Suppression of the Ewing's Sarcoma Phenotype by FLI1

Suppression of the Ewing’s sarcoma phenotype by FLI1/ ERF repressor hybrids

Meropi Athanasiou,1 Lionel LeGallic,2 Dennis K. Watson,3 Donald G. Blair,4 and George Mavrothalassitis2,5 1Intramural Research Support Program, Science Applications International Corporation, Frederick, Maryland 21702; 4Basic Research Laboratory, Division of Basic Sciences, National Cancer Institute, Frederick Cancer Research and Development Center, Frederick, Maryland 21702; 3Center for Molecular and Structural Biology, Hollings Cancer Center, Medical University of South Carolina, Charleston, South Carolina 29425; 2Institute for Molecular Biology and Biotechnology-Foundation for Research and Technology, Hellas, Greece; and 5Medical School, University of Crete, Heraklion, Crete, Greece.

Fusion of the 5Ј half of the Ewing’s sarcoma (ES) gene EWS with the DNA-binding domain of several transcription factors has been detected in many human tumors. The t(11;22)(q24;q12) chromosomal translocation is specifically linked to ES and primitive neuroectodermal tumors and results, in the majority of cases, in the fusion of the amino terminus of the EWS gene to the carboxyl-terminal DNA-binding domain of the FLI1 gene. The chimeric protein has been shown to be oncogenic, a potent transcriptional activator, and necessary for the maintenance of the Ewing’s phenotype, making it an attractive target for gene therapy. In this study, we demonstrate that the ES transformed phenotype can be suppressed by chimeric transcriptional repressors containing the DNA-binding domain of FLI1 and the regulatory and repressor domain of ERF, a transcription suppressor and member of the ets gene family. The hybrid repressor is expressed at levels comparable with EWS/FLI1, does not affect EWS/FLI1 expression, and exhibits similar DNA-binding specificity but suppresses transcriptional activity. The FLI1/ERF repressor, like the wild-type ERF, is regulated by mitogen-activated protein kinase-dependent subcellular localization. Our data suggest that transformation by EWS/FLI1 may partially be due to activation of specific EWS/FLI1-regulated genes involved in cell proliferation. Cancer Gene Therapy (2000) 7, 1188–1195

Key words: FLI1; ERF; Ewing’s sarcoma; tumor suppression.

berrant expression and/or structural alteration of family.8,9 In all of these fusions, the transactivating Atranscription factors have been suggested to be domain of the EWS gene is fused to the DNA-binding critical events in tumorigenic transformation.1,2 The domain of an ets gene. It has been shown that the t(11;22)(q24;q12) chromosomal translocation, detected EWS/FLI1 fusion protein is a more potent transcrip- in the majority of Ewing’s sarcoma (ES) and primitive tional activator than FLI110,11 and, in contrast to FLI1,is neuroectodermal tumors, fuses the amino terminus of a transforming gene.10 Both the transactivation domain the EWS gene to the carboxyl terminus of the ets of EWS and the ets-DNA-binding domain are required 3 transcription factor family member FLI1. In other ES for transformation,10 and the transactivation efﬁciency and primitive neuroectodermal tumors, analogous trans- of different classes of EWS-FLI1 fusions has been locations fuse the EWS gene to other members of the ets reported to correlate with the behavior of tumors in 4 5 6 7 family, including ERG, ETV-1, E1AF, and FEV, vivo.12 Furthermore, the transforming phenotype can be members of the ERG and PEA3 subclasses of the suppressed by blocking EWS-FLI1 production.13–15 The parental EWS protein contains an RNA-binding Received December 23, 1999; accepted April 15, 2000. domain and a transactivation domain and is similar to The publisher or recipient acknowledges the right of the US Govern- the TLS/FUS and hTAFII68 members of the TET family ment to retain a non-exclusive, royalty-free license in and to any copyright of proteins that are able to interact with both TFIID and covering the article. RNA polymerase II.16 The EWS-FLI1 hybrid, however, The content of this publication does not necessarily reﬂects the views or does not appear to be capable of similar functions,17 policies of the Department of Health and Human Services, nor does suggesting that different mechanisms may be involved in mention of trade names, commercial products, or organizations imply endorsement by the United States Government. EWS-FLI1-induced tumorigenesis. Address correspondence and reprint requests to Dr. George Mavroth- The second partner in the gene fusion, FLI1, belongs 18,19 alassitis, University of Crete, Heraklion, Crete, 714 09, Greece. E-mail to the ets family of oncogenic transcription factors, address: [email protected] whose members have been implicated in a number of

1188 Cancer Gene Therapy, Vol 7, No 8, 2000: pp 1188–1195 ATHANASIOU, LEGALLIC, WATSON, ET AL: SUPPRESSION OF ES 1189 human malignancies8,20,21 and have been reported to mined using anti-ERF-23,24 and anti-FLI1-specific Abs (rabbit mediate ras signaling.22 Ets genes share a conserved polyclonal against the full-length FLI1 protein; D.K.W., un- 26 DNA-binding domain that recognizes similar DNA se- published data), as described previously. The electrophoretic mobility shift assay was performed as described previously23 quences, but they have distinct expression patterns as 27 well as other unique functional domains. However, in using protein produced in vitro and the ETS1–3 -labeled oligonucleotide as a probe. ES, where the DNA-binding domain appears to be an important determinant, only members of the ERG and Cell line transfection and transformation PEA3 subclasses have been found to rearrange with EWS, perhaps indicating that important specificity de- The RD-ES and SK-ES-1 cell lines were obtained from the terminants lie within the conserved DNA-binding do- American Type Culture Collection (ATCC) (Manassas, Va) main. and maintained in 85% RPMI 1640 medium, 15% fetal bovine EWS/FLI1 sera (FBS), and 85% McCoy’s 5a medium 15% FBS, respec- The implication that the gene fusion is tively. NIH/3T3 and CRE cells were maintained in 92% responsible for the neoplastic phenotype in ES has led to Dulbecco’s modified Eagle’s medium and 8% bovine sera; the attempts to control its expression and function. Anti- PA317 cells were maintained in 90% Dulbecco’s modified sense oligonucleotides have been used to minimize the Eagle’s medium and 10% FBS. Helper-free viral stocks were 13,15 expression levels of the oncogene, and dominant- prepared by transfecting the ecotropic packaging cell line negative plasmids have been used to block EWS/FLI1 CRE28, harvesting the transiently expressed virus, and infect- function as a transactivator.14 Recently we have identi- ing the amphotropic packaging cell line PA317.29 Individual fied ERF, a member of the ets family of transcription G418-resistant colonies of PA317 cells were isolated, ex- factors and a transcriptional repressor.23 ERF is a panded, and screened for the production of virus by NIH/3T3 infection. Stocks from the higher expressing PA317 clones, mitogen-activated protein kinase-regulated protein that ϫ 3 ϫ 4 v-ets with titers of 5 10 to 5 10 colony-forming units/mL, were can suppress -induced tumorigenicity. Furthermore, used for the infection of the EWS cell lines. Cell line infections phosphorylation-deficient ERF mutations can suppress were performed as described previously.30 Infected or trans- ras-induced tumorigenicity and can arrest cells at the ␮ 24 fected SK-ES-1 and RD-ES cells were selected with 400 g/mL G0/G1 phase of the cell cycle. In this report, we G418, and pools of colonies or individual clones were ex- describe our results using retroviral vectors expressing panded and analyzed. The tumorigenic potential was deter- FLI1-ERF hybrid transcriptional repressors to block the mined after subcutaneous injection of SK-ES-1 cells in athymic transcriptional activation of potential EWS-FLI1 fusion mice. Animals were monitored twice a week for tumor devel- protein targets. Our data indicate both that retroviral opment and general health conditions. The effect on transcrip- gene transfer of a FLI1-ERF hybrid can suppress the ES tion was analyzed in transient transfection assays as described previously.23 A total of 0.5 ␮g of the TK-GATA.CAT reporter phenotype and that important specificity elements re- ␮ ets plasmid was cotransfected into NIH/3T3 cells with 0.5 gof quired for this suppression lie within the DNA- the pSG5-based expression construct described above and/or binding domain. These results suggest that use of ERF- the ⌬EB-EWS/FLI1 expression plasmid11 provided by Dr. J. FLI1, and other ERF chimeric proteins, may be a viable Ghysdael, along with 1 ␮g Rous sarcoma virus (RSV)-␤- approach to blocking the oncogenic potential not only in galactosidase, to normalize for transfection variation. ES, but also in other tumors in which the phenotype is dependent upon transcriptional inducers. RESULTS MATERIALS AND METHODS Generation of FLI1-ERF chimeras Plasmids, proteins, and antibodies (Abs) The EWS-FLI1 fusion product has been shown to be a potent transcriptional activator that can induce ets- The FLI1/ERF hybrid was generated by inserting an in-frame dependent promoters. We postulated that if the contin- ATG via an NcoI linker at the StyI site (amino acid 243) of the 19 uous presence of this activator is required for develop- FLI1 gene. The 439-bp EcoRI-RsaI fragment of the linker- ment of the ES phenotype, we might be able to suppress containing FLI1 sequence was fused at the XmnI site (amino acid 101) of the ERF sequence23 to generate the FLI1/ERF the phenotype by introducing an ets inhibitor into the hybrid. The ERF/FLI1 hybrid was generated by inserting an cells. To test this hypothesis, we used ERF, an ets- in-frame ATG via an NcoI linker at the SmaI site (amino acid domain protein that is a potent transcriptional repressor. 472) of the ERF gene. The 227-bp AvaI-HinfI fragment of the ERF is capable of repressing transcription from promot- linker, containing the ERF sequence, was fused at the EcoRI ers containing an ets-binding site (EBS) and of suppress- site (amino acid 233) of the FLI1 sequence to generate the ing ets-induced tumorigenesis in the NIH/3T3 system.23 ERF/FLI1 hybrid. The hybrids as well as the wild-type (wt) Initial experiments with ERF alone failed to suppress ERF and FLI1 genes were cloned into pSG5 (Stratagene, La the Ewing’s phenotype in ES-derived cell lines (see Jolla, Calif), pRc/RSV (Invitrogen, San Diego, Calif), and 25 below), suggesting that the presence of the FLI1 DNA- pLRNL vectors. The in-frame fusion of the proteins was binding domain was important for both transformation verified by sequencing as well as by in vitro protein production using the TnT௡ (transcription and translation) cell-free system and suppression. We had shown previously that the ERF repressor domain maintains its activity when transferred (Promega, Madison, Wis). The 650-bp EcoRI-HindIII frag- 23 ment of FLI1 was used as the 3Ј-specific probe; the 480-bp to a heterologous DNA-binding domain; therefore, to BamHI-EcoRI fragment of FLI1 was used as the 5Ј-specific directly address the specificity of the FLI1 DNA-binding probe. The subcellular localization of the proteins was deter- domain in the ES phenotype, we generated two hybrids

Cancer Gene Therapy, Vol 7, No 8, 2000 1190 ATHANASIOU, LEGALLIC, WATSON, ET AL: SUPPRESSION OF ES

tion factor activity of the two hybrids were initially tested in parallel with the parental proteins. The in vitro- produced hybrid proteins had the expected sizes of ϳ80 kDa for FLI1-ERF and ϳ40 kDa for ERF-FLI1 (Fig 1B), confirming the in-frame fusion and the absence of dramatic structural changes. The in vitro-produced wt and hybrid proteins could bind the ETS1–3 oligonucleotide,27 which contained the optimal EBS, with similar affinities (Fig 1C), suggesting that fusion did not affect the structure or accessibility of the ets-DNA-binding domain. Furthermore, both wt ERF and the FLI1-ERF hybrid could be super-shifted by an anti-ERF-specific Ab (Fig 1C, first three lanes), confirming the integrity of the hybrid. This Ab is directed against the amino terminus of the ERF protein23 and does not recognize the ERF-FLI1 hybrid because the epitope is destroyed in this fusion. Consistent with previous observations,10,11 transient transfection assays in NIH/3T3 cells using an EBS-containing reporter construct indicated that EWS/ FLI1 was a much more potent transcriptional activator than FLI-1 (Fig 1D). In contrast, both ERF/FLI1 hybrids, like ERF,23 exhibited transcriptional repressor activity and could antagonize the EWS/FLI1-induced transcriptional activation of the recombinant promoter more effectively than wt ERF (Fig 1D). The repression of the EWS/FLI1-induced activation was proportional to the relative amounts of the plasmids as well as to their affinity for the specific EBS present in the recombinant Figure 1. Hybrid generation. A: A schematic representation of the promoter (data not shown). The transactivation data in FLI1/ERF and the ERF/FLI1 hybrid proteins, together with the combination with the in vitro expression of the proteins parental FLI1 and ERF proteins, is shown. The EWS/FLI1 fusion also confirm the functional and structural integrity of the protein is also shown for comparison. B: The FLI1-ERF and ERF- chimeras with regard to their ability to produce the FLI1 hybrid proteins were synthesized in vitro, together with the parental FLI1 and ERF proteins, to verify the production of the expected protein in vitro and in vivo. correctly sized protein. C: Proteins from (B) were used in electrophoretic mobility shift assays to verify their DNA-binding ability. The Expression of transgenes in ES cells anti-ERF-specific Ab was used to identify the ERF- and ERF hybrid-containing complexes. D: The effect on the transcription of Two ES cell lines, RD-ES (HTB-166; ATCC) and the plasmids described in (A) was tested in transient transfections in SK-ES-1 (HTB-86; ATCC) were used as model systems NIH/3T3 cells using 0.5 ␮g of the TK-GATA.CAT reporter plasmid to test the effects of the expression of the ERF and and 0.5 ␮g of the indicated effector plasmids. Activities were ERF-hybrid proteins on the EWS phenotype. We veri- normalized for transfection activity variation. fied that both cell lines carry the EWS-FLI1 rearrangement by Northern blot analysis. Cell lines contained (Fig 1A). In the first (FLI1-ERF), we maintained the messages that hybridized to the 3Ј FLI1 probe, but not ERF structure and replaced the ERF DNA-binding the 5Ј FLI1 probe, contained in the EWS-FLI1 rear- domain (amino acids 1–101) with the FLI1 DNA-bind- rangement, (Fig 2); this was verified by reverse tran- ing domain (amino acids 245–380). This construct could scriptase-polymerase chain reaction with primers spe- address the contribution of DNA-binding specificity to cific for the EWS-FLI1 mRNA (data not shown). the possible suppression of the ES-transformed pheno- To test the effects of expression of the transcriptional type. The second hybrid (ERF-FLI1) retained the FLI1 repressors on the EWS phenotype, cells were infected part of the EWS-FLI1 hybrid (amino acids 233–451 of either with the control retrovirus or with one carrying FLI1) and replaced the EWS portion with the repressor the ERF, FLI1,orFLI1/ERF hybrid genes. Stable, domain of ERF (amino acids 473–546), allowing us to infected cells were selected with G418, and both pools test the role of the EWS portion of the EWS/FLI1 and individual G418-resistant colonies were isolated and hybrid in the development of the Ewing’s phenotype. analyzed. The expression level of the introduced genes, The hybrid genes, as well as the parental ERF and FLI1, as well as that of the relevant endogenous genes, was were cloned into expression vectors under the control of determined by Northern blot analysis. As probes, we either the simian virus 40 (SV40) promoter (pSG5 used a 3Ј FLI1 fragment (see Materials and Methods), vector), the RSV promoter (pRc/RSV), or the retroviral which can detect the wt FLI1 mRNA, the EWS/FLI1 long terminal repeat (pLRNL). chimeric transcript, and the FLI1-ERF hybrid, as well as The structural integrity, DNA binding, and transcrip- a5Ј FLI1 probe, which can detect only the wt FLI1

Cancer Gene Therapy, Vol 7, No 8, 2000 ATHANASIOU, LEGALLIC, WATSON, ET AL: SUPPRESSION OF ES 1191

many non-tumor cells and was an indication that moderate expression of the FLI1/ERF hybrid could suppress some aspects of the ES phenotype. The other ES- derived cell line, SK-ES-1, grows as a monolayer; we did not observe any significant morphological changes when these cells were infected with any of the three recombinant retroviruses (data not shown). A stronger indication of the tumorigenic potential of a cell, however, is tumor development after injection into athymic mice. Although neither RD-ES cells nor those cells overexpressing ERF, FLI1, or FLI1/ERF exhibited any tumorigenic potential in this assay (data not shown), the parental SK-ES-1 cells were highly tumorigenic when injected into nude mice. In addition, vector- infected, FLI1-infected, and ERF-infected SK-ES-1 cells were equally tumorigenic, and the tumors grew at similar rates (Table 1), suggesting that, as in the case of the RD-ES cell morphology (see above), expression of the wt ERF and FLI1 genes was not able to suppress the Figure 2. Analysis of RNA expression. RD-ES and SK-ES-1 cells tumorigenic phenotype. However, the SD-ES-1 cells were infected with the indicated recombinant retroviruses. After expressing the FLI1/ERF hybrid developed tumors G418 selection, individual clones (#1, #2) or pools of clones (P, and much more slowly, requiring almost twice as much time all lanes in SK-ES-1 cells), were expanded; the expression of the for palpable tumors to appear (Table 1). All of the cell transgene was detected by Northern blot analysis with the indicated lines had similar proliferation rates in culture, so this probes. The 3Ј FLI1 probe also detects the endogenous EWS/FLI1 delay could not be explained by growth rate effects. transcript, and its position is indicated. Endogenous FLI1 transcripts Furthermore, it was difficult to establish cell lines from Ј (detectable by the 5 FLI1 probe) are absent. the tumors that did arise, and we observed extensive cell death during the drug selection process, suggesting that many of the cells forming the tumor did not contain the mRNA. We also used an ERF probe, which can detect transgene. In all cases we were able to finally establish both the wt ERF and the ERF-hybrids. The RNA cell lines expressing the introduced genes, but signifi- expression level of the introduced genes, which was cantly, the level of fusion gene mRNA in the FLI1/ERF- determined by densitometric analysis of the autoradio- infected SK-ES-1 cells was much lower than that seen in grams, suggested a moderate overexpression. In general, other transgene-expressing cell lines, and was also lower the level of expression was 1- to 10-fold greater than than the level of endogenous EWS/FLI1 expression. both the level of endogenous EWS/FLI1 mRNA (Fig 2, This low level of expression, although sufficient to 3Ј FLI1 probe, lower bands) and the endogenous ERF decrease the tumorigenic potential of the SK-ES-1 cells, transcript (Fig 2, ERF probe, lower bands). It is of may be the cause of the incomplete suppression of the interest that the expression of the FLI1/ERF hybrid in tumorigenic phenotype in these cells. SK-ES-1 cells is 2-fold lower than that of endogenous The incomplete suppression of the SK-ES-1 tumori- EWS/FLI1 (Fig 2, last lane, FLI1/ERF), but is still genicity could also be a result of the subcellular local- sufficient to suppress the tumorigenic phenotype (see ization of the FLI1-ERF hybrid. We have shown that the below). The EWS/FLI1, ERF, FLI1, and FLI1-ERF intracellular distribution of ERF is controlled by Erk- protein levels, which were determined by Western blot- dependent phosphorylation in response to the prolifer- ting using the specific FLI1 and ERF Abs, were consis- ation stage of the cell.24 We examined the subcellular tent with the mRNA levels, suggesting minimal, if any, distribution of the FLI1/ERF hybrid and found it to be posttranslational effects (data not shown). identical with wt ERF. Thus FLI1/ERF is localized in the cytoplasm of proliferating cells and in the nucleus of FLI1-ERF fusion suppresses the ES phenotype serum-arrested cells (Fig 4). This pattern is unique to The ES tumor-derived RD-ES cells normally grow as the FLI1-ERF hybrid, because both EWS/FLI110 and loosely attached clusters that form primitive cell junc- FLI1 are nuclear proteins and their localization has not tions. Cells expressing exogenous FLI1 or ERF exhibited been reported to change with the cell cycle stage. The a morphology similar to the parental RD-ES cells, localization of the FLI1-ERF hybrid is also consistent although their clustering was occasionally more evident, with its transcription repressor and suppressor activity depending upon cell density and clonal variation. In (Table 1 versus Table 2) and suggests that even the contrast, expression of the FLI1/ERF hybrid induced a temporary presence of this hybrid protein in the nucleus drastic change in cell morphology. FLI1/ERF-infected was adequate to suppress the Ewing’s phenotype. cells grew as a firmly attached monolayer, suggesting To overcome the potential localization and low ex- that these cells had acquired an increased tendency to pression level problems and to generate a hybrid repres- adhere to the substrate (Fig 3). This is a characteristic of sor that would interfere more directly with EWS/FLI1

Cancer Gene Therapy, Vol 7, No 8, 2000 1192 ATHANASIOU, LEGALLIC, WATSON, ET AL: SUPPRESSION OF ES

Table 1. Suppression of EWS Tumorigenicity by FLI1/ERF Hybrids*

Week 2 Week 3 Week 4 Week 5 Week 6 Week 7

pLRNL 0/10 10/10 FLI1 0/10 10/10 ERF 0/10 9/10 10/10 FLI1/ERF 0/10 0/10 0/10 2/10 5/10 9/10 *A total of 500,000 G418-selected SK-ES-1 cells, generated after infection with the indicated virus, were injected subcutaneously into athymic mice. The animals were monitored twice a week for tumor development and general health.

function, we tested constructs expressing the ERF/FLI1 hybrid (Fig 1) under the control of two potent promoters derived fromSV40 and RSV (see above and Fig 1). In this hybrid, the FLI1 structure of the EWS/FLI1 protein is maintained, but the ERF repressor domain replaces the EWS activation domain. Signiﬁcantly, this construct also lacks the region responsible for the control of ERF localization. We introduced this hybrid into the two ES-derived cell lines, using cotransfection with a selectable marker in the case of the SV40-promoter driven construct. (The RSV promoter-driven construct contains an internal neo-selectable marker.) In replicate- independent experiments, transfections with the plasmids carrying the hybrid gene yielded 10- to 20-fold fewer colonies than the vector control (116 colonies for ERF/FLI1 constructs versus 1938 colonies for the vectors; Table 2). Furthermore, we were unable to obtain any colonies of SK-ES-1 or RD-ES cells that express this

Figure 4. Subcellular localization. ERF-expressing plasmids (a,a’), FLI1-expressing plasmids (b,b’), and FLI1/ERF-expressing plasmids (c,c’) were transfected into NIH/3T3 cells. The subcellular Figure 3. FLI1/ERF increases the attachment properties of ES cells. localization of the protein was detected by indirect immunofluores- RD-ES cells, uninfected or infected with the indicated retroviruses cence in exponentially growing cells (a–c) or serum-arrested cells and selected for G418 resistance, were grown in complete culture (a’–c’) with anti-ERF-specific (a,a’;c,c’) and anti-FLI1 specific (b,b’) media and photographed with a 10ϫ phase contrast lens. Abs.

Cancer Gene Therapy, Vol 7, No 8, 2000 ATHANASIOU, LEGALLIC, WATSON, ET AL: SUPPRESSION OF ES 1193

Table 2. Inhibition of Colony Formation in EWS Cells by transition, and has been shown to enter the nucleus at ERF/FLI1* specific stages of the cell cycle.24 The FLI1/ERF protein Promoter/ Vector ERF/FLI1 ERF/FLI1 demonstrates a similar mitogen-activated protein ki- vector Experiment clones clones protein nase-dependent subcellular localization, suggesting that temporary entry of the FLI1/ERF protein into the SV40/pSG5 1 256 25 0/25 nucleus is sufficient to suppress the ES phenotype. 2 128 12 0/12 3 114 5 0/5 Because FLI1/ERF and EWS/FLI1 have opposite effects RSV/pRcRSV 1 987 56 0/36 on transcription, the oncogenic activity of the EWS/FLI1 2 453 18 0/18 fusion protein is likely to involve activation of genes that can promote or allow inappropriate progression into the Total 5 1938 116 0/96 G1 phase of the cell cycle. However, the temporary entry *A total of 3 ␮g of the indicated plasmids under the control of of FLI1/ERF into the nucleus may also contribute to the the indicated promoter were transfected into SK-ES-1 and RD-ES incomplete suppression we observed. Indeed, ERF/ cells and selected for resistance to G418. ERF/FLI1-transfected FLI1, a hybrid that contains only the repressor domain clones were expanded and tested for the expression of the hybrid of ERF and the DNA-binding domain of FLI1 and protein using anti-ERF and/or anti-FLI1-specific Abs. whose localization in the nucleus, is independent of the cell cycle; in addition, ERF/FLI1 appeared to totally inhibit cell growth in ES, NIH/3T3, and HeLa cells. This hybrid protein (none of 96 clones tested by immunoblot- is consistent with our observation that this fusion is a ting; Table 2). These two observations suggested that more potent transcriptional repressor than the FLI1/ this potent transcriptional repressor hybrid totally blocks ERF hybrid. However, quantitative effects may also the ability of the cells to form colonies in tissue culture. contribute to these responses, because transfection may This was also true when the ERF/FLI1 constructs were have introduced high levels of this hybrid in the trans- introduced in NIH/3T3 and HeLa cells, suggesting that fected cells, and the hybrid transcript is driven by the nonregulatable form of this repressor may be toxic. stronger promoters. In contrast, the FLI1/ERF hybrid was successfully ex- Although ets proteins constitute one of the more expressed in a number of cell types (HeLa, K562, FDC-P2, tended families of transcription factors in mammalian cells, NIH/3T3) without any adverse effects (data not shown). only a limited number of ets domain proteins appear to be involved in rearrangements with the EWS gene. This could be due to the localization of other ets genes in regions with DISCUSSION lower recombination frequencies. However, ets genes have been found to recombine with other genes,31–34 suggesting The EWS/FLI1 fusion gene product is necessary for the that they are capable of recombination and rearrangement. maintenance of the transformed phenotype of ES tu- It is more likely that only specific ets-EWS recombinations mors; in addition, it is highly tumorigenic in model generate hybrid proteins that recognize specific targets that systems.10,11 For these reasons, its repression has be- subsequently promote tumor formation.35–38 It is likely that come an attractive target for the control of ES tumori- specific ets domains recognize unique cellular targets. This genesis.13–15 In this report, we show that it is possible, hypothesis is also supported by our finding that the FLI1/ using retroviral gene transfer, to suppress the ES phe- ERF hybrid, but not wt ERF, is able to suppress the ES notype by inducing the expression of a specific transcrip- phenotype, despite their similarities in expression level and tional repressor that can antagonize EWS/FLI1 activity. subcellular compartmentalization. Alignment of the ets We also show that this suppression occurs even at low genes found rearranged with the EWS genes (Fig 5) reveals levels of transgene expression. that two amino acids within the conserved DNA-binding ERF, a member of the ets gene family, is an active domain are unique among them (methionine 348 and repressor, and we have shown previously that it contains tyrosine 356 in the FLI1 sequence; indicated by open a domain that can exhibit transcriptional repressor ac- arrows in Fig 5). Random mutagenesis has shown that tivity when attached to a heterologous DNA-binding mutations at these residues drastically affect the DNA- domain.23 Furthermore, substoichiometric amounts of binding activity.39 These amino acids are not directly the ERF repressor can effectively repress transcription involved in protein-DNA contact,40 and it is likely that they in the presence of an activator that recognizes the same affect interactions with other proteins important for target targets. Here we have shown that introduction of a recognition. The contribution by the DNA-binding domain hybrid protein containing the FLI1 DNA-binding do- may be detectable in our study because the levels of protein main linked to the regulatory and repressor domains of expression were similar to normal cellular levels. Under ERF resulted in suppression of the tumorigenic pheno- these more physiological conditions, we may have been type of ES. In suppressed cells, the level of endogenous able to detect small, but important, differences that have EWS/FLI1 transcript (and protein; our unpublished been obscured previously by overexpression. This is one of observations) was unaffected, suggesting that the sup- the few cases in which functional distinctions can be made pression was not due to decreased oncogene levels, but among ets-DNA-binding domains, and it is consistent with rather to the decrease of its transcriptional effect on its in vivo studies indicating that small differences in the targets. ERF has been shown to be important for the G1 structure and activity of the fusion protein can be corre-

Cancer Gene Therapy, Vol 7, No 8, 2000 1194 ATHANASIOU, LEGALLIC, WATSON, ET AL: SUPPRESSION OF ES

Figure 5. DNA-binding domain alignment. The alignment of the DNA-binding domain of the ets family genes, which have been found to be rearranged with the EWS gene, as well as the DNA-binding domain of ERF are shown. Black areas indicate identical amino acids; gray areas indicate conservative substitutions. Filled arrows indicate amino acids unique to the ERF subfamily; open arrows indicate amino acids unique to mammalian ets genes rearranged with EWS.

lated with phenotypic differences.12,41,42 Other studies, ACKNOWLEDGMENTS however, have failed to detect DNA-binding-related differences.43,44 This paper is dedicated to the late T. S. Papas for his help and Previous attempts to use molecular methods to sup- support of our work throughout the years. We thank X.-K. press the EWS/FLI1 tumorigenic activity have involved Zhang for Abs and G. Beal Jr. for technical support. This work either antisense approaches or the introduction of con- was supported in part by European Union Grants FMRX- structs expressing high levels of FLI1 DNA-binding CT96-0041 and BMH4-CT96-1355 (to G.M.) and by Federal funds from the National Cancer Institute, National Institutes domains, which can act as competitors by occupying the of Health, under Contract No. NO1-CO-56000. same targets as the EWS/FLI1 protein.13–15 Both approaches resulted in decreased tumorigenicity, support- ing the hypothesis that the ES phenotype results from the inappropriate activation of genes controlled by FLI1 REFERENCES or related members of the ets family of transcription factors. However, these approaches involve overexpres- 1. Hurst HC. Transcription factors as drug targets in cancer. sion of the antagonizing factor, which is not easy to Eur J Cancer. 1996;32A:1857–1863. achieve under in vivo conditions, and which may also 2. Lewin B. Oncogenic conversion by regulatory changes in impair the normal physiological function of the FLLI transcription factors. Cell. 1991;64:303–312. gene and produce generalized toxic effects. Similar 3. Delattre O, Zucman J, Plougastel B, et al. Gene fusion effects are probably responsible for our inability to with an ETS DNA-binding domain caused by chromosome translocation in human tumours. Nature. 1992;359:162– obtain stable clones overexpressing the ERF/FLI1 fu- 165. sion protein, which is a very potent transcriptional 4. Sorensen PH, Lessnick SL, Lopez-Terrada D, Liu XF, Triche repressor, not only in ES cells, but also in other normal TJ, Denny CT. A second Ewing’s sarcoma translocation, or transformed cell types (unpublished data). t(21;22), fuses the EWS gene to another ETS-family tran- In contrast, we were successful with an approach using scription factor, ERG. Nat Genet. 1994;6:146–151. an active repressor that can effectively suppress the ES 5. Jeon IS, Davis JN, Braun BS, et al. A variant Ewing’s phenotype at expression levels lower than the transform- sarcoma translocation (7;22) fuses the EWS gene to the ing gene and that employs retroviral-mediated transfer ETS gene ETV1. Oncogene. 1995;10:1229–1234. for more effective gene delivery. Furthermore, the active 6. Urano F, Umezawa A, Hong W, Kikuchi H, Hata J. A hybrids contain the ERF domain that controls subcellu- novel chimera gene between EWS and E1A-F, encoding the adenovirus E1A enhancer-binding protein, in ex- lar localization, allowing for the generation of repressor traosseous Ewing’s sarcoma. Biochem Biophys Res Com- proteins whose functional activity can be modulated at mun. 1996;219:608–612. several levels. Both inhibition of Erk activity as well as 7. Peter M, Magdelenat H, Michon J, et al. Sensitive detec- introduction of mutations that affect the ERF subcellu- tion of occult Ewing’s cells by the reverse transcriptase- lar localization can generate a range of repressors with polymerase chain reaction. Br J Cancer. 1995;72:96–100. increased nuclear localization and thus increased repres- 8. Dittmer J, Nordheim A. Ets transcription factors and sor activity.24 It has been suggested that “the ideal human disease. Biochim Biophys Acta. 1998;1377:F1–F11. cancer therapy would accommodate the speciﬁc biology 9. de Launoit Y, Baert JL, Chotteau A, et al. Structure- of a tumor and be based on understanding the mecha- function relationships of the PEA3 group of Ets-related nisms of malignancy.”45 In this sense, ERF repressor transcription factors. Biochem Mol Med. 1997;61:127–135. 10. May WA, Lessnick SL, Braun BS, et al. The Ewing’s hybrids, with the unique potential to be regulated both sarcoma EWS/FLI-1 fusion gene encodes a more potent quantitatively and by extracellular signals, may be an transcriptional activator and is a more powerful transform- approach that could be valuable in the ES system, as well ing gene than FLI-1. Mol Cell Biol. 1993;13:7393–7398. as other systems that involve inappropriate transcription 11. Bailly RA, Bosselut R, Zucman J, et al. DNA-binding and factor activation. transcriptional activation properties of the EWS-FLI-1

Cancer Gene Therapy, Vol 7, No 8, 2000 ATHANASIOU, LEGALLIC, WATSON, ET AL: SUPPRESSION OF ES 1195

fusion protein resulting from the t(11;22) translocation in protein: the determination of the ETS binding motif. Ewing sarcoma. Mol Cell Biol. 1994;14:3230–3241. Oncogene. 1991;6:2249–2254. 12. Lin PP, Brody RI, Hemelin AC, Bradner JE, Healey JH, 28. Danos O, Mulligan RC. Safe and efﬁcient generation of Ladanyi M. Differential transactivation by alternative EWS- recombinant retroviruses with amphotropic and ecotropic FLI1 fusion proteins correlates with clinical heterogeneity in host ranges. Proc Natl Acad Sci USA. 1988;85:6460–6464. Ewing’s sarcoma. Cancer Res. 1999;59:1428–1432. 29. Miller AD, Buttimore C. Redesign of retrovirus packaging 13. Ouchida M, Ohno T, Fujimura Y, Rao VN, Reddy ES. cell lines to avoid recombination leading to helper virus Loss of tumorigenicity of Ewing’s sarcoma cells expressing production. Mol Cell Biol. 1986;6:2895–2902. antisense RNA to EWS-fusion transcripts. Oncogene. 30. Athanasiou M, Clausen PA, Mavrothalassitis GJ, Zhang 1995;11:1049–1054. XK, Watson DK, Blair DG. Increased expression of the 14. Kovar H, Aryee DN, Jug G, et al. EWS/FLI-1 antagonists ETS-related transcription factor FLI-1/ERGB correlates induce growth inhibition of Ewing tumor cells in vitro. Cell with and can induce the megakaryocytic phenotype. Cell Growth Differ. 1996;7:429–437. Growth Differ. 1996;7:1525–1534. 15. Tanaka K, Iwakuma T, Harimaya K, Sato H, Iwamoto Y. 31. Buijs A, Sherr S, van Baal S, et al. Translocation (12;22) EWS-Fli1 antisense oligodeoxynucleotide inhibits prolifer- (p13;q11) in myeloproliferative disorders results in fusion ation of human Ewing’s sarcoma and primitive neuroecto- of the ETS-like TEL gene on 12p13 to the MN1 gene on dermal tumor cells. J Clin Invest. 1997;99:239–247. 22q11. Oncogene. 1995;10:1511–1519. 16. Bertolotti A, Lutz Y, Heard DJ, Chambon P, Tora L. 32. Golub TR, Barker GF, Bohlander SK, et al. Fusion of the hTAF(II)68, a novel RNA/ssDNA-binding protein with TEL gene on 12p13 to the AML1 gene on 21q22 in acute homology to the pro-oncoproteins TLS/FUS and EWS, is lymphoblastic leukemia. Proc Natl Acad Sci USA. 1995;92: associated with both TFIID and RNA polymerase II. 4917–4921. EMBO J. 1996;15:5022–5031. 33. Golub TR, Goga A, Barker GF, et al. Oligomerization of 17. Bertolotti A, Melot T, Acker J, Vigneron M, Delattre O, the ABL tyrosine kinase by the Ets protein TEL in human Tora L. EWS, but not EWS-FLI-1, is associated with both leukemia. Mol Cell Biol. 1996;16:4107–4116. TFIID and RNA polymerase II: interactions between two 34. Lacronique V, Boureux A, Valle VD, et al. A TEL-JAK2 members of the TET family, EWS and hTAFII68, and fusion protein with constitutive kinase activity in human subunits of TFIID and RNA polymerase II complexes. Mol leukemia. Science. 1997;278:1309–1312. Cell Biol. 1998;18:1489–1497. 35. Hahm KB, Cho K, Lee C, et al. Repression of the gene 18. Ben-David Y, Giddens EB, Letwin K, Bernstein A. Eryth- encoding the TGF-␤ type II receptor is a major target of roleukemia induction by Friend murine leukemia virus: the EWS-FLI1 oncoprotein. Nat Genet. 1999;23:222–227. insertional activation of a new member of the ets gene 36. Kovar H. Ewing’s sarcoma and peripheral primitive neu- family, Fli-1, closely linked to c-ets-1. Genes Dev. 1991;5: roectodermal tumors after their genetic union. Curr Opin 908–918. Oncol. 1998;10:334–342. 19. Watson DK, Smyth FE, Thompson DM, et al. The ERGB/ 37. May WA, Arvand A, Thompson AD, Braun BS, Wright M, Fli-1 gene: isolation and characterization of a new member Denny CT. EWS/FLI1-induced manic fringe renders NIH of the family of human ETS transcription factors. Cell 3T3 cells tumorigenic. Nat Genet. 1997;17:495–497. Growth Differ. 1992;3:705–713. 38. Arvand A, Bastians H, Welford SM, Thompson AD, Ruder- 20. Hromas R, Klemsz M. The ETS oncogene family in man JV, Denny CT. EWS/FLI1 up regulates mE2-C, a development, proliferation, and neoplasia. Int J Hematol. cyclin-selective ubiquitin conjugating enzyme involved in 1994;59:257–265. cyclin B destruction. Oncogene. 1998;17:2039–2045. 21. May WA, Denny CT. Biology of EWS/FLI and related fusion 39. Mavrothalassitis G, Fisher RJ, Smyth F, Watson DK, genes in Ewing’s sarcoma and primitive neuroectodermal Papas TS. Structural inferences of the ETS1 DNA-binding tumor. Curr Top Microbiol Immunol. 1997;220:143–150. domain determined by mutational analysis. Oncogene. 22. Wasylyk B, Hagman J, Gutierrez-Hartmann A. Ets tran- 1994;9:425–435. scription factors: nuclear effectors of the Ras-MAP-kinase 40. Liang H, Mao X, Olejniczak ET, et al. Solution structure signaling pathway. Trends Biochem Sci. 1998;23:213–216. of the ets domain of Fli-1 when bound to DNA. Nat Struct 23. Sgouras DN, Athanasiou MA, Beal GJ Jr, Fisher RJ, Blair Biol. DG, Mavrothalassitis GJ. ERF, an ETS domain protein 1994;1:871–875. with strong transcriptional repressor activity, can suppress 41. Amann G, Zoubek A, Salzer-Kuntschik M, Windhager R, ets-associated tumorigenesis and is regulated by phosphor- Dovar H. Relation of neurological marker expression and ylation during cell cycle and mitogenic stimulation. EMBO EWS gene fusion types in MIC2/CD99-positive tumors of J. 1995;14:4781–4793. the Ewing family. Hum Pathol. 1999;30:1058–1064. 24. Le Gallic L, Sgouras D, Beal G, Mavrothalassitis G. 42. Thompson AD, Teitell MA, Arvand A, Denny CT. Diver- Transcriptional repressor ERF is a Ras/mitogen-activated gent Ewing’s sarcoma EWS/ETS fusions confer a common protein kinase target that regulates cellular proliferation. tumorigenic phenotype on NIH3T3 cells. Oncogene. 1999; Mol Cell Biol. 1999;19:4121–4133. 18:5506–5513. 25. Xu L, Yee J-K, Wolff JA, Friedmann T. Factors affecting 43. Ginsberg JP, de Alava E, Ladanyi M, et al. EWS-FLI1 and long-term stability of Moloney murine leukemia virus- EWS-ERG gene fusions are associated with similar clinical based vectors. Virology. 1989;171:331–341. phenotypes in Ewing’s sarcoma. J Clin Oncol. 17:1999:1809. 26. Zhang L, Eddy A, Teng YT, et al. An immunological renal 44. Jaishankar S, Zhang J, Roussel MF, Baker SJ. Transform- disease in transgenic mice that overexpress Fli-1, a mem- ing activity of EWS/FLI is not strictly dependent upon ber of the ets family of transcription factor genes. Mol Cell DNA-binding activity. Oncogene. 1999;18:5592–5597. Biol. 1995;15:6961–6970. 45. Kovar H, Aryee D, Zoubek A. The Ewing family of tumors 27. Fisher RJ, Mavrothalassitis G, Kondoh A, Papas TS. High and the search for the Achilles’ heel. Curr Opin Oncol. afﬁnity DNA:protein interactions of the cellular ETS1 1999;11:275–284.

Cancer Gene Therapy, Vol 7, No 8, 2000