Alternative Splicing Determines the Intracellular Localization of the Novel Nuclear Protein Nop30 and Its Interaction with the Splicing Factor SRp30c*

We report on the molecular cloning of a novel human cDNA by its interaction with the splicing factor SRp30c in a yeast two-hybrid screen. This cDNA is predominantly expressed in muscle and encodes a protein that is present in the nucleoplasm and concentrated in nucleoli. It was therefore termed Nop30 (nucleolar protein of 30 kDa). We have also identified a related cDNA with a different carboxyl terminus. Sequencing of the NOP gene demonstrated that both cDNAs are generated by alternative 5′ splice site usage from a single gene that consists of four exons, spans at least 1800 nucleotides, and is located on chromosome 16q21-q23. The alternative 5′ splice site usage introduces a frameshift creating two different carboxyl termini. The carboxyl terminus of Nop30 is rich in serines and arginines and has been found to target the protein into the nucleus, whereas its isoform is characterized by proline/glutamic acid dipeptides in its carboxyl terminus and is predominantly found in the cytosol. Interaction studies in yeast, in vitro protein interaction assays, and co-immunoprecipitations demonstrated that Nop30 multimerizes and binds to the RS domain of SRp30c but not to other splicing factors tested. Overexpression of Nop30 changes alternative exon usage in preprotachykinin and SRp20 reporter genes, suggesting that Nop30 influences alternative splice site selection in vivo.

We report on the molecular cloning of a novel human cDNA by its interaction with the splicing factor SRp30c in a yeast two-hybrid screen. This cDNA is predominantly expressed in muscle and encodes a protein that is present in the nucleoplasm and concentrated in nucleoli. It was therefore termed Nop30 (nucleolar protein of 30 kDa). We have also identified a related cDNA with a different carboxyl terminus. Sequencing of the NOP gene demonstrated that both cDNAs are generated by alternative 5 splice site usage from a single gene that consists of four exons, spans at least 1800 nucleotides, and is located on chromosome 16q21-q23. The alternative 5 splice site usage introduces a frameshift creating two different carboxyl termini. The carboxyl terminus of Nop30 is rich in serines and arginines and has been found to target the protein into the nucleus, whereas its isoform is characterized by proline/glutamic acid dipeptides in its carboxyl terminus and is predominantly found in the cytosol. Interaction studies in yeast, in vitro protein interaction assays, and co-immunoprecipitations demonstrated that Nop30 multimerizes and binds to the RS domain of SRp30c but not to other splicing factors tested. Overexpression of Nop30 changes alternative exon usage in preprotachykinin and SRp20 reporter genes, suggesting that Nop30 influences alternative splice site selection in vivo.
In eukaryotes, gene expression is controlled at several levels. Due to the presence of cell-specific factors, genes can be transcribed in a cell type or developmentally regulated manner. Moreover, primary transcripts undergo maturation processes such as pre-mRNA splicing (1) and capping. A growing number of proteins containing an N-terminal RNA recognition motif and C-terminal clusters of serines and arginines (SR proteins) 1 have been shown to be involved in constitutive and alternative splicing (2,3). Using the family member SRp30c (4) as an interactor in a two-hybrid screen, we have previously shown that the matrix attachment region element-binding protein SAF-B has the potential to link matrix attachment region elements, RNA polymerase II, and SR proteins (5). This supports growing evidence that gene expression achieved through RNA biosynthesis, RNA processing, and RNA transport is a highly coordinated process mediated by a network of proteins that has been termed the RNA "factory" (6) or transcriptosomal complex (7). For example, it has been shown that the carboxylterminal domain of RNA polymerase II binds to proteins involved in pre-mRNA processing (8,9) and that 3Ј processing of pre-mRNA is coupled to components of the splicing machinery (6).
RNA processing is not just confined to polymerase II transcripts, since rRNA generated by polymerase I undergoes maturation that removes externally and internally transcribed spacers in various steps to yield 28, 18, and 5.8 S ribosomal RNA in mammalian cells (10). This process is confined to nucleoli, the sites of ribosome biosynthesis. Three morphologically distinct areas of the nucleolus can be distinguished: the nucleolar fibrillar center, the dense fibrillar region, and the granular region. Functional studies have indicated that the fibrillar center is the region where ribosomal RNA genes, RNA polymerase I, and transcription factors are localized. The dense fibrillar region consists of unprocessed nascent ribosomal RNA and associated proteins. Mature 18 and 28 S rRNA together with intermediates in ribosome assembly are found in the granular region. Thus, the morphological and functional compartmentalization of the nucleolus reflects specialized areas involved in rRNA synthesis, processing, and export (11,12).
We are currently using yeast two-hybrid screens to identify novel proteins involved in RNA processing. Employing the splicing factor SRp30c (4) as a bait to screen a human library, we identified a cDNA coding for a nuclear protein that is enriched in nucleoli and migrates around 30 kDa on SDSpolyacrylamide gel electrophoresis. We therefore termed this protein Nop30. We identified an isoform of Nop30, which we call Myp. Myp has been previously isolated from rats in a screen aimed to find genes induced by nerve growth factor (13). Analysis of the NOP gene revealed that both isoforms are generated by alternative splicing. In contrast to the nuclear localization of Nop30, its isoform Myp is found in the cytosol. Overexpression of Nop30 with a cotransfected SRp20 or preprotachykinin minigene (14) affects splicing of the corresponding alternative exons in the reporter genes, suggesting a potential role of Nop30 in pre-mRNA splicing. Since Nop30 binds specifically to SRp30c and its mRNA is highly expressed in muscle, we suggest that this protein acts as a regulator of SR protein function in muscle. * This work was supported by the Max Planck Society, Human Frontier Science Program Grant RG562/96 (to S. S.), and National Institutes of Health Grant HL45565 (to T. A. C.). 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 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF064598, AF064599, and AF064600.

EXPERIMENTAL PROCEDURES
Two-hybrid Screening-A yeast two-hybrid screen using SRp30c as a bait in pGBT9 and a HeLa library was performed as described by Fields and Song (15). DNA from the autotrophic bacteria was sequenced using an ABI sequencer and analyzed with the GCG Wisconsin package (16). To test the interaction between proteins, 1 g of a prey construct fused to the Gal4 activation domain in pADGal4 and 1 g of bait-construct fused to the Gal4 binding domain were simultaneously co-transformed and plated onto double dropout plates lacking leucine and tryptophan. Surviving colonies were restreaked onto triple dropout plates lacking leucine, tryptophan, and histidine and onto triple dropout plates supplemented with 5 and 10 mM 3-aminotriazole (Sigma).
Genomic Cloning-Using the cDNA of Nop30 as a probe, a human BAC library was screened (kindly performed by the German Human Genome Project). Two PstI fragments containing the complete clone were subcloned in pBluescript-SK(ϩ) (Stratagene), and positive clones were identified by colony hybridization (17).
Computer Analysis-Calculation of splice site scores was as described by Stamm et al. (18). 2 Chromosomal Localization-Chromosomal localization was performed with the Stanford radiation hybrid panel 4.0 (Research Genetics Inc.). For PCR, the following primers were used: PEintrof1 (5Ј-GGTC-CGGGTGAGCGCGCGGG-3Ј; located in intron B) and PEint1 (5Ј-GAGATGACGGGAACAGTGGTCAAAG-3Ј; located in intron C). For data processing and mapping, World Wide Web servers were used. 3 Northern Blot-Nop30 full-length cDNA, SRp30c cDNA, and a ␤-actin probe (CLONTECH) were labeled with radioactive [␥-32 P]dCTP using the megaprime DNA labeling system from Amersham Pharmacia Biotech. A human multiple tissue Northern blot (CLONTECH) was probed according to the manufacturer's instructions.
In Situ Hybridization-0.2 ng of the cloned Nop30 probe was PCRamplified using primers specific for Nop30 anchored by either the T7 or Sp6 RNA polymerase recognition sequence and purified using QiaQuick (Qiagen). Transcription using either T7 or Sp6 RNA polymerase was done according to the manufacturer's protocol (Boehringer Mannheim) using 250 ng of the respective PCR product as template. In situ hybridization histochemistry and the following washing steps were performed as described earlier (19). 20-m-thick dried cryostat sections were incubated with 2 ϫ 10 6 cpm of the [ 35 S]UTP (Amersham Pharmacia Biotech)-labeled probe. After the washing steps, sections were dehydrated through graded ethanols containing 0.3 M ammonium acetate, air-dried, covered with a thin sheet of film emulsion (NBT-2, Eastman Kodak Co.), and exposed at 4°C for 7 days. They were developed in Kodak D19 developer and fixed in 24% sodium thiosulfate. Sections were counterstained with Cresyl Violet, dehydrated, and mounted in DePeX (Gurr).
Immunoprecipitation and Western Blot-For immunoprecipitation, human embryonic kidney (HEK) 293 cells were grown to 50% confluency and transfected with 0.1-g DNA constructs as indicated. Immunoprecipitation of the EGFP constructs was performed using an anti-GFP antibody (Boehringer Mannheim) (5). The co-precipitating FLAG constructs were analyzed after 10 -15% SDS-polyacrylamide electrophoresis and Western blotting using an anti-FLAG antibody (Santa Cruz Biotechnology). SDS-polyacrylamide gel electrophoresis was performed as described by Laemmli (24), and protein was transferred onto ECL membranes (Amersham Pharmacia Biotech). Blocking of the membranes and protein detection was performed as described by Nayler et al. (23).
Splicing Assays-Transient transfection and RT-PCR was performed as described previously (5). The indicated amounts of pNop30-C2, pEGFP-C2 (CLONTECH) and pCG35 were co-transfected with 2 g of either the SRp20 (14) or PPT minigene in HEK293 cells or in fibroblasts, respectively. The next day, the transfection rate was assessed by fluorescence microscopy, and RNA was isolated from equally transfected cultures using the RNeasy Kit (Qiagen). For the SRp20 minigene, 30 cycles were used with annealing and extension temperatures of 55 and 72°C with the primer pair T7 (5Ј-TAATACGACTCACTATAGGG-3Ј) and X16R (5Ј-CCTGGTCGACACTCTAGATTTCCTTTCATTTGAC-C-3Ј). PCR products were resolved on a 1.5% agarose gel, stained with ethidium bromide, and quantified using the Herolab EASY system. For the PPT minigene, reverse transcription followed by radioactive PCR using the primers T7PRO (5Ј-TAATACGACTCACTATA-3Ј) and PPT-E5rev (5Ј-GTGAGAGATCTGACCATGCC-3Ј) with 18 cycles was performed using annealing and extension temperatures of 70 and 72°C, respectively. PCR products were resolved on a 5% nondenaturing polyacrylamide gel. Bands were quantified directly from the gel using a phosphor imager. The percentage of inclusion is the percentage of spliced RNA that contains the exon (cpm of inclusion band/(cpm of inclusion band ϩ cpm of exclusion band) ϫ 100.

RESULTS
Sequence-In order to find new proteins involved in RNA processing, we performed yeast two-hybrid screens with SRp30c (4) and identified a clone from a HeLa library, which we named Nop30. Its sequence is shown in Fig. 1A. The cDNA of Nop30 encodes a protein with a calculated molecular mass of 24.3 kDa. The protein contains a highly acidic N terminus (amino acids 1-95) with an isoelectric point of 4.9 and a basic C terminus, enriched with arginines, serines, and prolines (amino acids 96 -124) (pI ϭ 12.7). The total protein is highly basic (pI 11.7) (Fig. 1B). Probably due to the bipolar charge distribution, it has an apparent molecular mass of 30 kDa upon SDS gel electrophoresis. Inspection of the sequence shows that it contains a potential cAMP-dependent protein kinase phosphorylation site (amino acids 197-200), several potential protein kinase C phosphorylation sites, and a potential casein kinase II phosphorylation site (amino acids [41][42][43][44]. The casein kinase II site may be functional, since recombinant protein can be phosphorylated by casein kinase II in vitro (data not shown). Due to the presence of arginine and serine residues (Fig. 1D), the C terminus of Nop30 shows homology to some SR and SR-related proteins (human protein kinase CLK3 (23) and splicing factors 9G8, SRp55, SRp75, and SC35 (2)).
Data base searches also showed that the first 95 amino acids of Nop30 share 94% homology to a rat protein with an as yet unknown function that was previously found in a screen aimed to identify genes induced with nerve growth factor (13). However, due to a frameshift, the carboxyl domains encoded by the two cDNAs are completely different. We searched the expressed sequence tag data base (25) and identified a human clone (accession number AA085275) corresponding to the rat cDNA. This expressed sequence tag clone was entirely sequenced and found to encode the human homologue of the rat cDNA clone described by Geertman et al. (13). The carboxyl terminus of the encoded protein is characterized by clusters of prolines and glutamic acids (Fig. 1, A and B). We named this protein Myp (muscle-enriched cytosolic protein) because of its expression pattern and intracellular localization (see below). In addition to the different C terminus caused by the frameshift, this clone differs from our two-hybrid isolate by the use of a different polyadenylation site (PS-2, Fig. 1A).
Since both human sequences differ by only ten nucleotides in their coding region, they could derive from alternative splicing. We therefore screened a human BAC library and obtained a genomic NOP clone. Comparison of its sequence with Nop30 and its variant revealed that the gene is composed of four exons and three introns and generates the two isoforms by using an alternative 5Ј splice site in exon 2 (Fig. 1, A and C). The size of the gene is at least 1800 nucleotides. The same first start codon present in all expressed sequence tags, our two-hybrid isolate, and the rat cDNA is located in exon II. In contrast to rat, the start codon in humans deviates slightly from the Kozak consensus sequence (26). A remarkable feature of the gene is the small size of its introns (115 nucleotides for intron A, 155/165 nucleotides for intron B, and 102 nucleotides for intron C). This is significantly smaller than mammalian internal introns, which average around 1127 nucleotides (27). All intron/exon boundaries are in agreement with the mammalian consensus sequence. However, calculation of the splice site scores revealed a difference in the alternatively used 5Ј splice site of intron B. The splice site generating Nop30 (score 5.3) deviates from the average score of constitutive exons (score 8.1), which is typical for a subset of alternatively spliced exons (18). In contrast, the score of the variant splice site (score 7.9) is close to the average score of constitutive exons (score 8.1).
Using the Stanford radiation hybrid panel with the primers PEint1 and PEintrof1, we mapped the NOP gene to human chromosome 16. The NOP gene was found 3.15 cR from WI-5594, between markers WI-5594 and WI-9392 (Fig. 2). This most likely corresponds to the region between 81 and 84 cM on chromosome 16 (q21 to q23). In summary, we have shown that the NOP gene is located on chromosome 16q and generates two isoforms via an alternative splicing mechanism.
Expression-We next determined the expression of NOP and performed Northern blot analysis with various human tissues using the full-length cDNA fragment of Nop30 as a probe. Nop30 mRNA shows the highest expression in heart and skeletal muscle (Fig. 3A). However, a faint band is visible in other tissues examined. The two bands of 1.8 and 1.3 kb most likely correspond to the two polyadenylation sites found in the NOP gene (Fig. 1A). These expression data are in agreement with the tissue distribution found for the rat cDNA (13). The 1.8-kb band corresponds to the size of the longest cDNA clone, indicating that we have obtained the full-length Nop30 transcripts. By probing mouse tissue with a human cDNA probe, it was shown that SRp30c is predominantly expressed in kidney, FIG. 3. Northern blot analysis of Nop30 expression. A, Northern blot analysis of NOP transcripts on various human tissues (CLON-TECH) using a full-length Nop30 probe. The signals at 1.8 and 1.3 kb in heart and skeletal muscle correspond to NOP transcripts processed at the two polyadenylation sites. The faint band detected in nonmuscle tissues is indicated by a triangle. B, the same filer was rehybridized with a human SRp30c full-length probe. The highest expression was detected in pancreas, followed by kidney, placenta, and heart. C, the same filer was rehybridized with a ␤-actin control probe. The cytoskeletal ␤-actin isoform is detected at 2.0 kb. The probe also hybridizes to the muscle-specific ␣ and ␥ isoforms (1.8 kb) in heart and muscle. lung, and spleen, whereas only low levels were detected in muscle (4). This expression pattern would make an in vivo interaction between SRp30c and Nop30 questionable. We therefore reprobed our filter with human SRp30c and found considerable SRp30c expression in human striated muscle as well as in other tissues (Fig. 3B), suggesting that SRp30c is a possible interactor for Nop30 in muscle. The same blot was reprobed with a ␤-actin control (CLONTECH) to verify that equivalent amounts of RNA were loaded. In heart and skeletal muscle, the additional muscle-specific actin isoforms at 1.8 kb were detected (28).
We noticed a faint signal present in nonmuscle tissues using Nop30 as a probe (Fig. 3A) and compared Nop30 mRNA expression in heart and brain using in situ hybridization. We used a rat Nop30 cDNA fragment containing the whole coding region, except for the C-terminal repetitive sequence, as a probe. The Nop30 antisense probe gave a strong signal after in situ hybridization in heart (Fig. 4A). In brain, a specific signal was only seen in the pia mater (Fig. 4C). The pia mater is of mesenchymal origin, surrounds the brain (Fig. 4D), and contains blood vessels lined with smooth muscle cells (29) . In contrast, the signal derived from neuronal or glial cells did not exceed the background. No signal could be detected using a sense control probe (data not shown). In summary, Northern blot analysis and in situ hybridization demonstrate that Nop30 shows the highest expression in muscle cells. Moreover, using RT-PCR, we did not detect a change in expression during C2C12 cell differentiation into myoblasts or during embryonic development, suggesting that Nop30 mRNA and its isoform are constitutive parts of muscle cells in various stages of differentiation and development (data not shown).
Interaction with Other Proteins-In order to obtain information on the possible function of Nop30 and its isoform, we tested its interaction with other proteins in the yeast two-hybrid system. Of all proteins involved in pre-mRNA metabolism tested (SRp20, SF2/ASF, SC35 (2), SmN (30), htra-beta1 (21), SRp30c, SRp75, SRp55, (4) SAF-B (5), U2AF35 (31), and the carboxyl-terminal domain of RNA polymerase II (8)), only SRp30c and Nop30 itself showed interaction in yeast (Fig. 5A). Interestingly, its splice variant Myp did not interact with SRp30c and reacted only weakly with Nop30. This indicated that the different carboxyl termini of the two splice variants are responsible for differences in protein-protein interactions. We therefore tested them separately. The deletion constructs used are shown schematically in Fig. 5B. We found that deletion of the N terminus of Nop30 abolishes binding to Nop30, whereas a clone with a deleted C terminus was still able to interact. In addition, the two C-terminal Nop30 deletion variants (Nop30⌬C) were also found to interact. Thus, Nop30 dimerizes or multimerizes via its N terminus. In contrast, deleting either the C or N-terminal part of Nop30 resulted in a lack of interaction with SRp30c. In addition, deletion of the RS domain of SRp30c prevented binding of SRp30c to Nop30 (Fig.  5A). We conclude from this analysis that Nop30 interacts with the RS domain of SRp30c, using parts of both its N-and C-terminal domain.
In order to verify the two-hybrid interactions biochemically, Nop30 was translated and incubated in vitro with glutathione-Sepharose, GST, GST-SRp30c, and GST-ASF/SF2 (Fig. 6). In this assay, Nop30 bound only to SRp30c, but not to SF2/ASF. This confirmed our two-hybrid analysis and indicated that Nop30 can discriminate between the related RS domains of SF2/ASF and SRp30c.
Next, we tested Nop30 interaction under more physiological conditions in immunoprecipitation assays. We fused Nop30, Myp, SRp30c, and their deletion variants previously analyzed in yeast (Fig. 5B) to EGFP and co-transfected these constructs with FLAG-tagged Nop30, SRp30c, and ASF/SF2 in HEK293 cells. Complexes were precipitated with a monoclonal anti-EGFP antibody, and interacting proteins were detected with an anti-FLAG antibody. As shown in Fig. 7A, Nop30 bound to itself, more weakly to its amino terminus (Nop30⌬C), and to SRp30c. As in yeast, there was no detectable interaction between EGFP-Nop30 and FLAG-SF2/ASF. The variant EGFP-Myp did not bind to FLAG-SRp30c, and EGFP alone did not interact with FLAG-Nop30 and FLAG-SRp30c (Fig. 7A, lanes 9  and 10). Despite inclusion of protease inhibitors, a nonspecific background was repeatedly observed in our experiments (Fig.  7A, lanes 3, 5, 6, and 7), which might be due to proteolytic fragmentation of the antibody used in the immunoprecipitation. Fig. 7B demonstrates that all EGFP-tagged constructs were immunoprecipitated with the anti-GFP antibody and mi- FIG. 4. In situ analysis of Nop30 expression in heart and brain. A, dark field picture of rat heart tissue probed with a rat Nop30 antisense cRNA. B, cresyl violet counterstain of the same area as in A). The blood vessel present in this area is indicated. C, dark field picture of rat brain probed with Nop30 antisense cRNA. The signal is located in the pia mater. No tissue is present between the top of this section and the pia mater. The signal there represents the backgound. D, cresyl violet counterstain of the same area as in C.
grated with the expected size. However, there is a faint band below some GFP signals, which again might be due to degradation. Similarly, all FLAG constructs were expressed with the predicted size (Fig. 7C).
Taken together, our binding experiments indicate that Nop30 binds to itself and to SRp30c. Binding to SRp30c is RS domain-dependent, and Nop30 is able to discriminate between the RS domain of SRp30c and other SR proteins, which implies that the binding is due to protein-protein interaction and is not RNA-mediated. In addition, we were unable to detect binding of EGFP-Nop30 to RNA or DNA in pull down experiments (data not shown).
Intracellular Localization-Since the two splice variants of NOP had different affinities toward nuclear proteins, we determined their intracellular localization. First, we tagged Nop30 with EGFP and expressed the fusion proteins in BHK and HEK293 cells. As shown in Fig. 8, EGFP-Nop30 is expressed in both cell lines in the nucleus, where it is concentrated in the nucleoli, but is also present in the nucleoplasm (Fig. 8, A-C, BHK cells; Fig. 8, D-F, HEK 293 cells). The staining in the nucleoli is not uniform, but it shows several small foci (Fig. 8A, arrow) or one large one (Fig. 8G, arrow). Untagged EGFP is located throughout BHK or HEK293 cells (Fig. 8, H and I). To verify that the structures identified in phase contrast were nucleoli, we performed costaining with an anti-nucleolus antibody. These experiments confirmed the predominant nucleolar localization of Nop30 (Fig. 9, A-C).
Three morphologically distinct areas of the nucleolus were defined on the ultrastructural level. It is possible to distinguish them by immunocytochemistry and confocal microscopy, using markers that have been shown to be restricted to certain nucleolar regions by correlating light and electron microscopy  23,000 cpm of radiolabeled Nop30 was incubated with 1 g of GST, GST-SRp30c, or GST-SF2/ASF bound to glutathione-Sepharose 4B. Ϫ, no protein bound. 30% of Nop30 were specifically retained by GST-SRp30c. (32)(33)(34). We therefore performed localization experiments with the nucleolar protein B23, which is a marker for the granular component of nucleoli (34). As shown in Fig. 9, D-F, Nop30 and B23 colocalize in the granular component of nucleoli. It is also evident that the major part of Nop30 localizes in the area between the foci and the granular component, which is likely to be the fibrillar component according to the classical organization of the nucleolus (35).
To determine whether an interaction between Nop30 and SRp30c can take place in vivo, we analyzed the localization of EGFP-Nop30 and FLAG-SRp30c (Fig. 9, G-I) using an anti-FLAG antibody. SRp30c does not localize in nucleoli, but both proteins were detected in the nucleoplasm, indicating that the interaction between both proteins could occur in vivo.
We next tested Myp, the splice variant of Nop30. Transfection of the EGFP-tagged variant revealed that it is predominantly located in the cytoplasm. Only faint staining could be detected in the nucleus, and no staining was detected in nucleoli (Fig. 10A). Therefore, we named this protein Myp (muscleenriched cytoplasmic protein). Since the intracellular localization of Myp and Nop30 is different, we tested their common N terminus and different C termini separately. The common N terminus gave a predominant cytoplasmic staining pattern resembling Myp (Fig. 10C). Transfection of cells with the C terminus of Nop30 resulted in nuclear staining, with predominant staining in the nucleoli similar to that of full-length Nop30 (Fig. 10D). In contrast, when we tested the C terminus of Myp, we observed staining throughout the cell, excluding the nucleoli (Fig. 10B). The integrity of various EGFP constructs used for fluorescence was tested by Western blot (Fig. 10E), demonstrating that all constructs expressed protein of the expected size.
We conclude from these experiments that the two splice variants generated by the NOP gene have a different intracellular localization; Nop30 is a nuclear protein concentrated in the nucleoli, and the variant Myp is predominantly localized in the cytoplasm. The localization to nucleoli is an intrinsic property of the basic arginine-rich C terminus of Nop30, whereas the acidic N terminus seems to exclude the protein from the nucleus.
In Vivo Splicing-We next asked whether the binding properties of Nop30 to SRp30c are relevant in vivo. Since concentration changes of SR proteins modulate alternative splicing patterns in vivo (3,36,37), we reasoned that overexpression of Nop30 would change the SR protein composition through interaction with SRp30c, which would change splicing patterns of a reporter gene. First, we employed the SRp20 XB minigene. Using this construct, it has been shown that SRp20 can promote inclusion of the alternative exon 4 in its own mRNA, whereas SF2/ASF has the opposite effect (14). We examined the effect of Nop30 on exon 4 usage (Fig. 11A) and cotransfected the SRp20 XB minigene with an increasing ratio of pNop30-C2 to the parental vector pEGFP-C2 (Fig. 11, A-C, top). With increasing amounts of pNop30-C2, we detected a decrease of exon 4 inclusion from ϳ34 to 10% (Fig. 11A). In order to compare the effect of SRp30c on this minigene, we performed a similar experiment and found that EGFP-SRp30c also decreases exon 4 usage (Fig. 11B). However, less EGFP-SRp30c cDNA needs to be transfected to reduce exon 4 usage. To examine the combined effect of SRp30c and Nop30 on exon 4 splicing, we performed cotransfection experiments in which increasing amounts of pNop30c were added to a constant amount of SRp20 XB minigene and SRp30c (Fig. 11C). We found that, in the presence of EGFP-SRp30c, less pNop30-C2 was necessary to induce a visible decrease in exon 4 usage. Transfection of 5 g of pNop30-C2 abolished inclusion of the alternative exon (Fig.  11C), whereas this amount of pNop30-C2 did not change exon 4 usage significantly in the absence of SRp30c (Fig. 11A). Overexpression of EGFP alone (first lane in each experiment) did not induce exon skipping. To test the effect of Nop30 in another system, we used a PPT minigene (38). As shown in Fig.  11D, cotransfection of 3 and 10 g of pNop30-C2 promotes alternative exon inclusion in the PPT system, again indicating that concentration changes of Nop30 changes alternative exon usage in vivo. In contrast, overexpression of the splicing factor SC35 had no significant effect on exon 4 inclusion. We conclude from these experiments that Nop30 can influence alternative splicing in vivo. DISCUSSION We report on the molecular cloning and characterization of the NOP gene and its two products, which are generated by alternative splicing. Northern blot analysis indicates that the gene is predominantly expressed in muscle cells. In order to investigate a weak signal present in all tissues examined, we performed in situ hybridization with rat brain sections and found that the NOP gene is expressed in the pia mater, indicating a specific expression pattern. Therefore, we assume that the faint signal detected in nonmuscle tissues with Northern blot analysis might result from muscle-derived cells that line blood vessels in all tissues. Therefore, it is possible that Nop30 expression is muscle-specific. However, from our data we cannot exclude a weak basal NOP expression in other tissues. The NOP gene is located on chromosome 16 between 16q21 and 16q23. Although only a few genes are identified in this region, it is interesting that a 16q23.1 deletion leads to disorders in musculoskeletal systems (39). Further analysis will be necessary to investigate whether Nop30 is involved in one of these disorders.
The Nop30 cDNA was obtained in a two-hybrid screen for proteins that interact with SRp30c. Using the three independent methods of yeast two-hybrid assays, GST pull down exper-iments and immunoprecipitations, we were able to demonstrate that Nop30 protein can bind to itself and to SRp30c. The multimerization of Nop30 protein depends on its N terminus. The interaction with SRp30c is dependent on the RS domain of SRp30c and both N-and C-terminal parts of Nop30. Our results demonstrate that protein-protein interactions mediated by RS domains can be extremely specific. In order to test whether an interaction of Nop30 with SRp30c is possible in vivo, we determined the distribution of SRp30c in human tissues. We found that human SRp30c is found in all tissues, including muscle. Furthermore, we show that both SRp30c and Nop30 colocalized in the nucleoplasm (Fig. 9, G-I). Taking the expression and localization experiments together, it is likely that Nop30 interacts with SRp30c in the nucleoplasm of muscle cells.
Alternative splicing is a commonly used mechanism to create protein isoforms (1,40). A usual feature of alternatively used exons is the introduction of stop codons or frameshifts (18) that can either switch off genes (41) or create proteins with different C termini, which can result in a different intracellular localization. For example, the nuclear localization of the estrogen receptor (42) and the transcription factors E2F (43), p105 NF-B (44), and a protein-tyrosine phosphatase (45) as well as the intracellular localization of the NADH-cytochrome b 5 reductase (46) are controlled by alternative splicing. In the NOP gene, the intracellular localization of the gene products is dependent on alternative 5Ј splice site usage, causing a frameshift that creates two different carboxyl termini. Nop30 is found in the nucleoplasm and is concentrated in the nucleoli, whereas its variant Myp is distributed throughout the cytosol. Myp did not interact with SRp30c (Fig. 5A), indicating that alternative splicing determines the binding specificity of the NOP gene products. Therefore, in muscle, the amount of the interactor of the splicing factor SRp30c could be regulated by alternative splicing. This is reminiscent of other proteins involved in pre-mRNA splicing that are regulated by alternative splicing. For example, in mammals, splice variants of SRp55, SRp40 (4,47), and ASF/SF-2 (48), and the mammalian homologue of SWAP (49) and transformer-2 (21) have been described. In the case of transformer-2 (50), SWAP (51), and X16/SRp20 (14), the alternative splicing event seems to be autoregulated.
Nucleoli vary in size and shape, reflecting the cellular activity (52,53). In addition, nucleoli of transformed cells have been reported to be extremely irregular in number, shape, and size (54,55), as can also be seen in our Figs. 8 and 9. No significant difference between nucleoli of transfected and untransfected cells was observed by bright field and fluorescence microscopy using the B23 antibody (Fig. 9F), indicating that Nop30 overexpression does not change the nucleolar shape. Nop30 protein expression is restricted to specified areas in the nucleolus. Colocalization with B23 revealed that Nop30 is present in the granular component, where the later stages of ribosome bio-synthesis occur. In addition, it is detected in the more central region that could be the dense fibrillar component. Furthermore, we noticed that, based on their size, number, and shape (56,57), Nop30 is absent from discrete foci that are most likely the fibrillar centers.
The function of Nop30 proteins has yet to be demonstrated directly. Colocalization and binding to SRp30c suggest a role of Nop30 in pre-mRNA processing. Indeed, we found an effect of Nop30 on splicing of a PPT and a SRp20 reporter gene in vivo. As with all mammalian factors involved in pre-mRNA processing, the specific target gene is unknown (3). Nop30 has a different effect on the two minigenes tested; it promotes exon inclusion in the PPT minigene but causes exon skipping when tested with the SRp20 construct. This is reminiscent of opposing effects of splicing factors on natural minigenes in vivo. For example, SF2/ASF promotes exon EN inclusion when tested with the clathrin light chain B minigene (36) but causes exon 4 skipping with the SRp20 minigene (14).
The predominant nucleolar localization of Nop30 also points to a potential role in RNA processing in muscle nucleoli. There are several examples of intimate connections between nuclear compartments. For example, in S. cerevisiae rRNA processing and pre-mRNA splicing are coupled via the ribosomal protein L32, which influences both splicing of its own transcript and the processing of rRNA (58). Another link between splicing and nucleolus comes from the observation that the transcript of the mouse ribosomal protein S24 gene is alternatively spliced (59). Therefore, it seems likely that Nop30 has a dual role in splicing as well as in the nucleolus.
Several studies in vivo (3,36,37,60) and in vitro (3,4,48,61,62) have shown that SR protein concentration can influence alternative splice site selection. Specifically, an increase in the nuclear concentration of a transiently overexpressed SR protein changes alternative exon usage. Tissue-specific differences in the ratio of SR proteins and hnRNP A1 suggest a role in tissue-specific splicing (63), and alterations in the levels of individual SR proteins have been associated with changes in CD44 alternative splicing during T cell activation (4). Therefore, SR proteins are likely to be involved in tissue-specific regulation of alternative splicing. Due to its binding properties to SRp30c, its ability to influence splice site decisions, and its high expression in muscle, it is possible that Nop30 changes the free concentration of SRp30c by specifically sequestering this SR protein in muscle cells. This muscle-specific sequestration would change the SR protein composition in a cell type-specific way and might indirectly lead to muscle-specific alternative splicing decisions. Alternatively, similar to results obtained with SRm160 (64), it is possible that Nop30 enhances effects of SR proteins on certain genes, which would be consistent with the synergistic effect of Nop30 and SRp30c on exon 4 of the SRp20 minigene (Fig. 11C). It remains to be seen whether other tissue-specific proteins exist that can discriminate between different SR proteins and might influence alternative splicing. FIG. 11. In vivo splicing assays. A, a total amount of 10 g of an increasing ratio of pNop30-C2 (encoding EGFP-Nop30) to empty vector pEGFP-C2 was cotransfected with 2 g of the SRp20 XB minigene in HEK293 cells and analyzed by RT-PCR. The amount of transfected EGFP constructs is indicated at the top of the gel. PCR products were stained with ethidium bromide and quantified. The size difference between the upper and the lower band corresponds to the size of alternative exon 4 (435 nucleotides). The structure of the PCR products is indicated schematically on the right. Primers used are indicated as arrows. Exon 4 inclusion was calculated as follows: (amount of inclusion band/(amount of inclusion band ϩ amount of exclusion band) ϫ 100. The percentage of exon inclusion is shown under each gel. Error bars represent the S.D. from three independent experiments. M, pBR322 MspI digest. B, a similar experiment as in A was performed with pSRp30c-C2 instead of pNop30-C2. C, in vivo splicing was investigated using a total amount of 10 g of a mixture of three different EGFP constructs (pNop30-C2, pSRp30c-C2, and pEGFP-C2) and 2 g of the SRp20 minigene. The amount of transfected pSRp30c-C2 was 5 g in every case, whereas 5 g of an increasing ratio of pNop30-C2 (encoding EGFP-Nop30) to empty vector (pEGFP-C2) were cotransfected. D, chicken fibroblast cells were cotransfected with 2 g of PPT minigene and 0, 3, and 10 g of either pNop30-C2 or pCG35 (expressing SC35). Control, only the PPT minigene was transfected. RNA was isolated, and splicing of the minigene was analyzed by radioactive RT-PCR with the primer pair T7PRO and PPTE5rev. The RT-PCR products containing exon 4 were separated from those lacking exon 4 on a 5% polyacrylamide gel. The quantification of three independent experiments is shown. Exon inclusion was calculated as follows: (cpm of inclusion band/(cpm of inclusion band ϩ cpm of exclusion band) ϫ 100. White bar, control; shaded bars, pNop30-C2; black bars, pCG35.