features that did not pass filtering in the original study (gray).

Close modal

We isolated total RNA from PCA5 and used an exon-walking quantitative PCR strategy to confirm the overexpression of ETV4. Quantitative PCR showed that exons 3′ to exon 2 of ETV4 were markedly overexpressed in this case compared with pooled benign prostate tissue (CPP; ∼900-fold) and prostate cancers that did not overexpress ETV4 and were either TMPRSS2:ERG positive (PCA1-2) or negative (PCA3-4; Fig. 2A). However, we observed a dramatic decrease (>99%) in the expression of exon 2 of ETV4 relative to distal regions in PCA5, suggesting a possible fusion with TMPRSS2, as observed previously in TMPRSS2:ERG-positive and TMPRSS2:ETV1-positive cases (4).

Figure 2.

Fusion of TMPRSS2 and ETV4 loci in a prostate cancer case that overexpresses ETV4. A, quantitative PCR was used to determine the relative expression of the indicated exons or region of ETV4 in pooled benign prostate tissue (CPP), prostate cancers that did not overexpress ETV4 and were either TMPRSS2:ERG positive (PCA1-2) or negative (PCA3-4), and the prostate cancer case from our LCM cohort with ETV4 overexpression (PCA5). The amount of ETV4 was normalized to the amount of GAPDH, and the relative amount of ETV4 for each sample was calibrated to the amount in CPP. Results are presented on a log scale. B, RLM-RACE reveals fusion of sequences upstream of TMPRSS2 with ETV4 in PCA5. RLM-RACE was done to identify the 5′ end of the ETV4 transcript in PCA5 using a reverse primer in exon 7 of ETV4. Sequencing of the cloned product revealed two transcripts beginning with 47 bp (TMPRSS2:ETV4a, dark and light blue) or 13 bp (TMPRSS2:ETV4b, light blue) located ∼8 kb upstream of TMPRSS2: a single base pair that did not map to this region or ETV4 (yellow) and a contiguous region composed of the 19 terminal base pairs in the intron before exon 3 of ETV4 (red) and the reference sequence of ETV4 from exons 3 to 7 (green). For the structural diagrams of the TMPRSS2 and ETV4 loci, exons are represented by boxes, introns by horizontal lines, and regions not involved in the RLM-RACE products are in black. The same coloring for the structural diagrams is used for the nucleotide tracing shown for TMPRSS2:ETV4a, and the location of primers used to confirm the presence of the indicated fusion transcripts are shown below the tracing. C, expression of TMPRSS2:ETV4a and TMPRSS2:ETV4b in PCA5 by quantitative PCR. Quantitative PCR was done using forward primers located as shown in (B) for TMPRSS2:ETV4a or TMPRSS2:ETV4b and ETV4_exon4-r or GAPDH primers on the same samples as (A). As no amplification was detectable after 40 cycles for CPP and PCA1-4, the results could not quantified, and products were electrophoresed on a 1.2% agarose gel and visualized with ethidium bromide. D, interphase FISH on formalin-fixed, paraffin-embedded tissue confirms fusion of TMPRSS2 and ETV4 loci in PCA5. Probes for TMPRSS2 (red) and ETV4 (green) show fusion of the genomic loci (yellow arrows) in cancerous cells from PCA5.

Figure 2.

Fusion of TMPRSS2 and ETV4 loci in a prostate cancer case that overexpresses ETV4. A, quantitative PCR was used to determine the relative expression of the indicated exons or region of ETV4 in pooled benign prostate tissue (CPP), prostate cancers that did not overexpress ETV4 and were either TMPRSS2:ERG positive (PCA1-2) or negative (PCA3-4), and the prostate cancer case from our LCM cohort with ETV4 overexpression (PCA5). The amount of ETV4 was normalized to the amount of GAPDH, and the relative amount of ETV4 for each sample was calibrated to the amount in CPP. Results are presented on a log scale. B, RLM-RACE reveals fusion of sequences upstream of TMPRSS2 with ETV4 in PCA5. RLM-RACE was done to identify the 5′ end of the ETV4 transcript in PCA5 using a reverse primer in exon 7 of ETV4. Sequencing of the cloned product revealed two transcripts beginning with 47 bp (TMPRSS2:ETV4a, dark and light blue) or 13 bp (TMPRSS2:ETV4b, light blue) located ∼8 kb upstream of TMPRSS2: a single base pair that did not map to this region or ETV4 (yellow) and a contiguous region composed of the 19 terminal base pairs in the intron before exon 3 of ETV4 (red) and the reference sequence of ETV4 from exons 3 to 7 (green). For the structural diagrams of the TMPRSS2 and ETV4 loci, exons are represented by boxes, introns by horizontal lines, and regions not involved in the RLM-RACE products are in black. The same coloring for the structural diagrams is used for the nucleotide tracing shown for TMPRSS2:ETV4a, and the location of primers used to confirm the presence of the indicated fusion transcripts are shown below the tracing. C, expression of TMPRSS2:ETV4a and TMPRSS2:ETV4b in PCA5 by quantitative PCR. Quantitative PCR was done using forward primers located as shown in (B) for TMPRSS2:ETV4a or TMPRSS2:ETV4b and ETV4_exon4-r or GAPDH primers on the same samples as (A). As no amplification was detectable after 40 cycles for CPP and PCA1-4, the results could not quantified, and products were electrophoresed on a 1.2% agarose gel and visualized with ethidium bromide. D, interphase FISH on formalin-fixed, paraffin-embedded tissue confirms fusion of TMPRSS2 and ETV4 loci in PCA5. Probes for TMPRSS2 (red) and ETV4 (green) show fusion of the genomic loci (yellow arrows) in cancerous cells from PCA5.

Close modal

To identify the 5′ end of the ETV4 transcript in PCA5, we did RLM-RACE using a reverse primer in exon 7. RLM-RACE revealed two transcripts, each containing 5′ ends consisting of sequence located ∼8 kb upstream of TMPRSS2 fused to sequence from ETV4 (Fig. 2B). Specifically, the 5′ end of TMPRSS2:ETV4a consists of 47 bp from this region upstream of TMPRSS2, whereas the 5′ end of TMPRSS2:ETV4b consists of the same terminal 13 bp. These 5′ ends of both transcripts were fused to the same contiguous stretch consisting of the 9 bp of the intron immediately 5′ to exon 3 of ETV4 and the reported reference sequence of exons 3 through the reverse primer in exon 7 of ETV4.

We confirmed the existence of both transcripts in PCA5 and their absence in CPP and PCA1-4 using quantitative PCR; however, the results could not be quantified due to no detectable amplification after 40 cycles in CPP and PCA1-4 (Fig. 2C). To further exclude the presence of fusion transcripts involving known exons from TMPRSS2, we attempted quantitative PCR using a forward primer in exon 1 of TMPRSS2 and the ETV4 exon 4 reverse primer, and as expected, no product was detected in CPP or PCA1-5 (data not shown).

Whether other prostate cancers with ETV4 dysregulation might contain TMPRSS2:ETV4 fusion transcripts structurally more similar to TMPRSS2:ERG and TMPRSS2:ETV1 transcripts (which involve known exons from TMPRSS2) is unknown. It is important to note that the TMPRSS2:ETV4 fusions reported here would not contain the well-characterized AREs immediately upstream of TMPRSS2. However, evidence exists for androgen-responsive enhancers located upstream of the TMPRSS2 sequences present in the TMPRSS2:ETV4 transcripts described here.9

9

Q. Wang and M. Brown, personal communication.

Nevertheless, the marked overexpression of only ETV4 exons involved in the fusion transcript strongly suggests that the gene fusion is responsible for the dysregulation of ETV4. Together, the structure of the TMPRSS2:ETV4 fusion transcripts supports the conclusion that the regulatory elements upstream of TMPRSS2, rather than transcribed TMPRSS2 sequences, drive the dysregulation of ETS family members.

To confirm the fusion of the genomic loci surrounding TMPRSS2 (21q22) and ETV4 (17q21) as shown by RLM-RACE and quantitative PCR, we used interphase FISH. Using probes 5′ to TMPRSS2 and 3′ to ETV4, we identified fusion of TMPRSS2 and ETV4 loci in 65% of cancerous cells from PCA5 (Fig. 2D). As further confirmation of the rearrangement of ETV4, using probes 5′ and 3′ to ETV4, 64% of cancerous cells from PCA5 showed split signals (data not shown). We also did FISH on PCA5 using two probes 3′ to ETV4, ERG split signal probes and TMPRSS2:ETV1 fusion probes to exclude additional rearrangements, with negative results obtained for each hybridization (data not shown).

Taken together, the results from this study highlight the importance of carefully examining outlier profiles in tumor gene expression data, as most analytic methods discount profiles that do not show consistent deregulation (1113) and would thus fail to identify ETV4 in prostate cancer, which seems rare (2 of 98 cases). Combined with the identification of TMPRSS2:ERG and TMPRSS2:ETV1 fusions, the results presented here support the hypothesis that dysregulation of ETS family members mediated by subversion of AREs or enhancers upstream of TMPRSS2 is a hallmark of prostate tumorigenesis. Although the majority of ETS family members were represented in the profiling studies examined, other ETS family members that were not monitored may also be rearranged in prostate cancers for which gene fusions have not been ascribed. The reason for the observed frequencies of fusion partners with TMPRSS2 (ERG > ETV1 > ETV4), which are consistent across independent sample sets, is unclear, although a similar situation is present in Ewing's sarcoma, where EWS partners with ETS family members in unequal frequencies (FLI1 > ERG > ETV1; ref. 14). Lastly, these results establish a third molecular subtype of prostate cancer, which may have prognostic and/or therapeutic relevance in the future.

Note: S.A. Tomlins, R. Mehra, and D.R. Rhodes contributed equally to this report.

A.M. Chinnaiyan is a Pew Biomedical Scholar. S.A. Tomlins and D.R. Rhodes are Fellows of the Medical Scientist Training Program.

Genbank Accession numbers for TMPRSS2:ETU4a and TMPRSS2:ETU4b are DQ396625-6.

Grant support: Early Detection Research Network Biomarker Developmental Lab grant UO1 CA111275-01 (A.M. Chinnaiyan), NIH Prostate Specialized Program of Research Excellence grant P50CA69568 (A.M. Chinnaiyan, M.A. Rubin, J.T. Wei, and R.B. Shah), grants RO1 CA97063 (A.M. Chinnaiyan) and RO1 AG21404 (M.A. Rubin), American Cancer Society grant RSG-02-179-MGO (A.M. Chinnaiyan), Department of Defense grants PC040517 (R. Mehra) and PC020322 (A.M. Chinnaiyan), Cancer Center Bioinformatics Core Support Grant 5P30 CA46592, and Cancer Biology Training Program (D.R. Rhodes).

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.

We thank Anjana Menon for technical assistance and Qianben Wang and Myles Brown for communicating unpublished results.

1
Mitelman F. Recurrent chromosome aberrations in cancer.
Mutat Res
2000
;
462
:
247
–53.
2
Rabbitts TH. Chromosomal translocations in human cancer.
Nature
1994
;
372
:
143
–9.
3
Rowley JD. Chromosome translocations: dangerous liaisons revisited.
Nat Rev Cancer
2001
;
1
:
245
–50.
4
Tomlins SA, Rhodes DR, Perner S, et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer.
Science
2005
;
310
:
644
–8.
5
Afar DE, Vivanco I, Hubert RS, et al. Catalytic cleavage of the androgen-regulated TMPRSS2 protease results in its secretion by prostate and prostate cancer epithelia.
Cancer Res
2001
;
61
:
1686
–92.
6
Jacquinet E, Rao NV, Rao GV, Zhengming W, Albertine KH, Hoidal JR. Cloning and characterization of the cDNA and gene for human epitheliasin.
Eur J Biochem
2001
;
268
:
2687
–99.
7
Lin B, Ferguson C, White JT, et al. Prostate-localized and androgen-regulated expression of the membrane-bound serine protease TMPRSS2.
Cancer Res
1999
;
59
:
4180
–4.
8
Lapointe J, Li C, Higgins JP, et al. Gene expression profiling identifies clinically relevant subtypes of prostate cancer.
Proc Natl Acad Sci U S A
2004
;
101
:
811
–6.
9
Rhodes DR, Yu J, Shanker K, et al. ONCOMINE: a cancer microarray database and integrated data-mining platform.
Neoplasia
2004
;
6
:
1
–6.
10
Vandesompele J, De Preter K, Pattyn F, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes.
Genome Biol
2002
;
3
:
RESEARCH0034
.
11
Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns.
Proc Natl Acad Sci U S A
1998
;
95
:
14863
–8.
12
Golub TR, Slonim DK, Tamayo P, et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring.
Science
1999
;
286
:
531
–7.
13
Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response.
Proc Natl Acad Sci U S A
2001
;
98
:
5116
–21.
14
Arvand A, Denny CT. Biology of EWS/ETS fusions in Ewing's family tumors.
Oncogene
2001
;
20
:
5747
–54.