Cloning of human PRP4 reveals interaction with Clk1.

Prp4 is a protein kinase of Schizosaccharomyces pombe identified through its role in pre-mRNA splicing, and belongs to a kinase family including mammalian serine/arginine-rich protein-specific kinases and Clks, whose substrates are serine/arginine-rich proteins. We cloned human PRP4 (hPRP4) full-length cDNA and the antiserum raised against a partial peptide of hPRP4 recognized 170-kDa polypeptide in HeLa S3 cell extracts. Northern blot analysis revealed that hPRP4 mRNA was ubiquitously expressed in multiple tissues. The extended NH(2)-terminal region of hPRP4 contains an arginine/serine-rich domain and putative nuclear localization signals. hPRP4 phosphorylated and interacted with SF2/ASF, one of the essential splicing factors. Indirect immunofluorescence analysis revealed that endogenous hPRP4 was distributed in a nuclear speckled pattern and colocalized with SF2/ASF in HeLa S3 cells. Furthermore, hPRP4 interacted directly with Clk1 on its COOH terminus, and the arginine/serine-rich domain of hPRP4 was phosphorylated by Clk1 in vitro. Overexpression of Clk1 caused redistribution of hPRP4, from the speckled to the diffuse pattern in nucleoplasm, whereas inactive mutant of Clk1 caused no change of hPRP4 localization. These findings suggest that the NH(2)-terminal region of hPRP4 may play regulatory roles under an unidentified signal transduction pathway through Clk1.

In the fission yeast Schizosaccharomyces pombe, 14 temperature-and cold-sensitive pre-mRNA processing (prp) 1 mutants (prp1-prp14) have been isolated with a genetic approach (1-3). One of them, temperature-sensitive mutant strain prp4, does not grow and accumulates pre-mRNAs of intron-containing genes at the restricted temperature (2,4). prp4 ϩ gene encoding a protein kinase complemented the prp4 mutation, and the gene product was shown to phosphorylate the human SF2/ASF, one of the serine/arginine-rich (SR) proteins, in vitro (5). SR proteins compose a family of the factors required for constitutive splicing of pre-mRNA (6) and play important roles in alternative splicing in vitro. They are highly conserved in metazoa and are characterized by containing one or two RNA recognition motifs at the amino terminus and an arginine/serinerich domain (RS domain) at the carboxyl terminus (7,8). SR proteins are known to be phosphorylated on the RS domain (9). Although its physiological role is still unknown, phosphorylation of SR proteins affects their protein-protein and protein-RNA interactions (10) and alternative splicing of pre-mRNA (11). To date several kinases have been reported to phosphorylate SR proteins, such as small nuclear ribonucleoprotein particle-associated kinase (12), SR protein-specific kinase (SRPK) family kinases (9,13,14), Clk/Sty family kinases (15,16), DNA topoisomerase I (17), and p34 cdc2 kinase (18).
SRPK1 was purified and cloned on the basis of its ability to phosphorylate SC35 or other SR proteins in vitro (9). SRPK2 also specifically phosphorylates SF2/ASF as SRPK1 does (13,14). SRPKs can make the complex with SR proteins such as SF2/ASF, and the complex formation is affected by phosphorylation of the SR protein (19). SRPK is highly related to the fission yeast kinase Dsk1, which was isolated as a multicopy suppressor for a cold-sensitive dis1 mutant that affects chromosome segregation during mitosis (20,21). From budding yeast, an SR protein-specific kinase, Sky1p, was cloned and characterized (22). Phosphorylation by Sky1p promotes Npl3p shuttling and mRNA dissociation (23,24). In Caenorhabditis elegans, a homologue of SRPK1, SPK-1/RSK-1, plays essential roles at the embryonic stage (25,26) and is also required for germline development (25).
cdc2-like kinase (Clk1) was found to form complexes with and phosphorylate members of the SR protein family of splicing factors (15) and has been shown to affect the splicing of mRNA through phosphorylation of SR proteins (11,27). Clk1, also termed Sty, was initially isolated as a tyrosine kinase and shown to autophosphorylate on serine, threonine, and tyrosine residues (28 -30). An additional three members of the Clk family were isolated from mouse and human and were also shown to be dual specificity kinases (16,31,32). Clk homologues, darkener of apricot (DOA) and AFC1, were isolated in Drosophila and Arabidopsis, respectively (33,34). Any Clk homologues have not been found in yeast, although KNS1 is derived from Saccharomyces cerevisiae and belongs to the LAMMER kinase family, including Clks (35).
In mammals, only partial clones of putative prp4 homologues have been reported. The first cDNA sequence of hPRP4, human homologue of prp4, was reported by Kä ufer and their colleagues (5). The reported hPRP4 cDNA, cloned from HeLa cDNA library, encoded a 55-kDa kinase consisting of 496 amino FIG. 1. Primary structure of hPRP4. A, the predicted amino acid sequences of hPRP4 and its putative homologues. Multiple alignment was shown using the ClustalX software. To optimize homology, gaps were inserted as denoted by dots. Black boxes indicate the residues identical to the counterparts of hPRP4. The RS and SR dipeptides are indicated by black lines. The arrowhead shows the residue corresponding to the virtual first codon in the previous report (5). The lysine mutated in the construct K717R is indicated by an asterisk. The peptide sequence spanning residues 522-541, indicated by the gray bar, was used to generate a rabbit polyclonal antiserum against hPRP4. TPY sequence in the activation loop is shown by the dots over the residues. mPRP4 (35), cePRP4, and spPRP4 (4) are derived from mouse, C. elegans, and S. pombe, respectively. The NH 2 -terminal extent of mPRP4 is yet unknown. B, phylogenetic tree representing the relation among hPRP4-related kinases. Comparison of the kinases were based on the sequences of the catalytic domains. dmPRP4 and atPRP4 are putative proteins presumed from genomic sequences. The GenBank accession numbers of the sequences in A and B are as follows: hPRP4, AY029347; mPRP4, AF033663; dmPRP4, AE003467; cePRP4, Z71262; atPRP4, AL132966; spPRP4, L10739; dmMNB, X70799; hDYRK1Bc, XM_009210; hMIRK, AF205861; scYAK1, X16056; hHIPK3, XM_012033. C, Western blot (WB) analysis of endogenous hPRP4. Lysates derived from HeLa S3 cells were immunoprecipitated (IP) with preimmune serum (preimmune) or antiserum to hPRP4 (anti-hPRP4). After separation by 10% SDS-PAGE, immunoprecipitated samples were transferred on nitrocellulose membrane and probed with antiserum to hPRP4. The lysate from COS-7 cells transfected with the expression vector encoding HA-tagged full-length hPRP4 was also analyzed (HA-hPRP4 lysate). acids and had an ATG codon with the adequate context of Kozak consensus (36), although any in-frame stop codons were not found in the upstream sequence. 2 More recently, mouse prp4 homologue (herein called mPRP4) was captured with a gene trap approach (37). The mPRP4 predicted from the cloned cDNA had the longer NH 2 -terminal sequence, which possesses basic and RS motifs. However, ATG codon was not found at the NH 2 -terminal end of the mPRP4.
Many kinases are reported to be able to phosphorylate SR proteins as described above. However, it remains to be elucidated how SR protein kinases interact with each other. Here we report the cloning and the characterization of full-length hPRP4 cDNA encoding 170-kDa polypeptide. hPRP4 colocalizes with SR proteins in the subnuclear structures, termed speckles. One of the SR protein kinases, Clk1, interacts with hPRP4 COOH terminus, and phosphorylates the NH 2 -terminal region of hPRP4. Furthermore, cotransfection analysis showed that subnuclear localization of hPRP4 depends on the Clk1 kinase activity within the cells. These findings suggest that SR protein kinases regulate both SR proteins and themselves in vivo.

EXPERIMENTAL PROCEDURES
Antibodies-The monoclonal antibodies to the hemagglutinin (HA) epitope tag (12CA5, Roche Diagnostics), the Myc epitope tag (9E10, Santa Cruz Biotechnology), the FLAG epitope tag (M2, SIGMA), the His 6 tag (Qiagen) and the rabbit polyclonal antibodies to the HA tag (MBL) were purchased from the indicated supplier. The rabbit polyclonal antiserum to hPRP4 was produced by immunizing rabbits with a synthetic peptide NH 2 -CDNLEDFDVEEEDEEALIEQR-COOH, including residues 522-541 of hPRP4, coupled to keyhole limpet hemocyanin.
Isolation and Sequence Analysis of the Full-length Clone of hPRP4 -The total RNA was isolated from confluent HeLa S3 cells on a 100-mm dish, using Isogen (Nippon Gene, Tokyo). Using the oligonucleotides 5Ј-TGGAGACATACCCTCTTTATC-3Ј and 5Ј-ATCCTTGTCTTCTTCT-GAGGAATG-3Ј as gene-specific primers 1 and 2, respectively, 5Ј-rapid amplification of cDNA ends (RACE) procedure was carried out according to the 5Ј-RACE system instructions (Life Technologies, Inc.). The RT-PCR products were electrophoresed, subcloned into pGEM-T Easy vector (Promega), and sequenced with T7 promoter primer and M13 (Ϫ21) primer. Sequencing reaction and analysis were performed using DyeTerminator Cycle Sequencing Ready Reaction (PE Applied Biosystems) and ABI Prism 310 genetic analyzer (PE Applied Biosystems), respectively. Based on the analyzed sequence, the oligonucleotides were newly synthesized for isolation of full-length clone of hPRP4. RT-PCR was performed using oligo(dT) [12][13][14][15][16][17][18] primer (Life Technologies, Inc.) for reverse transcription, and the oligonucleotides 5Ј-GTCGACATGGC-CGCCGCGGAGACC-3Ј and 5Ј-GCGGCCGCTTAAATTTTTTCCTG-GATGAA-3Ј for PCR. The amplified fragment was ligated into pGEM-T Easy vector, and the clone containing the appropriate insert was confirmed by DNA sequencing.
Plasmid Constructions-The mutant cDNA, hPRP4 (K717R), was generated by substituting arginine for lysine at residue 717, by PCRbased method. The amplified fragment was ligated to pGEM-T Easy vector. The cDNA fragment corresponding to hPRP4⌬C (amino acids 1-497) or hPRP4⌬N (amino acids 498 -1007) was amplified by PCR using primers containing SalI or NotI recognition site and ligated to pGEM-T Easy vector. All constructs made by PCR were verified by DNA sequencing. The hPRP4 cDNA and its mutants were subcloned into mammalian expression vector pME-HA (38) between SalI and NotI sites. Mouse Clk1 cDNA was amplified by RT-PCR, followed by subcloning into pCRII vector (Invitrogen). The mutation in the ATP-binding site (K190R) was introduced using a PCR in vitro mutagenesis kit (TaKaRa) according to the manufacturer's instructions. Both pcDNA3FLAG-Clk1 and pcDNA3FLAG-Clk1 (K190R) were constructed by subcloning 1.4-kb XhoI fragment of the Clk1 and Clk1 (K190R) fragments into pcDNA3FLAG3 (gift from T. Fujino), respectively. The directions of the inserts were verified by sequencing. For expression of recombinant proteins in Escherichia coli, pET-hPRP4⌬N and pET-hPRP4⌬C were constructed by subcloning 1.5-kb SalI/NotI fragment of hPRP4⌬N and -⌬C into pET32b(ϩ). Construction of pGEX-SF2 and pGEX-SF2⌬RS, the prokaryotic expression vectors of GST-fused SF2/ ASF derivatives, has been described previously (19). For the Bac-to-Bac expression system (Life Technologies, Inc.), pFastBacHT-hPRP4 was constructed by subcloning 3-kb SalI/NotI fragment of hPRP4 cDNA into the SalI/NotI site of pFastBacHTc.
Preparation of Recombinant Proteins-GST-Clk1 and its derivative were expressed in E. coli (DH5␣). The culture was induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside at an absorbance of ϳ0.8 at 600 nm and incubated for 16 h at 20°C. The cells were harvested, resuspended in 20 ml of buffer A (100 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1 mM EDTA, 2 mM DTT, 2 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride) and disrupted by sonication at 0°C (Ultra Disruptor UD-201, Tomy). Triton X-100 was then added to 1% (v/v) final and rotated for 30 min at 4°C. Insoluble material was removed by centrifugation. The supernatant was incubated with glutathione-Sepharose 4B (Amersham Pharmacia Biotech) for 3 h or overnight at 4°C. The resin was collected, washed three times with washing buffer (1% Triton X-100 in buffer A), and eluted with elution buffer (30 mM reduced glutathione, 100 mM Tris-HCl, pH 7.5, 120 mM NaCl, 2 mM DTT, 2 mM benzamidine). Elution was performed three times, and the fractions were collected and dialyzed twice in dialysis buffer (20 mM HEPES-KOH, pH 8.0, 100 mM KCl, 0.2 mM EDTA, pH 8.0, 20% (v/v) glycerol, 1 mM DTT) overnight at 4°C. His-tagged proteins HisTrx-hPRP4⌬C and HisTrx-hPRP4⌬N were expressed in E. coli BL21(DE3) using pET32-derived vectors and purified using a nickel-nitrilotriacetic acid-agarose chromatography (Qiagen) according to manufacturer's instructions. For preparation of His-tagged hPRP4 (His-hPRP4), hPRP4 is expressed using the Bac-to-Bac expression system (Life Technologies, Inc.). DH10Bac, the host strain of E. coli, was transformed with pFastBacHT-hPRP4, and the recombinant bacmid was prepared. Sf9 cells were transfected with the bacmid, and viral supernatant was collected. The baculovirus for protein preparation was used after amplification twice. Sf9 cells were infected, lysed at 2 N. F. Kä ufer, personal communication.
Northern Blot Analysis and in Situ Hybridization-hPRP4 cDNA was amplified by PCR and labeled with [␣-32 P]dCTP using randomprimer DNA labeling kit (Roche Molecular Biochemicals). The labeled probe was hybridized to human multiple tissue Northern blot I or II (CLONTECH) in ExpressHyb hybridization solution (CLONTECH) according to the manufacturer's instructions. In situ hybridization was carried out as described (39). Frozen sections (14 m thick) were cut on a crystal and mounted onto poly-L-lysine-coated slides. Sections were fixed in 4% paraformaldehyde in PBS for 10 min and acetylated with 0.25% acetic anhydrate in 0.1 M triethanolamine. The sections were dehydrated in an ascending ethanol series and air-dried. The 35 Slabeled probes (3 ϫ 10 6 cpm/ml) were dissolved in a buffer containing 50% formaldehyde, 10% dextran, 1ϫ Denhardt's solution, 12 mM EDTA, pH 8.0, 10 mM DTT. 150 l of probe solution was applied to each slide at 65°C for 16 h. Coverslips were then removed, and the slides were rinsed in 4ϫ SSC and digested with RNase A (20 g/ml) for 30 min in 0.1ϫ SSC at 65°C, before finally being dehydrated again. The sections were exposed to x-ray films (Eastman Kodak Co.) for 2 days, and the films were developed.
In Vitro Kinase Assay-To measure the kinase activity of immunoprecipitates, cells were transfected with HA-tagged expression vectors using LipofectAMINE and lysed at 24 h in the modified Net2 lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.05% (w/v) Nonidet P-40, 2 mM EGTA, 20 mM ␤-glycerophosphate, 0.1 mM Na 3 VO 4 (V)). Lysates were clarified at 15,000 ϫ g for 30 min at 4°C and preabsorbed with protein G-Sepharose. Immunoprecipitations from cell lysates were carried out using monoclonal anti-HA antibody (12CA5) together with protein G-Sepharose (Amersham Pharmacia Biotech). After incubation for 1 h at 4°C, the complexes were washed three times with the lysis buffer and twice with kinase buffer (20 mM Tris-HCl, pH 7.5, 2 mM EGTA, 1 mM DTT, 0.1% (w/v) Triton X-100). The kinase assays were initiated by the addition of 3 g of the substrate and 100 M [␥-32 P]ATP, 10 mM MgCl 2 in a final volume of 100 l. The reactions were terminated after 30 min at 30°C by addition of 6ϫ Laemmli buffer. The reaction products were resolved by 15% SDS-PAGE and visualized by an imaging analyzer (Fujix, BAS2000). To measure the kinase activities of recombinant proteins, the kinase reactions containing 0.3 g of substrate and the indicated amounts of kinase in a final volume of 25 l were incubated for 20 min at 30°C and terminated, resolved by 10% SDS-PAGE, and detected as above.
In Vitro Binding Assays-GST or GST fusion proteins immobilized on glutathione-Sepharose were incubated with the indicated His-tagged proteins or the lysate from COS-7 cells expressing HA-tagged proteins for 3 h or overnight at 4°C. After extensively washed, glutathione-Sepharose precipitates were immunoblotted using mouse anti-His 6 antibody or rabbit anti-HA antibody. 10% or 1% of the proteins used in binding reactions was loaded as a positive control (input).

FIG. 2. Expression of hPRP4 or its homologue in tissues.
A, Northern blot analysis of hPRP4. Human multiple tissue Northern blot I or II (CLONTECH) was hybridized with a probe corresponding to full-length hPRP4 cDNA (upper panels) or glyceraldehyde-3-phosphatase dehydrogenase (G3PDH; lower panels). Molecular size markers are shown on the left. B, in situ hybridization of the mouse brain using hPRP4 as a probe. Sagittal brain section was hybridized with the antisense probe of hPRP4. Brain section hybridized with the sense probe showed no signal (data not shown).
Phosphopeptide Mapping-In vitro kinase assay was performed in the reaction containing recombinant His-hPRP4 in the absence or presence of 0.3 g of recombinant GST-Clk1, as above. After separation of the reactions by SDS-PAGE, the bands corresponding to hPRP4 were excised and two-dimensional phosphopeptide mapping was performed as described previously (18).

RESULTS
Cloning of Full-length cDNA Encoding hPRP4 -To isolate the full clone of hPRP4 from HeLa S3 cells, 5Ј-RACE was performed, and a 300-base pair major fragment was amplified. The sequence analysis of the 300-base pair fragment revealed that it contains the putative 5Ј end of hPRP4 cDNA and a putative first ATG codon in the open reading frame of hPRP4 with an adequate context of Kozak consensus (5Ј-AAGATGG-3Ј) at the downstream of the GC-rich sequence (GenBank TM accession no. AY029347). Based on our upstream sequence data, an ϳ3-kb fragment of hPRP4 cDNA was amplified by RT-PCR. The predicted open reading frame of hPRP4 consisting of 1007 amino acids was aligned with the prp4 homologues.
Both hPRP4 and mPRP4 have basic amino acids repeat, which may function as a nuclear localization signal (NLS), and Arg-Ser/Ser-Arg (RS/SR) dipeptides at the NH 2 terminus. Kinase domains I-XI were well conserved from yeast to human (Fig.  1A). As shown in Fig. 1B, the phylogenetic tree showed that hPRP4 and its homologues are structurally related to DYRK family and HIPK family. Although it is not clear that they share common biochemical features with hPRP4, at least some isoforms of DYRKs and HIPKs have been reported to localize in nucleus (40,41). We raised an antibody against hPRP4 by immunizing a hPRP4 peptide (amino acids 522-541). A protein band of 170 kDa was immunoprecipitated and detected by immunoblotting with this antiserum (Fig. 1C). The recombinant HA-hPRP4 protein expressed in COS-7 cells showed a molecular mass of 170 kDa on SDS-PAGE. To examine the expression pattern of hPRP4 mRNA in human tissues, we performed the Northern blot analysis using the full-length hPRP4 cDNA as a probe. The analysis revealed that the major transcript of ϳ4.4 kb was widely expressed in human tissues, although the expression levels in lung and leukocytes seemed to be low ( Fig. 2A). To check the cell type-specific expression of hPRP4, we performed in situ hybridization of mouse brain (Fig.  2B). Sagittal section of a brain hybridized with antisense probe of hPRP4 revealed that the higher signals were detected in the cerebellum, the hippocampus, and the olfactory balb, suggesting the differential expression of hPRP4 or its homologue in neuronal cells.
hPRP4 Phosphorylates SF2/ASF More Efficiently than Clk1 Does-In previous reports, phosphorylation of SF2/ASF, myelin basic protein (MBP), or Elk1 by mammalian prp4 homologues has been reported using the catalytic fragments of the enzyme (5,42). We here examined the substrate specificity of full-length hPRP4, in comparison with Clk1, which was known to phosphorylate SF2/ASF, MBP, and histone H1 (15). COS-7 cells transfected with the expression vector encoding HAtagged hPRP4, Clk1, or an empty vector were lysed at 24 h after transfection. Immunoprecipitated hPRP4 phosphorylated SF2/ASF more efficiently than Clk1, while they equally phosphorylated MBP. Histone H1 was not a good substrate of hPRP4 (Fig. 3A).
Among other SR protein kinases, SRPK1 and SRPK2 were known to be associated with a substrate, SF2/ASF (19). Therefore, we investigated whether hPRP4 also interacts with SF2/ ASF. In the binding assay, hPRP4 bound GST-SF2 but did not bind GST-SF2⌬RS (Fig. 3B). These findings indicate that hPRP4 is able to interact with SF2/ASF in vitro, and that the interaction needs the RS domain of SF2/ASF.
The NH 2 -terminal Region of hPRP4 Contains the Signal Sequence Targeting to Nuclear Speckles-To elucidate the localization of endogenous hPRP4, we immunostained HeLa S3 cells with the antiserum to hPRP4. The punctate pattern in nuclei was stained with the antiserum to hPRP4, whereas no staining was observed with preimmune serum (Fig. 4A). In addition, overexpressed, HA-tagged full-length hPRP4 was also localized in a similar pattern according to the immunofluorescence analysis of MCF-7 cells expressing HA-hPRP4 using mouse monoclonal anti-HA antibody (Fig. 4B, left panels). In contrast, HA-tagged hPRP4⌬N (amino acids 498 -1007), the mutant lacking the NH 2 -terminal region including putative NLS and RS/SR dipeptide repeats (see Fig. 1A), was localized in the cytoplasm (Fig. 4B, right panels).
SR proteins are known to localize in the subnuclear structures, called speckles (43). Next, we investigated the localization of endogenous hPRP4 and SR proteins. When we performed double staining of HeLa S3 cells with polyclonal antiserum to hPRP4 and either monoclonal anti-SF2/ASF or anti-SC35 antibody, the colocalization of hPRP4 with these SR proteins in nuclear speckles was observed (Fig. 5, A and B).
Clk1 Interacts with the COOH-terminal Region and Phosphorylates the NH 2 -terminal Region of hPRP4 -Inactive mutant of Clk1 (Clk1 K190R) is reported to be localized in a speckled pattern in nuclei of the transfected cells (15). To investigate whether hPRP4 interacts with Clk1, in vitro binding assay was performed using recombinant hPRP4 lacking NH 2 terminus (HisTrx-hPRP4⌬N) and GST-Clk1. As shown in the right panel of Fig. 6A, hPRP4⌬N bound GST-Clk1. Since hPRP4 was sug-gested to be a phosphoprotein from the in vivo labeling experiment (data not shown), we tested the autophosphorylation of hPRP4 and its exogenous phosphorylation by Clk1 in vitro. The recombinant full-length hPRP4 (His-hPRP4) purified from Sf9 cells was autophosphorylated when incubated with [␥-32 P]ATP (Fig. 6B, second lane). Moreover, phosphorylation of His-hPRP4 was enhanced with increased amounts of GST-Clk1 (Fig. 6B, third through seventh lanes), while obvious change was not observed with GST-Clk1 (K190R), which lacks kinase activity by a point mutation at the ATP-binding site (Fig. 6B,  9th through 14th lane). Next, to compare the phosphorylation site(s) of the autophosphorylation and Clk1-induced phosphorylation of hPRP4, we performed phosphopeptide mapping of hPRP4. Phosphopeptides 9,11,13,14,15, and 17 seemed to be catalyzed by Clk1, whereas autophosphorylated peptides were 1, 2, 3, 4, 5, and 6 in the upper panel, and 7, 8, 10, 12, 16, and 18 in the lower panel (Fig. 6C). To confirm this result (that the phosphorylation sites of hPRP4 by Clk1 are not identical to the autophosphorylation sites), we prepared the NH 2 -terminal peptide (HisTrx-hPRP4⌬C) and the COOH-terminal peptide (HisTrx-hPRP4⌬N) of hPRP4 and compared the phosphorylation. As we predicted, HisTrx-hPRP4⌬C, which has RS repeats but lacks kinase domain, was well phosphorylated by Clk1 in a dose-dependent manner without autophosphorylation (Fig.  6D). In contrast to this, HisTrx-hPRP4⌬N showed autophosphorylation without any enhancement by Clk1. Taken together, these findings showed that Clk1 interacts with the COOH-terminal region and phosphorylates the NH 2 -terminal region of hPRP4.
Subnuclear Localization of hPRP4 Depends on Clk1 Activity-It has been reported that Clk1 overexpression causes redistribution of nuclear speckles to a diffuse nucleoplasmic pattern (15,32). Therefore, to investigate the interaction between Clk1 and hPRP4 in vivo, we examined the effect of Clk1 on the localization of hPRP4. HA-hPRP4 was stained as a diffuse pattern in the nuclei of human breast carcinoma MCF-7 cells transfected together with FLAG-Clk1 (Fig. 7A, upper panels). However, when HA-hPRP4 was expressed in cells together with FLAG-Clk1 (K190R), the mutant lacking kinase activity, HA-hPRP4 was stained as the speckled pattern (Fig. 7A, middle panels). To determine whether kinase activity of hPRP4 affects its own localization, MCF-7 cells were also transfected with FLAG-Clk1 together with HA-hPRP4 (K717R), which has a mutation from lysine to arginine at position 717 in the kinase subdomain II. This result showed that HA-hPRP4 (K717R) was stained as a diffuse pattern (Fig. 7A, bottom). Furthermore, we confirmed this tendency by quantitative analysis. For quantification of localization change of hPRP4 mediated by Clk1, the population of cells exhibiting nuclear diffuse staining of HA-hPRP4 or its mutant was shown in Fig. 7B. As described in Fig.  7A, it was shown that wild-type Clk1 caused redistribution of hPRP4 from the speckled to the diffuse pattern, and inactive Clk1 did not affect the localization of hPRP4, but that no difference was observed between using wild-type hPRP4 and its mutant K717R. These findings suggest that the distribution of hPRP4 in the nucleus is not regulated by its own activity, but determined by the kinase activity of Clk1. DISCUSSION In this study, we found that hPRP4 has the extended NH 2 terminus. The clone encoded a 170-kDa polypeptide consisting of 1007 amino acids. Our polyclonal antibody raised against a peptide 522-541 of hPRP4 recognized the endogenous hPRP4 of HeLa S3 cells with the same molecular mass of 170 kDa. At position 15 of spPRP4, a putative NLS was detected and the deletion of 11 amino acids including the putative NLS (RRKRR) canceled the complementation of the prp4 -73 allele (5). The putative NLS of spPRP4 is not conserved in mammalian PRP4 and the mouse PRP4 (mPRP4) clone consisting of 496 amino acids could not complement the prp4 -73 phenotype, whereas the chimeric constructs, which contains the NH 2 terminus of spPRP4 and the partial kinase domain of mPRP4, rescued the mutation (5). The hPRP4 possesses basic and RS domains at the NH 2 terminus, and the hPRP4 deletion construct of these domains showed the cytosolic localization. The wild-type hPRP4 was located in subnuclear structures such as speckles. It has been reported that overexpression of Clk1 in FIG. 7. Effect of Clk1 on the subcellular localization of hPRP4. A, MCF-7 cells were cotransfected with the expression vector encoding HA-tagged wild-type hPRP4 or its mutant K717R together with the vector encoding FLAG-tagged wild-type Clk1 or its mutant K190R. The cells were fixed at 24 h after transfection, immunostained with rabbit anti-HA antibody and mouse monoclonal anti-FLAG antibody (M2), and then visualized by confocal microscopy. The images were superimposed (Merge). The cells were also stained with Hoechst 33258. The scale bar represents 5 m. B, quantification of the localization change of hPRP4 mediated by Clk1. Subcellular localization of HA-tagged wild-type hPRP4 or its mutant K717R was scored using the same coverslips as in A. The cells that were signal-positive in nuclei were categorized into the two classes: diffuse and speckled. The population of cells exhibiting nuclear diffuse staining is presented as the percentage of total signal-positive cells examined. More than 100 cells were examined. Data are representative of three independent experiments. mammalian cells caused disruption and rearrangement of speckles containing splicing factors (15). Overexpression of hPRP4 did not affect the localization of nuclear speckles, although hPRP4 is far more active than Clk1 when SF2/ASF was used as a substrate in vitro. Instead, overexpressed Clk1 released hPRP4 from nuclear speckles in a kinase activity-dependent manner. As Clk1 can directly bind hPRP4 and the NH 2 terminus of hPRP4 was phosphorylated by Clk1, we hypothesized that the effect of Clk1 on nuclear speckles is catalyzed through the activation of hPRP4. However, this hypothesis seemed to be denied because cotransfection of HA-hPRP4 (K717R), mutated at the ATP-binding site, with Clk1 also disrupted the nuclear speckles. We should still consider the possibility that Clk1 interacted with and activated endogenous hPRP4 in the MCF-7 cells.
The hPRP4 is classified based on sequence similarity of the kinase domain into a member of CMGC group, which includes cyclin-dependent kinase family, mitogen-activated protein kinase family, glycogen synthase kinase family, and cdc2-like kinase (Clk) family (44). The region between the kinase subdomains VII and VIII of hPRP4 shows significant similarity to mitogen-activated protein kinases (5,42,45). Residues Thr-847 and Tyr-849 in the subdomain VIII (see Fig. 1A) are in an equivalent position to the TPY sequence in JNK/SAPK. Although the idea seems to be attractive that hPRP4 can be phosphorylated and activated by some upstream kinases at the position of the TPY, the motif is not conserved and substituted by APY in a PRP4 homologue of C. elegans. Actually, a mutant that possesses the amino acids APF in place of the wild type TPY could complement the prp4 -73 phenotype (5), indicating that the motif is not essential for the physiological function of PRP4, at least, in yeast. It would be interesting to reexamine the regulatory mechanism of the TPY motif using the fulllength clone of hPRP4.
The comparative study of SRPK1 and Clk1 revealed that SRPK1 is a highly specific for SR proteins and that Clk1 has a broader specificity, and a much lower phosphorylating activity for SF2/ASF than SRPK1 in vitro (46), and we found that Clk1 is less active than hPRP4 to phosphorylate SF2/ASF. These notions may suggest that Clk1 is not the direct regulator of SR proteins. SRPK1 and SRPK2 are mainly localized in cytoplasm (13,14), and the yeast homologue of SRPK1, Sky1p, phosphorylates the COOH-terminal RS of Npl3p and promotes its shuttling and mRNA dissociation (23,24), suggesting that SRPKs may be cytosolic regulators of RNA-binding proteins. The notion that PRP4 is involved in pre-mRNA splicing is based on the observation that intron-containing genes accumulate pre-mRNA at the restrictive temperature when the prp4 -73 allele is in the genetic background (4). More recently, spp42, encoding a homologue of the splicing factor Prp8p, was identified as a suppressor of prp4 -73 (47). In addition, Prplp, another splicing factor in fission yeast, was shown to interact with and to be phosphorylated by Prp4p (48). Our observations that hPRP4 bound SF2/ASF and was localized in nuclear speckles suggest that hPRP4 is an excellent candidate for a kinase directly controlling the pre-mRNA splicing. Although we do not know the mechanism whereby the activity of PRP4 affects the pre-mRNA splicing, identification and characterization of the hPRP4 full clone invite further study of phosphorylation-dependent regulation of pre-mRNA processing.