The Splicing Factor U1C Represses EWS/FLI-mediated Transactivation*

EWS is an RNA-binding protein involved in human tumor-specific chromosomal translocations. In approximately 85% of Ewing's sarcomas, such translocations give rise to the chimeric geneEWS/FLI. In the resulting fusion protein, the RNA binding domains from the C terminus of EWS are replaced by the DNA-binding domain of the ETS protein FLI-1. EWS/FLI can function as a transcription factor with the same DNA binding specificity as FLI-1. EWS and EWS/FLI can associate with the RNA polymerase II holoenzyme as well as with SF1, an essential splicing factor. Here we report that U1C, one of three human U1 small nuclear ribonucleoprotein-specific proteins, interacts in vitro and in vivo with both EWS and EWS/FLI. U1C interacts with other splicing factors and is important in the early stages of spliceosome formation. Importantly, co-expression of U1C represses EWS/FLI-mediated transactivation, demonstrating that this interaction can have functional ramifications. Our findings demonstrate that U1C, a well characterized splicing protein, can also function in transcriptional regulation. Furthermore, they suggest that EWS and EWS/FLI may function both in transcriptional and post-transcriptional processes.

Chromosomal abnormalities such as deletions, inversions, or translocations are common genetic mechanisms to induce mutations that contribute to tumorigenesis. One important consequence of chromosomal translocations is the creation of novel in-frame fusion genes that often involve transcription factors (1). In virtually all cases of Ewing's sarcoma, a chromosomal translocation creates a fusion gene between EWS and a member of the ETS family of transcription factors, most commonly FLI-1 (2)(3)(4)(5)(6). The resulting chimeric protein retains the N terminus of EWS and replaces the RNA binding domains in its C terminus with the DNA binding domain of FLI-1. Similar fusion proteins between EWS or the related proteins TLS or TAF II 68 with other transcription factor DNA binding domains have been observed in many other tumor types (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17).
EWS, TLS, and TAF II 68 comprise a unique family of RNAbinding proteins termed the TET family (TLS/FUS, EWS, TAF II 68), which may play a role in multiple nuclear processes.
The RNP motifs in these proteins differ structurally from those in other RNA-binding proteins in that they contain an unusually long predicted loop structure following the first ␣-helix of the domain. This structural feature is limited to TET family members and may allow for unique interactions with RNA (18). The N-terminal domains of EWS, TLS, and TAF II 68 involved in tumor-derived fusion proteins are all rich in glutamine, serine, and tyrosine and are capable of transcriptional activation (19 -23). Both EWS and TAF II 68 have been identified in specific populations of the TFIID complex and associate with the human RNA polymerase II holoenzyme (18,24). EWS and TLS have also been shown to interact with splicing factors (25)(26)(27). In addition, TLS is capable of promoting homologous DNA pairing and D-loop formation, essential steps in double-strand DNA break repair through recombination (28,29). Perhaps due to loss of this repair activity, TLS knock-out mice show evidence of genetic instability (30,31), suggesting that TLS may play an important role in genomic integrity and maintenance. It is unclear which of the functions associated with the TET family contribute to tumorigenesis in the context of the fusion proteins.
We used a yeast two-hybrid screen to identify proteins that interact with the N-terminal tumor-associated portion of EWS. We identified U1C, one of three U1 small nuclear ribonucleoprotein (snRNP) 1 -specific proteins. snRNPs play an essential role in splicing through a large complex known as the spliceosome. All snRNPs are made up of snRNA and associated proteins. They share at least eight common Sm proteins and have specific protein associations as well (32). The U1 snRNP is composed of an RNA backbone, common Sm proteins, and three U1-specific proteins: U1A, U1C, and U1-70K (33). The U1snRNP binds to the 5Ј splice site on pre-mRNA to form a stable complex identified as the early or E complex in mammalian splicing extracts (34). U1A and U1-70K contain RNAbinding domains and interact with naked snRNA on their own (35,36). However, the binding of U1C to the U1snRNP particle is dependent on protein-protein interactions between U1C and U1-70K as well as U1C and the common Sm proteins (37).
We show that U1C interacts in vitro and in vivo with EWS and with higher affinity with EWS/FLI. This interaction can cause repression of EWS/FLI-mediated transactivation and may therefore affect EWS/FLI target gene regulation.

EXPERIMENTAL PROCEDURES
Plasmids-To generate t-EWS/LexA as bait for the yeast two-hybrid screen, a fragment corresponding to the tumor-associated portion of EWS (t-EWS, amino acids 1-264) was PCR-amplified from HL-60 cDNA and cloned into a modification of the yeast expression vector pSD.08 (38)  (amino acids 2-202). Expression of the fusion protein was under control of the GAL1-10/CYC1 promoter. The plasmid also contained a TRP1 ϩ marker. Absence of PCR mutations was verified by sequence analysis.
To create a U1C clone for in vitro transcription and translation, the U1C coding sequence was amplified by PCR from the clone isolated in the yeast two-hybrid screen and cloned into pBluescript KS containing a FLAG epitope tag between the BamHI and EcoRI sites as an EcoRI/ HindIII fragment. The absence of PCR mutations was confirmed by sequence analysis.
To generate epitope-tagged constructs, the mammalian expression vector pCB6ϩ (a gift from Dr. Frank Rauscher) was modified by the addition of two tandem FLAG or HA epitope tags cloned between the BglII and EcoRI sites. EWS and EWS/FLI were amplified by PCR and cloned in frame with the tags at the EcoRI site. EWS/FLI deletion clones EF-B, EF-C, and EF-H were made by ligating the EWS portion from pGEX EWS-(1-120), pGEX EWS-(1-133), and pGEX EWS-(210 -264), respectively, in frame with the FLAG tag at the 5Ј-end and the region of FLI from EWS/FLI at the 3Ј-end.
To create clones for the mammalian two-hybrid system, the vectors pCMVbd and pCMVad (Stratagene) were modified by the addition of FLAG or HA epitope tags cloned between the BamHI and EcoRI sites to create pbdFLAG, padFLAG, and padHA. Sequences encoding EWS, EWS/FLI, EWS/FLI-B, EWS/FLI-C, and EWS/FLI-H were excised from pCFLAG and cloned into padFLAG or padHA as EcoRI/HindIII fragments. U1C was excised from pBluescript and cloned into pbdFLAG as an EcoRI/HindIII fragment. The reporter gene plasmid, pFrLuc (Stratagene), contains five tandem Gal4 DNA binding sites upstream of a luciferase reporter.
To generate tkD2A-luc, an EWS/FLI-responsive luciferase reporter construct, the chloramphenicol acetyltransferase gene and its poly(A) sequences were excised from the tkD2A vector (20), a kind gift from Dr. Jacques Ghysdael, as an MluI/EcoRI fragment. The EcoRI site was blunted, and the luciferase gene and its poly(A) sequences from the pGL2 promoter vector (Promega) were inserted as an MluI/blunted BamHI fragment.
Yeast Two-hybrid Screening-A cDNA library was generated using random-primed cDNA from the Ewing's sarcoma cell line RD-ES (ATCC) that contains an EWS/FLI fusion gene. cDNA was ligated to BstXI adapters and cloned into plasmid pSD10A (a gift from Dr. Steven Dalton) downstream of sequences encoding the activation domain of VP16 (38). The library plasmid pSD10a also contains URA3 and ␤-lactamase genes for selection in Saccharomyces cereviseae and Escherichia coli, respectively. The yeast reporter strain S330 (39) was transformed with the plasmid pt-EWS/LexA to create the strain S330-EL. To identify proteins that interact with t-EWS, S330EL was then transformed by electroporation as described (40). Colonies were plated onto nylon filters overlaid on medium that contained 2% glucose and was deficient in tryptophan and uracil (trp Ϫ ura Ϫ ). Thirty-six hours later, filters were transferred to trp Ϫ ura Ϫ medium containing 2% galactose, and colonies were grown on galactose for 12 h. ␤-Galactosidase activity was assayed as described (38). To control for the possible background contributed by t-EWS/LexA alone, S330-EL and S330 transformed with VP16/LexA were subjected to X-gal staining in parallel with the library transformants. All positive colonies were picked within 25 min following the addition of X-gal, before background staining in the S330-EL cells appeared. One positive clone was identified as U1C and contained the full-length coding sequence as well as 36 base pairs of 5Ј-untranslated sequence and 306 base pairs of 3Ј-untranslated sequence. The U1C/ VP16 plasmid isolated in the library screen was reintroduced into S330 yeast, alone or in combination with 14 different LexA fusion proteins (gifts from Dr. Mark Osborne), to ensure that the intrinsic activation activity of the t-EWS bait did not cause sporadic false positives. U1C/ VP16 did not activate transcription of the reporter with any of these negative controls.
Expression of EWS/FLI GST fusion protein was induced under the same conditions for 5 h. The soluble fraction of these proteins was coupled to glutathione-Sepharose beads according to the manufacturer's instructions (Amersham Pharmacia Biotech). Equivalent quantities of GST fusion proteins used for pull-down assays were verified by SDS-polyacrylamide gel electrophoresis and Coomassie staining. In vitro transcribed and translated U1C protein was synthesized using the TNT-coupled reticulocyte lysates (Promega) and [ 35 S]methionine. Labeled proteins were incubated for 1 h in 300 l of GST-binding buffer (150 mM NaCl, 1% IGEPAL, 50 mM Tris, pH 8.0, 5 mM MgCl 2 , 0.5 mM DTT, and 10% glycerol) with 2 g of GST fusion proteins bound to glutathione-Sepharose beads. Bound proteins were washed three times with 500 l of GST binding buffer, and were analyzed by electrophoresis on a 4 -20% gradient gel followed by autoradiography.
Mammalian Two-hybrid Assays-293T cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. 3 ϫ 10 5 cells were seeded on six-well plates and were transfected 16 -18 h later. All transfections contained 500 ng of pFrLuc reporter (Stratagene), 100 l of CMV␤gal, and 1.7 g of pCMVad and pCMVbd plasmids to achieve equivalent protein expression. DNA was mixed with 6. In Vivo Transactivation Assays-HeLa cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Approximately 4 ϫ 10 5 cells were seeded on 60-mm plates and were transfected 16 -18 h later. Cells were transfected according to the manufacturer's recommendations using 5.6 l of Superfect (Qiagen) and 2.8 g of total DNA (800 ng of ptkdluc reporter, 400 ng of CMVBgal, and a total of 1.6 g from combinations of pCFLAG EWS/FLI construct, pCFLAG U1C, or empty vector). Transfections were incubated 3 h at 37°C. Medium was then aspirated from plates, and 6 ml of Dulbecco's modified Eagle's medium were added. Cells were washed with phosphate-buffered saline and lysed in 150 l of reporter lysis buffer (Promega) 48 h later. Luciferase assays were performed with 20 and 80 l of cell lysis supernatant, and values were normalized with ␤-galactosidase activity as described above. The average results of at least three experiments are shown.
Western Blotting-Cell extracts were sonicated briefly, resolved by SDS-polyacrylamide gel electrophoresis, and transferred to nitrocellulose. Epitope-tagged proteins were detected with anti-FLAG M2 monoclonal antibody (Babco) at a concentration of 10 g/ml or anti-HA mouse monoclonal antibody (clone 12CA5) (Roche Molecular Biochemicals) at a concentration of 7 g/ml and visualized with peroxidase-conjugated anti-mouse antibody at a dilution of 1:5000, followed by enhanced chemiluminescence (Amersham Pharmacia Biotech).

U1C Interacts with the N Terminus of EWS-
We performed a yeast two-hybrid screen to identify proteins that interact with the amino-terminal region of EWS found in several tumorspecific chromosomal translocations. A bait expression vector was constructed by fusing amino acids 1-264 of EWS (t-EWS) in frame with sequences encoding the dimerization and DNA binding domain of the bacterial repressor protein LexA. A cDNA library was generated from a Ewing's sarcoma cell line containing an EWS/FLI fusion to increase the likelihood that essential interacting partners contributing to the transforming ability of the fusion protein would be expressed. The library was cloned into a vector containing the transcriptional activation domain from the herpes simplex virus protein VP-16.
Because the region of EWS contained in the bait possesses a strong transcription activation domain, two steps were taken to minimize background activity contributed from the bait alone. First, a yeast reporter strain was chosen that contains six LexA binding sites upstream of the lacZ gene integrated into the ura3 locus (39). This decreased background signals because only a single copy of the reporter is integrated. Second, the level of t-EWS/LexA expression was regulated by a galactose-inducible promoter. If cells were grown on galactose for 24 h or more, little difference could be seen between t-EWS/LexA and VP-16/ LexA activity as detected by X-gal staining. However, following 8 -10 h of galactose induction, a VP-16/LexA fusion protein produced blue staining within 12-15 min, while the t-EWS/ LexA fusion required 3 h to produce the same degree of staining. Protein-protein interactions were detected within 15-25 min after the addition of X-gal.
One clone identified in the two-hybrid screen contained the full-length sequence encoding U1C, one of three human U1 snRNP-specific proteins. This clone was further tested to exclude the possibility that it represented a false positive or nonspecific interaction. The U1C/VP16 clone failed to activate the yeast reporter in the absence of t-EWS/LexA and also failed to interact with any of a panel of LexA fusion proteins used to screen for nonspecific interactions. Therefore, the interaction between U1C and t-EWS was specific in yeast.

U1C Interacts with EWS and EWS/FLI in Vitro through the N-terminal Domain of EWS-To confirm the interaction of U1C
with EWS outside of the yeast system, we tested the in vitro association between these proteins by GST pull-down assays. We fused t-EWS (amino acids 1-264), the C terminus of EWS (amino acids 245-656), EWS/FLI, or wild-type EWS with GST (Fig. 1A) and expressed the resulting chimeric proteins in E. coli BL21. Full-length U1C protein was transcribed and translated in vitro and incubated with each of the chimeric proteins and GST alone. U1C interacted with t-EWS alone or in the context of the full-length EWS protein or EWS/FLI. This interaction was specific, since U1C did not bind to GST alone or to GST fused to the C terminus of EWS containing the RNA binding domains (Fig. 1B).
To further delineate the amino acids in the N terminus of EWS responsible for the interaction with U1C, eight GST fusion proteins comprising deletions of t-EWS were made ( Fig.  2A). GST pull-down assays were performed on this series of EWS deletions with in vitro translated U1C. U1C was specifically retained on beads coupled with GST-EWS-(1-120), GST-EWS- (1-133), GST-EWS-(1-209), and GST-t-EWS (Fig. 2B). Therefore, the minimal region required for interaction between EWS and U1C comprised amino acids 1-120 in EWS.

U1C Interacts with EWS and EWS/FLI in Vivo-To demonstrate an intracellular association between EWS or EWS/FLI
and U1C, a mammalian two-hybrid system was employed. This system is similar to a yeast two-hybrid system in that it allows detection of protein-protein interactions by activation of reporter gene expression if a functional transcription factor is reconstituted as a result of interaction between two hybrid proteins. We fused full-length EWS or EWS/FLI to the strong transcriptional activation domain of NF-B (ad-EWS and ad-EWS/FLI), and U1C was fused to the DNA binding domain of GAL4 (bd-U1C). We transfected ad-EWS or ad-EWS/FLI alone or with bd-U1C to assay the ability of the proteins to interact and activate transcription of a GAL4-responsive luciferase reporter gene. Equivalent protein expression of EWS, EWS/FLI, and U1C was verified by Western blotting (data not shown). When EWS was expressed with U1C, a 4-fold increase in luciferase activity was seen (Fig. 3A). Interestingly, a 14-fold increase was seen when EWS/FLI was expressed with U1C ( Fig.  3A), suggesting that the tumor-associated fusion protein binds U1C with a higher affinity than wild-type EWS. To ensure that this difference was not due to a higher intrinsic activation activity of EWS/FLI as compared with EWS, both proteins were expressed as GAL4 DNA-binding fusion proteins, and luciferase activity was measured. No difference was seen (data not shown), indicating that the higher luciferase activity seen by EWS/FLI with U1C was specific to the interaction. The difference in the strengths of interactions of U1C with EWS and EWS/FLI may not have been reflected in the GST pull-down assay because factors critical to the stronger EWS/FLI interaction may have been limited or absent in the in vitro assay but were available in the environment of the mammalian cells. In addition, although the GST protein was soluble, it is possible that a greater proportion of protein is folded in the correct conformation when synthesized in mammalian cells compared with E. coli. Alternatively, the mammalian two-hybrid assay may simply be a more sensitive assay.
To confirm the specificity of these interactions in mammalian cells, EWS deletion constructs were made. EWS deletions B and C ( Fig. 2A) contained the region of EWS necessary for interaction as defined in the GST pull-down assay, and EWS deletion H ( Fig. 2A) lacked the complete interaction domain. These regions of EWS were cloned as EWS/FLI fusions in the NF-B activation domain vector and expressed in the mammalian two-hybrid system with U1C as described above. The first 120 or the first 133 amino acids of EWS fused to FLI (EF-B and EF-C) interacted with U1C to a similar degree as full-length EWS/FLI, while minimal or no interaction was seen between U1C and amino acids 210 -264 of EWS fused to FLI (EF-H) ( Fig. 3B and data not shown). Because there was no substantial difference in interaction between U1C and EF-B or EF-C, we used EF-B, which contained the minimal region necessary for interaction, for subsequent experiments.
U1C Represses EWS/FLI-mediated Transactivation-To determine the functional significance of the interaction between U1C and EWS/FLI, we tested the ability of U1C to affect EWS/FLI-mediated transactivation. An EWS/FLI-responsive reporter gene was cotransfected with EWS/FLI and vector or increasing amounts of U1C. EWS/FLI activated the reporter 9.7-fold over basal levels seen with vector alone. When 400 ng of U1C was added, a 35% repression of this activation was seen. Increasing the amount of U1C to 800 ng increased the repression to 75% (Fig. 4A). The increased amounts of U1C plasmid in the transfections gave a corresponding increase in U1C protein (Fig. 4B). Therefore, co-expression of U1C repressed EWS/FLImediated transactivation in a dose-dependent manner. U1C did not specifically repress transactivation of a GAL4-responsive reporter construct by a t-EWS/GAL4 fusion protein (data not shown). Thus, the repression mediated by the U1C-EWS interaction may be dependent upon specific promoter contexts or dependent upon the C-terminal motifs fused to EWS, or it may not have been evident due to the decreased affinity of U1C for EWS compared with EWS/FLI.
To confirm that the repression effect of U1C on EWS/FLImediated transactivation is dependent on its binding to the N terminus of EWS, equivalent amounts of the deletion constructs EWS/FLI-B and EWS/FLI-H were transfected in place of full-length EWS/FLI. Both deletion constructs were still capable of activating transcription of the reporter (Fig. 4C). U1C repressed activation of EWS/FLI-B by 77%, showing a specific interaction with the minimal interacting domain of EWS, whereas no significant repression was seen with EWS/ FLI-H (Fig. 4C). Thus, the interaction domain of EWS is necessary for U1C-mediated repression. DISCUSSION The recurrent combination of t-EWS with DNA binding motifs via chromosomal translocations in multiple human tumors indicates an important role for this domain. We hypothesized that the N-terminal domain of EWS may contribute unique properties to the chimeric proteins through interactions with other cellular proteins. We used a yeast two-hybrid screen to  Fig. 2A) were transfected as above in the mammalian two-hybrid assay to show that the minimal interaction domain in EWS is sufficient for interaction with U1C. Each transfection was repeated at least three times, and the average luciferase activity normalized with ␤-galactosidase activity is shown. Error bars indicate S.D. values.
identify proteins that interact with this tumor-associated portion of EWS to identify cellular pathways that may be altered in the tumors. In this report, we showed that t-EWS in the context of EWS and EWS/FLI interacts with the splicing protein U1C. Additionally, the essential splicing factor SF1, also termed ZFM1 (42), was identified in our yeast two-hybrid screen (data not shown) and in a previously published study as an EWS-interacting protein (25). These interactions suggest that EWS and perhaps the EWS/FLI fusion protein may play a role in splicing.
Several additional lines of evidence implicate EWS and the related protein TLS in splicing. TLS and EWS have both been co-immunoprecipitated with the heterologous nuclear proteins hnRNPA1 and hnRNPC1/C2 (21), which also participate in splicing (43,44). Further, TLS can interact with TASR, a member of the SR domain-containing family of splicing proteins (26). TLS can also modulate 5Ј splice site selection (27). The role of U1C and SF1 in splicing is well established. These two splicing factors are involved in early stages of spliceosome formation. The U1 snRNP is required for the formation of complex E of the spliceosome, the earliest stage of spliceosome assembly, which commits the pre-mRNA to the splicing pathway. The U1snRNP complex binds to the 5Ј splice site, and the U1C protein appears to potentiate the base pairing between the U1 snRNA and the 5Ј splice site (45). SF1 as well as several other protein components and ATP are required for the formation of the next step in spliceosome assembly, the presplicing complex A (42).
Our results indicate that U1C binds to EWS in mammalian cells and can bind to EWS/FLI with higher affinity than to wild-type EWS (Fig. 3A). Differential interactions between EWS and EWS/FLI have also been observed with TFIID components (24,46). Taken together, the selectivity of these interactions suggests that the folding or accessibility of t-EWS is FIG. 4. U1C represses EWS/FLI-mediated transactivation. A, increasing amounts of an U1C expression plasmid were co-transfected with plasmids containing EWS/FLI and an EWS/FLI-responsive luciferase reporter gene into HeLa cells. A ␤-galactosidase expression plasmid was included as an internal control for transfection efficiency. Numbers under the graph represent g of each plasmid in individual transfections. pCB6-FLAG vector was added to keep the total amount of DNA constant in each transfection mixture. Each transfection was repeated at least three times, and the average luciferase activity normalized to ␤-galactosidase activity is shown. Error bars indicate S.D. values. B, Western blot showing the expression of EWS/FLI and U1C in transfected cells. Cell extracts were resolved by 4 -20% SDS-polyacrylamide gel electrophoresis, and the expression of proteins was detected using M2 antibody, which recognizes the FLAG tag on the transfected EWS/FLI and U1C proteins. Numbers above blot indicate the g of each plasmid used in individual transfections. C, U1C and an EWS/FLI-responsive luciferase reporter gene were cotransfected with EWS/FLI or the EWS/ FLI deletion derivative EF-B or EF-H in HeLa cells. A ␤-galactosidase expression plasmid was included as an internal control for transfection efficiency. Each transfection was repeated at least three times, and the average luciferase activity normalized with ␤-galactosidase activity is shown. Error bars indicate S.D. values. somewhat dependent upon the C terminus of the protein. Thus, the regulation of this domain in chimeric fusion proteins found in tumors may be modulated by a subset of EWS-interacting proteins. Furthermore, EWS/FLI could potentially interfere with normal interactions of EWS with the splicing and transcriptional machinery by binding interacting partners more strongly.
Our results also show that U1C is able to repress EWS/FLImediated transactivation (Fig. 4A). This functional effect is dose-dependent and requires the region of EWS that interacts with U1C. Like U1C, SF1 was also a specific repressor of EWS-mediated transactivation (25). As expected, EWS/FLI, a strong transcriptional activator, increases expression of some downstream genes including stromelysin and Manic Fringe (47,48). Interestingly, EWS/FLI can repress the transcription of the transforming growth factor-␤ type II receptor, leading to decreased sensitivity to transforming growth factor-␤ in cells expressing the fusion protein (49). Because target genes may be activated or repressed by EWS/FLI, the U1C-mediated repression could enhance or suppress the transforming activity of the fusion protein.
Creation of a functional transcription factor through chromosomal translocations is an invariant theme among EWS, TLS, and TAF II 68 fusion proteins generated in human solid tumors; therefore, it is likely that aberrant gene regulation plays an important role in tumorigenesis. However, while EWS/FLI is a strong transcriptional activator, deletions within the EWS domain that significantly decrease the transactivation potential of the fusion protein cause only a moderate decrease in transforming activity (50). The region of EWS contributing the greatest transforming activity is included in the region required for interaction with U1C. In addition, an EWS/ FLI mutant that can no longer bind DNA is still transforming (51), suggesting that the fusion proteins may perform other roles in addition to deregulated transcription. Thus, the transcriptional activation and cellular transformation functions of EWS are not entirely concordant. Accordingly, the effect of interaction with U1C on transformation may not be solely determined by modulation of EWS/FLI-mediated transactivation and may also include alterations in splicing regulation.
Accumulating evidence suggests that the processes of RNA transcription and processing are closely coupled in vivo. In fact, the carboxyl-terminal domain of RNA polymerase II is required for efficient capping, cleavage at the polyadenylation site, and splicing of mRNA (reviewed in Ref. 52). The sequence of the t-EWS domain consists of a series of degenerate repeats that show similarity to the repeats of the carboxyl-terminal domain of RNA polymerase II (2). Indeed, t-EWS can function as a transcription activation domain, and it has also been shown to interact directly with components of TFIID and the RNA polymerase II complex (24,46). Interaction between t-EWS and the splicing machinery further extends the functional similarities between EWS and the carboxyl-terminal domain.
The identification of these protein-protein interactions and their physiological relevance may be critical in understanding the molecular mechanisms underlying Ewing's sarcoma and multiple other tumor types containing EWS, TLS, and TAF II 68 fusion proteins. Emerging data suggest that disruption of RNA splicing may be a novel pathway that contributes to tumorigenesis. For example, changes in expression patterns of splicing factors as well as alterations in alternative splicing increased with tumor progression in a mouse model of mammary tumorigenesis (53). Additionally, changes in expression patterns of splicing factors have been identified in some human colon adenocarcinomas (54), and alternatively spliced forms of important cell cycle regulators, such as p53 and cyclin D1, have been found in human cancer cell lines and tumors (55,56). The interaction of t-EWS with splicing factors as well as basal transcription factors strongly suggests that the fusion of EWS, or the related proteins TLS and TAF II 68, to different DNA binding domains may provide an efficient mechanism for the tumor cell to subvert normal regulation of transcription and post-transcriptional processing.