|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: DG HvK HD CdR JT. Performed the experiments: DG HvK HD CdR. Analyzed the data: DG AdK JT. Contributed reagents/materials/analysis tools: HD WMvW. Wrote the paper: DG JT.
ETV1 is overexpressed in a subset of clinical prostate cancers as a fusion transcript with many different partners. However, ETV1 can also be overexpressed as a full-length transcript. Full-length ETV1 protein functions differently from truncated ETV1 produced by fusion genes. In this study we describe the genetic background of full-length ETV1 overexpression and the biological properties of different full-length ETV1 isoforms in prostate cancer. Break-apart FISH showed in five out of six patient samples with overexpression of full-length ETV1 a genomic rearrangement of the gene, indicating frequent translocation. We were able to study the rearrangements in more detail in two tumors. In the first tumor 5′-RACE on cDNA showed linkage of the complete ETV1 transcript to the first exon of a prostate-specific two exon ncRNA gene that maps on chromosome 14 (EST14). This resulted in the expression of both full-length ETV1 transcripts and EST14-ETV1 fusion transcripts. In chromosome spreads of a xenograft derived from the second prostate cancer we observed a complex ETV1 translocation involving a chromosome 7 fragment that harbors ETV1 and fragments of chromosomes 4 and 10. Further studies revealed the overexpression of several different full-length transcripts, giving rise to four protein isoforms with different N-terminal regions. Even the shortest isoform synthesized by full-length ETV1 stimulated in vitro anchorage-independent growth of PNT2C2 prostate cells. This contrasts the lack of activity of even shorter N-truncated ETV1 produced by fusion transcripts. Our findings that in clinical prostate cancer overexpression of full-length ETV1 is due to genomic rearrangements involving different chromosomes and the identification of a shortened biologically active ETV1 isoform are highly relevant for understanding the mechanism of ETV1 function in prostate cancer.
Gene fusions are important in the development of many hematological malignancies and sarcomas, but are rare in most other tumor types . However, the frequent fusion between TMPRSS2 and the ETS gene ERG showed that gene fusion is a highly relevant event in prostate cancer , . Overexpression of the TMPRSS2-ERG fusion gene has been reported in 40–70% of prostate cancer cases –. Fusions of TMPRSS2 and three other genes encoding ETS transcription factors, ETV1, ETV4 and ETV5, which are located on different chromosomes, occur at low frequency in prostate cancer , , . However, for ETV1 at least 10 different fusion partners have been described –. Most of these have in common that, like TMPRSS2, they are prostate-specific and androgen-regulated expressed. The properties of fusion partners are key elements in explaining the androgen-regulated overexpression of an ETS oncogene in prostate cancer. However, a unique characteristic of ETV1 is that it can not only be overexpressed in prostate cancer as a fusion transcript but also as a full-length wild-type transcript.
The large family of ETS transcription factors is composed of 27 members –. All members have in common a highly homologous DNA binding domain, the ETS domain. The remaining regions of most ETS proteins show limited structural homology. ETV1, ETV4 and ETV5 are the members of a small subfamily of structurally related ETS proteins. These proteins contain in the N-terminal region a conserved short acidic transactivation domain (TAD) that is absent in ERG. ETS proteins regulate many target genes that modulate biological processes like cell growth, angiogenesis, migration, proliferation and differentiation. However, which of the many molecular and biological functions of ETS proteins are most important in prostate cancer is not known.
Following ERG, ETV1 is the most frequently overexpressed ETS gene in prostate cancer (~10% of the tumors) . The ETV1 protein translated from most fusion transcripts is truncated, lacking the 131 N-terminal amino acids (dETV1). Approx. half of the tumors with ETV1 overexpression express a fusion transcript, the others show a high level of full-length ETV1 expression. The in vitro biological and molecular properties of dETV1 seem different from those of the full-length 477 amino acid protein . This observation suggests that tumors with overexpression of full-length ETV1 are different from tumors expressing dETV1.
Little is known about the mechanism of full-length ETV1 overexpression and its function in clinical prostate cancer. Our present results show that overexpression of full-length ETV1 is correlated with rearrangement of the ETV1 chromosomal region. Moreover, we identified a novel, N-truncated ETV1 isoform with the same activity as full-length ETV1.
Previously, we reported ETV1 overexpression in eight out of 84 prostate cancer samples. In four cases this was caused by a gene fusion, in the other four a full-length ETV1 transcript was overexpressed . In the present study we investigated overexpression of ETV1 by quantitative reverse transcriptase reaction (QPCR) in a novel cohort of 66 prostate cancers. In six RNAs ETV1 overexpression was detected. Samples G51, G59, G233, G270 and G268 were derived from primary tumors and G210 was derived from a recurrent tumor. The six samples were further studied by QPCR with primer pairs at the 5′-end of ETV1 mRNA (exon 1F and exon 4R) and at the 3′-end of the mRNA (exon 11F and exon 12R) (Figure 1A). A high exon 11–12 to exon 1–4 ratio is indicative for a fusion gene; a 11 ratio indicated overexpression of full-length ETV1 mRNA. Based on these criteria, tumors G51, G59, G233 and G270 expressed full-length ETV1 whereas G210 and G268 expressed a fusion transcript (see also control PC374 that expresses TMPRSS2-ETV1). HNRPA2B1-ETV1 was found as the fusion transcript in sample G210 (data not shown); the fusion transcript in G268 has not been identified as yet. These two samples were not investigated further in this study.
In the next experiments we focused on the elucidation of the mechanism of overexpression of full-length ETV1. Recently, it has been shown that in two prostate cancer cell lines overexpressing full-length ETV1, LNCaP and MDA PCa2b, ETV1 is translocated . We now addressed the question whether in clinical samples translocation of the complete ETV1 gene might occur. To detect genomic rearrangements tissue slides of all four prostate cancer samples with full-length ETV1 overexpression were analyzed by break-apart interphase FISH with two labeled BAC probes, one BAC recognized ETV1 and the second one the flanking gene DGKB (Figure 1B). The series was supplemented with two samples harboring full-length ETV1 overexpression from our previous study, G89 and G308 , for which frozen tissues of sufficient morphological quality were available. Figure 1B shows the results of the FISH experiments. Interestingly, we found split signals in all samples except for G59, indicating frequent ETV1 rearrangements in prostate cancers that overexpress full-length ETV1. Absence of a split signal in G59 suggests absence of translocation, although we cannot exclude a breakpoint outside the investigated region. Its observed high frequency indicates genomic rearrangement as an important mechanism of full-length ETV1 overexpression in clinical prostate cancer. Obviously, the chromosomal region to which ETV1 is translocated can contribute to elucidation of the mechanism of ETV1 overexpression. We were able to identify more details of the rearrangements in samples G270 and G89.
It has been shown in LNCaP cells that ETV1 is translocated to 14q13.3-q21.1. The whole gene is integrated in the last intron of MIPOL1. In MDA PCa2b, ETV1 is translocated to the same region, although the precise position is unknown . Moreover, we previously described insertion of truncated ETV1 in the intron of a two exon gene encoding a ncRNA, denoted EST14, giving rise to an EST14-ETV1 fusion transcript that contains ETV1 exon 5–12 sequences (sample G342 in ref. 10; Figure 2A). Importantly, EST14 maps directly adjacent to MIPOL1 on 14q. Like most ETV1 fusion partners, EST14 is an androgen-regulated prostate-specific gene. To investigate whether the same chromosomal region was involved in full-length ETV1 translocation in our novel cohort, interphase FISH was performed with the ETV1 BAC (Figure 1B) in combination with a MIPOL1 BAC (Figure 1C). A merging yellow signal was detected in sample G270 (Figure 1C) but in none of the other tumors. These data indicate that although there seems a preference for chromosome 14q13.3-21.1, other genomic regions will also contribute to rearrangement and overexpression of full-length ETV1.
Additional information of ETV1 rearrangement in G270 came from 5′-RACE of tumor cDNA (data not shown). Remarkably, we did not only detect as expected the full-length ETV1 transcript but also a fusion transcript (Figure 2A). Such a result was not found for any of the other tumors overexpressing full-length ETV1. Sequencing showed that the fusion transcript in G270 was composed of ETV1 exon 1–12 preceded by the first exon of EST14 (Figure 2A). Scanning the EST14 intron and ETV1 flanking region by long-range PCR and sequencing mapped the breakpoints in G270 ~1.9 Kbp upstream of ETV1 exon 1 and ~5 Kbp downstream of EST14 exon 1. The breakpoint in EST14 is only 180 bp apart from the breakpoint in G342 (Figure 2B and ref. 10).
Further information of ETV1 rearrangement was also collected for sample G89. Previously, a xenograft propagated on male nude mice had been generated from this tumor (PC135). Like tumor G89, PC135 overexpressed full-length ETV1 . The availability of the xenograft allowed the preparation of metaphase chromosome spreads. In multicolor FISH a complex chromosomal rearrangement pattern was found (data not shown). Individual chromosome paints were used to validate the multicolor FISH data. Painting of chromosome 7 indicated the presence of multiple chromosome 7 fragments (Figure 3). Hybridization with an ETV1 BAC identified the presence of three gene copies: two in apparently normal chromosomes 7 and one in a complex marker chromosome. Follow-up experiments showed that the marker chromosome contained fragments of chromosomes 4, 7 and 10, as first indicated by multicolor FISH (Figure 3). The ETV1 BAC hybridized at the junction of the chromosome 7 and the chromosome 4 fragment, strongly suggesting that the 4;7 translocation was instrumental in overexpression of ETV1. The precise positions of the breakpoints in 4 and 7 remain to be determined. Our data predict that multiple chromosomal regions are involved in overexpression of full-length ETV1. At least one of these regions is on chromosome 14 and a second one on chromosome 4. The chromosome 14 region is also involved in ETV1 gene fusion. Deep sequencing technology could be instrumental in identification of other ETV1 rearrangements.
Detailed characterization of the full-length ETV1 transcripts in the various tumors by 5′-RACE and sequencing showed that not only ETV1 exon 1- exon 12 transcripts were present but also various other full-length ETV1 transcripts, resulting from alternative promoter usage. These transcripts were denoted as ETV1, ETV1-1a, ETV1-1b and ETV1-1c. Figure 4A and Supplementary Figure S1 show the positions of the different first exons in the gene and indicate the various ATG start codons. Of both ETV1-1a and ETV1-1b two splice variants were found (data not shown). QPCR experiments using transcript-specific primers on RNA from all six clinical prostate cancer samples that overexpressed ETV1 showed that the level of expression of the different transcripts was variable in the various tumors (see Figure S2). ETV1 was hardly expressed in control benign prostate hyperplasia sample G277. Figure 4B schematically represents the predicted composition of the various ETV1 protein isoforms that will be produced. Note that ETV1-1c is by far the shortest, lacking the N-terminal 60 amino acids, including the major part of the conserved acidic TAD. In dETV1 that is expressed by most fusion genes, the N-termimal 131 amino acids are absent. ETV1, ETV1-1b1 and -1b2 were of similar size, as shown in Western blots of lysates from transfected HEK293T cells (Figure 4B).
Previously, we have shown that ETV1 and dETV1 differed in stimulation of in vitro anchorage-independent growth . PNT2C2 cells infected with all novel ETV1 constructs induced anchorage-independent growth in a similar manner as ETV1 (Figure 4C). Remarkably, ETV1-1c, although expressed at a lower level and much smaller, is as active as the longer ETV1 isoforms. Thus, the full N-terminal TAD was not needed but amino acids 61–131 seem essential for biological activity of ETV1 (Figure 4B, C).
In summary, the data presented reveal two important novel aspects of the role of ETV1 in prostate cancer. First, it is shown that in clinical prostate cancers a subgroup of ETV1 positive patients show full-length ETV1 overexpression due to translocations of the whole gene to different chromosomes. This novel observation complements the well-described mechanism of overexpression of truncated ETV1 caused by gene fusions where expression regulation is determined by the promoter and enhancers of the fusion partners. Secondly, in contrast to dETV1 produced by gene fusions, a short isoform of full-length ETV1, ETV1-1c, lacking most of the N-terminal TAD, is as active as longer ETV1 isoforms, containing the complete N-terminal acidic TAD. This finding pinpoints the anchorage-independent growth to a small region that is absent in truncated ETV1 expressed by fusion genes.
It is highly relevant to extend the number of clinical samples in order to be able to compare tumor progression in the two subgroups of prostate cancers showing overexpression of truncated vs. full-length ETV1, and to determine the molecular mechanisms involved in their different biological behavior.
Use of the samples was approved by the Erasmus MC Medical Ethics Committee according to the Medical Research Involving Human Subjects Act in protocol MEC-2004-261, entitled “The use of human normal and cancer residual tissue from a tissue bank for characterization of DNA, RNA and protein”.
Mice were housed according to guidelines of the Erasmus Medical Centre, and procedures were carried out in compliance with standards for use of laboratory animals. Animal experiments performed in this manuscript have been approved by the animal experimental committee of the Erasmus Medical Centre (DEC-consult Erasmus MC project 102-10-01).
Snap-frozen prostate cancers were obtained by radical prostatectomy or transurethral resection. Hematoxilin/eosin (HE) stained tissue sections were histologically evaluated by a pathologist (G. van Leenders). All samples contained at least 50% tumor cells.
RNA from clinical specimens was isolated using RNA-Bee (Campro Scientific, Berlin, Germany). DNA was isolated using the DNeasy DNA Extraction kit (Qiagen, Valencia, CA, USA). RNA from cell lines was isolated using the RNeasy RNA Extraction kit (Qiagen).
The position of the fusion point in tumor 270 was mapped by long-range PCR on genomic DNA using a forward primer in the EST14 intron and reverse primer upstream of ETV1 exon 1. PCR products were separated on a 1% agarose gel and sequenced in an ABI 3100 genetic analyzer (Applied Biosystems, Carlsbad, CA, USA).
mRNA expression was analyzed by QPCR. cDNA was prepared with MMLV-RT (Invitrogen, Carlsbad, CA, USA) and oligo(dT)12 primer. QPCR was performed in Power SYBR Green PCR Master Mix on an ABI Prism 7700 Sequence Detection System (Applied Biosystems). Amplified products were quantified relative to porphobilinogen deaminase (PBGD) by the standard curve method. For primers see Table S1.
5′-RLM-RACE was performed using the GeneRacer kit of Invitrogen. cDNA was amplified using the GeneRacer 5′-primer and a reverse gene-specific primer (ETV1 exon 6). PCR products were analyzed on a 1.5% agarose gel, and bands were excised, purified, and sequenced.
FISH was done on 5-µm frozen tissue sections according to standard protocols with minor modifications . BAC clones RP11-124L22 (ETV1), RP11-460G19 (MIPOL1), RP11-1149J13 (DGKB) were purchased from BacPac Resources (bacpac.chori.org). BACs were either digoxigenin-11-dUTP or biotin-16-dUTP (Roche, Basel, Schweiz) labeled and visualized with anti-digoxigenin FITC (Roche) or streptavidin-Alexa 594 (Invitrogen). Tissue sections were counterstained with DAPI. Images were collected on an epifluorescence microscope (Leica DM, Wetzlar, Germany) equipped with a charge-coupled device cooled camera (Photometrics, Tuscon, AZ, USA).
For preparation of metaphase spreads, xenograft PC135 was propagated on male nude mice. Single cells were collected by mincing and filtration. Metaphase preparation and hybridization were essentially as described . Chromosome paints 4, 7 and 10 were from Euro-Diagnostica (Malmö, Sweden). Metaphases were analyzed with an Axioplan 2 Imaging microscope (Carl Zeiss, Oberkochen, Germany) and images were captured using Isis software (MetaSystems, Altiussheim, Germany).
cDNAs of the different ETV1 isoforms were PCR amplified and cloned into pGEM-TEasy (Promega). Inserts were sequence verified and cloned into the pcDNA3 expression vector (Invitrogen) or the lenti-viral vector pWPXLd (Didier Trono, University of Geneva).
For Western blot analysis, HEK293T cells were transfected with the different pcDNA3-ETV1 expression constructs using the calcium phosphate precipitation method. Cells were harvested 48 h after transfection. Western blot analysis was carried out using standard procedures with antibody directed to the ETV1 C-terminus (Santa Cruz, Santa Cruz, CA, USA). β-actin was used as loading control (Sigma, St Louis, MO, USA). Proteins were visualized by chemiluminescence (Pierce, Rockford, IL, USA).
HEK293T cells were cotransfected with pWPXLd-ETV1 expression vectors, or pWPXLd-GFP (control), and pPAX2 and pMD2.G (Trono) using the calcium phosphate precipitation method. Virus was harvested from the supernatant and used for infection of PNT2C2 cells. Pools of infected cells were propagated.
A layer of 0.6% low-melting agarose in standard culture medium was prepared in six-well plates. On top, a layer of 0.3% agarose containing 1×104 PNT2C2 cells infected with various ETV1 expressing viruses or control PNT2C2-GFP cells were plated. At day 14, cells were stained with crystal violet and colonies were counted.
Part of genomic sequence of ETV1. The different exons are highlighted in yellow. Translation start codons are underlined. Note that the transcript starting at exon 1a may or may not include exon 1a1 but the translation start is the same as that of the transcript starting at exon 1. The end of exon 1b1 is indicated in red. ETV1 transcripts starting at exon 1c lack exons 1, 2 and 3 which results in a truncated TAD in the translated protein.
Expression of the different ETV1 transcripts was determined by QPCR in the 7900HT Fast Real-Time PCR system from Applied Biosystems using the power SYBR-green master mix (Applied Biosystems). Expression levels are relative to the housekeeping gene PBGD. ETV1 and PBGD primers are listed in Supplementary Table S1. Sample G277 is a BPH. It has very low or no expression of all of the different ETV1 transcripts. Samples G51, G59, G89, G233, G270 and G308 all overexpress ETV1. The different transcripts are expressed in variable levels.
Competing Interests: The authors have declared that no competing interests exist.
Funding: This study was supported by a Marie Curie Grant (Cancure) provided by the European Union. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.