Identification of novel TMPRSS2:ERG mechanisms in prostate cancer metastasis: involvement of MMP9 and PLXNA2.

Prostate cancer (PCa) is one of the major public health problems in Western countries. Recently, the TMPRSS2:ERG gene fusion, which results in the aberrant expression of the transcription factor ERG, has been shown to be the most common gene rearrangement in PCa. Previous studies have determined the contributions of this fusion in PCa disease initiation and/or progression in vitro and in vivo. In this study on TMPRSS2:ERG regulation in PCa, we used an androgen receptor and TMPRSS2:ERG fusion double-negative PCa cell model: PC3c. In three cell clones with different TMPRSS2:ERG expression levels, ectopic expression of the fusion resulted in significant induction of cell migration and invasion in a dose-dependent manner. In agreement with this phenotype, high-throughput microarray analysis revealed that a set of genes, functionally associated with cell motility and invasiveness, were deregulated in a dose-dependent manner in TMPRSS2:ERG-expressing cells. Importantly, we identified increased MMP9 (Metalloproteinase 9) and PLXNA2 (Plexin A2) expression in TMPRSS2:ERG-positive PCa samples, and their expression levels were significantly correlated with ERG expression in a PCa cohort. In line with these findings, there was evidence that TMPRSS2:ERG directly and positively regulates MMP9 and PLXNA2 expression in PC3c cells. Moreover, PLXNA2 upregulation contributed to TMPRSS2:ERG-mediated enhancements of PC3c cell migration and invasion. Furthermore, and importantly, PLXNA2 expression was upregulated in metastatic PCa tumors compared with localized primary PCa tumors. This study provides novel insights into the role of the TMPRSS2:ERG fusion in PCa metastasis.

Identification of amino acid residues in the ETS transcription factor Erg that mediate Erg-Jun/Fos-DNA ternary complex formation.

Jun, Fos, and Ets proteins belong to distinct families of transcription factors that target specific DNA elements often found jointly in gene promoters. Physical and functional interactions between these families play important roles in modulating gene expression. Previous studies have demonstrated a direct interaction between the DNA-binding domains of the two partners. However, the molecular details of the interactions have not been investigated so far. Here we used the known three-dimensional structures of the ETS DNA-binding domain and Jun/Fos heterodimer to model an ETS-Jun/Fos-DNA ternary complex. Docking procedures suggested that certain ETS domain residues in the DNA recognition helix alpha3 interact with the N-terminal basic domain of Jun. To support the model, different Erg ETS domain mutants were obtained by deletion or by single amino acid substitutions and were tested for their ability to mediate DNA binding, Erg-Jun/Fos complex formation, and transcriptional activation. We identified point mutations that affect both the DNA binding properties of Erg and its physical interaction with Jun (R367K), as well as mutations that essentially prevent transcriptional synergy with the Jun/Fos heterodimer (Y371V). These results provide a framework of the ETS/bZIP interaction linked to the manifestation of functional activity in gene regulation.

Erg proteins, transcription factors of the Ets family, form homo, heterodimers and ternary complexes via two distinct domains.

The ets genes family encodes a group of proteins which function as transcription factors under physiological conditions. We report here that the Erg proteins, members of the Ets family, form homo and heterodimeric complexes in vitro. We demonstrate that the Ergp55 protein isoform forms dimers with itself and with the two other isoforms, Ergp49 and Ergp38. Using a set of Erg protein deletion mutants, we define two distinct domains independently involved in dimerization. The first one is located in the amino-terminal part of the protein containing the pointed domain (PNT), conserved in a subset of Ets proteins. The second one resides within the ETS domain, the DNA-binding domain. We also show that the Erg protein central region behaves as an inhibitory domain of dimerization and its removal enhances the Ergp55 transactivation properties. Furthermore, Ergp55 forms heterodimers with some other Ets proteins. Among the latter, we show that Fli-1, Ets-2, Er81 and Pu-1 physically interact with Erg. Finally, we show that the formation of the previously described ternary complex Ergp55/Fos/jun is mediated by ETS domain and Jun protein, while the ternary complex Ergp49/Fos/Jun is mediated by Fos protein.

Isolation and characterization of a chicken homologue of the Spi-1/PU.1 transcription factor.

Spi-1/PU.1 is a member of the Ets family of transcription factors important in regulation of hematopoiesis. We have isolated a chicken cDNA homologuous to the mammalian Spi-1/PU.1 gene with an open reading frame of 250 amino acids (aa). The chicken Spi-1/PU.1 protein is 14 aa and 16 aa shorter than its human and mouse counterparts but is extremely well conserved with 78.8% and 75.2% identity respectively. The carboxy terminal DNA binding region, or ETS binding domain, is 100% identical to that of human and mouse. Some differences with the mammalian homologues are seen in the N-terminal part of the protein and in the PEST connecting domain. However, the differences are mainly conservative and all the features underlying functional aspects seem preserved. The major discrepancy lies in a 12 aa deletion in an already poorly conserved part of the PEST sequence. Spi-1/PU.1 transcripts were detected at high levels in spleen and Fabricius bursa of chick embryos by Northern blot and in situ hybridization. Our results show that the chicken Spi-1/PU.1 protein behaves like a bonafide Spi-1/PU.1 transcription factor in its DNA binding and transactivating properties.

Molecular characterization of mouse 17 beta-hydroxysteroid dehydrogenase IV.

17 beta-hydroxysteroid dehydrogenases (17 beta-HSD) catalyze the conversion of estrogens and androgens at the C17 position. The 17 beta-HSD type I, II, III and IV share less than 25% amino acid similarity. The human and porcine 17 beta-HSD IV reveal a three-domain structure unknown among other dehydrogenases. The N-terminal domains resemble the short chain alcohol dehydrogenase family while the central parts are related to the C-terminal parts of enzymes involved in peroxisomal beta-oxidation of fatty acids and the C-terminal domains are similar to sterol carrier protein 2. We describe the cloning of the mouse 17 beta-HSD IV cDNA and the expression of its mRNA. A probe derived from the human 17 beta-HSD IV was used to isolate a 2.5 kb mouse cDNA encoding for a protein of 735 amino acids showing 85 and 81% similarity with human and porcine 17 beta-HSD IV, respectively. The calculated molecular mass of the mouse enzyme amounts to 79,524 Da. The mRNA for 17 beta-HSD IV is a single species of about 3 kb, present in a multitude of tissues and expressed at high levels in liver and kidney, and at low levels in brain and spleen. The cloning and molecular characterization of murine, human and porcine 17 beta-HSD IV adds to the complexity of steroid synthesis and metabolism. The multitude of enzymes acting at C17 might be necessary for a precise control of hormone levels.

A model for gene evolution of the ets-1/ets-2 transcription factors based on structural and functional homologies.

The chicken c-ets-1 locus encodes two transcription factors, p54c-ets-1 and p68c-ets-1 that differ in their N-termini, encoded respectively by the I54 and alpha beta exons. p68c-ets-1 equivalents are only found in birds and reptiles while p54c-ets-1 is widely conserved in vertebrates, from amphibians to mammals. Thus, the classical view concerning the evolution of the c-ets-1 gene has been to consider that I54 is of ancient origin whereas alpha and beta, which provide an additional activating domain in p68c-ets-1, would have been acquired much more recently. Sequencing the alpha and beta exons in various species pinpointed a highly conserved region of 13 amino acids which is rich in acidic and hydrophobic residues, a feature of some other transactivating domains. Strikingly, this subdomain is also present in the otherwise unrelated N-terminal activating region of p58c-ets-2 and was thus named BEC for Ets-1-beta/Ets-2-Conserved sequence. Moreover, the two N-termini share the BEC sequence at a homologous position in their highly similar genomic organization indicating a common origin. This structural homology underlies a functional similarity since fusion of the heterologous GAL4 DNA-binding domain with either of the two isolated domains demonstrates that BEC is essential in both cases for the transactivating activity. The function of the alpha beta domain in the context of p68c-ets-1 also strictly depends on the presence of the BEC sequence. Finally, the whole N-terminus of p58c-ets-2 can functionally substitute for its counterpart in p68c-ets-1 further demonstrating that p68c-ets-1 and p58c-ets-2 are structurally and functionally more closely related than previously thought. Besides, we also found BEC in the N-terminus of the Drosophila pointed gene which may be considered as closely related to the uncommitted 'ets1/2' common ancestor. These data demonstrate that the alpha and beta exons are not a recent and specific acquisition but stem, like the p58c-ets-2 N-terminus, from the invertebrate unduplicated 'ets 1/2' gene. This work unravels a new model for the ets-1/ets-2 gene's evolution, based for the first time on both structural and functional evidences. Accordingly, p68c-ets-1 and p58c-ets-2 are the direct descendants of the ancestral 'ets1/2' gene whereas I54 may have been acquired as a second promoter in the c-ets-1 gene after the duplication. Indeed, I54 is not found in the Drosophila pointed gene. The high degree of similarity, and hence of functional redundancy, between p68c-ets-1 and p58c-ets-2 may have led to the rapid divergence (and even loss in mammals) of alpha and beta during evolution whereas I54, which provided a novel function unique to c-ets-1, was maintained within the presently widespread p54c-ets-1 version.

Identification of two ets related genes in a marine worm, the polychaete annelid Nereis diversicolor.

The ETS family includes a growing number of transcription factors with a highly conserved DNA-binding domain, the ETS domain. We have used PCR amplification with degenerated oligonucleotides to isolate two putative ETS DNA-binding coding domains in a primitive form of coelomate, the polychaete annelid Nereis diversicolor. These sequences are highly related to the ETS and ERG groups of the ets gene family. For the erg sequence an adjacent region encoding for 91 amino acids has been characterized after library screening, and we show an expression in cells isolated from the coelomic cavity of the animal. A phylogenic analysis confirms that the ets-1/ets-2 and the erg/fli dichotomy arose specifically in the vertebrate lineage.

A residue of the ETS domain mutated in the v-ets oncogene is essential for the DNA-binding and transactivating properties of the ETS-1 and ETS-2 proteins.

The c-ets-1 locus encodes two transcription factors, p54c-ets-1 and p68c-ets-1 that recognize purine-rich motifs. The v-ets oncogene of the avian retrovirus E26 differs from its cellular progenitor p68c-ets-1 by two amino acid substitutions (alanine 285 and isoleucine 445 in c-ets-1 both substituted by valine in v-ets, mutations A and B respectively) and its carboxy-terminal end (mutation C). The B mutation affects a well conserved residue in the carboxy-terminal 85 amino acids, ETS DNA-binding domain. To address the biological relevance of the B mutation found between v-ets and c-ets-1, we have randomly mutagenized isoleucine 445 of p68c-ets-1 by polymerase chain reaction. Using in vitro gel mobility shift assays, we show that this residue is crucial for the binding properties of c-ets-1 since the 12 mutations we have generated at this position, all diminish or even abolish the binding, to the 'optimized' Ets-1 binding site (EBS), of 35 kDa proteins corresponding to the 311 carboxy-terminal residues of c-ets-1. Among them, substitutions of isoleucine to glutamic acid, glycine or proline have the highest inhibitory effects. Similar results were obtained when the same mutations were introduced either in full-length p68c-ets-1 protein or into a carboxy-terminal polypeptide of 109 amino acids encompassing the ETS-domain which has previously been shown to display a very high binding activity as compared with the full-length protein. Consistent with the in vitro results, point mutations in p68c-ets-1 that decrease binding activity to EBS abrogate its ability to transactivate reporter plasmids carrying either the TPA Oncogene Response Unit of the Polyoma virus enhancer (TORU) or a sequence derived from the HTLV-1 LTR. Furthermore, as this isoleucine residue is rather well-conserved within the ETS gene family, we show that mutation of the corresponding isoleucine of c-ets-2 into glycine also abrogates its DNA-binding and hence, transactivating properties. Thus, the v-ets B mutation highlights the isoleucine 445 as an essential amino acid of the c-ets-1 and c-ets-2 DNA-binding domains.

A new mechanism of oncogenic activation: E26 retroviral v-ets oncogene has inverted the C-terminal end of the transcription factor c-ets-1.

The v-ets-encoded domain in the P135gag-myb-ets transforming protein of the E26 retrovirus differs mainly from its cellular progenitor, p68c-ets-1 by two point mutations and by the replacement of the 13 last C-terminal amino acids present in c-ets-1 by 16 unrelated residues of previously unknown origin in v-ets. Here, we demonstrate that these v-ets C-terminal specific residues are in fact encoded by the opposite strand of the c-ets-1 C-terminus.

The two functionally distinct amino termini of chicken c-ets-1 products arise from alternative promoter usage.

The chicken c-ets-1 locus gives rise to two distinct transcription factors differing by structurally and functionally unrelated N-termini. p54c-ets-1 shows a striking phylogenetic conservation from Xenopus to humans, while p68c-ets-1, the cellular counterpart of the E26-derived v-ets oncogene, is apparently restricted to avian and reptilian species. In the chick embryo, both mRNAs are expressed in a wide array of tissues of mesodermal origin; however, in the embryo and after hatching, p68c-ets-1 is excluded from lymphoid cells where p54c-ets-1 accumulates. In this report, we define the basis of the differential expression of the chicken c-ets-1 products to assess their different potentials as transcription factors. We demonstrate that the two distinct N-termini arise from alternative promoter usage within the chicken c-ets-1 locus. Examination of both promoters reveals that transcription initiates from multiple sites, consistent with the absence of TATA and CAAT elements. Of these two regulatory regions, only the one that initiates the p54c-ets-1 mRNA synthesis is of the G + C-rich type, and its organization is conserved in humans. The avian-specific p68c-ets-1 promoter activity was enhanced by its own product. In addition, we identify numerous potential binding sites for lymphoid-specific transcription factors that might contribute to a tight repressor effect in lymphoid tissues.

Phylogeny of the p68c-ets-1 amino-terminal transactivating domain reveals some highly conserved structural features.

The chicken c-ets-1 locus gives rise to two distinct transcription factors differing only in their structurally and functionally unrelated N-termini. One of these transcription factors, p54c-ets-1, contains a specific, short (27 amino acids), hydrophilic N-terminus encoded by a single exon, I54, that is widely conserved among vertebrates. The other one, p68c-ets-1, the cellular counterpart of the viral ets oncogene product, differs in the replacement of the I54 by two exons, termed alpha and beta, encoding a larger (71 amino acids), hydrophobic N-terminus which, in contrast to I54, exhibits properties of a transactivating domain. To date the alpha and beta exons have only been found in chicken. Here, we demonstrate the existence of the alpha and beta exons in other avian species (quail and duck) and the existence of the alpha exon in reptiles (turtle). However, none of them could be detected in mammals. Our results strongly suggest that, in contrast to the phylogenetically well-conserved I54 exon, the alpha exon is restricted to reptilian species (birds and 'true' reptiles), whereas the beta exon is detectable so far only in birds. Comparison of their amino acid sequences reveals that the alpha exon and to a much greater extent the beta exon have diverged faster than the I54 exon. In addition, we show that the N- and C-terminal thirds of the alpha exon and the highly hydrophobic nature of the alpha beta-encoded sequence are heavily conserved features and thus likely to be required for function as a transactivating domain in p68c-ets-1 and possibly in the viral P135gag-myb-ets transforming protein.

Alternative splicing within the chicken c-ets-1 locus: implications for transduction within the E26 retrovirus of the c-ets proto-oncogene.

Two overlapping c-ets-1 cDNA clones were isolated which contained the alpha and beta genomic sequences homologous to the 5' end of v-ets not detected in the previously described c-ets RNA species or proteins. Nucleotide sequencing demonstrated that these cDNAs corresponded to the splicing of alpha and beta to a common set of 3' exons (a through F) already found in the p54c-ets-1 mRNA. They contained an open reading frame of 1,455 nucleotides which could encode a polypeptide of 485 amino acids with a predicted molecular mass of 53 kilodaltons. However, when expressed in COS-1 cells, the cDNAs directed the synthesis of a protein with an apparent molecular mass in sodium dodecyl sulfate-polyacrylamide gel electrophoresis of 68 kilodaltons, p68c-ets-1, comigrating with a protein expressed at low levels in normal chicken spleen cells. These two proteins were shown to be identical by partial digestion with protease V8. Northern (RNA) blot hybridization analysis with the p68c-ets-1 -specific sequence and RNase protection experiments showed that the corresponding mRNA was expressed in normal chicken spleen and not in normal chicken thymus or in various T lymphoid cell lines. Thus, two closely related proteins, having distinct amino-terminal parts, are generated within the same locus by alternative addition of different 5' exons, alpha and beta or I54, respectively, onto a common set of 3' exons (a to F). Finally, we demonstrate that an aberrant splicing event between a cryptic splice donor site in c-myb exon E6 and the normal splice acceptor site of c-ets-1 exon alpha involved in the genesis of the E26 myb-ets sequence.

Cloning and expression of chicken p54c-ets cDNAs: the first p54c-ets coding exon is located into the 40.0 kbp genomic domain unrelated to v-ets.

We have isolated cDNA clones of chicken c-ets mRNA the longest of which, designated pCk E54A, contained approximately 2.0 kb of a c-ets mRNA species. Nucleotide sequencing of this clone revealed a single long open reading frame, extending from the first ATG codon (nucleotide +1) to a TGA termination codon at nucleotide 1324. The predicted translation product contains 441 amino acid residues and its molecular weight is 48 kd. Expression in COS-1 cells of this clone resulted in the synthesis of polypeptides immunologically indistinguishable from the authentic p54c-ets after one-dimensional gel electrophoresis. Comparison of the nucleotide sequence of this cDNA to that of v-ets of avian acute leukemia virus E26 showed that both sequences are almost colinear with the exception of five point mutations but present striking differences in their 5' and 3' parts. 79 nucleotides downstream of the first ATG codon in c-ets cDNA are not found in the 5' part of v-ets where they are replaced by 223 different nucleotides. The 3' parts of v-ets and the coding region of the chicken c-ets cDNAs are also different: the last 13 codons of the cDNA are replaced by 16 different codons in v-ets. Thus our results precisely define the structural differences between the ets encoded domain of E26 viral transforming protein (P135 gag-myb-ets) and the normal cellular protein p54c-ets expressed at high levels in chicken thymocytes and bursal lymphocytes. They also suggest the possibility of alternative splicing of different 5' exons to a common set of 3' exons.

Multiple domains for the chicken cellular sequences homologous to the v-ets oncogene of the E26 retrovirus.

We have investigated the structure of chicken genomic DNA homologous to v-ets, the second cell-derived oncogene of avian retrovirus E26. We isolated a c-ets locus spanning ca. 30.0 kilobase pairs (kbp) in the chicken genome with homologies to 1,202 nucleotides (nt) of v-ets (total length, 1,508 nt) distributed in six clusters along 18.0 kbp of the cloned DNA. The 5'-distal part of v-ets (224 nt) was homologous to chicken cellular sequences contained upstream within a single 16.0-kbp EcoRI fragment as two typical exons but not found transcribed into the major 7.5-kb c-ets (or 4.0-kb c-myb) RNA species. Between these two v-ets-related cellular sequences we found ca 40.0 kbp of v-ets-unrelated DNA. Finally, the most 3' region of homology to v-ets in the cloned DNA was shown to consist of a truncated exon lacking the nucleotides coding for the 16 carboxy-terminal amino acids of the viral protein but colinear to one of the two human c-ets loci, c-ets-2.