Oncogene (2009) 28, 270–278 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE The myxoid liposarcoma FUS-DDIT3 fusion oncoprotein deregulates NF-jB target by interaction with NFKBIZ

MGo¨ ransson1, MK Andersson1, C Forni2, A Sta˚ hlberg3, C Andersson1, A Olofsson1, R Mantovani2 and P A˚ man1

1Lundberg Laboratory for Cancer Research (LLCR), Department of Pathology, Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden; 2Dipartimento di Scienze Biomolecolari e Biotecnologie, Universita` degli Studi di Milano, Milano, Italy and 3Department of Clinical Neuroscience and Rehabilitation, Center for Brain Repair and Rehabilitation (CBR), Sahlgrenska Academy at University of Gothenburg, Gothenburg, Sweden

FUS (also called TLS), EWSR1 and TAF15 (also called Introduction TAF2N) are related genes involved in tumor type-specific fusion oncogenes in human malignancies. The FUS- FUS, EWSR1 and TAF15 are three closely related genes DDIT3 fusion oncogene results from a t(12;16)(q13;p11) encoding RNA-binding with N-terminal parts translocation and has a causative role in the sharing stretches of degenerated SYGQQS repeats initiation of myxoid/round cell liposarcomas (MLS/ (A˚ man et al., 1996; Morohoshi et al., 1998; Law et al., RCLS). The FUS-DDIT3 induces increased 2006). All three genes formfusion oncogenes, generated expression of the CAAT/enhancer-binding protein by chromosome translocations in a large group of (C/EBP) and nuclear factor-jB (NF-jB)-controlled human sarcomas and leukemias (Perez-Losada et al., IL8, and the N-terminal FUS part is required for this 2000b; Kovar, 2005). The fusion proteins consist of activation. Chromatin immunoprecipitation analysis N-terminal parts of FUS, EWS or TAF15, which are showed that FUS-DDIT3 binds the IL8 promoter. juxtaposed to the DNA-binding parts of various Expression studies of the IL8 promoter harboring a transcription factors, and the chimeric proteins have C/EBP–NF-jB composite site pinpointed the importance been shown to act as abnormal transcription factors of NF-jB for IL8 expression in FUS-DDIT3-expressing (Ohno et al., 1993; Sanchez Garcia and Rabbitts, 1994; cells. We therefore probed for possible interaction of Zinszner et al., 1994). In some tumor types, FUS, FUS-DDIT3 with members of the NF-jB family. The EWSR1 and TAF15 replace each other as 50 fusion nuclear factor NFKBIZ colocalizes with FUS-DDIT3 in partners showing that the N-terminal parts of their nuclear structures, and immunoprecipitation experiments protein products share features and functions important showed that FUS-DDIT3 binds the C-terminal of for cell transformation and tumor development in these NFKBIZ. We also report that additional NF-jB-con- groups of sarcomas and leukemias. trolled genes are upregulated at the mRNA level in In the present investigation, we focused on myxoid/ FUS-DDIT3-expressing cell lines and they can be induced round cell liposarcoma (MLS/RCLS). More than 90% by NFKBIZ. Taken together, the results indicate that of these tumors carry a t(12;16)(q13;p11) chromosome FUS-DDIT3 deregulates some NF-jB-controlled genes translocation that results in a fusion between FUS and through interactions with NFKBIZ. Similar mechanisms the transcription factor encoding DDIT3 (also called may be a part of the transformation process in other CHOP and GADD153) (Crozat et al., 1993; Rabbitts tumor types carrying FUS, EWSR1 and TAF15 contain- et al., 1993). A minority of MLS/RCLS tumors carry a ing fusion oncogenes. variant fusion gene in which FUS is replaced by EWSR1 Oncogene (2009) 28, 270–278; doi:10.1038/onc.2008.378; (Panagopoulos et al., 1996). DDIT3 encodes a DNA- published online 13 October 2008 binding transcription factor of basic leucine zipper type (CAAT/enhancer-binding protein (C/EBP) family) that Keywords: MLS/RCLS; FUS; DDIT3; NF-kB; binds to DNA as a heterodimer with other C/EBP NFKBIZ; sarcoma transcription factors. DDIT3 cannot form homodimers, but C/EBPb, which is natively expressed in the here investigated cells, is considered as a major dimerization partner to DDIT3 and the dimer-forming function is maintained in the FUS-DDIT3 fusion protein (Crozat et al., 1993). DDIT3 and FUS-DDIT3 may therefore Correspondence: Professor P A˚ man, Lundberg Laboratory for Cancer interfere with C/EBPb and its binding to transcription Research (LLCR), Department of Pathology, University of Gothen- regulatory C/EBP sites. burg, Gula Straket 8, Gothenburg SE-413 45, Sweden. E-mail: [email protected] The causative role of the FUS-DDIT3 fusion gene in Received 21 August 2008; accepted 1 September 2008; published online the initiation of MLS/RCLS, and its role for the MLS- 13 October 2008 specific tumor morphology has been demonstrated in FUS-DDIT3 deregulates NF-jB target genes MGo¨ransson et al 271 transgenic mice and xenografts, and more recently in FUS-DDIT3 carrying mesenchymal stem cells (Perez- Losada et al., 2000b; Engstro¨ m et al., 2006; Riggi et al., 2006). The importance of the FUS N-terminal part for oncogenic activity of FUS-DDIT3 was further con- firmed by NIH-3T3-based transformation assays (Zinsz- ner et al., 1994) and in transgenic mice (Perez-Losada et al., 2000b). We have recently shown that the C/EBP and nuclear factor-kB (NF-kB)-controlled gene IL8 is upregulated in FUS-DDIT3-expressing cell lines, whereas expression of the normal DDIT3 inhibits IL8 transcription (Go¨ rans- son etal., 2005). These observations provided an opportunity to investigate the molecular mechanism behind aberrant gene regulation caused by the FUS- DDIT3 fusion protein. We here report that interaction with an NF-kB- binding site in the IL8 promoter is crucial for the divergent regulation by DDIT3 and FUS-DDIT3. Our results provide an important insight in the molecular mechanism behind the abnormal transcription factor activity of FUS-DDIT3. Figure 1 (a) Luciferase assays of the IL8 promoter in HT1080 cell lines. Bars represent mean IL8 promoter activity relative to the intact IL8 promoter activity in each cell line. Graphed values were Results obtained by normalizing measured luciferase activity by b-galactosidase activity and background signal fromthe empty control vector. Error bars represent 95% confidence intervals. Data NF-kB is a major factor controlling IL8 transcription are pooled from two independent experiments performed in in FUS-DDIT3-expressing cells biological triplicates. Asterisks indicate statistical significance by DDIT3 and FUS-DDIT3 give opposite effects on the independent samples t-test where ***Po0.001. (b) Cartoon IL8 transcription in HT1080 cells (Go¨ ransson etal., depicting the IL8 promoter constructs used. 2005). To study the mechanism behind this difference, we expressed luciferase reporter constructs containing the C/EBP–NF-kB composite site of the IL8 promoter, et al., 1993). The observed effects of FUS-DDIT3 and or mutant forms of this construct in genetically modified DDIT3 on IL8 expression could be explained by direct HT1080 cells (Figure 1). binding to this composite site. Chromatin immunopre- Deletion of the C/EBP site resulted in reduced IL8 cipitation (ChIP) experiments with anti-DDIT3 and promoter activity in HT1080 showing that transcription anti-FUS serumsamples on MLS 402-91 cells verified fromthe IL8 promoter depends on the C/EBP site in that FUS-DDIT3 bound to the IL8 promoter (Figure 2). these cells (Figure 1). Similar results were obtained for DDIT3-expressing HT1080 cells in which the activity FUS-DDIT3 and NFKBIZ colocalize in nuclear was strongly reduced. In contrast, FUS-DDIT3-expres- structures sing cells were not dependent on the C/EBP-binding site We have reported earlier that FUS-DDIT3 localizes to for IL8 promoter activity. specific nuclear structures (Go¨ ransson et al., 2002; Mutation of the NF-kB site led to a significantly Thelin-Ja¨ rnum et al., 2002). We used this feature to increased promoter activity in the HT1080 control cells. search for interactions of FUS-DDIT3 with members of In DDIT3-expressing cells, this increase was more than the NF-kB/Rel family in FUS-DDIT3-GFP transfected 14-fold. These results indicate that the NF-kB site HT1080 cells. Transfected cells were stained with Cy3- functions as a negative regulator of the IL8 promoter in labeled antisera directed against NF-kB1 (p50), NF-kB2 control cells and DDIT3-expressing cells. In contrast, (p52), RELA (p65), RELB and REL, and the cellular the NF-kB site functions as a strong activator in FUS- distribution of the red and green fluorescence was DDIT3-expressing cells because mutation of the NF-kB studied. We found no evident colocalization between site drastically decreased IL8 promoter activity in these the FUS-DDIT3 protein and these NF-kB/Rel family cells. Thus, the fusion of FUS N-terminal parts to members (data not shown). No useful antibodies were DDIT3 results in a switch froma suppressive to a available for the NFKBIZ. However, the nuclear stimulatory effect that is dependent on the NF-kB site. pattern of green fluorescent protein (GFP)-tagged NFKBIZ (Kitamura et al., 2000) closely resembled FUS-DDIT3 binds to the IL8 immediate promoter region the localization pattern of FUS-DDIT3-GFP (Thelin- The immediate promoter region of IL8 contains a well- Ja¨ rnum et al., 2002), prompting us to probe for a characterized C/EBP-NF-kB composite site (Libermann colocalization between these proteins. Confocal laser- and Baltimore, 1990; Mukaida et al., 1990; Matsusaka scanning microscopy of HT1080 cells coexpressing

Oncogene FUS-DDIT3 deregulates NF-jB target genes MGo¨ransson et al 272 and the wild-type DDIT3 construct were able to translocate NFKBIZ-C-term-DsRed1 to the nucleus (Figure 3b, IV). This effect was weaker compared with the full-length fusion protein. Similarly, the full-length FUS protein and the N-terminal part of FUS, involved in the FUS-DDIT3 fusion, were found only partially associated with the NFKBIZ mutant protein (Figure 3b, V and VI).

Physical interaction between NFKBIZ and FUS-DDIT3 To confirma binding between NFKBIZ and FUS- DDIT3, we attempted to precipitate the two proteins together using anti-GFP serum. The C-terminal NFKBIZ-GFP and GFP control vectors were coex- pressed with FUS-DDIT3-DsRed1. GFP-tagged pro- teins were immunoprecipitated, and the co-precipitation of FUS-DDIT3-DsRed1 was analysed. A band of approximately 80 kDa corresponding to FUS-DDIT3- DsRed1 was detected in the immunoprecipitated frac- tion of C-terminal NFKBIZ-GFP. Coprecipitation of FUS-DDIT3-DsRed1 could not be observed by the GFP control alone (Figure 4). These results suggest that NFKBIZ binds directly to FUS-DDIT3 or associates indirectly with FUS-DDIT3 in a larger complex through its C-terminal domain.

The NF-kB-controlled genes MMP1 and LCN2 are upregulated in FUS-DDIT3-expressing cell lines MMP1 and LCN2, both genes involved in extracellular matrix remodeling, were found to be upregulated in FUS-DDIT3-GFP cells, whereas DDIT3 had very little effect (Go¨ ransson etal., 2005). The expression patterns were confirmed here by real-time PCR (Table 1). In addition, three investigated myxoid liposarcoma cell lines expressed LCN2 and MMP1 (data not shown).

Figure 2 Chromatin immunoprecipitation (ChIP) of the proximal IL8, IL6 and LCN2 are transcriptional targets of NFKBIZ IL8 promoter in myxoid liposarcoma MLS 402-91 cells. Antibodies In addition to IL8, MMP1 has also been shown to be specific for FUS, DDIT3 and Flag (negative control) were used for regulated by the NF-kB system, and both IL6 and precipitations. A satellite 11 (Sat11) sequence that is repeated multiple times in the genome was used as a control for nonspecific LCN2 have been reported to be downstreamtargets of precipitation and loading. The cell line MLS 402-91 expresses FUS- NFKBIZ (Vincenti et al., 1998; Motoyama et al., 2005; DDIT3 but not the normal DDIT3. FUS-DDIT3 was precipitated Oikawa et al., 2006). To test whether these genes were by the anti-FUS and the anti-DDIT3 serum. Sequence of the transcriptionally controlled by NFKBIZ in our system, proximal IL8 promoter is shown with C/EBP sites boxed and the NF-kB site shaded. Locations of primer sequences are underlined. we used a tamoxifen-regulated vector system (Little- wood et al., 1995). Human HT1080 cells were transi- ently transfected by MOR-GFP or NFKBIZ-MOR- GFP construct for 24 h. The transfection efficiency was GFP-tagged NFKBIZ and DsRed1-tagged FUS- approximated to 100% for the MOR-GFP construct DDIT3 revealed that these factors colocalize in nuclear and to 30% for the NFKBIZ-MOR-GFP construct by structures (Figure 3a). Next, we constructed an counting GFP-positive cells and by western blot analysis N-terminally truncated version of NFKBIZ lacking the (Figure 5b). The expression levels of IL8, IL6, MMP1 nuclear localization signal and tagged the C-terminal and LCN2 were measured by real-time PCR 4 h after the with DsRed1. When this construct was expressed alone, addition of tamoxifen to the cell culture. The results the protein localized to the cytoplasm(Figure 3b, I). In showed that mRNA levels of IL8, IL6 and LCN2 were contrast, when coexpressed with FUS-DDIT3-GFP, this significantly raised in response to nuclear expression of mutant localized to the nuclear structures, suggesting NFKBIZ (Figure 5a). The MMP1 expression was not that FUS-DDIT3 binds and translocates the C-terminal significantly altered. These findings confirmprevious NFKBIZ to the nucleus (Figure 3b, III). Furthermore, if reports that IL6 and LCN2 are downstreamtargets of strongly expressed, both the FUS-DDIT3-DLZ-GFP NFKBIZ and further suggest that IL8 but not MMP1 is construct lacking the leucine zipper dimerizing domain a transcriptional target of the NFKBIZ protein.

Oncogene FUS-DDIT3 deregulates NF-jB target genes MGo¨ransson et al 273

Figure 3 The affinity for NFKBIZ is enhanced by fusion of the N-terminal of FUS to DDIT3. Confocal laser-scanning microscopy showing the localization of fluorescently tagged NFKBIZ and FUS-DDIT3 proteins. Coloring indicates GFP (green), DsRed1 (red) and the DNA dye DAPI (blue). Areas of overlap between red and green channels, that is, colocalization, are yellow. (a) Full-length NFKBIZ-GFP and FUS-DDIT3-DsRed1 colocalize in nuclear structures. Scale bar in panel a ¼ 5 mm.(b, I–VI) Expression of the C- terminal part of NFKBIZ-DsRed1 alone results in cytoplasmic localization. (b, I). Co-expression with FUS-DDIT3-GFP translocates the N-terminally truncated NFKBIZ-DsRed1 mutant to the nucleus. (b, II). To a lesser extent, nuclear translocation of C-terminal NFKBIZ could also be achieved by high coexpression with FUS-DDIT3-DLZ-GFP (b, III) or DDIT3-GFP (b, IV). The full-length FUS protein and the N-terminal part of FUS tagged with GFP associate with the N-terminally truncated NFKBIZ (b, V and VI) and colocalization could be observed in the cytoplasm. Scale bars in panel b ¼ 10 mm.

Table 1 Expressions of MMP1 and LCN2 in DDIT3-GFP and FUS-DDIT3-GFP-containing HT1080 cells Gene Method HT1080 DDIT3 HT1080 FUS-DDIT3

MMP1 Quantitative PCR 1.1±0.1 17±1.2 MMP1 Microarray 0.5±0.1 4.4±0.1 LCN2 Quantitative PCR 2.1±0.1 230±87 Figure 4 Physical interaction between C-terminal NFKBIZ and LCN2 Microarray 1.7±1.2 14±1.0 FUS-DDIT3 revealed by co-immunoprecipitation. HT1080 cells were co-transfected with FUS-DDIT3-DsRed1 and either Values represent fold change compared with the original HT1080 cell NFKBIZ-cterm-GFP or empty green fluorescent protein (GFP) line as the mean of three independent biological replicates ±s.e.m. vector. Immunoprecipitation was performed using an antibody directed against GFP. Input represents the relative amount of tagged protein present in the cell lysate. IP shows the amount of GFP-tagged constructs bound to Protein A agarose. Co-IP shows response to overexpression of mutated variants of FUS- the amount of FUS-DDIT3 that was co-precipitated together with DDIT3 and NFKBIZ in wild-type HT1080 cells and in the NFKBIZ-Cterm-GFP construct or the empty GFP vector. IB, FUS-DDIT3-expressing HT1080 cells (Figure 6) was protein detected by western blot. studied using quantitative PCR. The results showed that although target gene expression in wild-type HT1080 FUS-DDIT3-expressing cells show resistance cells were altered by most constructs, such effects were to DDIT3-mediated downregulation of target genes smaller or absent in FUS-DDIT3-expressing cells. To further investigate the effects of FUS-DDIT3 on the Interestingly, DDIT3 (and FUS-DDIT3) with a deleted regulation of IL8, IL6 and LCN2, their expression in leucine zipper dimerization region also induced a

Oncogene FUS-DDIT3 deregulates NF-jB target genes MGo¨ransson et al 274

Figure 5 (a) Relative mRNA levels of IL8, IL6, MMP1 and LCN2 upon nuclear NFKBIZ-MOR-GFP and MOR-GFP expres- sions. After 24 h transfection of MOR-LBD constructs, human HT1080 fibrosarcoma cells were stimulated by 4-hydroxy-tamox- ifen to induce nuclear translocation of ectopically expressed proteins. Cells were harvested 4 h after tamoxifen stimulation, and the relative expression levels of IL8, IL6, MMP1 and LCN2 between NFKBIZ-MOR-GFP- and MOR-GFP-transfected cells were measured using quantitative real-time PCR. GAPDH and ACTB were used as endogenous controls for normalization. The bars represent mean relative expression of six biological replicates fromtwo independent experiments. Error bars represent 95% confidence intervals. Asterisks indicate statistical significance by independent samples t-test with *Po0.05. (b) Western blots showing the MOR-GFP and NFKBIZ-MOR-GFP protein pro- ducts. The MOR-LBD and GFP proteins add approximately 35 kDa to the NFKBIZ protein. IB, protein detected by western blot.

decrease in IL8, IL6 and LCN2 gene expressions. Transient transfection of the N-terminal parts of NFKBIZ showed low transfection efficiency and no detectable changes in target gene expression (data not shown). The transfection efficiency was judged by fluorescence microscopy to around 95% for all con- structs shown. The large variation in LCN2 expression Figure 6 Relative mRNA expression in transiently transfected FUS-DDIT3-GFP and HT1080 cells. Cells were transiently depended on differences between the biological repli- transfected with the corresponding constructs and allowed to cates. proliferate for 48 h where after total RNA was isolated. The bars represent mean relative expression of three biological replicates. Error bars show 95% confidence intervals. Adjacent bars with nonoverlapping error bars are significantly different at a P-level of Discussion 0.05. Asterisks indicate means that are significantly different from the GFP control with *Po0.05. The contribution of FUS, EWSR1 and TAF15 to the oncogenic activity of fusion oncogenes is well documen- ted (Kovar et al., 1994; Zinszner et al., 1994, 1997; Perez-Losada et al., 2000a). The three genes may replace DDIT3 downregulates IL8, whereas the expression of each other as fusion partners with specific transcription FUS-DDIT3 results in an upregulation of IL8 in factors, indicating that the N-terminal parts of the three genetically modified HT1080 fibrosarcoma cells (Go¨ r- proteins share important functions. In the present ansson etal., 2005). Here, we show that the C/EBP site investigation, we focused on the FUS-DDIT3 fusion in the composite C/EBP–NF-kB site of the promoter is oncogene and the differential effects of DDIT3 and important for IL8 transcription in control HT1080 cells, FUS-DDIT3 on IL8 transcription. as mutation of the C/EBP site resulted in a strongly

Oncogene FUS-DDIT3 deregulates NF-jB target genes MGo¨ransson et al 275 reduced transcriptional activity (Figure 1). This suggests (Trinh et al., 2008) but we could not detect NF-kB1 in that the C/EBP site binds a factor, probably the natively complex with FUS-DDIT3 and NFKBIZ. expressed C/EBPb, which promotes transcription. On A previous report showed that the N-terminal of the contrary, the NF-kB site is important for negative NFKBIZ was important for nuclear localization (Ya- regulation of IL8 because mutation of this site leads to mazaki et al., 2001). Our results confirmed these an increased transcription. Thus, in wild-type HT1080 findings, as the N-terminally truncated mutant localized cells, the NF-kB site probably binds a protein that to the cytoplasm. However, coexpression with FUS- suppresses IL8 transcription. DDIT3 showed that FUS-DDIT3 can translocate the Introduction of constitutively expressed GFP-tagged NFKBIZ C-terminal to the nucleus. Taken together, the DDIT3 caused a decreased IL8 transcription. According FUS-DDIT3-mediated nuclear transport of NFKBIZ, to the current models of DDIT3 action (Ron and the co-immunoprecipitation and the nuclear co-localiza- Habener, 1992), DDIT3 forms a dimer with C/EBPb tion of FUS-DDIT3 and NFKBIZ indicate that the two and changes the binding properties of this protein to proteins bind directly or in a common complex. DNA or to other promoter-interacting proteins. How- NFKBIZ promotes transcription (Kitamura et al., ever, in contrast to this conventional model of IL8 2000; Motoyama et al., 2005) but may also act as an regulation and DDIT3-C/EBP activity, our experiments inhibitor of p50 containing NF-kB complexes (Yama- showed that a deletion mutated DDIT3 lacking the zaki et al., 2001; Totzke et al., 2006). In this study, we leucine zipper dimerization domain caused downregula- report that the MMP1 and LCN2 genes encoding matrix tion of IL8 transcription in wild-type HT1080 cells. metalloproteinase 1 and lipocalin 2 (a cofactor for Thus, this effect could not be a result of DDIT3-forming MMP9), both involved in extracellular matrix remodel- dimers with C/EBP proteins. Instead, our results ing, are strongly expressed in FUS-DDIT3 carrying cell suggest that part of DDIT3’s inhibitory effect may be lines. MMP1 is reported to be regulated by NF-kB mediated through interactions with a factor binding to through a distal promoter element (Vincenti et al., 1998) the NF-kB site as mutation of the NF-kB site strongly and LCN2 has been reported to be transcriptionally counteracted the inhibitory effects of DDIT3 on IL8 controlled by NFKBIZ (Oikawa et al., 2006). These transcription. findings are supported by our experiments with an Contrary to DDIT3, FUS-DDIT3 caused an inducible NFKBIZ expression vector showing that the increased IL8 transcription and this was maintained mRNA levels of LCN2, but not MMP1, are significantly in the absence of the C/EBP site. The C/EBP site raised by the nuclear expression of NFKBIZ. The is therefore not needed for FUS-DDIT3-induced IL8 augmented MMP1 transcription in FUS-DDIT3- transcription. Instead, FUS-DDIT3 depends on the NF- expressing cells probably results froma different mechan- kB site to stimulate transcription. Data from the ChIP ism. Elevated expressions of MMP1 and LCN2 have experiment showed that FUS-DDIT3 indeed binds to been reported to be associated with advanced cancers, the IL8 promoter. This binding could be indirect metastasis and invasiveness (Nakopoulou et al., 1999; through a complex with an NF-kB site-binding protein. Bratt, 2000; Brinckerhoff et al., 2000), suggesting a role Whether FUS-DDIT3 binds and blocks the hypothetical for these genes also in the development of MLS/RCLS. suppressor protein at this site or recruits a new factor Extensive attempts to knock down FUS-DDIT3 and that replaces the suppressor remains unknown. NFKBIZ by RNA interference were made with methods Systematic studies of interactions between FUS- that worked successfully in other cell systems but these DDIT3 and NF-kB factors failed to detect any failed in FUS-DDIT3-carrying cells. Thus, we could not interactions with NF-kB1 (p50), NF-kB2 (p52), RELA verify the activating effects FUS-DDIT3 or NFKBIZ on (p65), RELB or REL, but revealed an association target gene expression through such experiments. between NFKBIZ and FUS-DDIT3. Taken together Instead, transient overexpression of recombinant with the ChIP and reporter assay results, we conclude DDIT3, deletion mutants of DDIT3 and FUS-DDIT3, that FUS-DDIT3 binds to NFKBIZ and that this -repeat containing C-terminal NFKBIZ and the complex binds the NF-kB site to induce IL8 transcrip- N-terminal of FUS were attempted to test the roles of tion. Furthermore, overexpression of NFKBIZ in FUS-DDIT3 and NFKBIZ in target gene transcription. HT1080 cells resulted in elevated IL8 transcription, These recombinant proteins could potentially act as indicating a role for NFKBIZ. This interpretation dominant negatives, competing out binding between is further supported by a recent report showing FUS-DDIT3 and NFKBIZ. Our results show that the that NFKBIZ binds and acts on a composite or adjacent FUS-DDIT3/NFKBIZ target genes IL8, IL6 and LCN2 C/EBP–NF-kB site almost identical to the site investi- are more resistant to downregulation by the mutated gated in this study (Matsuo et al., 2007). recombinant proteins in FUS-DDIT3-carrying cells and FUS has been reported earlier to interact with the further support the hypothesis that the fusion oncopro- RelA/p65 component of the NF-kB complex (Uranishi tein imposes a strong transcription regulatory activity et al., 2001). However, this interaction is mainly on the target genes (Figure 6). Part of the mechanism mediated through the C-terminal of FUS, which is lost behind this effect may be a tight binding of FUS-DDIT3 in the FUS-DDIT3 fusion protein. This might explain to NFKBIZ and the NF-kB site of target gene why we were unable to detect colocalization between promoters. The results from the N-terminal FUS and FUS-DDIT3 and RelA/p65 in this study. NF-kB1 was C-terminal NFKBIZ were, however, inconclusive with recently reported to interact with NFKBIZ and DNA inhibitory effects only in some conditions. These results

Oncogene FUS-DDIT3 deregulates NF-jB target genes MGo¨ransson et al 276 may depend on the observed low nuclear concentrations tamoxifen was added to the medium at a final concentration of the recombinant proteins, both lacking a nuclear of 100 nM (Littlewood et al., 1995). localization sequence. Involvement of the NF-kB signal transduction and Cell lines transcription regulatory systemhas been reported in The MLS cell line MLS 402-91 (A˚ man et al., 1992; Thelin- many human cancers (Sun and Xiao, 2003). Deregula- Ja¨ rnum et al., 1999) and human fibrosarcoma cell lines HT1080, tion of normal NF-kB functions by the FUS-DDIT3 HT1080 DDIT3-GFP and HT1080 FUS-DDIT3-GFP (En- fusion oncoprotein may affect vital functions of the gstro¨ m et al., 2006) were kept frozen in liquid nitrogen or cultured at 37 1C and 5% CO2 in RPMI 1640 medium with normal cell and contribute to the development of MLS/ HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) RCLS. Growth-regulating genes, such as CCND1, buffer supplemented with 2 mML-glutamine, 50 U/ml penicillin, PTX3, JUNB and CCNE, are aberrantly or strongly 50 mg/ml streptomycin and 8% fetal calf serum (Invitrogen, expressed in FUS-DDIT3-expressing cells (Olofsson Carlsbad, CA, USA). G418 (200 mg/ml, Invitrogen) was et al., 2004; Schwarzbach et al., 2004), and these genes constantly added to cell lines HT1080 FUSA-GFP, HT1080 have been shown to be regulated by NF-kB factors DDIT3-GFP and HT1080 FUS-DDIT3-GFP to ensure stable (Biegel et al., 1993; Diffin et al., 1994; Basile et al., 1997; expression of GFP constructs in the cell population. Hinz et al., 1999). Interaction between FUS-DDIT3 and NFKBIZ may thus lead to transcriptional deregulation Luciferase reporter assay of these and many other NF-kB-controlled target genes. A PCR fragment corresponding to nucleotides À109 to À1of the IL8 promoter was generated using the IL8PromU-XhoI In conclusion, this study shows that the differential 0 0 effect of normal DDIT3 and oncogenic FUS-DDIT3 (5 -ATACTCGAGGGGATGGGCCATCA-3 ) and IL8- PromL-XhoI(50-ATACTCGAGTGCTCCGGTGGCTT-30) on IL8 transcription is linked to interactions with the primer pair, and subsequently cloned into the pGL2 Enhancer NF-kB site of the promoter. Binding of FUS-DDIT3 vector (Promega Corporation, Madison, WI, USA). Mutants to NFKBIZ may be an important part of the FUS-DDIT3 were generated by PCR using IL8PromDCEBP-U-XhoI(50- oncogenic activity and could contribute to the develop- ATACTCGAGCGTGGAATTTCCTC-30) or IL8PromDNF- ment of this tumor type through the deregulation of kB-U-XhoI(50-ATACTCGAGTTGCAAATCGTGGTTTTT additional NF-kB-controlled genes. Similar mechanisms CC-30) together with IL8PromL-XhoI. In the C/EBP mutant, may be involved in other tumor types carrying FUS, the CAGTTGCAAT sequence was deleted fromthe C/EBP- EWSR1 and TAF15-associated fusion oncogenes. NF-kB composite site (50-CAGTTGCAATCGTGGAATT TCC-30), and in the NF-kB mutant, the GGAATTCC element was changed to GGTTTTTCC. All constructs were confirmed by sequencing. The pCI b-gal expression vector was a kind gift fromDr Y Arvidsson. Cells were co-transfected using 1 mg Materials and methods pGL2 enhancer construct and 0.4 mg pCI b-gal vector in 12- GFP and DsRed1 vector constructs well plates and harvested in 100 ml reporter lysis buffer The full-length coding regions of DDIT3, FUS-DDIT3 cDNA (Promega) 48 h after transfection. Luciferase activity was type II (Panagopoulos et al., 1994) and NFKBIZ transcript detected using Promega’s Luciferase Assay System and variant 1 were cloned into the pEGFP-N1 and/or pDsRed1- normalized against b-galactosidase activity in each sample. N1 vectors (Clontech Laboratories, Mountain View, CA, Each experiment was conducted twice in biological triplicates. USA) in frame with the GFP or DsRed1 sequence. Similarly, partial cDNAs encoding the C-terminal (amino acids 357–718) Colocalization studies part of NFKBIZ, the N-terminal part of FUS involved in the Colocalization studies were performed (as described earlier FUS-DDIT3 fusion (FUSA) and DDIT3-DLZ/FUS-DDIT3- Go¨ ransson et al., 2002) in flaskettes (Nalgene Nunc Thermo DLZ lacking the 38 C-terminal amino acids containing a Fisher Scientific, Rochester, NY, USA) using 1 mg of each leucine zipper domain were cloned into the pEGFP-N1 and/or plasmid construct for transfection. HT1080 cells were washed pDsRed1-N1 vectors. MOR-GFP and NFKBIZ-MOR-GFP twice in phosphate-buffered saline (PBS) of 0.01 M, pH 7.2 vectors were constructed by an in-frame ligation of the MOR- (PBS) and fixed in 4% formaldehyde in PBS 24 h after LBD construct immediately upstream of the gene encoding transfection. For immunofluorescence studies, the following GFP. All constructs were confirmed by sequencing. antisera were used: NF-kB1 (06-886; Millipore, Billerica, MA, The mouse estrogen receptor ligand-binding domain (MOR- USA), NF-kB2 (C-5), Rel A (F-6), RelB (C-19), c-Rel (N-466) LBD) construct was made by mutating the wild-type mouse and bcl3 (C-14) (Santa Cruz Biotechnology, Santa Cruz, CA, estrogen receptor (a kind gift fromDr M Parker). The ligand- USA). Cells were mounted in Prolong Gold with DAPI (40,6- binding domain (DNA encoding amino acids 290–599) of the diamidino-2-phenylindole; Invitrogen) and let to cure over- receptor was cloned using the primer set: MORLBD BamHI-U night. The intracellular localization of fluorescently tagged (50-TATGGATCCAGGAGACATGAGGGCTGCCAACCT proteins was recorded using a Zeiss LSM 510 META system TTG-30) and MORLBD BamHI-L (50-TATGGATCCATCGT with LSM-5 software supplied by the manufacturer. Settings GTTGGGGAAGCCCTCT-30). The G525R point mutation for confocal microscopy were as follows: DAPI, 405 nm was introduced by PCR mutagenesis and amplification of excitation and emission BP 420–480 filter; GFP, 488 nm circular DNA in vitro (Chen and Ruffner, 1998) using the excitation and emission BP 505–530 filter; and DsRed1, primer set (50–30): MORLBD mut-U (GGCACATGAGTAA 561 nm excitation and emission META detector 572–625 nm. CAAACGCATGG) and MORLBD mut-L (ATGTTGTAGA GATGCTCCATGCGTTTGTT). The MOR-LBD G525R Immunoprecipitation and western blot mutant is unable to bind estrogen, yet it retains affinity for a ChIP studies were performed as described elsewhere (Caretti synthetic ligand, 4-hydroxy-tamoxifen. For nuclear transloca- et al., 2003) with antibodies FUS (H76), and DDIT3 tion of mutant MOR-LBD-fused constructs, 4-hydroxy- (GADD153 R20) fromSanta Cruz Biotechnology. Satellite

Oncogene FUS-DDIT3 deregulates NF-jB target genes MGo¨ransson et al 277 sequences repeated at multiple positions in the genome were Biosystems, Foster City, CA, USA) with SYBR Green used as a control for unspecific precipitation and as a loading detection (Qiagen). Formation of expected PCR products control. The PCR primers used for detection of the IL8 were confirmed by agarose gel electrophoresis and melt curve proximal promoter sequences were (50–30): forward: analysis. Gene expression data were normalized against AGGTTTGCCCTGAGGGGATG and reverse: GCTTGTG ACTB and GAPDH by geometric averaging (Vandesompele TGCTCTGCTGTCTC. Control satellite sequences were et al., 2002). The expressions of ACTB, GAPDH, IL8, amplified using forward CAATTATCCCTTCGGGGAATCGG IL6, MMP1 and LCN2 were monitored using the following and reverse GGCGACCAATAGCCAAAAAAGTGAG primers. primer pairs (50–30): ACTBU (TCATGAAGTGTGTGACGTT Immunoprecipitation experiments for FUS-DDIT3 and GACATCCGT), ACTBL (CCTAGAAGCAATTTGCGGT NFKBIZ were carried out on human fibrosarcoma HT1080 GCACGATG); GAPDHU (GTGAAGGTCGGAGTCAAC cells transiently co-transfected with NFKBIZ-GFP and FUS- G), GAPDHL (GGTGAAGACGCCAGTGGACTC); IL8U DDIT3-DsRed1. The cells were harvested 24 h after transfec- (TACTCCAAACCTTTCCACCCC), IL8L (CCTTGGCCTC tion and lysed in lysis buffer (50 mM Tris-HCl pH 7.4, 1% NP- AATTTTGCTAT; IL8_133F (CAGCCTTCCTGATTTCT 40, 50 mM NaCl, 0.1 mM EDTA, 10% protease inhibitor GC), IL8_375R (ACTTCTCCACAACCCTCTGC); IL6U cocktail (no. 1836153; Roche, Basel, Switzerland). The lysates (CAGCCCTGAGAAAGGAGACAT), IL6L (AATCTGA were incubated overnight with 4 mg anti-GFP peptide antibody GGTGCCCATGCTAC); IL6_187F (CACACAGACAGCCA (no. 8367-2; Clontech Laboratories) and precipitated using CTCACC), IL6_364R (CATCCATCTTTTTCAGCCATC); 50 ml protein A agarose beads (Millipore). The precipitates MMP1U (TACAACTTACATCGTGTTGCG), MMP1L (G were washed three times in lysis buffer and eluated in LDS GTTTTTCCAGTGTTTTCCTC); LCN2U (GGGAAGTG western sample buffer (Invitrogen) containing 10% protease GTATGTGGTAGGC), LCN2L (GATTGGGACAGGGA inhibitor cocktail. Input and immunoprecipitated material AGACGAT). ACTB and GAPDH expressions were also were separated on 4–12% SDS–polyacrylamide gels (Invitro- measured using primers contained in the Human Endogenous gen), blotted onto Immobilon-P PDVF (polyvinylidene Control Gene Panel supplied by TATAA Biocenter, Gothen- fluoride) membranes (Millipore) and detected using antibodies burg, Sweden. All experiments were performed in at least three directed against GFP (8367-2; Clontech Laboratories) or independent biological replicates. DDIT3 (GADD153 R20; Santa Cruz Biotechnology), followed by alkaline phosphatase-conjugated secondary anti- Statistical methods bodies (Dako, Glostrup, Denmark) and CDP-Star (Tropix, t Bedford, MA, USA) as substrate. The chemiluminescent signal Independent samples -test (using the SPSS software (SPSS, was detected using electrochemiluminescence film (Kodak, Chicago, IL, USA)) was used to compare luciferase activity IL8 Rochester, NY, USA). Similarly, the GFP antibody was used between different promoters in luciferase reporter assays, IL8 IL6 MMP1 LCN2 to detect the MOR-GFP construct, and rabbit antisera raised as well as differences in , , and P against the N-terminal part of NFKBIZ isoform A was used to expressions were quantified by real-time PCR. -values of detect the NFKBIZ-MOR-GFP construct in transiently less than 0.05 were considered significant. All statistical tests transfected HT1080 cells. were two-tailed.

Real-time PCR analysis Acknowledgements Cells were washed twice in cold PBS, and total RNA was isolated using the QIAshredder and RNeasy Mini Kit (Qiagen, We thank Ulric Pedersen for image processing. This work was Hilden, Germany). cDNA was generated from 5 mg total RNA supported by grants fromthe Inga-Britt and Arne Lundberg using oligo dT primers and Superscript III reverse transcrip- Research Foundation, the Swedish Cancer Society, Assar tase (Invitrogen) or using a QuantiTect Reverse transcription Gabrielssons Research Foundation and the Johan Jansson kit (Qiagen). Real-time PCR was measured in the LightCycler Foundation for Cancer Research. RM was supported by an (Roche) as described earlier (Sta˚ hlberg et al., 2004) or AIRC grant. AS is supported by a postdoctoral fellowship alternatively using a 7500 Fast real-time PCR system (Applied award fromthe Swedish Research Council.

References

A˚ man P, Panagopoulos I, Lassen C, Fioretos T, Mencinger M, Caretti G, Salsi V, Vecchi C, Imbriano C, Mantovani R. (2003). Toresson H et al. (1996). Expression patterns of the human Dynamic recruitment of NF-Y and histone acetyltransferases on sarcoma-associated genes FUS and EWS and the genomic structure cell-cycle promoters. J Biol Chem 278: 30435–30440. of FUS. Genomics 37: 1–8. Chen Z, Ruffner DE. (1998). Amplification of closed circular DNA A˚ man P, Ron D, Mandahl N, Fioretos T, Heim S, Arheden K et al. in vitro. Nucleic Acids Res 26: 1126–1127. (1992). Rearrangement of the transcription factor gene CHOP in Crozat A, A˚ man P, Mandahl N, Ron D. (1993). Fusion of CHOP to a myxoid liposarcomas with t(12;16)(q13;p11). Genes novel RNA-binding protein in human myxoid liposarcoma. Nature Cancer 5: 278–285. 363: 640–644. Basile A, Sica A, d’Aniello E, Breviario F, Garrido G, Castellano M Diffin F, Porter H, Mott MG, Berry PJ, Brown KW. (1994). Rapid et al. (1997). Characterization of the promoter for the human long and specific diagnosis of t(11;22) translocation in paediatric Ewing’s pentraxin PTX3. Role of NF-kappaB in tumor necrosis factor-alpha sarcoma and primitive neuroectodermal tumours using RNA-PCR. and interleukin-1beta regulation. J Biol Chem 272: 8172–8178. J Clin Pathol 47: 562–564. Biegel JA, Conard K, Brooks JJ. (1993). Translocation Engstro¨ mK, Willen H, Ka˚ bjo¨ rn-Gustafsson C, Andersson C, Olsson (11;22)(p13;q12): primary change in intra-abdominal desmoplastic M, Go¨ ransson M et al. (2006). The myxoid/round cell liposarcoma small round cell tumor. Genes Chromosomes Cancer 7: 119–121. fusion oncogene FUS-DDIT3 and the normal DDIT3 induce a Bratt T. (2000). Lipocalins and cancer. Biochim Biophys Acta 1482: 318–326. liposarcoma phenotype in transfected human fibrosarcoma cells. Am Brinckerhoff CE, Rutter JL, Benbow U. (2000). Interstitial collage- J Pathol 168: 1642–1653. nases as markers of tumor progression. Clin Cancer Res 6: Go¨ ransson M, Elias E, Sta˚ hlberg A, Olofsson A, Andersson C, A˚ man 4823–4830. P. (2005). Myxoid liposarcoma FUS-DDIT3 fusion oncogene

Oncogene FUS-DDIT3 deregulates NF-jB target genes MGo¨ransson et al 278 induces C/EBP beta-mediated interleukin 6 expression. Int J Cancer CHOP fusion protein specifically induces liposarcomas in transgenic 115: 556–560. mice. Oncogene 19: 2413–2422. Go¨ ransson M, Wedin M, A˚ man P. (2002). Temperature-dependent Perez-Losada J, Sanchez-Martin M, Rodriguez-Garcia MA, Perez- localization of TLS-CHOP to splicing factor compartments. Exp Mancera PA, Pintado B, Flores T et al. (2000b). Liposarcoma Cell Res 278: 125–132. initiated by FUS/TLS-CHOP: the FUS/TLS domain plays a critical Hinz M, Krappmann D, Eichten A, Heder A, Scheidereit C, Strauss role in the pathogenesis of liposarcoma. Oncogene 19: 6015–6022. M. (1999). NF-kappaB function in growth control: regulation of Rabbitts TH, Forster A, Larson R, Nathan P. (1993). Fusion of the cyclin D1 expression and G0/G1-to-S-phase transition. Mol Cell dominant negative transcription regulator CHOP with a novel gene Biol 19: 2690–2698. FUS by translocation t(12;16) in malignant liposarcoma. Nat Genet Kitamura H, Kanehira K, Okita K, Morimatsu M, Saito M. (2000). 4: 175–180. MAIL, a novel nuclear I kappa B protein that potentiates LPS- Riggi N, Cironi L, Provero P, Suva ML, Stehle JC, Baumer K et al. induced IL-6 production. FEBS Lett 485: 53–56. (2006). Expression of the FUS-CHOP fusion protein in primary Kovar H. (2005). Context matters: the hen or egg problem in Ewing’s mesenchymal progenitor cells gives rise to a model of myxoid sarcoma. Semin Cancer Biol 15: 189–196. liposarcoma. Cancer Res 66: 7016–7023. Kovar H, Zoubek A, Pfleiderer C, Jug G, Auinger A, Aryee D et al. Ron D, Habener JF. (1992). CHOP, a novel developmentally regulated (1994). The EWS gene rearrangement in Ewing tumors: key to the nuclear protein that dimerizes with transcription factors C/EBP and disease. Klin Padiatr 206: 196–200. LAP and functions as a dominant-negative inhibitor of gene Law WJ, Cann KL, Hicks GG. (2006). TLS, EWS and TAF15: a transcription. Genes Dev 6: 439–453. model for transcriptional integration of gene expression. Brief Funct Sanchez Garcia I, Rabbitts TH. (1994). Transcriptional activation by Genomic Proteomic 5: 8–14. TAL1 and FUS-CHOP proteins expressed in acute malignancies as Libermann TA, Baltimore D. (1990). Activation of interleukin-6 gene a result of chromosomal abnormalities. Proc Natl Acad Sci USA 91: expression through the NF-kappa B transcription factor. Mol Cell 7869–7873. Biol 10: 2327–2334. Schwarzbach MH, Koesters R, Germann A, Mechtersheimer G, Littlewood TD, Hancock DC, Danielian PS, Parker MG, Evan GI. Geisbill J, Winkler S et al. (2004). Comparable transforming (1995). A modified oestrogen receptor ligand-binding domain as an capacities and differential gene expression patterns of variant improved switch for the regulation of heterologous proteins. Nucleic FUS/CHOP fusion transcripts derived fromsoft tissue liposarco- Acids Res 23: 1686–1690. mas. Oncogene 23: 6798–6805. Matsuo S, Yamazaki S, Takeshige K, Muta T. (2007). Crucial roles of Sta˚ hlberg A, Ha˚ kansson J, Xian X, Semb H, Kubista M. (2004). binding sites for NF-kappaB and C/EBPs in IkappaB-zeta-mediated Properties of the reverse transcription reaction in mRNA quantifi- transcriptional activation. Biochem J 405: 605–615. cation. Clin Chem 50: 509–515. Matsusaka T, Fujikawa K, Nishio Y, Mukaida N, Matsushima K, Sun SC, Xiao G. (2003). Deregulation of NF-kappaB and its upstream Kishimoto T et al. (1993). Transcription factors NF-IL6 and kinases in cancer. Cancer Metastasis Rev 22: 405–422. NF-kappa B synergistically activate transcription of the inflammatory Thelin-Ja¨ rnumS, Go¨ ransson M, Burguete AS, Olofsson A, A˚ man P. , interleukin 6 and interleukin 8. Proc Natl Acad Sci USA (2002). The myxoid liposarcoma specific TLS-CHOP fusion protein 90: 10193–10197. localizes to nuclear structures distinct fromPML nuclear bodies. Int Morohoshi F, Ootsuka Y, Arai K, Ichikawa H, Mitani S, Munakata N J Cancer 97: 446–450. et al. (1998). Genomic structure of the human RBP56/hTAFII68 Thelin-Ja¨ rnumS, Lassen C, Panagopoulos I, Mandahl N, A˚ man P. and FUS/TLS genes. Gene 221: 191–198. (1999). Identification of genes differentially expressed in TLS-CHOP Motoyama M, Yamazaki S, Eto-Kimura A, Takeshige K, Muta T. carrying myxoid liposarcomas. Int J Cancer 83: 30–33. (2005). Positive and negative regulation of nuclear factor-kappaB- Totzke G, Essmann F, Pohlmann S, Lindenblatt C, Janicke RU, mediated transcription by IkappaB-zeta, an inducible nuclear Schulze-Osthoff K. (2006). A novel member of the IkappaB family, protein. J Biol Chem 280: 7444–7451. human IkappaB-zeta, inhibits transactivation of p65 and its DNA Mukaida N, Mahe Y, Matsushima K. (1990). Cooperative interaction binding. J Biol Chem 281: 12645–12654. of nuclear factor-kappa B- and cis-regulatory enhancer binding Trinh DV, Zhu N, Farhang G, KimBJ, Huxford T. (2008). The protein-like factor binding elements in activating the interleukin-8 nuclear I kappaB protein I kappaB zeta specifically binds NF- gene by pro-inflammatory cytokines. J Biol Chem 265: 21128–21133. kappaB p50 homodimers and forms a ternary complex on kappaB Nakopoulou L, Giannopoulou I, Gakiopoulou H, Liapis H, Tzonou DNA. J Mol Biol 379: 122–135. A, Davaris PS. (1999). Matrix metalloproteinase-1 and -3 in breast Uranishi H, Tetsuka T, Yamashita M, Asamitsu K, Shimizu M, Itoh M cancer: correlation with progesterone receptors and other clinico- et al. (2001). Involvement of the pro-oncoprotein TLS (translocated pathologic features. Hum Pathol 30: 436–442. in liposarcoma) in nuclear factor-kappa B p65-mediated transcription Ohno T, Rao VN, Reddy ES. (1993). EWS/Fli-1 chimeric protein is a as a coactivator. J Biol Chem 276: 13395–13401. transcriptional activator. Cancer Res 53: 5859–5863. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Oikawa K, Ishida T, Imamura T, Yoshida K, Takanashi M, Hattori H Paepe A et al. (2002). Accurate normalization of real-time et al. (2006). Generation of the novel monoclonal antibody against quantitative RT-PCR data by geometric averaging of multiple TLS/EWS-CHOP chimeric oncoproteins that is applicable to one of internal control genes. Genome Biol 3: RESEARCH0034. the most sensitive assays for myxoid and round cell liposarcomas. Vincenti MP, Coon CI, Brinckerhoff CE. (1998). Nuclear factor Am J Surg Pathol 30: 351–356. kappaB/p50 activates an element in the distal matrix metalloprotei- Olofsson A, Willen H, Go¨ ransson M, Engstro¨ mK, Meis-Kindblom nase 1 promoter in interleukin-1beta-stimulated synovial fibroblasts. JM, Stenman G et al. (2004). Abnormal expression of cell cycle Arthritis Rheum 41: 1987–1994. regulators in FUS-CHOP carrying liposarcomas. Int J Oncol 25: Yamazaki S, Muta T, Takeshige K. (2001). A novel IkappaB protein, 1349–1355. IkappaB-zeta, induced by proinflammatory stimuli, negatively Panagopoulos I, Ho¨ glund M, Mertens F, Mandahl N, Mitelman F, regulates nuclear factor-kappaB in the nuclei. J Biol Chem 276: A˚ man P. (1996). Fusion of the EWS and CHOP genes in myxoid 27657–27662. liposarcoma. Oncogene 12: 489–494. Zinszner H, Albalat R, Ron D. (1994). A novel effector domain from Panagopoulos I, Mandahl N, Ron D, Ho¨ glund M, Nilbert M, Mertens the RNA-binding protein TLS or EWS is required for oncogenic F et al. (1994). Characterization of the CHOP breakpoints and transformation by CHOP. Genes Dev 8: 2513–2526. fusion transcripts in myxoid liposarcomas with the 12;16 transloca- Zinszner H, Immanuel D, Yin Y, Liang FX, Ron D. (1997). A tion. Cancer Res 54: 6500–6503. topogenic role for the oncogenic N-terminus of TLS: nucleolar Perez-Losada J, Pintado B, Gutierrez-Adan A, Flores T, Banares- localization when transcription is inhibited. Oncogene 14: Gonzalez B, del Campo JC et al. (2000a). The chimeric FUS/TLS- 451–461.

Oncogene