Manipulation of Alternative Splicing by a Newly Developed Inhibitor of Clks* □ S

The regulation of splice site usage provides a versatile mechanism for controlling gene expression and for the generation of proteome diversity, playing an essential role in many biological processes. The importance of alternative splicing is further illustrated by the increasing number of human diseases that have been attributed to mis-splicing events. Appropriate spatial and temporal generation of splicing variants demands that alternative splicing be subjected to extensive regulation, similar to transcriptional control. The Clk (Cdc2-like kinase) family has been implicated in splicing control and con-sists of at least four members. Through extensive screening of a chemical library, we found that a benzothiazole

The regulation of splice site usage provides a versatile mechanism for controlling gene expression and for the generation of proteome diversity, playing an essential role in many biological processes. The importance of alternative splicing is further illustrated by the increasing number of human diseases that have been attributed to mis-splicing events. Appropriate spatial and temporal generation of splicing variants demands that alternative splicing be subjected to extensive regulation, similar to transcriptional control. The Clk (Cdc2-like kinase) family has been implicated in splicing control and consists of at least four members. Through extensive screening of a chemical library, we found that a benzothiazole compound, TG003, had a potent inhibitory effect on the activity of Clk1/Sty. TG003 inhibited SF2/ASFdependent splicing of ␤-globin pre-mRNA in vitro by suppression of Clk-mediated phosphorylation. This drug also suppressed serine/arginine-rich protein phosphorylation, dissociation of nuclear speckles, and Clk1/ Sty-dependent alternative splicing in mammalian cells. Consistently, administration of TG003 rescued the embryonic defects induced by excessive Clk activity in Xenopus. Thus, TG003, a novel inhibitor of Clk family will be a valuable tool to dissect the regulatory mechanisms involving serine/arginine-rich protein phosphorylation signaling pathways in vivo, and may be applicable for the therapeutic manipulation of abnormal splicing.
Recent whole genome sequence analyses revealed that a high degree of proteomic complexity is achieved with a limited number of genes. This surprising finding underscores the importance of alternative splicing, through which a single gene can generate multiple structurally and functionally distinct protein isoforms (1). Based on genome-wide analysis, 35-60% of hu-man genes are thought to encode at least two alternatively spliced isoforms (2). The regulation of splice site usage provides a versatile mechanism for controlling gene expression and for the generation of proteome diversity, playing essential roles in many biological processes, such as embryonic development, cell growth, and apoptosis. Splicing mutations located in either intronic or exonic regions frequently cause hereditary diseases (reviewed in Refs. [3][4][5]. More than 15% of mutations that cause genetic disease affect pre-mRNA splicing (6). Pre-mRNA splicing is also regulated in a tissue-specific or developmental stagespecific manner. Indeed, the selection of splice site can be altered by numerous extracellular stimuli, including growth factors, cytokines, hormones, depolarization, osmotic shock, and UVC irradiation through synthesis, phosphorylation, and a change in localization of serine/arginine-rich (SR) 1 proteins (7). SR proteins are a family of essential factors required for constitutive splicing of pre-mRNA (8) and play an important role in modulating alternative splicing (9). They are highly conserved in eukaryotes and are characterized by having one or two RNA-recognition motifs at the amino terminus and an RS domain at the carboxyl terminus (10,11). RS domains consist of multiple consecutive RS/SR dipeptide repeats and differ in length among different SR proteins. Extensive phosphorylation of serines in the RS domain occurs in all SR proteins (12,13). Although its precise physiological role is still unknown, phosphorylation of SR proteins affects their protein-protein and protein-RNA interactions (14), intracellular localization and trafficking (15,16), and alternative splicing of pre-mRNA (17). Spliceosome assembly may be promoted by phosphorylation of SR proteins that facilitate specific protein interactions, while preventing SR proteins from binding randomly to RNA (14). Once a functional spliceosome has formed, dephosphorylation of SR proteins appears to be necessary to allow the transesterification reactions to occur (18). Therefore, the sequential phosphorylation and dephosphorylation of SR proteins may mark the transition between stages in each round of the splicing reaction. To date, several kinases have been reported to phosphorylate SR proteins, including SRPK family kinases (19,20), hPRP4 (21), and topoisomerase I (22), and a family of kinases termed Clk (Cdc2-like kinase), or LAMMER kinases from the consensus motif, consisting of four members (Clk1/Sty and Clk2-4) (23,24).
Mammalian Clk family kinases contain an SR domain and are demonstrated to phosphorylate SR proteins in vitro and SF2/ASF in vivo (24). Clks are shown to be dual-specificity kinases that autophosphorylate on tyrosine, serine, and threonine residues in overexpression systems and in vitro (24 -26). When overexpressed, the catalytically inactive mutant kinases localize to nuclear speckles where splicing factors are concentrated, whereas the wild-type enzymes distribute throughout the nucleus and cause speckles to dissolve (23). The overexpression of Clks also affects splicing site selection of pre-mRNA of both its own transcript and adenovirus E1A transcripts in vivo (17). These results have led us to the current model that Clk family members regulate alternative splicing by phosphorylation of SR proteins, although their signal pathways and biological functions are largely unknown in vertebrates.
Here we hypothesized that pharmacological inhibition of Clk kinases might provide a useful way to modulate alternative splicing, and we set out to screen a chemical library to look for compounds that affect the regulation of alternative splicing. In this paper, we report a novel compound, TG003, that inhibits the kinase activity of Clks and affects the regulation of alternative splicing mediated by phosphorylation of SR proteins in vitro and in vivo. Furthermore, TG003 also suppressed defects in early Xenopus development induced by excess level of Clk activity, suggesting its potential use of TG003 for manipulation of alternative splicing in vivo.
In Vitro Splicing-m 7 GpppG-capped and 32 P-labeled pre-mRNA substrates were made by runoff transcription of linearized human ␤globin template DNA with SP6 RNA polymerase (28). HeLa cell S100 extract and purified SF2/ASF were prepared as described (29). In vitro splicing reaction mix containing the HeLa S100 extract, purified SF2/-ASF, and 20 fmol of 32 P-labeled pre-mRNA was incubated with/without TG003 or TG009 at 30°C for 3-4 h (29). The RNA products were analyzed by electrophoresis on a 5.5% polyacrylamide, 7 M urea gel and autoradiography.
In Vitro Kinase Assay-Kinase activity of Clks and SRPKs was assayed in a reaction mixture, containing 200 mM Tris-HCl (pH 7.5), 12.5 mM MgCl 2 , 8 mM dithiothreitol, 4 mM EGTA, 1-20 M ATP, 1 Ci of [␥-32 P]ATP, 1 g of synthetic peptide of SF2/ASF RS domain (NH 2 -RSPSYGRSRSRSRSRSRSRSRSNSRSRSY-OH), and 0.1-1 g of puri-fied kinases in a final volume of 40 l. cAMP-dependent protein kinase activity was assayed in a reaction mixture containing 80 mM Tris-HCl (pH 7.5), 12.5 mM MgCl 2 , 8 mM dithiothreitol, 4 mM EGTA, 10 M ATP, 1 Ci of [␥-32 P]ATP, 5 g of histone H1 (Sigma), and 1 g of catalytic subunit of rat cAMP-dependent protein kinase purified as described (30). Protein kinase C activity was assayed in a reaction mixture containing 200 mM Tris-HCl (pH 7.5), 12.5 mM MgCl 2 , 1 mM CaCl 2 , 80 g/ml phosphatidylserine, 8 g/ml diolein, 10 M ATP, 1 Ci of [␥-32 P]ATP, 5 g of histone H1, and 2 l of partially purified rat protein kinase C (Seikagaku Kogyo). The final concentration of Me 2 SO was adjusted to 1% regardless of inhibitor concentration. The reaction mixture was incubated at 30 or 25°C for mammalian or Xenopus recombinant proteins, respectively, for 10 min, and a half-portion was spotted on P81 phosphocellulose membrane (Whatman). The kinase assay conditions, including the incubation period and concentration of kinases and substrates, were optimized to maintain the linearity during incubation. The membrane was washed with 5% phosphoric acid solution (SF2/ASF RS domain) or 5% trichloroacetic solution (histone H1) at least over 15 min. The radioactivity was measured using a liquid scintillation counter. The net radioactivity was deduced by subtracting the background count from the reaction mixture without kinase, and the data are expressed as the percentage to the control sample containing the solvent.
Immunofluorescence Staining-HeLa cells grown on coverslips in a 12-well dish were transfected with Clk1/Sty expressing vectors (0.5 g; pME-HA-mClk1/Sty or -mClk1/Sty K190 ) (21) using GeneJuice (Novagen; 1.5 l) and further incubated for 36 h. All following procedures were performed at room temperature. Cells were fixed with 4% paraformaldehyde in 250 mM Hepes-NaOH (pH 7.4) for 20 min, permeabilized with 1% Triton X-100 in PBS for 20 min, and washed four times in PBS. The cells were incubated in blocking solution (1% bovine serum albumin, 0.2% gelatin, and 0.05% Tween 20 in PBS, pH 8.0) for 30 min and incubated with rabbit anti-HA tag antibody (Santa Cruz Biotechnology; 1:1000) and mouse mAb1H4 recognizing phosphorylated SR proteins (ATCC; 1:5 of hybridoma supernatant) or mouse anti-SC35 antibody (Sigma; 1:4000) in blocking solution for 2 h. After washing several times over 1 h in PBST (PBS containing 0.05% Tween 20), the coverslips were incubated with donkey anti-mouse IgG (HϩL) (Jackson Laboratories; 1:200) conjugated with Alexa 488 (Molecular Probes) and Cy3-conjugated donkey anti-rabbit IgG (HϩL) (Jackson Laboratories; 1:200) in blocking solution for 2 h. After washing several times over 1 h in PBST and three times with PBS, the coverslips were mounted in Vectashield (Vector Laboratories). The images were taken using a confocal microscope (Olympus FV500 or Carl Zeiss LSM510 META). The subnuclear distribution of HA-Clk1/Sty was classified into three patterns (diffuse, intermediate, and speckle), and the number of cells showing each pattern was counted independently by four individuals for semi-quantitation.
Effects of TG003 on Cell Growth-2 ϫ 10 5 HeLa cells or 1.5 ϫ 10 5 COS-7 cells resuspended in 2 ml of medium were plated on 6-well dishes, and 2 l of 10 mM TG003 dissolved in Me 2 SO (final concentration at 10 M), or 2 l of Me 2 SO, was added to some wells. Cells were trypsinized, and the density was counted every 24 h for 3 days. Cells were then fixed with 1 ml of ice-cold 70% ethanol, washed with PBS, incubated in 1 ml of PBS containing 1 g/ml DNase-free RNase A (Roche Applied Science) and 50 g/ml propidium iodide (Sigma) for 20 min at 37°C, and proceeded to cell cycle analysis by FACSCalibur (BD Biosciences).
In Vivo Splicing Assay-COS-7 cells grown in a 60-mm dish were transfected with Myc-tagged Clk minigene (CMV-Clk1 or -Clk1 K190R (17); Fig. 4A) or adenovirus E1A minigene (pMT-E1A) (31) in combination with the Clk expression vector (Fig. 4B), using LipofectAMINE (Invitrogen) according to the manufacturer's instructions. Twenty four hours after transfection, the total RNA was extracted using ISOGEN (Nippon Gene); for Fig. 4A, cells were also lysed in SDS-gel loading buffer (0.1 M Tris-HCl (pH 6.8), 0.2 M dithiothreitol, 4% SDS, 20% glycerol) to prepare total cellular protein extract. Five micrograms of RNA was used for reverse transcription (RT), and then 1:50 was used for PCR amplification (94°C for 5 min, (94°C for 30 s, 57°C for 30 s, and 72°C for 1 min) ϫ 25 cycles, 72°C for 5 min). PCR conditions, including the number of cycles and template concentrations, were optimized to maintain the linearity during amplification. PCR products were separated in agarose gel and stained with ethidium bromide. Total protein was separated in SDS-PAGE and transferred to PVDF membrane. To detect Myc-tagged Clk protein (31), the membrane was incubated with mouse anti-Myc tag antibody (MBL, Co., LTD, Nagoya, Japan) followed by alkaline phosphatase-conjugated anti-mouse IgG ϩ A ϩ M (HϩL) (Bio-Rad). For splicing assay for endogenous genes in Fig.  5, mouse embryonic fibroblasts (STO cells) were incubated in the pres-ence or absence of 10 M TG003 for 4 h, and total RNA was extracted using TRIzol (Invitrogen) before RT-PCR using primers for SC35 and Clk1/Sty designed as per Pilch et al. (32). The PCR conditions were as follows: 94°C for 5 min (94°C for 15 s, 55°C for 30 s, and 68°C for 1 min) ϫ 25 cycles (SC35) or 30 cycles (Clk1/Sty).
Isolation and Sequence Analysis of Xenopus Clk-The total mRNA was extracted from Xenopus embryos at stage 2, 12, 18, and 40 using TRIzol (Invitrogen) according to the manufacturer's instruction. cDNA encoding Xenopus Clk was amplified by RT-PCR using SuperScriptII (Invitrogen; 42°C for 30 min) and High Fidelity PCR Master (Roche Applied Science; 95°C for 5 min (94°C for 30 s, 55°C for 30 s, 68°C for 2 min) ϫ 25 cycles, 72°C for 10 min) with primers designed according to the IMAGE clone of xClk (BC043963; 5Ј-ATGCCTCACTCCAGACGT-TACGGTTCGTCA-3Ј for the 5Ј primer and 5Ј-TCATCGGCTTATGTC-CCGGCCAGTGTCCCA-3Ј for the 3Ј primer). The PCR products were cloned into pGEM-T Easy (Promega), and the nucleotide sequence was verified. To make the mRNA expression vector, the resulting plasmid (i.e. pGEM-T Easy containing xClk) was digested with NotI, blunted with Klenow enzyme, digested with SpeI, and ligated into pCS2ϩ (33) digested with XbaI and StuI. For bacterial expression, the pGEM-T Easy containing xClk was digested with NotI and inserted into NotI digested pGEX-5X-3 (Amersham Biosciences).
Xenopus Embryo Manipulation-Xenopus laevis embryos were obtained from in vitro fertilization of eggs with testes homogenates as described (34), dejellied with 3% cysteine, and washed several times with water. Embryos were staged according to Nieuwkoop and Faber (35). Embryos were cultured at 22°C for 2 or 5 days with TG003 or its solvent (Me 2 SO) in dark.
Microinjection of Synthetic mRNA-Capped mRNA was synthesized from linearized xClk/CS2ϩ vectors using the mMessage Machine kit (Ambion). Synthesized mRNA was injected into the dorsal blastomeres of four-cell stage embryos, which were further cultured in Steinberg's buffer containing 3% Ficoll with TG003 or Me 2 SO for 2 or 5 days, and phenotypes were scored on the 2nd day.

RESULTS
TG003 Inhibits Clk1/Sty and Clk4 in Vitro-Through extensive screening of 100,000 chemical compounds in a chemical library by in vitro phosphorylation assay, we found that a benzothiazole compound had a potent inhibitory effect on the activity of Clk1/Sty. We therefore synthesized a series of benzothiazole derivatives, as shown in Fig. 1A. Among these compounds, (Z)-1-(3-ethyl-5-methoxy-2,3-dihydrobenzothiazol-2ylidene)propan-2-one, designated TG003, showed the most potent effect on Clk1/Sty and Clk4 (IC 50 , 15-20 nM) and lesser on Clk2 (200 nM) (Fig. 1B). This result is consistent with the amino acid sequence similarity; Clk1/Sty and Clk4 are more closely related to each other (69% identity) than to Clk2 or Clk3 (43% identity) (24). No inhibitory effect was observed on Clk3, SRPK1, SRPK2, cAMP-dependent protein kinase, or protein kinase C up to 1 M. The double-reciprocal Lineweaver-Burk plot indicated that TG003 acts on Clk1/Sty competitively with ATP (K m 3.35 M) with a K i value of 0.01 M (Fig. 1C). TG009 is a structurally analogous compound with 500 -1000 times weaker effect on Clk1/Sty and Clk4 and was used as a negative control throughout the following experiments.
TG003 Inhibits SF2/ASF-and Clk-dependent Splicing in Vitro-Because phosphorylation of SR proteins is known to be the critical regulatory step for alternative splicing (14,36), we tested if TG003 can block the phosphorylation of recombinant His-tagged SF2/ASF (rSF2/ASF) by HeLa cytosolic S100 extract (29) or Clk1/Sty. SR proteins are phosphorylated at multiple serine residues within their RS domains, and the electrophoretic mobility of SF2/ASF and SC35, well known SR proteins, on SDS-PAGE is affected by their phosphorylation state (14,37); phosphorylated protein shows more reduced mobility shift than unphosphorylated proteins. rSF2/ASF purified from E. coli is thought to be unphosphorylated. rSF2/ASF was incubated with HeLa cytosolic S100 extract (29) or Clk1/ Sty as the kinase source in the splicing condition in the absence or presence of TG003 or its negative control TG009. rSF2/ASF exhibited reduced electrophoretic mobility in the presence of Clk1/Sty (Fig. 2A, lane 2) compared with the rSF2/ASF alone (Fig. 2A, lane 1), and TG003 (1 M) completely blocked the mobility shift (Fig. 2A, lane 3), whereas TG009 had no effect ( Fig. 2A, lane 4). TG003 blocked the mobility shift of rSF2/ASF induced by S100 (Fig. 2A, lane 6), whereas TG009 again had no effect. Considering the inhibition spectrum of TG003, this result suggests that the major SR protein kinase activity in the HeLa S100 extract is either Clk1/Sty or Clk4.
We next examined if TG003 has an effect on splicing reaction in vitro by complementation assay. Human ␤-globin pre-mRNA FIG. 2. TG003 inhibits SF2/ASF-dependent splicing in vitro by suppression of Clk1/Sty-mediated phosphorylation. A, phosphorylation of SF2/ASF was inhibited by TG003 in HeLa cytosolic S100 extract. Recombinant SF2/ASF (rSF2/ASF; 0.2 g) in splicing buffer was incubated for 4 h at 30°C with either recombinant Clk1/Sty (0.5 g) (lanes 2-4) or HeLa S100 extract (lanes 4 -7) in the absence (lanes 2 and 5) or presence of 1 M TG003 (lanes 3 and 6) or TG009 (lanes 4 and 7). Aliquots were fractionated by SDS-PAGE and analyzed by Western blotting with monoclonal antibody AK103 (37). Positions of phosphorylated and unphosphorylated rSF2/ASF are indicated on the right of the panel as P-rSF2/ASF and rSF2/ASF, respectively. Without any kinase sources, mobility of SF2/ASF is not changed during the incubation (lane 1). B, TG003 altered the pattern of the SF2/ASFdependent splicing of human ␤-globin in vitro. m 7 GpppG-capped and 32 P-labeled human ␤-globin pre-mRNA was incubated with cytosolic S100 extract complemented with SF2/ASF purified from HeLa cells (lanes 1-3) or recombinant SF2/ASF (lanes 5-7). The solvent (DMSO) (lanes 2 and 6) or TG003 (1 M) (lanes 3 and 7) was added to reaction mixtures before starting splicing reaction. The RNA products were analyzed by electrophoresis on a 5.5% polyacrylamide, 7 M urea gel and autoradiography.
Positions of the pre-mRNA, spliced product, and intermediates are depicted by symbols on the right. was incubated in HeLa S100 extract (29) and supplemented with either human SF2/ASF (hSF2/ASF) purified from HeLa cells or rSF2/ASF (Fig. 2B). As expected, TG003 attenuated the splicing of ␤-globin pre-mRNA in S100 extract complemented with rSF2/ASF (Fig. 2A, lane 7); in contrast, it had no effect when complemented with hSF2/ASF (Fig. 2A, lane 3). It is likely that the unphosphorylated rSF2/ASF needs to become phosphorylated during the incubation to support splicing reaction, which was inhibited by TG003, whereas hSF2/ASF is already phosphorylated when it was purified from HeLa cells (37). As an ATP-regenerating system and magnesium are usually used in splicing assays, SR proteins should be maintained in a phosphorylated state throughout the splicing reaction, as long as the extract contains the kinase activity and protein phosphatases are not in excess. Indeed, at the end of the splicing reaction, the majority of rSF2/ASF displayed reduced electrophoretic mobility (Fig. 2A, lane 5).
TG003 Inhibits Clk1/Sty Kinase Activity in Mammalian Cells-Many splicing factors including small nuclear ribonucleoproteins and SR proteins are found to be localized in nuclear structures termed speckles, proposed to act as storage/ assembly/modification sites for splicing components (reviewed in Ref. 38). Overexpression of Clk kinases can modulate the subnuclear localization of SR proteins and Clk itself from speckles to nucleoplasm (17,23), suggesting that Clk kinase phosphorylates SR proteins and Clk itself to promote their release from storage sites and increases its effective nucleoplasmic concentration and availability to participate in the splicing reaction (17). To address whether TG003 can inhibit the kinase activity of Clk1/Sty in living cells, we first assessed if the compound inhibits the hyperphosphorylation of SR proteins and its redistribution from speckles to a diffuse nucleoplasmic pattern induced by overexpression of HA-tagged Clk1/ Sty. Even in the presence of the negative control drug TG009 (10 M), transfected wild-type HA-Clk1/Sty caused a redistribution of splicing factor SC35 (not shown) and of Clk1/Sty itself from a speckled to a diffuse pattern with enhanced staining by mAb1H4, which specifically recognizes phosphorylated SR proteins (39) (Fig. 3A, panels a, c, e, and g). When we administered 10 M TG003 into the culture media, HA-Clk1/Sty was localized in nuclear speckles in HA-Clk1/Sty-overexpressing HeLa cells with suppressed phosphorylation of SR proteins (Fig. 3A, panels b and f), as observed in cells expressing catalytically inactive HA-Clk1/Sty (Clk1/Sty K190R ) (Fig. 3A, panels d and h). The inhibition of Clk1/Sty-induced phosphorylation by TG003 was further supported by Western blotting analysis (Fig. 3B). COS-7 cells were transfected with HA-Clk1/Sty, HA-Clk1/ Sty K190R , or mock vector as above and incubated in the absence or presence of 10 M TG003 or TG009 for 12 h. Total cellular protein was prepared, fractionated in 8% SDS-polyacrylamide gel, and immunoblotted with mAb104, which also recognizes phospho-SR proteins (40) (Fig. 3B) or mAb1H4 (not shown). When wild-type Clk1/Sty was overexpressed, the band at ϳ75 kDa showed reduced mobility with increased intensity (Fig. 3B, lane 2), compared with mock or Clk1/Sty K190R -transfected cells (Fig. 3B, controls, lanes 1 and 5), suggesting hyperphosphorylation of SRp75 by Clk1/Sty. Administration of TG003, but not TG009, inhibited such effect in Clk1/Sty-overexpressed cells (Fig. 3B, lanes 3 and 4). These data imply that TG003 penetrates into cells and inhibits the kinase activity of Clks in vivo.
To analyze if the TG003 effect is reversible, HeLa cells transfected with HA-Clk1/Sty expression vector were incubated for 12 h with TG003 and then washed and further incubated in fresh medium (Fig. 3C). After the release from TG003 administration, the distribution of HA-Clk1/Sty became diffuse in most cells in 1 h (Fig. 3C, panel c) and almost all cells in 2 h (Fig. 3C, panel d). The level of SR phosphorylation in HA-Clk1/ Sty-positive cells also increased in 2 h (Fig. 3C, panels g and h). Thus, the inhibitory effects of the drug on SR protein phosphorylation and relocalization appeared to be reversible. It should be noted that TG003 appears to have no toxic effect on growth of HeLa and COS-7 cells at 10 M concentration for a few days, because the growth rate and cell cycle profile of TG003-treated and -untreated cells were similar (Supplemental Material Fig. 1).
TG003 Alters Clk1/Sty-regulated Alternative Splicing in Vivo-We next tested if TG003 affects Clk1/Sty-regulated alternative splicing in vivo. Mouse Clk1/Sty isoforms are translated from two alternatively spliced transcripts encoding either a full-length catalytically active protein (Clk1/Sty) or a truncated protein lacking the catalytic domain (Clk1/Sty T ) (17) (Fig.  4A, upper panel). It is reported that Clk1/Sty regulates splicing of its own pre-mRNA according to its kinase activity; increased expression of the catalytically active Clk1/Sty influences splicing to generate the splicing variant that lacks exon 2 and thus encodes the kinase-negative Clk1/Sty T . We assessed the effect of the compound on the kinase activity-mediated exon skipping of Clk1/Sty pre-mRNA by RT-PCR and Western blotting. As shown in Fig. 4A, TG003 suppressed the exon skipping and increased the levels of full-length form (Fig. 4A, lane 3), as observed in cells transfected with the kinase-negative one (Fig.  4A, lane 5). The effect of TG003 on a different type of alternative splicing was further tested (Fig. 4B). The adenovirus E1A pre-mRNA is spliced into three predominant mRNA variants termed 13 S, 12 S, and 9 S mRNAs, through the use of three alternative 5Ј splice sites and a single 3Ј splice site (41). COS-7 cells were transfected with a reporter adenovirus E1A gene (31). Co-transfection of Clk1/Sty increased the use of the most distal 5Ј splice site, which gives rise to the 9 S isoforms (31) (Fig. 4B, lane 2). TG003 also inhibited the production of the 9 S isoform (Fig. 4B, lane 5). Thus, the alteration of splicing site selection induced by Clk kinase activity was suppressed by TG003 in mammalian cells.
TG003 Affects the Alternative Splicing of Endogenous Genes-We wondered whether TG003 induces changes in the splicing profile of endogenous genes, and we analyzed those of Clk1/Sty and SC35, because the alteration of splicing pattern of these genes by drug treatment has been reported (32,42).
Among several mouse cell lines tested, RT-PCR revealed that immortal embryonic fibroblasts (STO cells) showed changes in splicing profiles of both genes by administration of 10 M TG003 for 4 h (Fig. 5). In untreated cells, PCR product corresponding to the short form (183 nt), which produces kinasenegative Clk1/Sty T , was observed in addition to the long form (274 nt) producing the full-length (kinase-positive) Clk1/Sty (Fig. 5A). This short form disappeared when cells were administered TG003, in good agreement with the feedback regulation of Clk expression (17) (Fig. 4A). The subtle change of SC35 splicing profile was also observed (Fig. 5B). In untreated cells, PCR products corresponding to the major (668 nt) and the minor (170 and 274 nt) transcripts for SC35 were detected. TG003 treatment increased the band intensity of 274 nt and decreased that of 668 nt. These results indicate that alternative splicing of endogenous genes could be controlled by TG003.
TG003 Suppresses Developmental Abnormality Induced by xClk-To evaluate the potential use of TG003 in whole animal body, we used X. laevis embryo as a model system. As it was reported that the Drosophila homologue of Clk1/Sty, DOA (darkener of apricot), is essential during early embryonic development (43,44), Xenopus Clk homologues could also play important roles during development. In a data base, we found a cDNA sequence of Xenopus Clk (xClk; GenBank TM accession number BC043963), whose amino acid sequence is most homologous to mammalian Clk2 (ϳ70% identity at the amino acid level) in the Clk family (Supplemental Material Fig. 2). By using RT-PCR, the expression of xClk mRNA during early development was analyzed, and it appears through all stages of Xenopus embryos (Fig. 6A). We prepared recombinant xClk protein and found that the kinase activity was sensitive to TG003 at similar dose ranges as mouse Clk2 (Fig. 6B and Fig.  1B). Dorsal injection of xClk mRNA induced morphological abnormalities in the dorsal mesoderm and ectoderm (Fig. 6C, panels b and f), suggesting that an increase in xClk kinase activity disturbs normal embryogenesis. Indeed, the abnormal development phenotype of the Xenopus embryos was rescued when they were incubated with 10 M TG003 (Fig. 6C, panels  d and h, and Fig. 6D).

DISCUSSION
Although a number of studies reported that Clk kinases can modulate the localization and function of SR proteins in the nucleus (17,23,24), the biological functions of mammalian Clk remain unknown. It is possible that Clk family kinases are key regulators of SR protein function, which in turn regulate alternative splicing, by phosphorylating SR proteins. Because the phosphorylation also affects the subcellular localization (15) and the stability of a particular SR protein (45), Clk kinase activity can regulate the balance of alternatively spliced forms in a developmental stage-and tissue-specific manner. Clk ho-mologues have been isolated from distantly related species, including Saccharomyces cerevisiae (46), Arabidopsis thaliana (47), and Drosophila melanogaster (43). These kinases share the conserved amino acid motif "EHLAMMERILG" in the kinase subdomain X, which has led these kinases to be dubbed "LAMMER" kinases (43). In Drosophila, Doa protein is required for segmentation and development of the nervous system, and Doa mutations are almost invariably recessive lethal (44). Du et al. (48) showed that mutations in the Doa locus affect sexual differentiation by specifically disrupting sex-specific splicing of doublesex pre-mRNA through a genetic interaction with the SR-like proteins TRA and TRA2. Thus, it is likely that the kinase activity of DOA is regulated depending on the developmental stage. Here, we observed stable expression of xClk in Xenopus embryo, but overexpression of xClk induced embryonic defects. This result suggests that the kinase activity of Clk is tightly regulated during vertebrate embryonic development. Although the regulatory pathways of Clks remain unknown, a regulatory protein that specifically binds to unphosphorylated Clk4 protein was cloned by two-hybrid  1 and 4). Top, diagram of the E1A mRNAs generated by alternative 5Ј splice site selection and a primer set used for RT-PCR (31). Bottom, RT-PCR. The position of different spliced products is indicated. screening (49). TG003, a specific inhibitor of Clk1/Sty and Clk4, will be a valuable tool to dissect the regulatory mechanisms involving SR protein phosphorylation in vivo and may be applicable for the therapeutic manipulation of abnormal splicing.
To date, a number of diseases caused by mis-splicing have been reported; in some cases, mutation(s) found around splice sites appear to be responsible for changing the splicing pattern of a transcript by unusual exon inclusion or exclusion and/or alteration of 5Ј or 3Ј sites (reviewed in Refs. [3][4][5]. A typical example is ␤-thalassemia, an autosomal recessive disease, which is often associated with mutations in intron 2 of the ␤-globin gene. The generation of aberrant 5Ј splice sites activates a common 3Ј cryptic site upstream of the mutations and induces inclusion of a fragment of the intron-containing stop codon. As a result, the amount of functional ␤-globin protein is reduced. For therapeutic modulation of alternative splicing, several trials with antisense oligonucleotide (reviewed in Ref. 50), peptide nucleic acid oligonucleotide (51), and RNA i (52, 53) have been reported. These approaches could be useful for manipulating a specific splice site selection of a known target sequence like ␤-globin (50). However, the aberrant splicing, found in the patients of breast cancer, Wilm's tumor, and amyotrophic lateral sclerosis (ALS), are not always accompanied with mutations around splice sites. In sporadic ALS patients, EAAT2 (excitatory amino acid transporters 2) RNA processing is often aberrant in motor cortex and in spinal cord, the regions specifically affected by the disease. As exon 9 is aberrantly skipped in some ALS patients without any mutation in the gene (54), the disorders could be attributed to abnormalities in regulatory factors of splicing. Actually the balance of alternative splicing products can be affected by changes in the ratio of heterogeneous nuclear ribonucleoprotein and SR proteins (28,31) and in the phosphorylation state and localization of SR proteins (17,23). Because the expression of Clk increases the level of SR phosphorylation and leads to exon skipping, suppression of the kinase activity by TG003 may rescue the splicing aberration produced by exon skipping as observed in EAAT2 mRNA. In addition to ALS, TG003 may be applicable for spinal muscular atrophy by increasing an exon inclusion in SMN2 (survival of motor neuron 2) gene to produce functional SMN2 if Clk is involved in SMN2 exon skipping. Some other small molecules (e.g. aclarubicin (55) and sodium butyrate (56)) have potency to increase an exon inclusion of SMN2 gene. However, the mechanisms of these effects remain to be unknown. Moreover, because aclarubicin and sodium butylate were found as an anti-cancer reagent and a histone deacetylase inhibitor affecting transcription, respectively, these compounds have obvious pleiotropic effects other than splicing.
As for the inhibitors of Clk family, 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole (DRB) was shown to influence endogenous Clk2 autophosphorylation levels and its subnuclear localization (57). Although DRB has been reported to inhibit the broad range of protein kinases, including casein kinase II (58) and P-TEFb (59), combination of DRB and the newly developed TG003, a specific inhibitor of Clk family kinases, may give us clues to clarify the Clks-mediated signal pathways and their biological functions.