EWS.Fli-1 fusion protein interacts with hyperphosphorylated RNA polymerase II and interferes with serine-arginine protein-mediated RNA splicing.

Ewing's sarcoma displays a characteristic chromosomal translocation that results in fusion of the N-terminal domain of the Ewing's sarcoma protein (EWS) to the C-terminal DNA-binding domain of the ETS family transcription factor Fli-1 (Friend leukemia integration-1). EWS possesses structural motifs suggesting a role in transactivation as well as RNA binding. We demonstrate that wild-type EWS protein functions as an adapter molecule coupling transcription to RNA splicing by binding to hyperphosphorylated RNA polymerase II through the N-terminal domain of EWS and recruiting serine-arginine (SR) splicing factors through the C-terminal domain of EWS. The oncogenic EWS.Fli-1 fusion protein retains the ability to bind to hyperphosphorylated RNA polymerase II but lacks the ability to recruit SR proteins because of replacement of the C-terminal domain of EWS by Fli-1. In an in vivo splicing assay, the EWS.Fli-1 fusion protein inhibits SR protein-mediated E1A pre-mRNA splicing in a dominant-negative manner. These results indicate that EWS.Fli-1 interferes with the normal function of EWS and implicate uncoupling of gene transcription from RNA splicing in the pathogenesis of Ewing's sarcoma.

The EWS gene was originally identified in Ewing's sarcoma with the t(11;22) chromosomal translocation, where it is fused to the DNA-binding domain of the ETS transcription factor Fli-1 1 (1). Subsequent studies revealed that in Ewing's sarcoma EWS may be fused to one of five different members of the ETS family, namely Fli-1 (1), ERG (2), ETV-1 (3), E1AF (4,5), and FEV (6). In addition to fusions with ETS transcription factors in Ewing's sarcoma, EWS has been shown to form fusion proteins with a number of other partners including ATF-1 in malignant melanoma of soft parts (7), WT-1 in desmoplastic small round cell tumors (8,9), TEC in extraskeletal myxoid chondrosarcoma (10), and CHOP in myxoid liposarcoma (11).
In EWS fusion proteins, the N-terminal domain (NTD) of EWS is retained, whereas the C-terminal domain (CTD) of EWS is replaced by the corresponding fusion partner.
Understanding the mechanism of transformation by EWS fusion proteins will probably require knowledge regarding the functions of the wild-type proteins. In this regard, the N-terminal domain of EWS is rich in glutamine, serine, and tyrosine, residues that are commonly found in transcriptional activation domains. EWS is known to associate with a specific subpopulation of the TFIID basal transcription factor and with certain subunits of the RNA polymerase II (Pol II) complex (12). However, the C-terminal domains of EWS contain ribonucleoprotein consensus sequence and multiple arginine-glycine-glycine (RGG) repeats, both of which are signatures of RNA-binding proteins (13).
The Ewing's sarcoma protein EWS (1), the translocation in liposarcoma protein (TLS) (14,15), and the TATA-binding protein associated factor (TAF II 68) (16) comprise a unique family of proteins with shared structural features. We previously reported that wild-type TLS not only binds to RNA Pol II, but it also interacts with two newly characterized serine-arginine (SR) splicing factors (17,18). The structural similarities between EWS and TLS led to an assessment of the interaction of EWS with RNA Pol II and these TLS-associated SR (TASR) splicing factors. We determined that both EWS and EWS⅐Fli-1 interact with the hyperphosphorylated largest subunit of the RNA Pol II complex (Pol IIo) through the N-terminal domain of EWS. However, EWS interacts with TASR proteins, whereas EWS⅐Fli-1 is unable to interact with TASR proteins because of replacement of its C-terminal domain by the Fli-1 fusion partner. These biochemical differences between EWS and EWS⅐Fli-1 have functional consequences because EWS⅐Fli-1 interferes with TASR-mediated splicing in an in vivo E1A splicing assay. These results suggest that the EWS⅐Fli-1 fusion protein may contribute to cellular transformation through an effect on the coupling of transcription to RNA splicing.

EXPERIMENTAL PROCEDURES
Plasmids-The cDNAs for EWS, EWS⅐Fli-1, and Fli-1 were cloned into the pSG5-FL vector with the Flag epitope at the N-terminal end. The EWS-NTD deletion mutant consists of amino acids 1-245 of the EWS protein, and the EWS-CTD deletion mutant contains amino acids 267-656 of the EWS protein. Myc-TASR-1 and -2 expression vectors were constructed by cloning full-length TASR cDNAs into pCS2-MT vector (Sigma) with the Myc epitope at the N-terminal end. For in vivo splicing assay, TASR cDNAs were inserted into pMH vector (Sigma) to generate pMH-TASR with one copy of the influenza hemagglutinin epitope at the C-terminal ends of TASR proteins. Reporter plasmid pCS3-MT-E1A was a kind gift from Dr. Moreau-Gachelin (19).
Immunoprecipitation and Western Blotting-For expression of Flagor Myc-tagged proteins, 10 g of the pSG5-Flag expression construct and 10 g of the pCS2-Myc expression construct were introduced into 3 ϫ 10 6 COS-7 cells by electroporation. 48 h after electroporation, the cells were lysed with 0.6 ml of lysis buffer A (10 mM Tris⅐HCl, pH 7.4, 2.5 mM MgCl 2 , 100 mM NaCl, 0.5% Triton X-100). Prior to cell lysis, 3 l of 9E10 mouse monoclonal anti-Myc antibody (Sigma) or 10 l of 8WG16 mouse monoclonal anti-RNA Pol II antibody (Research Diagnostics, Inc.) was incubated with 30 l of protein A/G plus agarose (Santa Cruz Biotechnology) for 50 min at 4°C in 0.3 ml of buffer A, and the antibodyprotein A/G-agarose complex was then incubated with 0.2 ml of fresh cell lysate for 20 min. After four washes with radioimmune precipitation buffer, the immunoprecipitates were separated by SDS-polyacrylamide gel electrophoresis in a 10% gel, and the proteins were detected with M2 mouse monoclonal anti-Flag antibody (Sigma), the anti-Myc antibody, or C-21 rabbit polyclonal anti-RNA Pol II antibody (Santa Cruz Biotechnology).
For immunoprecipitation of endogenous EWS and EWS⅐Fli-1 fusion proteins, 30 l of protein A/G plus agarose was pre-charged with 10 l of 8WG16 anti-Pol II or control mouse IgG and then incubated with 0.2 ml of fresh lysate from HeLa, SK-ES, or RD-ES cells as described above. The immunoprecipitates were blotted with N-18 rabbit polyclonal anti-EWS (Santa Cruz Biotechnology) and C119 rabbit polyclonal anti-Fli-1 (20). Protein bands were visualized using the ECL Western blotting analysis system (Amersham Pharmacia Biotech).
Identification of Hypo-and Hyperphosphorylated Forms of the Pol II Largest Subunit-For immunoprecipitation of hyperphosphorylated RNA Pol IIo, 15 l of mouse monoclonal H5 or H14 antibody (Research Diagnostics, Inc.) was coupled to 40 l of protein A/G plus agarose beads via 30 l of rabbit anti-mouse IgM (Zymed Laboratories Inc.). For co-immunoprecipitation of EWS-associated RNA Pol IIo, 30 l of rabbit polyclonal anti-EWS antibodies or goat polyclonal anti-Flag antibody (Santa Cruz Biotechnology) was first bound to 40 l of protein A/G plus agarose beads. HeLa cells from a 100-mm dish were lysed with 0.6 ml of cell lysis buffer (50 mM Tris⅐HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate, 10 mM ␤-glycerophosphate). Insoluble material was removed via centrifugation at 15,000 ϫ g for 5 min, and the immunoprecipitation was then carried out by incubating 0.2 ml of lysate with antibody pre-charged agarose beads at 4°C for 4 h. After being washed three times with ice-cold cell lysis buffer, the immunoprecipitates were separated by SDS-polyacrylamide gel electrophoresis in a 6% gel. The hypophosphorylated RNA Pol IIa was detected with the 8WG16 antibody, and the hyperphosphorylated RNA Pol IIo was detected with the H14 antibody as described previously (21).
Transfection of HeLa Cells and RT-PCR-For in vivo splicing of E1A pre-mRNA, 2 g of pCS3-MT-E1A and 2 g of pMH-TASR plus 6 g of pSG5-Flag construct were mixed with 60 l of DOTAP (Roche Molecular Biochemicals), and the DNA-DOTAP mixture was added to two 60-mm duplicate dishes with 65% confluent HeLa cells. 40 h after transfection, cells from one dish were lysed with 0.25 ml of radioimmune precipitation buffer for Western blotting, and cells from the other dish were used for RNA isolation with an RNeasy column (Qiagen).
To ensure adequate annealing, one-fifth of the total RNA was hybridized to 20 pmol of T7 primer overnight. The reverse transcriptase reaction was carried out with 200 units of SuperScript II reverse transcriptase (Life Technologies, Inc.) at 42°C for 1 h, terminated by heat inactivation at 80°C, and diluted to a final volume of 40 l containing 0.5 unit of RNase H. For PCR amplification of E1A splicing isoforms, 4 l of the reverse transcriptase reaction mixture was used as the DNA template in a 50-l reaction with RR67 (5ЈGAGCTTGGGCGAC-CTCA3Ј) and T7 (5ЈAATACGACTCACTATAG3Ј) as the primer pair. The PCR was carried out by a 2-min incubation at 94°C, then 20 cycles of 1 min at 94°C, 1 min at 55°C, 1 min at 72°C followed by a final extension of 7 min at 72°C.
RNase Protection Assay-In RNase protection assay, one-fifth of the total RNA was hybridized to 1 ϫ 10 6 cpm of 32 P-labeled antisense E1A RNA probe (covering bases 499 -1316 of the E1A gene). After overnight hybridization, excessive RNA probe was digested with a mixture of RNase A ϩ T1 (PharMingen RNase protection assay system) at 30°C for 1 h, and the protected antisense E1A RNA fragments were isolated according to the manufacturer's instructions. After separation on a 6% QuickPoint pre-cast gel (Novex), the protected antisense RNA fragments were visualized by exposure to x-ray films.
whereas Flag-EWS⅐Fli-1 and Flag-Fli-1 did not co-immunoprecipitate with either TASR protein (Fig. 1b, lanes 8 , 9, 11, and 12). Because the C-terminal domain of EWS is replaced by the Fli-1 sequence in the EWS⅐Fli-1 fusion protein, these results suggested that wild-type EWS might mediate interaction with the TASR proteins through its C-terminal domain. To test this possibility, the N-terminal domain of EWS (EWS-NTD) and the C-terminal domain of EWS (EWS-CTD) were cloned separately into the pSG5-FL expression vector for transfection (Fig. 1a). When expressed in COS-7 cells, Flag-EWS-CTD co-immunoprecipitated with both TASR proteins (Fig. 1c, lanes 9 and 12), whereas the Flag-EWS-NTD did not, demonstrating that EWS interacts with TASR-1 and -2 through the C-terminal domain of EWS.
The N-terminal Domain of EWS Associates with RNA Pol II-EWS, as well as the structurally related TLS and TAF II 68 proteins, has been shown to associate with TFIID and subunits of the RNA Pol II complex (12,16,18,22); however, EWS⅐Fli-1 has been reported to lack this ability to interact with RNA Pol II (12). Because RNA Pol II is known to couple gene transcription with RNA processing in vivo (23), the demonstration that the C-terminal domain of EWS binds to SR proteins raised the possibility that EWS might function as an adapter molecule recruiting SR splicing factors to the site of active gene transcription by RNA Pol II. To assess the interaction between RNA Pol II, EWS, and EWS⅐Fli-1, immunoprecipitation studies were carried out using a mouse monoclonal 8WG16 antibody recognizing the C-terminal heptapeptide repeat on the largest subunit of RNA Pol II. Western blotting of immunoprecipitates from COS-7 cells using an anti-Flag antibody demonstrated that both Flag-EWS and Flag-EWS⅐Fli-1, but not Flag-Fli-1, were present in the immunocomplexes with RNA Pol II (Fig.  2a, lanes 4 -6). Using deletion mutants of EWS, Flag-EWS-NTD but not Flag-EWS-CTD was demonstrated to be the domain responsible for association with RNA Pol II (Fig. 2b, lanes  10 -12).
To determine whether endogenous EWS⅐Fli-1 associated with RNA Pol II in Ewing's sarcoma cells, we used the same 8WG16 anti-Pol II antibody in co-immunoprecipitation experiments with lysate from HeLa cells, which lack the EWS⅐FLI-1 fusion protein, and with lysates from two Ewing's sarcoma cell lines, SK-ES and RD-ES, which are known to express the 68-kDa EWS⅐Fli-1 fusion protein. The immunoprecipitated proteins were separated and blotted with antibodies recognizing EWS and EWS⅐Fli-1 fusion protein. The anti-EWS antibody recognized the 90-kDa EWS protein in the immunoprecipitates (Fig. 2c, lanes 13-18, upper panel), and the anti-Fli-1 antibody detected the 68-kDa EWS⅐Fli-1 fusion protein in the immunoprecipitates from the two Ewing's sarcoma cell lines but not from the HeLa cells (Fig. 2c, lanes 13-18, lower panel). These results indicated that both endogenous EWS and EWS⅐Fli-1 are associated with RNA Pol II in vivo.
EWS and EWS⅐Fli-1 Associate with Hyperphosphorylated RNA Pol II-The largest subunit of RNA Pol II can exist either as a hypophosphorylated form (Pol IIa) that migrates at 220 kDa or as a hyperphosphorylated form (Pol IIo) that migrates at 240 kDa. The phosphorylation sites on this largest Pol II subunit have been localized to the C-terminal domain. Hypophosphorylated RNA Pol IIa is believed to be recruited to pre-initiation complexes. Concomitant with initiation of transcription, the CTD of Pol IIa becomes phosphorylated, and elongating polymerase therefore has a hyperphosphorylated CTD (24). Because RNA processing takes place along with transcriptional elongation, RNA splicing factors are associated with hyperphosphorylated Pol IIo (25)(26)(27). If EWS indeed functions as an adapter molecule that recruits SR splicing factors to the site of transcription, EWS would be expected to interact with hyperphosphorylated Pol IIo.
In this study, three mouse monoclonal anti-Pol II antibodies (8WG16, H5, and H14) were used in immunoprecipitation and Western blotting to differentiate the phosphorylation status of the Pol II subpopulations associated with EWS and EWS⅐Fli-1. Initial experiments confirmed previous observations that 8WG16 recognizes both hypophosphorylated Pol IIa and hyperphosphorylated Pol IIo in immunoprecipitation studies (21,26); however, 8WG16 primarily detects the hypophosphorylated Pol IIa form in Western blotting (Fig. 3a, lane 2, upper and lower  panels). The H5 and H14 antibodies recognize only the hyper- phosphorylated Pol IIo form in both immunoprecipitation and Western blotting (21) (Fig. 3a, lanes 3 and 4). Repeated attempts to co-immunoprecipitate EWS with H5 and H14 were not successful (data not shown), and it is likely that binding by H5 or H14 causes Pol IIo to be released from the multi-protein complex as previously suggested (26). As an alternative approach, RNA Pol II was co-immunoprecipitated with a rabbit polyclonal antibody against the C terminus of EWS, whereas a rabbit polyclonal antibody against the N terminus of EWS failed to co-immunoprecipitate RNA Pol II (Fig. 3a, lanes 6 and  7). This different ability to co-immunoprecipitate RNA Pol II suggests that the rabbit antibody against the C terminus of EWS is less disruptive to the maintenance of the EWS-Pol II multi-protein complex than is the antibody against the N terminus of EWS. The RNA Pol II subpopulation associated with EWS appeared to be predominantly the hyperphosphorylated Pol IIo form because the co-immunoprecipitated RNA Pol II is detectable by H14 but not by 8WG16 in Western blotting (Fig.   3a, lane 6, upper and lower panels).
To co-immunoprecipitate RNA Pol II with the EWS⅐Fli-1 fusion protein, plasmids expressing Flag-tagged EWS, EWS⅐Fli-1, and Fli-1 were transfected into HeLa cells, and the lysates were used in co-immunoprecipitation with a goat polyclonal anti-Flag antibody. Under our experimental conditions, only Flag-EWS⅐Fli-1 co-immunoprecipitated with hyperphosphorylated Pol IIo (Fig. 3b, lane 10), whereas Flag-EWS and Flag-Fli-1 did not co-immunoprecipitate with Pol IIo (Fig. 3b,  lanes 9 and 11). Because the Flag epitope is tagged at the same N-terminal ends of EWS and EWS⅐Fli-1 and because the immunoprecipitation was carried out with the same goat anti-Flag, these results suggested that EWS and EWS⅐Fli-1 might associate differently with RNA Pol IIo. The EWS⅐Fli-1 fusion protein may have a higher affinity than EWS toward the same site on Pol IIo, and this high affinity may preserve the interaction between Pol IIo and EWS⅐Fli-1 despite interference by the anti-Flag antibody binding. Another potential explanation is that EWS⅐Fli-1 may bind to a different site on Pol IIo that is less prone to disruption by the antibody.
EWS⅐Fli-1 Inhibits RNA Splicing Mediated by SR Proteins-To investigate whether this interaction of EWS with RNA Pol IIo and the TASR proteins played a role in RNA splicing, we tested both wild-type EWS and the EWS⅐Fli-1 fusion protein in an E1A splicing assay in HeLa cells. The alternative splicing of E1A pre-mRNA in HeLa cells results in the generation of five different splicing isoforms designated 13 S, 12 S, 11 S, 10 S, and 9 S (Fig. 4a) (28), and co-expression of individual SR proteins is known to alter the splicing pattern of E1A isoforms (29).
Analysis of E1A splicing products by RT-PCR indicated that expression of TASR-1 promoted splicing to the 11 S-, 10 S-, and 9 S-isoforms ( Fig. 4b; compare lanes a and e), whereas expression of TASR-2 promoted splicing to the 9 S-isoform (Fig. 4b,  compare lanes a and i). Endogenous EWS is constitutively expressed in HeLa cells, and co-expression of EWS did not alter the E1A splicing profile (Fig. 4b, lanes f and j). Co-expression of EWS⅐Fli-1 resulted in a marked inhibition of E1A splicing by both TASR-1 and TASR-2 (Fig. 4b, compare lane f with lane g, and lane j with lane k); however, co-expression of Fli-1 had little effect on E1A splicing (Fig. 4b, lanes h and l). The band designated "reporter DNA" represents the PCR-amplified E1A genomic sequence from residual plasmid DNA.
To demonstrate that the results obtained by RT-PCR were not due to selective amplification of specific sequences, we developed an RNase protection assay to directly measure the profile of E1A alternative splicing in HeLa cells. In the RNase protection assay, TASR-1 expression increased the E1A fragments corresponding to the 11 S-, 10 S-, and 9 S-isoforms ( Fig.  4 c, compare lanes a and e), whereas TASR-2 expression increased the E1A fragments to the 9 S-isoform (Fig. 4c, compare  lanes a and i). Co-expression of Flag-EWS and Flag-Fli-1 did not significantly alter TASR-mediated E1A pre-mRNA splicing (Fig. 4c, lanes f, h, j, and l); however, co-expression of Flag-EWS⅐Fli-1 strongly inhibited E1A pre-mRNA splicing mediated by both TASR proteins (Fig. 4c, lanes g and k). Transfection efficiency among different samples was similar in that comparable levels of EWS, EWS⅐Fli-1, Fli-1, and TASR proteins were expressed in the transfected HeLa cells (Fig. 4c, bottom panels). Unspliced E1A pre-mRNA was not detected by the RNase protection assay, consistent with previous reports that the unprocessed E1A pre-mRNA molecule is unstable in HeLa cells (28).
The effects of EWS deletion mutants on TASR-mediated E1A splicing were tested to determine whether the N-terminal domain or the C-terminal domain of EWS alone was responsible for the alteration in E1A splicing. Even though in HeLa cells both EWS deletion mutants were expressed at levels comparable with EWS⅐Fli-1, neither the EWS-NTD nor the EWS-CTD inhibited TASR-mediated E1A pre-mRNA splicing (Fig. 4d,  lanes m-r), indicating that inclusion of Fli-1 in the fusion protein is necessary for EWS⅐Fli-1 inhibition of RNA splicing.  (lanes m-r). IP, immunoprecipitation.

EWS⅐Fli-1 Interferes with SR Protein-mediated Splicing 37616
Inclusion of the Fli-1 sequence in the fusion protein may result in changes in protein folding and/or accessibility of the EWS N-terminal domain to RNA Pol II, thus endowing the EWS⅐Fli-1 fusion protein with a higher affinity toward certain subunits of RNA Pol II such as hsRPB7 (22). This might enable the fusion protein to associate preferentially with the RNA Pol II complex, resulting in a dominant-negative effect on RNA splicing.
Although these results are consistent with the hypothesis that EWS⅐Fli-1 uncouples E1A reporter gene transcription from E1A pre-mRNA splicing and leads to degradation of the unprocessed E1A pre-mRNA transcripts in HeLa cells, the observed decrease in E1A splicing products could also be explained by EWS⅐Fli-1 inhibition of transcription from the pCS3-MT-E1A reporter or by EWS⅐Fli-1-induced degradation of the E1A splicing products. To test whether EWS⅐Fli-1 suppresses transcription from the pCS3-MT-E1A reporter or destabilizes the E1A splicing isoforms, we constructed an additional reporter plasmid, pCS3-MT-E1A-9S, which was generated from the same vector as pCS3-MT-E1A but contains a cDNA insert corresponding to the 9S.E.1A splicing isoform. The 9S.E.1A isoform is devoid of intron sequence; therefore its expression should be controlled primarily by gene transcription rather than by splicing. When analyzed in HeLa cells under the same conditions as described above, the resultant 9S.E.1A mRNA levels were not decreased by co-expression of EWS⅐Fli-1 or altered by TASR-1 or TASR-2 (Fig. 5). These results indicate that EWS⅐Fli-1 does not suppress transcription from the pCS3-MT-E1A vector or selectively destabilize the 9S.E.1A splicing isoform. Thus, the decrease in the steady-state level of E1A transcripts most likely results from EWS⅐Fli-1 inhibition of TASR-mediated splicing.

DISCUSSION
The EWS⅐Fli-1 fusion protein has been known to bind to specific DNA sequences and to transactivate target genes with Fli-1 binding sites in their promoter regions (30 -33). Because deletion of either the EWS or the Fli-1 domain in the fusion protein results in loss of transforming ability (34), the EWS⅐Fli-1 fusion protein is believed to lead to malignant transformation via inappropriate activation of Fli-1 target genes (35). However, increasing evidence suggests that an alternative mechanism may also be involved in transformation by EWS⅐Fli-1. First, although EWS⅐Fli-1 is a more potent transactivator than Fli-1, deletional studies indicate that the domain within EWS required for transactivation differs from the domain required for transformation (36). Second, a recent mutagenesis study demonstrated that a point mutation abolishing the DNA binding activity of EWS⅐Fli-1 did not abolish its transforming ability (37). Third, the N-terminal domain of EWS has been reported to interact with splicing factors SF1 (38) and U1C (39) to regulate gene expression. Fourth, the RNA binding motifs in the C-terminal domain of EWS are replaced in the EWS⅐Fli-1 fusion protein, thus implicating a potential loss of function in RNA processing by the fusion protein.
Our findings provide evidence for a pathway whereby wildtype EWS functions as a docking molecule that binds via its N-terminal domain to hyperphosphorylated RNA Pol IIo and recruits specific SR splicing factors through its C-terminal domain. EWS fusion proteins, on the other hand, bind to RNA Pol IIo but fail to recruit TASR splicing factors. This uncoupling of RNA processing from transcription would be expected to alter gene expression in tumor cells with EWS fusions. EWS fusion proteins might block splicing by SR proteins, leading to degradation of the unprocessed target pre-mRNA and downregulation of target gene expression, or lead to alternative splicing of the target pre-mRNA. In this regard, abnormal RNA splicing in Ewing's sarcoma cells has been reported to affect a variety of molecules such as the fragile histidine triad (FHIT) tumor suppressor (40), the p53-inducible P2XM ion channel (41), and the PAX3 and PAX7 regulators of myogenic and neural development (42).
Our findings also demonstrate that the EWS⅐Fli-1 fusion protein interferes with E1A pre-mRNA splicing despite the presence of endogenous EWS protein. These results appear to be relevant to the pathogenesis of Ewing's sarcoma because in Ewing's sarcoma cells one EWS allele remains intact (43). In our splicing assays the EWS fusion protein functions in a dominant-negative manner. This dominant-negative effect could be partly explained by the fact that EWS⅐Fli-1 is present exclusively in the nucleus, whereas EWS is detectable in both the nucleus and the cytoplasm (43). The exclusive nuclear localization of the EWS⅐Fli-1 fusion protein in effect increases the nuclear concentration of the EWS⅐Fli-1 fusion protein and enhances its chance to compete with EWS for interaction with the RNA Pol II complex. Alternatively, EWS and EWS⅐Fli-1 may each associate with different subunits of the RNA Pol II complex, and these associations may be mutually exclusive. This notion is supported by the findings that nuclear complexes containing EWS and EWS⅐Fli-1 have different sedimentation profiles when fractionated through a glycerol gradient (12) and that the EWS⅐Fli-1 fusion protein possesses a much higher affinity toward certain subunits of the RNA Pol II complex than does wild-type EWS, as shown in this study and by others (22). It is also possible that high affinity binding to certain subunits of RNA Pol II by EWS⅐Fli-1 dominant-negatively affects the recruitment of SR splicing factors by EWS to RNA Pol II, leading to disruption of normal splicing mediated by these EWS-associated SR proteins. The role of competition between EWS and EWS⅐Fli-1 will be investigated in future studies using inducible expression of EWS in Ewing's sarcoma cell lines.