Biochemical Characterization and Localization of the Dual Specificity

Journal of Cell Science 113, 3241-3253 (2000) 3241 Printed in Great Britain © The Company of Biologists Limited 2000 JCS1491

Biochemical characterization and localization of the dual speciﬁcity kinase CLK1

Harry J. Menegay, Michael P. Myers, Fred M. Moeslein and Gary E. Landreth* Alzheimer Research Laboratory, Department of Neurosciences, Case Western Reserve University, School of Medicine, 10900 Euclid Ave., Cleveland, Ohio 44106, USA *Author for correspondence (e-mail: [email protected])

Accepted 1 July 2000; published on WWW 22 August

SUMMARY

CLK1 was one of the first identified dual specificity kinases negatively regulate CLK1 activity. CLK1 activity is and is the founding member of the ‘LAMMER’ family of positively regulated by phosphorylation on either tyrosine kinases. We have established the substrate site specificity of residues or serine/threonine residues, and is negatively CLK1. We report here that truncation of the N terminus regulated by steric constraints mediated by the N-terminal of CLK1 resulted in a dramatic increase in CLK1 domain, as well as, by phosphorylation on a subset of enzymatic activity, indicating that the N terminus acts as a serine/threonine residues within the catalytic domain. negative regulatory domain. The N-terminal truncation CLK1 mRNA is expressed at low levels in all tissues and resulted in a 45-fold increase in Vmax, suggesting that this cell lines examined. The full-length and truncated splice domain does not contain a pseudo-substrate motif, but may forms are expressed at roughly equivalent levels in most act to conformationally constrain the catalytic activity of tissues. The ratio of the two splice variants of CLK1 can be CLK1. Tyrosine phosphorylation has been proposed to be altered by treatment with cycloheximide. CLK1 protein critical for CLK1 activity, however, CLK1 activity was expression is limited to a small subset of highly localized unaffected by exposure to tyrosine phosphatases. neuronal populations in the rat brain. Contrary to previous Treatment of CLK1 with the serine/threonine specific studies using overexpression systems, we show that CLK1 phosphatase PP2A, resulted in a 2- to 6-fold increase in protein is primarily found in the cytoplasm of these cells, enzymatic activity. Incubation of CLK1 with tyrosine with only a small fraction localized to the nucleus. phosphatases in combination with PP2A abolished CLK1 activity. These data suggest that CLK1 is regulated by three distinct mechanisms that serve to both positively and Key words: Phosphatase, Cytoplasmic, Neuronal

INTRODUCTION the dual specificity kinases is indistinguishable from that of the serine/threonine kinases. Traditionally, protein kinases have been divided into two non- One of the first dual specificity kinases to be discovered was overlapping families: the protein tyrosine kinases and the CLK1 (cdc2-like kinase (Ben-David et al., 1991; Howell et al., serine/threonine kinases (Hanks and Quinn, 1991; Hanks et al., 1991; Johnson and Smith, 1991). CLK1, also termed STY, was 1988). The protein tyrosine kinase family largely consists of initially isolated in a screen designed to identify tyrosine the receptor protein kinases and cytoplasmic tyrosine kinases kinases (Ben-David et al., 1991; Howell et al., 1991). CLK1 linked to intracellular signaling mechanisms (Ullrich and has been shown to autophosphorylate on serine, threonine and Schlessinger, 1990). The remainder of protein kinases belong tyrosine residues and phosphorylate exogenous substrates on to the serine/threonine family of protein kinases and are serine and threonine residues (Ben-David et al., 1991; Duncan involved in the regulation of a wide variety of cellular functions et al., 1995; Howell et al., 1991). Tyrosine phosphorylation of (Hanks and Hunter, 1995). These two distinct families were CLK1 has been proposed to be an important determinant of originally identified based on conserved amino acid motifs enzyme activity, as the majority of CLK1 activity can be within their catalytic domains (Hanks et al., 1988). Due to the immunoprecipitated with antibodies to phosphotyrosine absolute specificity for their respective substrate residues, it (Howell et al., 1991). CLK1 has recently been shown to belong was believed that protein kinases exclusively phosphorylated to a small subfamily of protein kinases which contains at least either tyrosine residues or serine/threonine residues. The recent four highly conserved isoforms CLK1, CLK2, CLK3, and discovery of protein kinases capable of phosphorylating all CLK4 (Hanes et al., 1994; Nayler et al., 1997). CLK three hydroxyl amino acids, termed the dual specificity kinases, homologues have been isolated from a number of organisms has forced a re-evaluation of protein kinase specificity including: Arabidopsis, Drosophila, mouse and human (Ben- (Lindberg et al., 1992). Surprisingly, the catalytic domain of David et al., 1991; Duncan et al., 1995; Hanes et al., 1994; 3242 H. J. Menegay and others

Howell et al., 1991; Johnson and Smith, 1991). All four amplified using a plasmid containing the full length CLK1 cDNA as isoforms have a highly conserved domain structure. The a template and the high fidelity Vent polymerase (New England conserved kinase domain is located at the C terminus of the Biolabs). Expression of CLK1 was induced by stimulating 500 ml of molecule and contains the signature amino acid motif mid-log phase bacteria with 1 mM IPTG for 4 hours at 30°C. The ‘EHLAMMERILG’ in subdomain X, which has led these cells were harvested by centrifugation, resuspended in 3 ml of sonication buffer (50 mM NaH2PO4, 10 mM Tris, 100 mM NaCl, pH kinases to be dubbed ‘LAMMER’ kinases (Hanes et al., 1994; 8.0) containing 20 µg/ml polymyxin B (Sigma) and incubated on ice Yun et al., 1994). The 160 amino acid N-terminal domain has for 10 minutes. The resuspended cells were sonicated 3× 1 minute and been proposed to comprise a putative regulatory domain and then centrifuged for 10 minutes to clear the lysate. Cleared lysates includes a bipartite nuclear localization signal (Duncan et al., from hexahistidine-CLK1 cells were bound in batch to 1 ml of Talon 1995). With the exception of the nuclear localization signals, Affinity Resin (Clontech), while GST-CLK1 lysates were bound to 1 the N-termini of the CLK isoforms are only distantly related ml of glutathione Sepharose (Pharmacia). The resins were washed 3 (Hanes et al., 1994). Surprisingly, mRNAs for all four CLK times with 10 ml of sonication buffer and then 3 more times with isoforms are alternatively spliced to produce proteins in which sonication buffer at pH 7.4. Hexahistidine-CLK1 was eluted with an the kinase domain is missing, resulting in the expression of imidizole elution buffer (250 mM imidizole, 10 mM Tris, 30% proteins which comprise only the putative N-terminal ethylene glycol, pH 8.0) while GST-CLK1 was eluted with a glutathione elution buffer (10 mM glutathione, 10 mM Tris, 30% regulatory sequences (Duncan et al., 1995; Hanes et al., 1994). ethylene glycol, pH 8.0). A CLK1 N-terminal truncation (trCLK1) Mutations in the Drosophila CLK homologue, darkener of was created by PCR using the following PCR primers apricot (DOA) suggest that CLK family members play an (5′CGGGATCCGGGGAAGAGTCACCGAAGG3′, 5′CCCGGAT- important role during development. Mutations in DOA are TCCTCACGTATGCTTTTTAAGTGG3′) which truncate the 128 N- embryonic lethal and lead to defects in differentiation, terminal amino acids leaving the entire CLK1 kinase domain intact. including abnormalities in segmentation, eye formation, and The PCR was performed using a plasmid containing the full length neuronal development (Yun et al., 1994). The importance of CLK1 and Vent polymerase. The PCR product was cloned into pGEX- CLK family members during neuronal development is 2T and the GST-trCLK1 fusion protein purified as described. supported by the finding that expression of CLK1 in PC12 cells Protein kinase assays causes the cells to undergo neuronal differentiation (Myers et The protein kinase assays employed in these studies utilized 0.5 µg al., 1994). The ability of CLK1 to differentiate PC12 cells was of recombinant CLK1 in the presence of 1 µg of myelin basic protein 32 associated with the CLK1-dependent activation of members of (MBP), 10 mM MgCl2, 2 mM MnCl2, 10 µM [γ- P]ATP (44 the MAP kinase cascade, suggesting that CLK1 normally dpm/fmol) in kinase buffer (10 mM Tris, 1 mM EGTA, 100 µM functions in signal transduction cascades (Myers et al., 1994). sodium ortho-vanadate, 20 mM p-nitrophenyl phosphate (PNP), pH Indeed, the Arabadopsis CLK1 homologue, AFC1, 7.4) in a reaction volume of 50 µl. The reactions were incubated at complements mutations in the yeast MAP kinases KSS1 and room temperature for 20 minutes and stopped with the addition of 3× FUS3 (Bender and Fink, 1994). Both CLK1 and CLK2 have Laemmli sample buffer and boiled. The reaction products were recently been shown to phosphorylate and activate the tyrosine separated by SDS-PAGE and visualized by autoradiography of the dried gels using a phosphoimager (Molecular Dynamics). phosphatase PTP1B (Moeslein et al., 1999). CLK1 may also Incorporation of radioactivity into MBP was quantitated by play a role in regulating RNA splicing, as CLK1 has been Cherenkov counting of MBP-containing gel fragments. MBP peptides found to form complexes with and phosphorylate members of (LC Laboratories), S6 Peptide (UBI), PKCζ substrate peptide (from the SR family of splicing factors (Colwill et al., 1996a,b) and Dr M. Wooten), Tau peptide (from Dr S. Feinstein), Myosin peptide has been shown to affect the splicing of mRNA (Duncan et al., (from Dr T. Egelhof), and Crebtide were used at a concentration of 1997). It has recently been demonstrated that this affect is the 50 µM under standard reaction conditions. Peptide kinase reactions result of the phosphorylation of SR proteins, but not other were stopped by the addition of BSA to 0.2% and TCA to 7% and essential splicing factors (Prasad et al., 1999). incubated on ice for 15 minutes. Following a 10 minute centrifugation at 10,000 g, aliquots of the supernatants were applied to Whatman A better understanding of the biochemical characteristics of × CLK1 will provide additional insights into the function and P81 paper which were then washed 3 5 minutes in 75 mM H3PO4 as previously described (Glass et al., 1978). The incorporated regulation of this potentially important family of kinases. radioactivity was determined by Cherenkov counting of the washed Previous studies have investigated the expression of CLK1 papers. mRNA in some tissues, but have not examined expression of the endogenous protein. This study describes the identification Sequencing of MBP peptides of the consensus phosphorylation site of CLK1 and examines Myelin basic protein (50 µM) was phosphorylated in a kinase reaction the role that phosphorylation and the N-terminal domain play containing 5 µg of trCLK1 in kinase buffer containing 10 µM ATP in regulating CLK1 activity. We also describe the expression (44 dpm/fmol), 10 mM MgCl2, and 2 mM MnCl2 for 2 hours. The pattern and localization of CLK1 in the rat tissues and in cell reaction products were separated by SDS-PAGE and then transferred lines. to PVDF as previously described (Myers et al., 1994). The phosphorylated MBP was excised from the membrane and digested with endoproteinase AspN. The digested peptides were separated by reverse phase HPLC and the phosphorylated peptides were sequenced MATERIALS AND METHODS by Dr C. Beach (University of Kentucky).

Expression of CLK1 and trCLK1 in bacteria Phosphatase sensitivity CLK1 was expressed either as a hexa-histidine or a GST fusion Full length and trCLK1 were treated for 20 minutes at room protein. Briefly, the full length CLK1 was amplified by PCR and temperature with 2 units of PP2A, YPTP1, or PTP-1B in phosphatase cloned into the BamHI sites of pQE10 (Qiagen Corp.) and pGEX-2T buffer (10 mM Tris, 1 mM EGTA, 1 mM EDTA, 1 mM DTT, 20 mM (Pharmacia) and transformed into the JM109 E. coli strain. CLK1 was MgCl2, pH 7.4). Phosphatase inhibitors were added to yield final Characterization and localization of CLK1 3243 concentrations of 5 µM okadaic acid, 100 mM ortho-vanadate, and 20 radiation (290-320 nm) with a Stratalinker (Stratagene) and then mM p-nitrophenylphosphate (PNP). The resulting mixture was incubated for an additional 60 minutes. RNA was then isolated and 32 supplemented with [ P]ATP (44 dpm/fmol), MnCl2 to 2 mM and 1 CLK1 and GAPDH message were amplified and analyzed as above. µg of MBP and assayed as above. Antibody production Phosphoamino acid analysis Members of the CLK family, while highly conserved in their kinase MBP, histone VIIIS (arginine rich), and casein were phosphorylated domain, are very divergent their N-terminal domain. A synthetic by trCLK1 under standard reaction conditions and incubated at 20°C peptide (KRTYCPDWDERD) was synthesized which corresponds to for 40 minutes. The reaction products were separated by SDS-PAGE a part of this N-terminal region that is not conserved among other and transferred to PVDF membrane. The membrane was stained with CLK family members and has no significant sequence homology with Coomassie blue and then destained in 10% glacial acetic acid, 15% other genes in the GenBank database. The peptide was conjugated to methanol and autoradiographs were obtained. The stained radioactive KLH using the Imject Supercarrier System for Peptides (Pierce). This proteins were excised from the gel, subjected to Cherenkov counting, conjugated peptide was injected into rabbits with Freund’s Complete submerged in 100 µl of 6 N HCl and incubated for 45 minutes at Adjuvant to generate a polyclonal antiserum in rabbits. Affinity 110°C and phosphoamino acid analysis was performed essentially as purified CLK1 antibody was prepared from the serum by passing the previously described (Boyle et al., 1991). Standards were visualized serum over two columns. The first column consisted of BSA by spraying the dried plates with 0.1% ninhydrin. Radioactive amino conjugated to CNBr-activated Sepharose 4B (Pharmacia Biotech) to acids were visualized by autoradiography. remove non-specific interacting antibodies (Porath and Axen, 1976). The second column was prepared by first conjugating Phosphopeptide analysis/two-dimensional CLK1-synthetic-peptide to BSA using 1-ethyl-3(3- phosphopeptide mapping dimethylaminopropyl)carbodiimide-HCl (Pharmacia Biotech). The TrCLK1 was autophosphorylated under standard reaction conditions product of this reaction was then conjugated to CNBr-activated and incubated at 20°C for 40 minutes. The reaction products were Sepharose 4B. CLK1 specific antibodies were eluted from the second separated by SDS-PAGE and transferred to nitrocellulose membrane. column with 100 mM glycine (pH 2.5), or 100 mM triethanolamine Autoradiographs were obtained from the membrane and a fragment (pH 11.5) and were then neutralized in 100 mM Tris (pH 7.4). The of the membrane, which contained the radioactive protein, was cut eluted antibodies were concentrated with a Centricon filter to working out. The 32P-labeled proteins were proteolytically digested as concentration (0.1 µg/µl). The affinity purified CLK1 antisera was described by Luo et al. (1991). Briefly, the immobilized proteins were immunoreactive by western analysis with both the synthetic CLK1- digested with either trypsin or chymotrypsin or double digested with peptide as well as recombinant full-length CLK1 protein expressed in both enzymes. The digestion was performed by addition of 10 µg of bacteria, however it was unable to immunoprecipitate the protein. enzyme/200 µl for 1.5 hours at 37°C, at which point 10 µg of enzyme/200 µl were added, and a second 1.5 hour digest at 37°C was Protein expression in rat tissues performed. The supernatants were transferred to another tube and Tissue lysates were made by freezing dissected adult rat tissues in dried in a Speed-vac. Radioactivity incorporated into the pellets of liquid nitrogen and pulverizing the tissue with a pestle and mortar. digested peptides was determined by Cherenkov counting, and the The powdered tissues were then resuspended in lysis buffer (1% NP- pellets were resuspended at 1000 cpm/µl. 40, 2 mM NaF, 500 µM sodium vanadate, 200 µg/ml The 32P-labeled phosphopeptides were resolved in the first phenylmethylsulfonyl fluoride, 2 µg/ml aprotinin, 5 µg/ml leupeptin, dimension by electrophoresis at pH 1.9 on TLC plates as described 150 mM NaCl, 50 mM Tris, pH 8.0) and sonicated for 3 times for 30 (Boyle et al., 1991). The plates were dried, and then subjected to seconds. Cell lysates were prepared by resuspending pelleted cells in ascending chromatography in the second dimension using a buffer lysis buffer followed by sonication for 30 seconds. Proteins from each composed of isobutanol, pyridine, acetic acid, and water tissue or cell line (50 µg) were separated on SDS-polyacrylamide gels, (75:15:50:60). The plates were dried and the phosphopeptides were and transferred to PVDF membrane by electroblotting. The blots were visualized by autoradiography. probed with the purified CLK1 antisera (0.5 µg), or CLK1 antisera (0.5 µg) that had been preincubated with the immunizing CLK1 RNA extraction and RT-PCR peptide (10 ng). CLK1 was detected with a horseradish peroxidase- RNA was isolated from cell lines and mouse tissues using the method conjugated goat anti-rabbit antibody and visualized by ECL reagents of Chomczynski and Sacchi (1987). Total RNA (1 µg) was used in a (Amersham). reverse transcription reaction using SuperScript II reverse transcriptase (Gibco) primed by oligo(dT). A 50-µl volume PCR Immunostaining of tissues reaction was carried out using a fraction (1/100) of the reverse Adult rats were perfused with Bouin’s fixative and dissected tissues transcription reaction using Taq DNA polymerase (Boehringer were then incubated for 24 hours in the same fixative (Humason, Mannheim) including 50 µCi [α-32P]dCTP. The CLK1 PCR was 1962). Following fixation the tissues were submersed for 72 hours in carried out for 30 cycles, while the GAPDH PCR was carried out for 18% sucrose/PBS. The tissues were frozen in OCT (Miles Inc.) on 25 cycles. Additional PCR reactions were done at higher and lower dry ice, and 10 µM sections were prepared on a cryostat. The tissue cycle numbers to assure the amplification analyzed was in the linear sections were then probed with the purified CLK1 antibody, or range. Primers used were: upstream primer 5′-ATG AGA CAT TCA purified CLK1 antibody which had been preincubated with the peptide AAG AGA ACT TAC TGT CCT-3′, and downstream primer 5′-CCG the antibody was generated against. Visualization was by either a AAT TCC TGC TAC ACG TCT ACC TCC CAC-3′. Products were Texas Red-conjugated or fluorescein-conjugated secondary antibody separated on polyacrylamide gels, dried, and imaged using (Jackson ImmunoResearch Laboratories Inc.) using a Leitz phosphorimager plates (Molecular Dynamics). The predicted size of microscope or Zeiss confocal microscope. the full-length kinase containing PCR product is 560 bp and the truncated kinase-less PCR product is 453 bp. Parallel PCR reactions Expression of CLK1 in PC12 cells were performed using primers to GAPDH as RNA loading controls. PC12 cells grown on coverslips were fixed in methanol at –20°C, or in 0.4% paraformaldehyde, 4% paraformaldehyde, Acrolein fixative, Analysis of CLK1 isoform mRNA or Bouin’s fixative on ice for 20 minutes. After extensive washing the PC12 cells were treated with 500 mM NaCl or 10 µg/ml cells were probed with the CLK1 antibody and visualized with Texas- cycloheximide for 60 minutes, or were exposed to 300 J/m2 of UV Red-conjugated secondary antibody on a confocal microscope. 3244 H. J. Menegay and others

Overexpression of CLK1 in 3T3 cells this region of MBP (MBP104-118) was efficiently pIND plasmid (Invitrogen) containing the CLK1 gene and pVgRXR phosphorylated by CLK1 (Fig. 4), verifying that Ser 109 was were cotransfected into NIH-3T3 cells with calcium phosphate. The phosphorylated by CLK1. This major site of phosphorylation pVgRXR plasmid constitutively expresses the ecdysone receptor, on MBP by CLK1 is different from that of another LAMMER while pIND contains an ecdysone-responsive promotor. It was found family member DOA, whose major site of phosphorylation on that steroid induction with ecdysone was not necessary for CLK1 MBP is serine 164 (SGSPMAR) (Lee et al., 1996). protein production in this system. Transfected cells were selected for three weeks in media containing Geneticin (G418) and zeocin to obtain stably expressing cells. The cells were then grown on Determination of CLK1 phosphorylation consensus coverslips. Localization of the CLK1 protein was observed by either site fixation in Bouin’s fixative for 20 minutes on ice, or in methanol at In order to determine the consensus site of phosphorylation, –20°C for 20 minutes, followed by extensive washing and probing the ability of CLK1 to phosphorylate other peptides was tested. with the CLK1 antibody. CLK1 was visualized using a Texas-Red CLK1 efficiently phosphorylated MBP104-118 (GKGRGLS- conjugated secondary antibody on a confocal microscope. LSRFSWGA), PKCζ substrate (ERMRPRKRKRQGSVRRR), Crebtide (REILSRRPSYRK), but did not efficiently phosphorylate kemptide (KLRRASLG), or a peptide derived RESULTS from myosin (KKRAARATSNVFA) (Fig. 3A). CLK1 was also unable to phosphorylate a casein kinase substrate peptide CLK1 substrate specificity (RREEETEEE), MBPtide (APRTPGGRR), a peptide derived A battery of substrates were tested for their ability to be from c-fos (RKGSSSNEPSSD) or a battery of tyrosine kinase phosphorylated by CLK1. CLK1 phosphorylated casein, peptides, including a src substrate peptide histone VIIIS (arginine rich histones), and MBP, with the (KVEKIGEGTYGVVYK), Raytide-ELtm (Oncogene histones and MBP being the best substrates (Fig. 1A). Sciences), or polyGluTyr (data not shown). These data Phosphoamino acid analysis of these substrates revealed that indicated that a basic residue is necessary at positions +3 and CLK1 phosphorylated MBP and Histone VIIIS predominantly −3 relative to the phosphorylated residue (Table 1) and that the on serine residues. Casein was phosphorylated on serine, and tyrosine kinase activity of CLK1 is likely to be highly at lower levels on threonine and tyrosine residues (Fig. 1B). restricted. This conclusion was confirmed by using mutant CLK1 did not phosphorylate histone IIIS (lysine rich histones), MBP104-118 peptides in which either arginine 107 or arginine lysozyme, or enolase (data not shown). 113 were changed to Alanine. As predicted by the peptide specificity data, these peptides were not efficiently Determination of the CLK1 phosphorylation site in phosphorylated by CLK1 (Fig. 3B). Mutation at arginine 107 MBP resulted in a 90% decrease in phosphorylation and increased Further biochemical characterization was performed using the Km 7.8 fold (Table 1). Similarly, mutation of arginine 113 MBP as a CLK1 substrate. The stoichiometry of resulted in a 92.5% decrease in phosphorylation and increased phosphorylation was determined by incubating 1 µM MBP the Km 11-fold (Table 1). The apparent decrease in activity was with CLK1 for increasing periods of time. The phosphorylation not due to a decrease in the peptides affinity for of MBP reached maximum levels within 1 hour and plateaued phosphocellulose, as similar results were obtained when the at 1 mole 32P/mole MBP (Fig. 2A), indicating that CLK1 phosphorylated peptides were analyzed by SDS-PAGE (data phosphorylated MBP at a single site. The site phosphorylated not shown). These data indicate that basic residues at both −3 by CLK1 was determined by radiosequencing of the and +3 are critical determinants of CLK1 substrate specificity. phosphorylated MBP. Following phosphorylation and The polybasic PKCζ substrate peptide was the best peptide digestion with endoproteinase LysC, the major radioactive substrate tested (Km=452 nM) indicating that basic residues at peptide was sequenced (Fig. 2B). The majority of the other positions flanking the phosphorylation site enhance radioactivity eluted in cycle 5, corresponding to Serine 109 in CLK1 substrate recognition and phosphorylation. However, the MBP. Very little radioactivity eluted in cycle 7 (Ser 111), cycle basic residues at −3 and +3 play the major role in forming a 10 (Ser 114), or cycle 27 (Ser 131), which identified Ser 109 CLK1 phosphorylation site. This consensus site for CLK1 is as the site of phosphorylation in MBP. A synthetic peptide from similar to that of CLK2 which prefers positively charged

Table 1. CLK1 substrate speciﬁcity

Peptide Sequence Relative activity Km (µM) Consensus R-X-X-S*-X-X-R MBP104-118 G-K-G-R-G-L-S*-L-S-R-F-S-W-G-A 100 61 MBP104-118Ala107 G-K-G-A-G-L-S*-L-S-R-F-S-W-G-A 10 480 MBP104-118Ala113 G-K-G-R-G-L-S*-L-S-R-F-S-W-G-A 7.5 674 PKC-zetatide E-R-M-P-R-K-R-Q-G-S*-V-R-R-R 239 0.45 Crebteide R-E-I-L-S-R-R-P-S*-Y-R-K 49 80 Erktide A-P-R-T*-P-G-G-R-R 13 n.d. Kemptide K-L-R-R-A-S*-L-G 9.7 n.d. Myosin peptide K-K-R-A-A-R-A-T*-S-N-V-F-A 7.5 n.d. The sequence of the tested peptides are listed, along with their Km’s for CLK1. The activity toward the individual peptides is normalized to that obtained using MBP104-118 as a peptide substrate. Amino acids in bold indicate the positions identiﬁed as important in forming the consensus phosphorylation site and the phosphorylated residue is indicated by an asterisk. Characterization and localization of CLK1 3245 A

Fig. 1. Substrate specificity of CLK1. (A) CLK1 was incubated alone (Auto) or with casein, histone IIIS, histone VIIIS, and MBP. The reaction’s products were separated by SDS-PAGE transferred to PVDF and autoradiographs were obtained. The positions of the Fig. 2. (A) Stoichiometry of MBP phosphorylation by CLK1. molecular mass standards are shown. (B) Phosphoamino acid Stoichiometry of MBP phosphorylation was determined by analysis of the phosphorylated casein, histone VIII and MBP. The incubating 1 µM MBP with 1 µg of CLK1 for increasing amounts of positions of phospho serine (S), threonine (T), and tyrosine (Y) are time under standard assay conditions. Each time point was performed indicated. in triplicate and the data is displayed as the mean ± s.d. (B) CLK1 phosphorylates MBP on Ser109. MBP was phosphorylated with CLK1 for 2 hours in the presence of [γ-32P]ATP. The reaction residues at –3 and +3 as well as additional positively charged products were separated by SDS-PAGE, transferred to PVDF − membranes and visualized by staining with Commassie blue. The residues immediately N-terminal of 3 (Moeslein et al., 1999). MBP band was excised, digested with endoproteinase AspN and the fragments separated by HPLC. The principal radioactive peak was N terminus of CLK1 acts as a regulatory domain sequenced and the radioactivity liberated in each cycle is indicated CLK1 is a 57 kDa protein kinase with the kinase domain along with the corresponding identified amino acid. located at the C terminus and a putative regulatory domain comprising 160 amino acids at the N terminus. The regulatory in an approximately 40-fold increase in CLK1 activity (Fig. 4). activity of the N-terminal domain was tested by creating a The Km of CLK1 and trCLK1 for MBP and ATP were truncated CLK1 protein in which the N-terminal 128 amino determined by performing Lineweaver-Burke analysis. Only acids have been deleted, leaving the kinase domain intact. The small differences were found in the Km of trCLK1 and flCLK1 resulting N-terminal truncation, termed trCLK1, was expressed for MBP and in the Km for ATP (Table 2). Significantly, the N- as a GST-fusion in bacteria and its ability to phosphorylate terminally truncated CLK1 exhibited a 45-fold increase in MBP was compared to full length CLK1 (flCLK1) expressed Vmax relative to the full-length molecule. The increase in Vmax as either a GST-fusion or a hexahistidine fusion. The specific largely accounts for the increased activity seen with the activities of the GST-flCLK1 and hexahistidine-flCLK1 fusion truncated allele. These data demonstrate that the N terminus of proteins were essentially identical. The hexahistidine fusion CLK1 acts to negatively regulate the C-terminal kinase domain protein, however, was produced with greater efficiency and was by altering the rate of catalysis, rather than by altering the used in these studies. The truncation of the N terminus resulted affinity of the enzyme for it’s substrates, MBP and ATP. 3246 H. J. Menegay and others

Fig. 4. CLK1 is activated by N-terminal truncation. The N terminus of the CLK1 contains a putative regulatory domain. The 140 amino acids comprising this domain were deleted, creating a truncated allele (trCLK1). Kinase activity of trCLK1 and the full length CLK1 (ﬂCLK1) were measured using MBP as a substrate. Each assay was performed in triplicate and is expressed as mean ± s.d.

antibodies to phosphotyrosine, suggesting that phosphorylation of tyrosine residues is an important mode of CLK1 regulation (Howell et al., 1991). In order to determine the relative importance of serine/threonine and tyrosine phosphorylation for CLK1 activity, CLK1 was dephosphorylated using phosphatases specific for phosphoserine/threonine (PP2A), or the tyrosine specific phosphatases YPTP1 and PTP1B. Phosphate was removed from all residues using either alkaline phosphatase or by simultaneous treatment with YPTP1 and PP2A. Treatment with the tyrosine phosphatase YPTP1 had no effect on flCLK1 or trCLK1 activity (Fig. 5A and B). Western blots with antibodies to phosphotyrosine indicated that YPTP1 and PTP1B were able to quantitatively remove tyrosine-bound phosphate from flCLK1 (Fig. 5C). Surprisingly, treatment with serine/threonine phosphatases resulted in the 2-fold activation of flCLK1 (Fig. 5A) and the 6-fold activation of trCLK1 (Fig. 5B). Treatment with a non-specific phosphatase, alkaline Fig. 3. (A) CLK1 peptide substrate specificity. The indicated phosphatase, or the combined treatment of PP2A and YPTP1 peptides were phosphorylated with CLK1 for 20 minutes under completely abolished both flCLK1 and trCLK1 activity (Fig. standard conditions. The incorporated radioactivity was determined 5A and B). These data indicate that the regulation of CLK1 by by Cherenkov counting and is expressed as mean ± s.d. of triplicate phosphorylation is complex, with CLK1 activity being determinations. (B) Sequence determinants of CLK1 substrate dependent on phosphorylation of serine/threonine or of specificity. MBP104-118, MBP104-118Ala107, or MBP104- tyrosine residues, but not both. The activation of CLK1 by 118Ala113 were incubated with 1 µg of CLK1 under the standard PP2A demonstrates that phosphorylation of a subset of kinase reaction conditions as described. Each assay was performed in serine/threonine residues within the kinase domain acts to triplicate and is expressed as mean ± s.d. negatively regulate CLK1 activity. Similar data were obtained using the tyrosine phosphatase PTP1B or the serine/threonine Phosphatase sensitivity phosphatase PP1 (data not shown), indicating that effects seen are not specific to YPTP1 and PP2A. Previous reports have shown that CLK1 undergoes We have attempted the further analysis of the autophosphorylation on serine, threonine and tyrosine residues phosphorylation sites in CLK1 by peptide mapping. Tryptic and that the active form of the recombinant enzyme could be phosphopeptide maps of autophosphorylated CLK1 yield nine quantitatively recovered by immunoprecipitation with major and more than fifteen minor phosphopeptides (Fig. 6). The large number of phosphopeptides is reflective of the Table 2. Kinetic analysis of flCLK1 and trCLK1 complex regulation of the enzyme through phosphorylation at flCLK trCLK Fold change multiple sites. Phosphorylation is involved in both the regulation of the activity and the intracellular localization of KmATP (µM) 80 40 −2 KmMBP (µM) 20 75 +3.7 the enzyme (Nayler et al., 1998). 3 4 Vmax (mole/min/µg) 1.85×10 8.21×10 +45 RT-PCR of CLK1 transcripts The Km of flCLK1 and trCLK1 for ATP and MBP were determined by performing Lineweaver Burke analysis. The Vmax of flCLK1 and trCLK1 CLK1 mRNAs have been detected in a number of tissues by were determined using MBP as the substrate. northern analysis (Ben-David et al., 1991; Hanes et al., 1994; Characterization and localization of CLK1 3247

The full-length kinase-containing form and truncated kinase- less form of the CLK1 transcripts were expressed at similar ratios in all tissues examined. Most tissues had a near 1:1 ratio of the two splice forms. However, hindbrain, lung, and PC12 cells were notable for the much higher levels of the full-length splice form expressed. Stress-induced attenuation of CLK1 mRNA splicing CLK1 has been shown to be capable of regulating the splicing of it’s own mRNA (Duncan et al., 1997). We tested whether expression of CLK1 splice products was subject to regulation by cellular stressors. We examined several stress conditions for a possible induced shift in the splicing of CLK1. PC12 cells were treated with either 500 mM NaCl, 10 µM cyclohexamide, for 60 minutes or exposed to UV light (290-320 nm, 300 J per m2) then incubated for 60 minutes. The cells were then harvested and analyzed for CLK1 mRNA expression. UV exposure or high salt conditions had no effect on the ratio of full-length to truncated spice forms of CLK1. Cycloheximide however had a large effect, changing the ratio dramatically in favor of the truncated kinase-less form of mRNA (Fig. 8). It has previously been shown that inhibition of protein synthesis may cause a superinduction of short-lived transcripts and has been suggested to be due to stabilization of the transcripts or increased gene expression (Greenberg et al., 1986; Meijlink et al., 1985; Ringold et al., 1984; Shaw and Kamen, 1986). Western blot analysis CLK1 mRNA is detectable in all tissues examined however the expression pattern of CLK1 protein is unknown. Tissue mRNA expression does not always correlate with protein expression. The analysis of the intracellular localization has been entirely reliant on analysis of CLK1 distribution following the overexpression of epitope tagged molecules. In order to determine the actual pattern of CLK1 protein expression, tissue lysates were analyzed for CLK1 protein. A CLK1 specific polyclonal antiserum was produced against a peptide corresponding to a part of the N terminus of CLK1. This antiserum was affinity purified using a column containing the peptide the antisera was generated against. The resulting Fig. 5. Phosphatase sensitivity of CLK1. The activity of flCLK1 (A) purified antiserum was immunoreactive against bacterially or trCLK1 (B) following treatment with tyrosine phosphatases expressed CLK1, and could detect 10 ng of the purified (YPTP1), treatment with PP2A, treatment with both YPTP1 and recombinant protein on dot blots. This immunoreactivity was PP2A, or nonspecific alkaline phosphatase. (C) CLK1 was incubated blocked by pre-incubation with the synthetic CLK1 peptide. in the absence or presence of YPTP1 or PTP1b and then analyzed by Lysates of rat tissues and cell lines were examined for the western blotting using PY20 anti-phosphotyrosine antibodies and presence of CLK1 protein using this antibody. No CLK1 visualized with ECL reagents. protein, of either the predicted 57 kDa full-length form or the 16 kDa kinase-less truncated form were detected in any of the tissue lysates examined. Abundant full-length, but not Howell et al., 1991; Johnson and Smith, 1991; Nayler et al., truncated CLK1 protein was observed in PC-12 cell lysates 1997). Northern analysis, however, does not differentiate (Fig. 9). between mRNAs containing the two known splice variants of CLK1. CLK1 mRNAs exist in two alternative spliced forms, Immunostaining of rat brain encoding either the full length CLK1 or a transcript encoding The examination of whole tissue lysates may only provide a kinase-deficient form comprised of only the N-terminal information about expression of proteins that are expressed in regulatory domain (Duncan et al., 1995). We have surveyed a a large proportion of the cells in a tissue and may not detect number of tissues to determine if both mRNA splice variants expression in specific cell types within the tissue. The observed were expressed and in what ratio the transcripts were expression of CLK1 mRNA in all tissues examined and the expressed, by RT-PCR on total RNA. All tissues examined failure to detect CLK1 protein within lystes of these whole (forebrain, hindbrain, heart, lung, liver, kidney, spleen, and tissues lead us to examine rat brain for possible expression by muscle) contained both CLK1 mRNA splice forms (Fig. 7). CLK1 in sub-populations of cells within this tissue. Most 3248 H. J. Menegay and others

Fig. 6. Phosphopeptide analysis. HA-Tagged CLK1 protein expressed in bacteria was autophosphorylated and digested with cyanogen bromide, trypsin, chymotrypsin, or both trypsin and chymotrypsin. The digestion products were separated on a 2-D gel. tissues examined (heart, lung, liver, kidneys, spleen, muscle, CLK1. In cells fixed in 4% paraformaldehyde or Acrolein skin, bone, blood vessles) had no detectable staining (data not fixative, the majority of the CLK1 staining was cytoplasmic, shown), however there was robust staining in a subpopulation and the nuclei exhibited low levels of immunoreactivity of cells in the adult rat brain. Cells exhibiting a neuronal compared to the cytoplasm. Fixation in Bouin’s fixative phenotype were stained throughout the brain (Fig. 10). Notably resulted in cytoplasmic staining, however much of the CLK1 Purkinje cells in the cerebellum, cells of the inferior olive, epitope appeared to be localized to perinuclear spots in the raphe nucleus, hippocampus, and frontal cortex a showed cytoplasm. expression of CLK1 protein. Significantly, CLK1 was found principally within the cytoplasm of the neurons. This staining Immunostaining of CLK1 transfected 3T3 cells could be blocked with the preincubation of the antisera with To confirm the results from the PC-12 cells, and to verify that the CLK1 peptide the antibody was generated against. the protein the antibody was reacting with was indeed CLK1, NIH 3T3 cells were transfected with a CLK1 expression Immunostaining of PC12 cells construct. Upon fixation in Bouin’s fixative staining in CLK1- The immunohistochemical detection of CLK1 within the expressing NIH-3T3 cells was limited to the cytoplasm. cytoplasm of neurons with little or no nuclear staining was in Fixation with methanol resulted in staining mainly in the sharp contrast to previous studies examining the localization of nucleus, although there was staining in the cytoplasm of these CLK family members within cells. These previous studies have cells (Fig. 12). The nuclear staining was mainly localized to shown that the protein is principally localized to ‘speckles’ in ‘speckles’ as have been previously reported in other the nucleus, presumptive sites of storage for splicing factors overexpresssion studies. Fixation in methanol precipitates but (Huang and Spector, 1996; Jimenez-Garcia and Spector, 1993; does not cross-link proteins and may allow for their movement Mattaj, 1994). However, these studies have only been done in during the fixation process. Bouin’s fixative, however, acts to CLK overexpression systems that could misrepresent both precipitate and cross-links proteins and may provide a localization of the native protein by expressing the protein at much higher levels than the endogenous protein. CLK1 protein contains nuclear localization motifs within its N terminus and expression in excess of any cytoplasmic binding protein may result in artifactual nuclear localization of the protein. We have recently shown that another CLK family member, CLK3 is localized to the cytoplasm, rather than the nucleus of cells (Menegay et al., 1999). Although the localization of CLK3 to the nucleus may lead to inferences about the localization of a family member, their regulation may be very different. While the kinase domains of CLK1 and CLK3 are fairly well conserved with 62% identity, their N-terminal domains, which contain their putative nuclear localization are very poorly conserved, with only 33% identity (Hanes et al., 1994). In order to determine the localization of native CLK1 protein we Fig. 7. CLK1 mRNA expression in adult rat tissues and cell lines detected by RT-PCR. (A) Total RNA was isolated from cell lines and examined the distribution of CLK1 in PC12 cells. PC12 cells µ were fixed in a variety of fixatives and examined with the rat tissues. The RNA (1 g) was reverse transcribed, then amplified by PCR. The PCR of full-length kinase domain containing CLK1 antibody. Fixation in methanol resulted in detection of transcripts results in a 560 bp product, while the truncated, kinase- nuclear staining of the overexpressed and epitope tagged CLK1 defidient transcripts yields a 453 bp product. Lanes: F, forebrain; protein (Fig. 11). Fixation in 0.4% paraformaldehyde also Hb, Hindbrain, H, Heart; Lg, Lung; Li, Liver; K, Kidney; S, Spleen; resulted in a similar pattern of staining. Fixation in faster acting M, Muscle; 3T3 cells; PC-12 cells. (B) PCR of GAPDH controls fixatives however resulted in cytoplasmic localization of from each sample to control for RNA loading. Characterization and localization of CLK1 3249

Fig. 8. Shift in PCR products of splice forms with cycloheximide. Control or PC12 cells treated with 10 µg/ml cycloheximide for 60 minutes were harvested, RNA was extracted, and RT-PCR was performed. (A) PCR products of the 560 bp full-length form or the 453 kinase-less form of CLK1 message are shown. (B) PCR of GAPDH controls from each sample to control for RNA loading. more accurate image of localization of the protein. We conclude from these experiments that the majority of the endogenous CLK1 protein is localized to the cytoplasm. Fig. 10. Immunostaining of rat brain sections. CLK1 expression in rat brain. Sections of brain from adult rats were examined for the presence of CLK1 protein. The sections were immunostained with DISCUSSION the affinity puriﬁed anti-CLK1 antibody and Texas-Red conjugated goat-anti-rabbit secondary antibody (B,D,F,H) or secondary antibody Historically, protein kinases have been divided into two alone (A,C,E,G). A confocal microscope was used to image the distinct groups, the tyrosine kinases and the serine/threonine sections. (A-B) Cells in the area of the raphe nucleus. kinases. Recently, however, a new class of kinases has been (C-D) CA1 region of hippocampus. (E-F) Frontal cortex. (G-H) Cerebellum. Bar, 50 µm.

discovered that phosphorylates both serine/threonine and tyrosine residues. Based on conserved amino acid motifs, these dual specificity kinases are indistinguishable from the serine/threonine kinases. The lack of any distinguishable amino acid motifs and the fact that, with few exceptions, most dual specificity kinases, including CLK1, have weak tyrosine kinase activity has led to the conclusion that dual specificity is a property that may not be physiologically relevant (Lindberg et al., 1992). The robust tyrosine kinase activity of the dual specificity kinases MEK and wee1, however, is restricted to their specific, physiological substrates (Crews et al., 1992; Featherstone and Russell, 1991; Russell and Nurse, Fig. 9. Western analysis of CLK1 protein in rat tissues and cell lines. 1987). This suggests that specificity toward tyrosine residues Protein (50 µg) from tissue or cell line lysates or were separated on SDS-polyacrylamide gels, transferred to PVDF membrane. The blots is far more stringent than for serine/threonine residues. The were then probed with the affinity purified CLK1 antibody and tyrosine kinase activity of many of the dual specificity kinases visualization by ECL reagents. The position of the molecular mass is detectable only in autophosphorylation reactions and markers and full-length (57 kDa) product is indicated. Lanes: 3T3 tyrosine phosphorylation of exogenous substrates may cells; PC-12 cells; Br, brain; H, heart; Lg, lung; Li, liver; K, kidney; therefore only be achieved when assayed with the appropriate S, spleen; M, muscle. substrate. 3250 H. J. Menegay and others

peptide substrates revealed that the consensus site for CLK1 phosphorylation is R-X-X-(S/T)-X-X-R with the basic residues at both the +3 and −3 playing approximately equivalent roles in determining substrate specificity (Table 1). Basic residues at other positions may also play a lesser role in determining substrate specificity, as the poly-basic PKCζ substrate peptide was the most efficiently utilized peptide substrate (Fig. 4 and Table 1). The requirement of multiple basic residues would explain why SR proteins, which are serine-arginine rich splicing factors, are so efficiently phosphorylated by CLK1 (Colwill et al., 1996a). Somewhat surprisingly, CLK1 does not contain an acidic residue at an analogous position, demonstrating that the substrate specificity determinants of CLK1 are distinct from those of other kinases. CLK1 exhibits a domain structure in which the conserved catalytic domain is positioned at the C terminus, while the N-terminal domain contains a functional nuclear localization signal which may direct CLK1 into the nucleus (Duncan et al., 1995). The N terminus may also act to anchor CLK1 to its substrates, as well as acting to regulate kinase activity. The presence of a regulatory domain in a number of kinases, including the EGF receptor (Vojtek and Cooper, 1995), c-raf (Li et al., 1991), src (Superti-Furga, 1995), and pp70S6K (Crowley et al., 1994; Dennis et al., 1998; Smeyne et al., 1994), has been demonstrated by an increase in kinase activity upon truncation of the regulatory domain. Truncation of the 128 N-terminal amino acids of CLK1 results in an approximately 40-fold increase in enzymatic activity, indicating that the N terminus of CLK1 encodes a regulatory domain. Although the consensus phosphorylation site predicts the presence of a number of potential phosphorylation sites in the N terminus, it is unlikely that the truncation removes an inhibitory pseudo-substrate domain. The removal of a pseudosubstrate, which acts as a competitive inhibitor, would be predicted to result in an increase in Km for exogenous substrates. Truncation of the N Fig. 11. Immunostaining of PC12 cells fixed with a variety of fixatives. PC12 terminus of CLK1 resulted in only a small decrease cells were fixed for 20 minutes with methanol, 0.4% paraformaldehyde, 4% in Km for ATP and a small increase in Km for MBP paraformaldehyde, Acrolein fixative, or Bouin’s fixative and immunostained (Table 2). The truncation did, however, result in a 45- with the CLK1 antibody and Texas-Red-conjugated secondary antibody. fold increase in Vmax, suggesting that the truncation Images were captured on a confocal microscope. Images are individual 0.5 allows CLK1 to adopt a conformation which allows µM confocal image sections or composite projections of 0.5 µM confocal sections throughout the thickness of the cells. for a faster rate of phospho-transfer rather than affecting enzyme-substrate interactions. The cellular regulation of CLK1 activity is currently unclear. We have provided evidence Traditionally, proteins, such as MBP, histones, or casein, indicating that the control of CLK1 kinase activity is complex. have been used as exogenous substrates in vitro to evaluate CLK1 has been shown to autophosphorylate on serine, substrate specificity. We subsequently determined that CLK1 threonine, and tyrosine residues (Howell et al., 1991). Howell phosphorylated a single site in MBP (Fig. 2). Phospho-peptide et al. reported that CLK1 activity can be quantitatively sequencing revealed that the site phosphorylated by CLK1 is immunoprecipitated with antibodies to phosphotyrosine, which Ser109 (Fig. 3). This site has also been shown to be suggested that tyrosine phosphorylation is necessary for CLK1 phosphorylated by cAMP-dependent kinase, Cam kinase II and activity (Howell et al., 1991). Paradoxically, treatment with PKC (Crews et al., 1992; Li et al., 1991; Smeyne et al., 1994; tyrosine specific phosphatases had no discernable effect on Vojtek and Cooper, 1995). Unlike these kinases whose either full length or truncated CLK1 activity. Treatment with consensus site is R-R-X-X-(S/T) or R-R-X-S, a survey of 14 serine/threonine specific phosphatases, on the other hand, Characterization and localization of CLK1 3251

Fig. 12. Immunostaining of NIH 3T3 cells overexpressing CLK1 with Bouin’s or methanol. NIH 3T3 cells which were transfected with the control pIND vector or transfected with the CLK1-pIND construct were fixed with Bouin’s fixative or methanol. The fixed cultures were then immunostained with the CLK1 antibody and visualized with a Texas- Red-conjugated secondary antibody imaged with a confocal microscope. resulted in a two- to sixfold activation of CLK1. Simultaneous unclear but it does not appear to contain a pseudo-substrate treatment with both classes of phosphatases, or with alkaline motif and may act to conformationally constrain catalysis by phosphatase, a non-selective phosphatase, abolished CLK1 another mechanism. It is has been speculated that the N- activity. These data indicate phosphorylation at either terminal splice variant, which lacks a kinase domain, functions serine/threonine residues or tyrosine residues is sufficient to to regulate CLK1 function. Perhaps by binding to substrates or activate CLK1. The activation of CLK1 by serine/threonine by directly interacting with the full-length allele. phosphatases indicates that phosphorylation of some Here we report that the ratio of full-length to truncated CLK serine/threonine sites acts to specifically inhibit CLK1 activity mRNA products can be altered by cycloheximide treatment. It or that the enzyme phosphorylated exclusively on tyrosine has previously been demonstrated that CLK1 can regulate it’s residues is more active than the enzyme phosphorylated on own splicing and is capable of phosphorylating SR proteins serine/threonine and tyrosine residues. Given the high specific (Colwill et al., 1996b; Duncan et al., 1997). It has also been activity of the untreated enzyme, the latter case appears to be shown that members of the CLK family are capable of more likely. regulating mRNA splicing of an adenovirus E1A splicing The MAP kinases are also regulated by phosphorylation at reporter construct in vivo, as well as causing the redistribution tyrosine and serine/threonine residues, but unlike CLK1, MAP of SR proteins from nuclear speckles to a diffuse kinase activation requires phosphorylation of both nucleoplasmic localization (Colwill et al., 1996b; Duncan et serine/threonine and tyrosine residues (Crews et al., 1992; al., 1997, 1998). This observation has been broadened to Payne et al., 1991; Schanen-King et al., 1991). The residues include other members of the LAMMER family (Du et al., responsible for MAP kinase activation lie in kinase subdomain 1998; Wang et al., 1998). It has recently been demonstrated VIII (Payne et al., 1991). Phosphorylation of similarly located that CLK1’s effect on splicing is due to it’s phosphorylation of residues has been shown to increase the activity of a number SR proteins rather than other members of the splicing complex of serine/threonine kinases (Hanks and Hunter, 1995; Taylor et (Prasad et al., 1999). al., 1995). Some tyrosine kinases are also regulated by tyrosine A subset of SR proteins, including ASF/SF2, shuttle phosphorylation at a distinct site in kinase subdomain VIII. between the nucleus and the cytoplasm based upon their CLK1 has a number of potential phosphorylatable residues in phosphorylation state (Caceres et al., 1998). CLK1 has been this region, which correspond to those found in activity- shown to directly interact with ASF/SF2, and the expression of dependent phosphorylation sites from both tyrosine kinases myc-tagged CLK1 protein causes the accumulation of and from serine/threonine kinases. The presence of potential ASF/SF2 to the cytoplasm (Caceres et al., 1998; Colwill et al., regulatory sites from both classes of protein kinases may 1996b; Prasad et al., 1999). These data suggest the possibility explain why CLK1 activity is unaffected upon treatment with that the majority of CLK1 may be normally held in a the tyrosine phosphatases or serine/threonine phosphatases cytoplasmic reservoir, but may also shuttle between the nucleus tested and why CLK1 activity is lost when treated with both and cytoplasm with ASF/SF2. phosphatases. The identity of the regulatory phosphorylation The cytoplasmic localization of the bulk of CLK1 protein in sites are currently unknown but they must be located in the vivo suggests that it may have cytoplasmic targets in addition catalytic domain, as the truncated CLK1 is activated upon to its role in the regulation of splicing. Interestingly, CLK1 has treatment with serine/threonine phosphatases. In addition to recently been shown to be capable of phosphorylating and phosphorylation, CLK1 is also negatively regulated by its N activating the tyrosine phosphatase PTP1B (Moeslein et al., terminus. The specific mechanism of inhibition remains 1999). This phosphatase is can be found in the cytoplasm of 3252 H. J. Menegay and others cells, and may be a cytoplasmic target of CLK1. Although Protein phosphorylation plays an essential role in the regulation of alternative CLK1 and another CLK family member, CLK3, have been splicing and sex determination in Drosophila. Mol. Cell 2, 741-750. Duncan, P. I., Howell, B. W., Marius, R. M., Drmanic, S., Douville, E. M. shown to be localized to the nucleus in overexpression systems, and Bell, J. C. (1995). Alternative splicing of STY, a nuclear dual specificity we recently demonstrated that in vivo the CLK3 protein is kinase. J. Biol. Chem. 270, 21524-21531. localized to the cytoplasm of developing spermatozoa and to Duncan, P. I., Stojdl, D. F., Marius, R. M. and Bell, J. C. (1997). In vivo the acrosome of mature sperm (Menegay et al., 1999). The N- regulation of alternative pre-mRNA splicing by the Clk1 protein kinase. Mol. Cell Biol. 17, 5996-6001. terminal domains of the CLK family, containing their nuclear Duncan, P. I., Stojdl, D. F., Marius, R. M., Scheit, K. H. and Bell, J. C. (1998). localization signals, are not conserved (Nayler et al., 1997). The Clk2 and Clk3 dual-specificity protein kinases regulate the intranuclear Their variability in this region suggests that their localization distribution of SR proteins and influence pre-mRNA splicing. Exp. Cell Res. may be regulated very differently, however we believe that both 241, 300-308. Featherstone, C. and Russell, P. (1991). Fission yeast p107wee1 mitotic of these CLK family members are generally cytoplasmic inhibitor is a tyrosine/serine kinase. Nature 349, 808-811. proteins. Glass, D. B., Masaracchia, J. R. and Kemp, B. E. (1978). Anal. Biochem. 87, We demonstrate that the apparent subcellular distribution of 566-575. CLK1 protein varies dramatically based upon the fixation Greenberg, M. E., Hermanowski, A. L. and Ziff, E. B. (1986). Effect of protein synthesis inhibitors on growth factor activation of c-fos, c-myc, and agent. These differences are likely due to the rate of fixation actin gene transcription. Mol. Cell Biol. 6, 1050-1057. and cross-linking between these fixing agents. Methanol Hanes, J., von der Kammer, H., Klaudiny, J. and Scheit, K. H. (1994). precipitates proteins, while Bouin’s fixative both precipitates Characterization by cDNA cloning of two new human protein kinases. and cross-links. A light fixative such as 0.4% Evidence by sequence comparison of a new family of mammalian protein kinases. J. Mol. Biol. 244, 665-672. paraformaldehyde gives a similar pattern of staining as Hanks, S. K. and Hunter, T. (1995). Protein kinases 6. The eukaryotic protein methanol, while a faster fixation in 4% paraformaldehyde kinase superfamily: kinase (catalytic) domain structure and classification. preserves a pattern similar to that of Bouin’s. This suggests to FASEB J. 9, 576-596. us that the pattern observed with Bouin’s fixative may be more Hanks, S. K. and Quinn, A. M. (1991). Protein kinase catalytic domain sequence database: identification of conserved features of primary structure representative of the pattern of CLK1 localization in vivo. and classification of family members. Meth. Enzymol. 200, 38-62. Hanks, S. K., Quinn, A. M. and Hunter, T. (1988). The protein kinase family: This work was supported by grants from the National Institutes of conserved features and deduced phylogeny of the catalytic domains. Science Health (NS31987) and M.P.M. was supported by a training grant from 241, 42-52. the NIH (HD 07204-25). We are indebted to Dr T. Egelhof, Dr M. Howell, B. W., Afar, D. E., Lew, J., Douville, E. M., Icely, P. L., Gray, D. A. Wooten, Dr R. Erikson, and Dr M. Greenberg for kindly providing and Bell, J. C. (1991). STY, a tyrosine-phosphorylating enzyme with sequence homology to serine/threonine kinases. Mol. Cell Biol. 11, 568-572. reagents. We are also indebted to Dr C. Beach for providing excellent Huang, S. and Spector, D. L. (1996). Dynamic organization of pre-mRNA technical guidance and for performing the peptide sequencing. splicing factors. J. Cell. Biochem. 62, 191-197. Humason, G. L. (1962). Animal Tissue Techniques. W. H. Freeman and Company. REFERENCES Jimenez-Garcia, L. F. and Spector, D. L. (1993). In vivo evidence that transcription and splicing are coordinated by a recruiting mechanism. Cell 73, 47-59. Ben-David, Y., Letwin, K., Tannock, L., Bernstein, A. and Pawson, T. (1991). Johnson, K. W. and Smith, K. A. (1991). Molecular cloning of a novel human A mammalian protein kinase with potential for serine/threonine and tyrosine cdc2/CDC28-like protein kinase. J. Biol. Chem. 266, 3402-3407. phosphorylation is related to cell cycle regulators. EMBO J. 10, 317-325. Lee, K., Du, C., Horn, M. and Rabinow, L. (1996). Activity and Bender, J. and Fink, G. R. (1994). AFC1, a LAMMER kinase from Arabidopsis autophosphorylation of LAMMER protein kinases. J. Biol. Chem. 271, 27299- thaliana, activates STE12-dependent processes in yeast. Proc. Nat. Acad. Sci. 27303. USA 91, 12105-12109. Li, P., Wood, K., Mamon, H., Haser, W. and Roberts, T. (1991). Raf-1: a Boyle, W. J., van der Geer, P. and Hunter, T. (1991). Phosphopeptide mapping kinase currently without a cause but not lacking in effects. Cell 64, 479-482. and phosphoamino acid analysis by two-dimensional separation on thin-layer Lindberg, R. A., Quinn, A. M. and Hunter, T. (1992). Dual-specificity protein cellulose plates. Meth. Enzymol.201, 110-149. kinases: will any hydroxyl do? Trends Biochem. Sci. 17, 114-119. Caceres, J. F., Screaton, G. R. and Krainer, A. R. (1998). A specific subset Luo, K. X., Hurley, T. R. and Sefton, B. M. (1991). Cyanogen bromide of SR proteins shuttles continuously between the nucleus and the cytoplasm. cleavage and proteolytic peptide mapping of proteins immobilized to Genes Dev. 12, 55-66. membranes. Meth. Enzymol.201, 149-252. Chomczynski, P. and Sacchi, N. (1987). Single-step method of RNA isolation Mattaj, I. W. (1994). RNA processing. Splicing in space [news; comment]. by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Nature 372, 727-728. Biochem. 162, 156-159. Meijlink, F., Curran, T., Miller, A. D. and Verma, I. M. (1985). Removal of Colwill, K., Feng, L. L., Yeakley, J. M., Gish, G. D., Caceres, J. F., Pawson, a 67-base-pair sequence in the noncoding region of protooncogene fos T. and Fu, X. D. (1996a). SRPK1 and Clk/Sty protein kinases show distinct converts it to a transforming gene. Proc. Nat. Acad. Sci. USA 82, 4987-4991. substrate specificities for serine/arginine-rich splicing factors. J. Biol. Chem. Menegay, H., Moeslein, F. and Landreth, G. (1999). The dual specificity 271, 24569-24575. protein kinase CLK3 is abundantly expressed in mature mouse spermatozoa. Colwill, K., Pawson, T., Andrews, B., Prasad, J., Manley, J. L., Bell, J. C. and Exp. Cell Res. 253, 463-473. Duncan, P. I. (1996b). The Clk/Sty protein kinase phosphorylates SR splicing Moeslein, F. M., Myers, M. P. and Landreth, G. E. (1999). The CLK family factors and regulates their intranuclear distribution. EMBO J.15, 265-275. kinases, CLK1 and CLK2, phosphorylate and activate the tyrosine Crews, C. M., Alessandrini, A. and Erikson, R. L. (1992). The primary phosphatase, PTP-1B. J. Biol. Chem. 274, 26697-26704. structure of MEK, a protein kinase that phosphorylates the ERK gene product. Myers, M. P., Murphy, M. B. and Landreth, G. (1994). The dual-specificity Science 258, 478-480. CLK kinase induces neuronal differentiation of PC12 cells. Mol. Cell Biol. Crowley, C., Spencer, S. D., Nishimura, M. C., Chen, K. S., Pitts-Meek, S., 14, 6954-6961. Armanini, M. P., Ling, L. H., MacMahon, S. B., Shelton, D. L., Levinson, Nayler, O., Stamm, S. and Ullrich, A. (1997). Characterization and comparison A. D. et al. (1994). Mice lacking nerve growth factor display perinatal loss of of four serine- and arginine-rich (SR) protein kinases. Biochem. J. 326, 693- sensory and sympathetic neurons yet develop basal forebrain cholinergic 700. neurons. Cell 76, 1001-1011. Nayler, O., Schnorrer, F., Stamm, S. and Ullrich, A. (1998). The cellular Dennis, P. B., Pullen, N., Pearson, R. B., Kozma, S. C. and Thomas, G. localization of the murine serine/arginine-rich protein kinase CLK2 is (1998). Phosphorylation sites in the autoinhibitory domain participate in regulated by serine 141 autophosphorylation. J. Biol. Chem. 273, 34341- p70(s6k) activation loop phosphorylation. J. Biol. Chem. 273, 14845-14852. 34348. Du, C., McGuffin, M. E., Dauwalder, B., Rabinow, L. and Mattox, W. (1998). Payne, D. M., Rossomando, A. J., Martino, P., Erickson, A. K., Her, J. H., Characterization and localization of CLK1 3253

Shabanowitz, J., Hunt, D. F., Weber, M. J. and Sturgill, T. W. (1991). Lira, S. A. and Barbacid, M. (1994). Severe sensory and sympathetic Identiﬁcation of the regulatory phosphorylation sites in pp42/mitogen- neuropathies in mice carrying a disrupted Trk/NGF receptor gene [see activated protein kinase (MAP kinase). EMBO J.10, 885-892. comments]. Nature 368, 246-249. Porath, J. and Axen, R. (1976). Immobilization of enzymes to agar, agarose, Superti-Furga, G. (1995). Regulation of the Src protein tyrosine kinase. FEBS and Sephadex supports. Meth. Enzymol. 44, 19-45. Lett. 369, 62-66. Prasad, J., Colwill, K., Pawson, T. and Manley, J. L. (1999). The protein Taylor, S. S., Radzio-Andzelm, E. and Hunter, T. (1995). How do protein kinase Clk/Sty directly modulates SR protein activity: both hyper- and kinases discriminate between serine/threonine and tyrosine? Structural hypophosphorylation inhibit splicing. Mol. Cell Biol. 19, 6991-7000. insights from the insulin receptor protein-tyrosine kinase. FASEB J. 9, 1255- Ringold, G. M., Dieckmann, B., Vannice, J. L., Trahey, M. and McCormick, 1266. F. (1984). Inhibition of protein synthesis stimulates the transcription of human Ullrich, A. and Schlessinger, J. (1990). Signal transduction by receptors with beta-interferon genes in Chinese hamster ovary cells. Proc. Nat. Acad. Sci. tyrosine kinase activity. Cell 61, 203-212. USA 81, 3964-3968. Vojtek, A. B. and Cooper, J. A. (1995). Rho family members: activators of Russell, P. and Nurse, P. (1987). Negative regulation of mitosis by wee1+, a MAP kinase cascades. Cell 82, 527-529. gene encoding a protein kinase homolog. Cell 49, 559-567. Wang, H. Y., Lin, W., Dyck, J. A., Yeakley, J. M., Songyang, Z., Cantley, L. Schanen-King, C., Nel, A., Williams, L. K. and Landreth, G. (1991). Nerve C. and Fu, X. D. (1998). SRPK2: a differentially expressed SR protein- growth factor stimulates the tyrosine phosphorylation of MAP2 kinase in speciﬁc kinase involved in mediating the interaction and localization of pre- PC12 cells. Neuron 6, 915-922. mRNA splicing factors in mammalian cells. J. Cell Biol. 140, 737-750. Shaw, G. and Kamen, R. (1986). A conserved AU sequence from the 3′ Yun, B., Farkas, R., Lee, K. and Rabinow, L. (1994). The Doa locus encodes untranslated region of GM-CSF mRNA mediates selective mRNA a member of a new protein kinase family and is essential for eye and degradation. Cell 46, 659-667. embryonic development in Drosophila melanogaster. Genes Dev. 8, 1160- Smeyne, R. J., Klein, R., Schnapp, A., Long, L. K., Bryant, S., Lewin, A., 1173.