The Prooncoprotein EWS Binds Calmodulin and Is Phosphorylated by Protein Kinase C through an IQ Domain*

A growing family of proteins is regulated by protein kinase C and calmodulin through IQ domains, a regulatory motif originally identified in neuromodulin (Alexander, K. A., Wakim, B. T., Doyle, G. S., Walsh, K. A., and Storm, D. R. (1988) J. Biol. Chem. 263, 7544–7549). Here we report that EWS, a nuclear RNA-binding prooncoprotein, contains an IQ domain, is phosphorylated by protein kinase C, and interacts with calmodulin. Interestingly, PKC phosphorylation of EWS inhibits its binding to RNA homopolymers, and conversely, RNA binding to EWS interferes with PKC phosphorylation. Several other RNA-binding proteins, including TLS/FUS and PSF, co-purify with EWS. PKC phosphorylation of these proteins also inhibits their binding to RNAin vitro. These data suggest that PKC may regulate interactions of EWS and other RNA-binding proteins with their RNA targets and that IQ domains may provide a regulatory link between Ca2+ signal transduction pathways and RNA processing.

Neuromodulin (GAP-43, B-50, F1) and neurogranin (RC3, BICKS) are neurospecific calmodulin (CaM) 1 -binding proteins that are phosphorylated by protein kinase C (PKC) (for reviews see Refs. 8 and 42). These proteins share a nearly identical sequence of 20 amino acids designated the IQ motif containing a CaM-binding domain and a PKC phosphorylation site (1)(2)(3)(4). The interaction between CaM and the IQ domain of neuromodulin or neurogranin has been characterized extensively in vitro (2,(5)(6)(7)(8)(9). The affinity of neuromodulin or neurogranin for CaM is higher in the absence than in the presence of Ca 2ϩ , and CaM binding is inhibited by PKC phosphorylation within the IQ domain. Furthermore, CaM binding inhibits PKC phosphorylation of neuromodulin and neurogranin at a specific serine in the IQ domain. These observations led to the hypothesis that neuromodulin may bind and concentrate CaM at specific sites in neurons and release CaM in response to PKC phosphorylation or increases in Ca 2ϩ (5,10). Recent data from several laboratories have confirmed that neuromodulin and CaM interact in cells. For example, elevation of free Ca 2ϩ or PKC activation in primary hippocampal neuron cultures inhibits CaM/neuromodulin interactions in vivo (11). Neuromodulin and CaM have also been shown to interact in vivo using the yeast two-hybrid system (12).
The IQ domain may serve as a general regulatory domain in proteins for CaM binding and PKC phosphorylation. It has been found in other proteins including the Ca 2ϩ vector protein target from amphioxus (13), the neurospecific peptide PEP-19 (14), the early endosome-associated protein EEA1 (15) and the guanine nucleotide exchange factor p140/Ras-GRF (16). The conventional and unconventional myosins (17), the igloo protein from Drosophila (18), the GTPase-activating protein IQ-GAP (19,20), and the docking protein of insulin receptor IRS-1 (21) all contain repeat IQ motifs. Some of these proteins, e.g. unconventional myosins and PEP-19 (14), interact with CaM in a Ca 2ϩ -independent manner, whereas the Ca 2ϩ vector protein target and p140/Ras-GRF bind CaM in the presence of Ca 2ϩ (16,22). The IQ domain may also mediate interactions with other Ca 2ϩ -binding proteins belonging to the EF-hand family. The repeat IQ motifs of conventional myosins bind myosin light chains (23), and neuromodulin and neurogranin can also interact with the S100b protein (24,25). The conservation of the IQ regulatory sequence in a variety of proteins strongly suggests that the molecular function of these proteins may be regulated by Ca 2ϩ signal transduction pathways through IQ domain(s), although these proteins have very different biochemical functions.
Using an antibody against the IQ domain of neurogranin, we discovered a new member of the IQ protein family, P68-RNA helicase, the first nuclear IQ domain protein identified (26). Human P68-RNA helicase is an RNA-dependent ATPase that belongs to the family of putative helicases known as the DEAD box proteins (27). These proteins are implicated in some aspects of RNA functions including translation initiation, splicing, and ribosome assembly. PKC phosphorylation and CaM binding both inhibit P68-RNA helicase ATPase activity, suggesting that its RNA unwinding activity or binding to RNA may be regulated by Ca 2ϩ signal transduction pathways (26). Other lines of evidence suggest that RNA processing may be controlled by Ca 2ϩ signal pathways. Recently, Bachs and colleagues (28) discovered that heterogeneous nuclear ribonucleoproteins (hnRNPs) A2 and C interact with CaM in the presence of Ca 2ϩ . Furthermore, hnRNP A1 is phosphorylated in vivo by PKC- (29), and PKC phosphorylation of hnRNP A1 regulates its RNA binding activity in vitro (29,30).
Since CaM is also present in the nucleus (31) and P68-RNA helicase contains an IQ domain, we examined nuclear extracts for other RNA-binding proteins containing IQ domains. We found that the human nuclear RNA-binding prooncoprotein EWS (Ewing sarcoma protein) contains an IQ domain, binds to * This work was supported by National Institutes of Health Grant NS 20498. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  CaM, and is phosphorylated by PKC. The EWS gene is implicated in specific chromosomal translocations occurring in peripheral primitive neuroectodermal tumors (32,33). We also report that EWS co-purifies with another RNA-binding protein and PKC substrate, TLS/FUS (34 -36) as well as the splicing factor PSF (37). Interestingly, PKC phosphorylation of EWS, TLS/FUS, and PSF inhibits their RNA binding activity in vitro.

EXPERIMENTAL PROCEDURES
Materials-The alkaline phosphatase-conjugated substrate kit and prestained SDS-PAGE standards were from Bio-Rad. The nitrocellulose membrane was from Schleicher and Schuell. The polyvinylidene difluoride membrane for protein sequencing (Immobilon P SQ ) was from Millipore Corp. (Bedford, MA). High resolution hydroxylapatite resin was from Calbiochem. Fibrous cellulose CF11 was from Whatman. Sequencing grade trypsin and V8 protease were from Promega (Madison, WI). ssDNA-cellulose, dsDNA-cellulose, CaM-agarose, L-␣-phosphatidyl-L-serine, and diolein were from Sigma. CNBr-activated Sepharose 4B and ribonucleotide homopolymers were from Pharmacia Biotech Inc. [␥-32 P]ATP (3000 Ci/mmol) was from NEN Life Science Products. Dulbecco's modified Eagle's medium, bovine calf serum, fetal calf serum, and biotinylated calmodulin were from Life Technologies, Inc. PKC was purified from rabbit brains as described by Alexander et al. (5) or was purchased from Boehringer Mannheim.
Protease Inhibitors-Leupeptin, aprotinin, pepstatin A, and phenylmethylsulfonyl fluoride were purchased from Sigma and used at final concentrations of 5 g/ml, 10 g/ml, 5 g/ml, and 1 mM, respectively. The addition of protease inhibitors in solutions and buffers is indicated as mixture inhibitor.
Isolation of Subcellular Fractions-All extraction steps were performed at 4°C. HEK-293 or 3T3-L1 cells (100-mm plates) were washed three times with 4°C phosphate-buffered saline (PBS), scraped off, and collected by centrifugation for 5 min at 200 ϫ g. Pelleted cells were then suspended in 5 volumes of hypotonic buffer containing 10 mM Tris-HCl, pH 6.8, at 25°C, 0.5 mM DTT, 100 mM NaCl, 2.5 mM MgCl 2 , 0.5% Triton X-100, and mixture inhibitor and were lysed by 10 strokes of a Kontes all glass Dounce homogenizer (type B pestle). The homogenate was centrifuged for 10 min at 800 ϫ g to pellet nuclei. The supernatant was again centrifuged 10 min at 1000 ϫ g to eliminate debris and then for 30 min at 100,000 ϫ g to pellet membranes. The supernatant was designated as the cytosolic fraction. The pelleted membranes were solubilized in alkaline buffer containing 25 mM Tris-HCl, pH 10.0, at 25°C, 0.5 M NaCl, 5 mM DTT, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, and mixture inhibitor. The membrane fraction was clarified by 12,000 ϫ g for 20 min. The pelleted nuclei obtained from the low speed spin were washed twice with 5 volumes of wash buffer containing 40 mM Tris-HCl, pH 6.8, at 25°C, 250 mM sucrose, 5 mM MgCl 2 , and mixture inhibitor and were centrifuged for 10 min at 800 ϫ g. The pelleted nuclei were then resuspended in 1.5 volumes of the above alkaline buffer with a Kontes all glass Dounce homogenizer (20 strokes with type A and type B pestle) and kept on ice for 30 min. The nuclear homogenate was then centrifuged for 30 min at 100,000 ϫ g. The pellet was discarded, and the supernatant was designated as the nuclear extract.
Isolation of Subnuclear Fractions-Nuclei were isolated from HEK-293 cells as described above and incubated in wash buffer containing 250 g/ml RNase A and 250 g/ml DNase I for 1 h at 4°C. Triton X-100 and NaCl were added to a final concentration of 0.5% and 1.5 M, respectively. The nuclei were then vigorously vortexed and incubated for a further 30 min on ice. RNase and DNase reactions were stopped by the addition of 5 mM EDTA, and the homogenate was centrifuged at 12,000 ϫ g for 20 min. The supernatant was designated as the S1 fraction. The pellet was washed extensively in the wash buffer containing 0.5% Triton X-100, and the nuclear matrix proteins were solubilized in the alkaline buffer. This fraction was designated as the nuclear matrix fraction.
Partial Purification of EWS and Co-purifying Proteins-Nuclear extract from 36 plates (100 mm) of 14-day-old HEK-293 cells was prepared in alkaline buffer as described above. The pH of the nuclear extract (10 ml containing about 50 mg of total protein) was adjusted to 8.0 with 1 M Tris-HCl buffer, pH 7.4, at 25°C (500 l for 10 ml of nuclear extract) and then loaded on a hydroxylapatite/CF11 column (mixture of 1 and 0.5 g, respectively), which was preequilibrated with a potassium buffer containing 40 mM KH 2 PO 4 /K 2 HPO 4 , pH 8.0, 1 mM phenylmethylsulfonyl fluoride, and 5 mM DTT. The column (4-ml bed volume) was washed with 10 ml of alkaline extraction buffer adjusted to pH 8.5, and the flow-through fraction containing unbound proteins was collected to a final volume that was 1.4 times the loaded volume. The flow-through fraction was brought to 15% ammonium sulfate, incubated for 1 h at 4°C with agitation, and centrifuged for 20 min at 12,000 ϫ g. The pellet, containing approximately 500 g of total protein, was solubilized in urea buffer containing 25 mM Tris HCl, pH 7.4, at 4°C, 2 M urea, 50 mM NaCl, 1 mM EGTA, 2.5 mM MgCl 2 , 5 mM DTT, and mixture inhibitor and was analyzed by SDS-PAGE. Coomassie Blue-stained gels showed that it was possible to obtain two different kinds of ammonium sulfate fractions containing either five or three major proteins, designated as type A or type B fractions (see "Results").
DNA-cellulose affinity chromatography was performed essentially as described by Pinõl-Roma et al. (38). The type A fraction (50 g) was applied to an ssDNA-or dsDNA-cellulose column (0.5 ml each) in the urea buffer described above. Columns were washed with the same solution (40 bed volumes), and bound proteins were first eluted with the urea buffer containing heparin (1 mg/ml) and then were sequentially eluted with 0.5, 1, and 2 M NaCl.
RNA Binding Activity-RNA homopolymers were immobilized on CNBr-activated Sepharose 4B beads as described in the Pharmacia instructions. The type A fraction (10 g) was incubated with RNA homopolymer-Sepharose beads in urea buffer for 1 h at 4°C. Beads were washed, drained, and then boiled in Laemmli sample buffer.
In Vitro Phosphorylation of Type A and B Fractions by PKC-Phosphorylation assays were carried out as described by Alexander et al. (5). The type A or B fraction (1 l, 40 g/ml) was incubated with purified PKC in 20 mM Hepes, pH 7.4, 5 mM MgCl 2 , 5 mM DTT in the presence or absence of 1.5 mM CaCl 2 , 60 g/ml L-␣-phosphatidyl-L-serine, 6 g/ml diolein, and 70 nM bisindolylmaleimide for the indicated times at 30°C. The reaction was initiated by the addition of 500 M [␥-32 P]ATP. The total volume of the reaction was 50 l, and the reaction was terminated by the addition of Laemmli sample buffer or urea. The proteins were separated by SDS-PAGE. The gels were stained with Coomassie Blue, dried, and autoradiographed. The autoradiograms were scanned with a Scan Jet II CX scanner and analyzed by densitometry using NIH Image software.
Protein Cleavage and Amino Acid Sequencing-Peptide mapping was carried out as described by Cleveland et al. (39) with some modifications. Briefly, the type A fraction (five lanes loaded with 40 g) was run on 8% SDS-PAGE and stained with Coomassie Blue. The band corresponding to IQ85 was excised from the gel and reelectrophoresed on a 12% SDS-PAGE gel in the presence of trypsin (0.5 g in 0.125 mM Tris-HCl, pH 7.8, at 25°C, 1 mM CaCl 2 , and 0.1% SDS). To increase the efficiency of cleavage by trypsin, the pH of the stacking gel was brought to 7.8, and 1 mM CaCl 2 was added. Digested peptides were electrophoretically transferred to a polyvinylidene difluoride membrane and stained with Coomassie Blue. Selected peptide fragments were sequenced with an automated amino acid sequencer (Applied Biosystems, model 477A).
Western Blots-Proteins were electrophoresed by SDS-PAGE and transferred electrophoretically to nitrocellulose membrane as described by Towbin et al. (40). After transfer, the membrane was blocked for 1 h with 3% cold fish gelatin in PBS buffer. After washing in PBS, the blot was incubated overnight at 4°C with primary antibodies in TPBS (PBS plus 0.05% Tween 20) as indicated. Bound antibody was detected using alkaline phosphatase-conjugated secondary antibody and an alkaline phosphatase-conjugated substrate kit according to the manufacturer's instructions.
Immunoprecipitation-Flow-through fractions from a hydroxylapatite column (100 l) were precleared with 20 l of protein A-Sepharose beads (Pierce) for 1 h. The supernatant was incubated with IQ antibody (2 g of IgG) or EWS 677 antibody (10 l) overnight with agitation, and then 20 l of protein A-Sepharose beads was added for 1 h. Beads were pelleted and washed with PBS plus 1% Triton X-100 (5 ϫ 1 ml). Immunoprecipitated proteins were analyzed by Western blot using EWS and IQ antibodies.
Other Methods-SDS-PAGE was performed as described by Laemmli et al. (41), and proteins were quantified by the Bradford method (42). The BCA protein assay kit (Pierce) was used to determine the concentration of neurogranin and neuromodulin. CaM binding was monitored by biotinylated calmodulin overlay (43).

RESULTS
Characterization of IQ85-To identify nuclear proteins that contain IQ domains, we performed Western analysis of subcellular fractions from HEK-293 and mouse 3T3-L1 fibroblast cell lines using a polyclonal antibody (IQ Ab) that recognizes the IQ domain of neuromodulin, neurogranin, and P68-RNA helicase (26). In both cell lines, the IQ Ab recognized P68-RNA helicase and a larger protein of 85 kDa (Fig. 1A). An additional protein with an apparent molecular mass of 100 kDa was detected in the HEK-293 cell nuclear extracts. These three proteins were localized in the nuclear extract (NE) but were not found in cytoplasmic (Ct) or membrane fractions (Mb). To determine if the 85-and 100-kDa nuclear proteins contained IQ domains, the IQ Ab was incubated with an excess of pepP68, a peptide containing the IQ domain of P68-RNA helicase, and the proteins were analyzed by Western blots. PepP68 totally suppressed interactions between the IQ Ab and P68-RNA helicase or the 85-kDa protein. In contrast, the immunoreactivity observed toward P100 was unchanged (Fig. 1B). These data suggested that the 85-kDa protein, but not the 100-kDa protein, may contain an IQ epitope. Consequently, it was designated IQ85.
Bachs and Carafoli (31) reported that CaM is associated with the nuclear matrix fraction that contains nuclear membrane proteins, proteins implicated in DNA transcription and replication, and proteins involved in RNA metabolism (44). Furthermore, P68-RNA helicase may also be associated with the nuclear matrix (45). Therefore, we used the IQ Ab and a P68-RNA helicase monoclonal antibody, PAb204 (46), to investigate the subnuclear localization of IQ85 and P68-RNA helicase in HEK-293 cell nuclei. Purified nuclei were treated with DNase I and RNase A, followed by solubilization using 0.5% Triton X-100 and 1.5 M NaCl. The solubilized proteins (S1) and the pellet containing the nuclear matrix fraction (Mx) were analyzed for IQ domain proteins. Western blot analysis with PAb204 and IQ Ab showed that IQ85 and P68-RNA helicase are partially associated with matrix, whereas P100 is predominantly in the S1 fraction (Fig. 1C).
Partial Purification of IQ85-To identify IQ85, fractions from purified nuclei of quiescent HEK-293 cells were prepared as described under "Experimental Procedures," and IQ85 was partially purified. The first purification step was similar to that used for isolation of P68-RNA helicase (45). An alkaline buffer was used to extract protein from the nuclear matrix, which was then applied to a hydroxylapatite column. Hydroxylapatite fractions were analyzed on 8% SDS-PAGE by Coomassie Blue staining ( Fig. 2A) and immunoblotting with the IQ Ab (Fig.  2B). Unlike P68-RNA helicase, which absorbed to the column and was concentrated in the eluate (El), IQ85 was not retained on hydroxylapatite and was found exclusively in the flowthrough (FT) fraction (Fig. 2B). Additional purification of IQ85 was achieved by ammonium sulfate (AS) precipitation of the hydroxylapatite FT fraction. IQ85 was one of the major Coomassie Blue-stained proteins in this fraction ( Fig. 2A). The ammonium sulfate pellets were insoluble in buffers lacking detergents and could only be resuspended in the presence of 1% SDS or 2 M urea, suggesting that the precipitated proteins were strongly aggregated. We obtained two types of ammonium sulfate fractions, depending upon the number of HEK-293 cell passages (Fig. 2C). The type A fraction was obtained from HEK-293 cells grown less than 10 passages, whereas the type B fraction was obtained when cell passage was greater than 10. The type A fraction contained five major Coomassie Bluestained proteins designated P98, IQ85, P70, P60, and P48. P98 and P60 were absent from type B preparations. Densitometry analysis of Coomassie Blue-stained gels showed that IQ85 represented about 20 and 40% of total protein in type A and type B fractions, respectively.
Since P68-RNA helicase and IQ85 are both found in the nuclear matrix fraction and hnRNP proteins are also major components of this fraction (44), we considered the possibility that IQ85 may interact with RNA and bind to ssDNA. P68-RNA helicase has a high affinity for ssDNA but not for dsDNA (45), and most hnRNP proteins can be purified using ssDNA columns. hnRNP proteins are eluted from ssDNA resin with heparin or salt at variable concentrations (38). To determine if IQ85 and co-purifying proteins bind RNA, we carried out affinity chromatography on ssDNA-cellulose and dsDNA-cellulose in 2 M urea. Proteins were eluted from the ssDNA or dsDNA columns by sequential elution with 1 mg/ml heparin, 0.5 NaCl, and 1 M NaCl (Fig. 3). All major proteins from the type A fraction bound to the ssDNA resin (Fig. 3A) but not to the dsDNA resin (Fig. 3B). P98, IQ85, P71, and P60 were eluted from the ssDNA resin with heparin, whereas P48 was eluted with 0.5 M NaCl. The RNA binding activity of IQ85 and copurifying proteins in fraction A were investigated using polyribonucleotides attached to Sepharose beads (Fig. 3C). IQ85, P98, P71, and P60 displayed a similar preference for poly(U)and poly(G)-Sepharose (Fig. 3C). IQ85, P98, and P71 showed FIG. 1. Expression and subcellular localization of IQ85. A, IQ85 is a nuclear protein. Nuclear extract from HEK-293 cells and nuclear extract (lanes NE, 15 g), cytosolic fraction (lane Ct, 15 g), and membrane fraction (lane Mb, 15 g) from 3T3-L1 cells were separated by 8% SDS-PAGE, transferred to nitrocellulose membrane, and immunoblotted with the IQ domain antibody. B, IQ85 contains an IQ epitope. Western blot analysis of HEK-293 cell nuclear extract (lanes NE, 15 g) using the IQ Ab in the absence or presence of 0.3 mg/ml pepP68. C, IQ85 and P68-RNA helicase are present in the nuclear matrix. Extraction of nuclear matrix-associated proteins (lanes Mx, 10 g) after treatment of HEK-293 nuclei with RNase and DNase, 1.5 M NaCl, and 0.5% Triton X-100 (lanes S1, 10 g) was performed as detailed under "Experimental Procedures." Samples were analyzed by Western blot using the IQ Ab and PAb204 Ab, which both recognize P68-RNA helicase (P68). Standard molecular masses (in kDa) are indicated to the right and the positions of P100, P68-RNA helicase, and IQ85 are shown to the left of the blots .   FIG. 2. Partial purification of IQ85. IQ85 was partially purified from HEK-293 nuclear extract as described under "Experimental Procedures." Samples from the load (lanes S100, 15 g), flow-through fraction ( lanes FT, 15 g), elution fraction (lanes El, 15 g), and ammonium sulfate pellet (lanes AS, 5 g) were run on 8% SDS-PAGE, stained with Coomassie Blue (A), or transferred and immunoblotted with the IQ Ab (B). Molecular weight standards, IQ85, and P68 are indicated beside the blots. C, type A (6 g) and type B (3 g) ammonium sulfate pellets were analyzed by 8% SDS-PAGE. The major Coomassie Blue-stained proteins are indicated to the left, and molecular weight standards are indicated to the right. low binding for poly(A)-or poly(C)-Sepharose beads, whereas P60 interacted weakly with poly(A)-and poly(C)-Sepharose. P48 was only retained on poly(U)-Sepharose beads. These data indicate that IQ85 and co-purifying proteins are RNA-binding proteins.
Identification of IQ85, P98, P71, and P60 -To identify IQ85, the protein was subjected to peptide mapping using a modified Cleveland method (39). We sequenced three tryptic fragments from IQ85 (Fig. 4A), which corresponded exactly to sequences (32) found in human EWS. To confirm the sequencing data, IQ85 was immunoprecipitated with an antibody raised against the N-terminal domain of EWS (EWS 677 Ab), and Western analysis was carried out using the IQ Ab. The IQ Ab recognized the protein immunoprecipitated by the EWS antibody (Fig.  4B). Conversely, IQ85 immunoprecipitated with IQ Ab was also recognized by the EWS antibody. These data indicated that IQ85 is the prooncoprotein EWS, a putative RNA-binding protein expressed in normal cells (32,47,48). To identify other RNA-binding proteins that co-purified with EWS, we sequenced one V8 protease fragment from P98 (Fig. 4A). This sequence identified P98 as PSF, another RNA-binding protein that is essential for splicing of pre-mRNA (37,49). Furthermore, P98 was recognized by the ␣PSF antibody (37) and was present only in the type A fraction (Fig. 4C). In addition, P60 was recognized by a monoclonal antibody directed against p54 nrb and co-migrated with the human p54 nrb recombinant protein (Fig. 4C). p54 nrb is a protein first described as the human homologue of S. cerivisiae splicing factor PRP18; it shares 70% homology with PSF (50). We also assayed for other members of the EWS family in type A and B fractions using an antibody (R␣ TLS Ab) raised against the N-terminal domain of TLS/FUS (47). TLS/FUS is a prooncoprotein product of a gene that, like the EWS gene, can be translocated in human sarcomas (34,35). P71 was recognized by the TLS/FUS antibody (Fig. 4C). The co-purification of these other RNA-binding proteins with EWS is interesting but not unexpected, since the chromatography columns used select for RNA-binding proteins.
To determine if the IQ Ab recognizes EWS through the IQ domain described above (EWS-(259 -277)), we synthesized a peptide (pepEWS) corresponding to this sequence (Fig. 4A). We monitored the ability of the peptides pepEWS and pepP68 to suppress the immunoreactivity of the IQ Ab toward EWS, P68-RNA helicase, neuromodulin, and GST-neurogranin (GNg). An excess of pepP68 or pepEWS was preincubated with the IQ Ab prior to Western blotting. These peptides completely suppressed the reactivity of the IQ Ab toward EWS and P68- RNA helicase from HEK-293 nuclear extract (NE), EWS from the type B fraction, and purified rat neuromodulin (Fig. 5B). PepEWS and pepP68-RNA helicase also partially decreased the immunoreactivity of the IQ domain antibody against GSTneurogranin. These data indicate that the EWS amino acid sequence corresponding to pepEWS is the recognition site for the IQ Ab.
Phosphorylation of EWS by PKC-Because EWS contains the conserved PKC phosphorylation site found in other IQ domain proteins, we examined the phosphorylation of EWS by PKC. The phosphorylation of EWS by PKC was dependent on the presence of both Ca 2ϩ and phospholipids (Fig. 6A). Furthermore, bisindolylmaleimide, a specific inhibitor of PKC, totally inhibited phosphorylation of EWS. Interestingly, TLS/FUS and PSF were also phosphorylated by PKC in the presence of Ca 2ϩ and phospholipids, whereas p54 nrb and P48 were poor PKC substrates (Fig. 7B and data not shown).
An analysis of the kinetics for phosphorylation by PKC showed that phosphorylation of EWS resulted in a mobility shift on SDS-PAGE gels (Fig. 6B). Within 15 min, two bands with equal intensity were apparent on the Coomassie Bluestained gel, demonstrating that approximately 50% of the protein was phosphorylated. Furthermore, superposition of the autoradiogram with the Coomassie Blue-stained gel showed that only the upper band incorporated phosphate. These fractions were also analyzed by Western analysis using an antibody specific to EWS (EWS 677 Ab) and the IQ Ab (Fig. 6B). Although the EWS 677 Ab recognized both forms of EWS, the IQ Ab only recognized the unphosphorylated form of EWS. Furthermore, the peptide corresponding to the IQ domain of EWS, pepEWS, was phosphorylated by PKC in vitro (data not shown). The major PKC phosphorylation site of EWS is most likely Ser 266 within the putative IQ domain.
CaM Binding to EWS Inhibits PKC Phosphorylation-The interaction between EWS and CaM was analyzed using a CaM gel overlay assay (Fig. 7A). Electrotransferred proteins were incubated with biotinylated CaM in the presence or absence of Ca 2ϩ . EWS bound to CaM in the presence, but not in the absence, of Ca 2ϩ . TLS/FUS also interacted with CaM in the presence of Ca 2ϩ , but this interaction was apparently weaker. Since PKC phosphorylation of neuromodulin or neurogranin is inhibited by CaM (2, 5), we monitored the effect of CaM on EWS phosphorylation. If EWS binds CaM through its IQ domain, CaM should also inhibit phosphorylation of EWS by PKC. PKC phosphorylation of EWS was reduced by increasing concentrations of CaM with maximal inhibition at approximately 4.5 M CaM (Fig. 7, B and C). Calmodulin also inhibited PKC phosphorylation of PSF and TLS/FUS but did not affect autophosphorylation of PKC (data not shown). These data are consistent with the hypothesis that the major PKC phosphorylation site of EWS is within the IQ domain, which includes or overlaps with the CaM binding domain.
PKC Phosphorylation Inhibits Binding of RNA to EWS-To determine whether the phosphorylation of EWS modulates RNA binding, we first investigated the effect of soluble RNA homopolymers on the phosphorylation of EWS, PSF, and TLS/ FUS. Increasing concentrations of all RNA homopolymers inhibited the phosphorylation of all three proteins by PKC (Fig.  8A). Poly(U) was the strongest inhibitor of EWS phosphorylation (Fig. 8B). The RNA homopolymers did not affect phosphorylation of the GST-neurogranin fusion protein or the autophosphorylation of PKC (data not shown).
To determine whether phosphorylation of EWS by PKC affects RNA binding, proteins from the type A fraction were phosphorylated by PKC for 1 h in the presence or absence of bisindolylmaleimide, the PKC inhibitor. Phosphorylation of the type A fraction for 1 h induced a complete mobility shift of EWS on SDS gels (Fig. 9A, part a), indicating that it was totally phosphorylated by PKC. PKC phosphorylation was blocked by bisindolylmaleimide (Fig. 9A, part d). The same amount of protein was incubated with poly(G)-or poly(U)-Sepharose beads, which were then washed with 2 M urea in the presence of 50, 100, or 200 mM NaCl. Proteins bound to the poly(U)-or poly(G)-Sepharose beads were separated by SDS-PAGE (Fig.  9A, parts b and c), and phosphoproteins were detected after autoradiography (Fig. 9A, parts e and f). EWS had comparable affinity for poly(G) and poly(U) (Fig. 9A, parts b and c), whereas TLS/FUS had a stronger affinity for poly(G) and PSF for poly(U).
When EWS was phosphorylated by PKC (Fig. 9A, ϪbisInd) its affinity for poly(G) and poly(U) was significantly reduced compared with the unphosphorylated protein (ϩbisInd). This was particularly evident when interactions between poly(U) and phosphorylated or unphosphorylated EWS were compared with increasing salt concentration. At 100 mM NaCl, binding of phosphorylated EWS to poly(U)-Sepharose was completely inhibited, but unphosphorylated EWS was still absorbed (Fig.  9A, part c). Interactions between the RNA homopolymers and PSF as well as TLS/FUS were also lowered by PKC phosphorylation. Phosphorylated forms of EWS and PSF apparently have a lower affinity for poly(U) than do the unphosphorylated forms of these proteins. Phosphorylation of EWS, PSF, and TLS/FUS was also examined when these proteins were absorbed to RNA homopolymers (Fig. 9B). PKC phosphorylation of EWS, TLS/FUS, and PSF was significantly reduced when they were first bound to poly(U)-Sepharose beads. However, phosphorylation of p54 nrb was not affected by poly(U)-Sepharose beads. This suggests that there are two distinct interaction sites for poly(U) and poly(G) on these proteins and that poly U masks the phosphorylation site(s) of PKC. Alternatively, poly(U) binding may induce a conformational change that affects the ability of PKC to phosphorylate PSF, EWS, and TLS/ FUS. Collectively, these data suggest that PKC phosphorylation of EWS and interactions with RNA may be mutually exclusive.

DISCUSSION
The rapid redistribution of Ca 2ϩ throughout intracellular compartments, including the nucleus, is a general signaling mechanism for transfer of information conveyed by growth factors and hormones. Several nuclear events including DNA replication, DNA repair, and cell cycle progression are regulated by Ca 2ϩ through the activation and nuclear translocation of CaM or PKC isoforms (51)(52)(53)(54)(55)(56)(57)(58).
Since the IQ domain may be a general regulatory element for Ca 2ϩ signal transduction, it was important to determine if the nucleus contains IQ domain proteins other than P68-RNA helicase (26). In this study we show that the nuclear RNA-binding protein EWS contains an IQ domain and that its phosphorylation by PKC inhibits RNA binding. In addition, EWS copurifies with several other RNA-binding proteins that are also PKC substrates.
Sequence analysis indicates that EWS shares four common amino acids with the IQ domain of neurogranin, neuromodulin, and P68-RNA helicase, the residues 264 QXSFR. Since the IQ Ab immunoreacted only with the unphosphorylated form of EWS, the site of PKC phosphorylation is very likely Ser 266 , which is within the IQ domain. Ser 266 corresponds to the PKC phosphorylation sites in neuromodulin and neurogranin (2,6). Furthermore, EWS contains Phe 267 and Arg 268 , two amino acids adjacent to Ser 266 that are conserved in the IQ domain of neurogranin, neuromodulin, and P68-RNA helicase. Substitution of these residues with other amino acids greatly diminishes phosphorylation of neurogranin by PKC (4). Collectively, these data indicate that EWS contains an IQ domain with Ser 266 acting as the primary site for PKC phosphorylation.
The EWS IQ domain is located in the C-terminal end of the N-terminal domain of EWS (Fig. 10), which contains multiple copies of the sequence SYGQQS (32). The N-terminal domain was first described as the EWS domain that fuses with the DNA-binding domain of the ETS transcription factor FLi-1 after chromosomal translocation in Ewing's sarcoma (32). Depending on the breakpoint positions of genes coding for EWS or FLi-1, three different types of chimeric oncoproteins are produced (32,59). The IQ domain is lost in type 1 and type 2 chimeric oncoproteins, whereas it is retained in type 3 EWS/ FLi-1. The N-terminal domain acts as a transcriptional activator when it fuses with the Fli-1 DNA-binding domain of Fli-1 (60 -62). The fusion protein mediates the transformation activity of EWS/FLi-1, suggesting that the N-terminal domain interacts with the basal transcription machinery through protein-protein interactions (62,63). It would be interesting to determine if PKC or CaM regulate transcription and transformation activities of the type 3 EWS/FLi-1 through its IQ domain.
The inhibition of RNA binding to EWS by PKC phosphorylation is consistent with other data in the literature indicating that interactions between RNA and RNA-binding proteins are regulated by phosphorylation. For example, a dynamic phosphorylation/dephosphorylation cycle of C hnRNP protein modulates its binding to pre-mRNA (64). Recently, Municio et al. (49) discovered that the phosphorylation of hnRNPA1 by PKCseverely impairs its ability to bind to RNA through its RNAbinding consensus sequence. PKC phosphorylation of the IRPs RNA-binding proteins also regulates their affinity for RNA (65). Although we did not determine the mechanism for PKC inhibition of RNA binding to EWS, it is probably not due to direct phosphorylation of the RNA interaction site. Using truncated forms of EWS-b, an EWS spliced form (Fig. 10), Ohno et al. (48) showed that the full-length EWS-b binds to poly(G) and poly(U) homopolymers, whereas the C-terminal domain (CTD) of EWS-b only binds to poly(G) homopolymers. The N-terminal domain of EWS contains the PKC phosphorylation site, and it has no RNA binding activity of its own. Consequently, inhibition of RNA binding by PKC is not due to direct phosphorylation of the RNA binding site. PKC inhibition of EWS binding to RNA may reflect an indirect role of the N-terminal domain in regulating RNA binding.
Several lines of evidence classify EWS as a member of the hnRNP family, proteins that are associated with hnRNAs (including pre-mRNA) and are implicated in RNA metabolism (66,67). EWS shares structural homology with hnRNP proteins (Fig. 10) (32,68). Native EWS (Fig. 9) and the EWS-b spliced form (48) are RNA-binding proteins in vitro. EWS co-immunoprecipitates with hnRNPs A1, C (47), and TLS/FUS, a protein structurally and functionally related to EWS, has been identified as hnRNP P2 in the protein complex assembled on pre-mRNA (36). Furthermore, we demonstrated that EWS is associated with the nuclear matrix, a structure that contains many hnRNP proteins (44) and plays an important role in several nuclear functions including RNA processing (69 -71).
While purifying EWS, we found that it co-purifies with other RNA-binding proteins implicated in RNA metabolism including PSF, TLS/FUS, and p54 nrb . These basic proteins all contain Pro/Gln-rich sequences and RGG boxes, which have been implicated in the RNA binding activity (48,72). They share the same RNA-binding preference for poly(U) and poly(G) homopolymers and bind ssDNA-cellulose in a heparin-sensitive manner. Two of them, PSF and p54 nrb , bind to the polypyrimidine tract of mammalian introns (sequence rich in uridine bases) and are implicated in splicing of pre-mRNA. These observations suggest that EWS and TLS/FUS may play a role in the post-transcriptional modifications of pre-mRNA.
The nuclear matrix contains CaM and PKC␣ (31, 73, 74) as well as EWS and other RNA-binding proteins whose interactions with RNA may be regulated by PKC. Our data support the general hypothesis that Ca 2ϩ signal transduction pathways may control the activity of RNA-binding proteins through PKC or CaM. The presence of IQ domains in two RNA binding proteins, EWS and p68 RNA helicase, raises the interesting possibility that this domain may coordinate general regulation of RNA processing by PKC or CaM.