Identification and characterization of the CLK1 gene product, a novel CaM kinase-like protein kinase from the yeast Saccharomyces cerevisiae.

The CLK1 gene of Saccharomyces cerevisiae encodes a 610-residue protein kinase that resembles known type II Ca2+/calmodulin-dependent protein kinases (CaM kinases), including the CMK1 and CMK2 gene products from the same yeast. The Clk1 kinase domain is preceded by a 162-residue N-terminal extension, followed by a 132-residue C-terminal extension (which contains a basic segment resembling known calmodulin-binding sites) and is as similar to mammalian CaM kinase (38% identity to rat CaM kinase α) as it is to yeast CaM kinase (37% identity to Cmk2). However, Clk1 shares 52% identity with Rck1, another putative protein kinase encoded in the S. cerevisiae genome. Clk1 tagged with a c-myc epitope (expressed in yeast) and a GST-Clk1 fusion (expressed in bacteria) underwent autophosphorylation and phosphorylated an exogenous substrate (yeast protein synthesis elongation factor 2), primarily on Ser. Neither Clk1 activity was stimulated by purified yeast calmodulin (CMD1 gene product), with or without Ca2+; no association of Clk1 with Cmd1 was detectable by other methods. C-terminally truncated Clk1(Δ487-610) was growth-inhibitory when overexpressed, whereas catalytically inactive Clk1(K201R Δ487-610) was not, suggesting that the C terminus is a negative regulatory domain. Using immunofluorescence, Clk1 was localized to the cytosol and excluded from the nucleus. A clk1Δ mutant, a clk1Δ rck1Δ double mutant, a clk1Δ cmk1Δ cmk2Δ triple mutant, and a clk1Δ rck1Δ cmk1Δ cmk2Δ quadruple mutant were all viable and manifested no other overt growth phenotype.

Cmk2. Ample precedent for such a situation in S. cerevisiae exists. For example, the yeast genome contains three different genes (TPK1, TPK2, and TPK3) that encode isotypes of the catalytic subunit of cyclic 3Ј,5Ј-AMP-dependent protein kinase (19). Cells lacking any one or any pair of these genes are viable, but cells lacking all three are not.
It was initially brought to our attention that the nucleotide sequence upstream of the YEF3 gene on chromosome 12 that was originally deposited in GenBank TM (20) contained a partial open reading frame with significant similarity to CaM kinases. 3 To determine if this genomic segment actually contained an authentic CaM kinase-related gene, we cloned and sequenced this region, delineated the open reading frame that encodes this CaM kinase-like protein kinase, which we designated CLK1, and examined in detail the biochemical properties and cellular localization of this enzyme.
Isolation of Genomic DNA Containing the CLK1 Gene-Oligonucleotides were synthesized in the DNA Synthesis Facility (University of California, Berkeley, CA) or by IDX, Inc. (Coralville, IA). To screen an S. cerevisiae genomic DNA library for segments derived from the portion of chromosome 12 upstream of the YEF3 gene (20), a probe based on the published sequence of this region (27) (5Ј-CAATGTTATGT-GGGTTCCGCCATT-3Ј) was prepared, end-labeled with polynucleotide kinase (Boehringer Mannheim) and [␥-32 P]ATP, and separated from unincorporated nucleotide using a NICK TM column (Pharmacia Biotech Inc.) (26). The radiolabeled probe was hybridized to five nitrocellulose filter replicas (15-cm diameter) of plates (ϳ8,000 colonies/plate) containing a yeast genomic DNA library in the plasmid vector, pSB32 (gift of F. Spencer and P. Hieter, Dept. of Molecular Biology and Genetics, Johns Hopkins University School of Medicine, Baltimore, MD). The bacterial cells were grown, transferred to filters, and screened by hybridization using standard procedures (26). Two independently isolated clones (YCpMM1.4 and YCpMM1.6), which reproducibly hybridized to the probe in subsequent secondary and tertiary tests, were recovered.
Plasmid Constructions-Restriction endonucleases and other DNAmodifying enzymes were purchased from Boehringer-Mannheim, New England Biolabs, or Stratagene and were used as recommended by the manufacturer. Conventional methods were used for the construction, propagation, and purification of all plasmids (26). DNA fragments separated on agarose gels were recovered from the gels via adsorption to powdered glass (GeneClean, Bio 101, Inc.). For certain purposes, it was necessary to eliminate the EcoRI site in the vector Bluescript SK (Stratagene), which was accomplished by cleaving with EcoRI, filling in the resulting 5Ј-overhangs by treatment with T4 DNA polymerase in the presence of all four deoxyribonucleoside 5Ј-triphosphates (26), and incubation with T4 DNA ligase and ATP. A 3-kb ApaI fragment was excised from YCpMM1.6 and inserted into the ApaI site of this modified version of Bluescript SK, yielding plasmid pMMR2. The same fragment was ligated in the reverse orientation in the standard Bluescript SK, producing pMMA3.
To generate a version of Clk1 containing a 16-residue epitope from the c-Myc oncoprotein (-LEEQKLISEEDLLRKR-COOH) attached in frame to the C-terminal end of Clk1, the polymerase chain reaction and a three-primer scheme (28) were used. For this purpose, the template DNA was YCpMM1.6, the 5Ј-primer was 5Ј-AAGAAGACCGTATGGGG-3Ј, the joiner oligonucleotide was 5Ј-TTTCTGCTCCTCGAGTTCCCT-GATAGTGGC-3Ј, and the 3Ј-primer encoding the tag was 5Ј-TTACCT-CTTCCTGAGGAGGTCCTCTTCGCTGATTAATTTCTGCTCCTCGAG--3Ј. Amplification was achieved using the following program: four cycles of denaturation (94°C, 30 s), annealing (40°C, 30 s), and synthesis (72°C, 30 s) followed by 16 cycles under the same conditions, except that annealing was carried out at 45°C. The resulting product was ligated into the vector, pCR1000 (Promega), according to the manufacturer's instructions, to generate plasmid pCR-MM1. A NarI-KpnI fragment of pCR-MM1 was excised and used to replace the corresponding NarI-KpnI segment in plasmid pMMR2, thereby creating plasmid pMM15. To express full-length c-Myc-tagged CLK1 from its own promoter, a PstI-KpnI fragment of pMM15 was ligated into YEp352 (29) that had been previously digested with PstI and KpnI, yielding pMM16.
Polymerase chain reaction was also used to construct a derivative of CLK1 containing a C-terminal truncation. For this purpose, the template DNA was pMMR2, the 5Ј-primer was 5Ј-GGGGGATCCCCAT-GCTTAAAATAAAGGCC-3Ј, which included a BamHI site (boldface type) 7 base pairs upstream of the sequences corresponding to the ATG (underlined) at the start of the CLK1 open reading frame, and the 3Ј-primer was 5Ј-CCCGTCGACTCATTACTTTGGTAAACAATCGAA-3Ј, which included two stop codons (underlined) after codon 486 of the CLK1 coding sequence immediately followed by SalI site (boldface type). Amplification was achieved using the following program: 22 cycles of denaturation (94°C, 30 s), annealing (37°C, 30 s), and synthesis (72°C, 30 s). To remove any overhanging 3Ј-extensions, the resulting product was incubated with T4 DNA polymerase (26) and then ligated into Bluescript SK that had been cleaved with EcoRV, yielding plasmid pBS-⌬C. A BamHI-SalI fragment was excised from pBS-⌬C and inserted immediately downstream of the GAL1 promoter in the vector, YEp351GAL (30), that had been cleaved with BamHI and SalI, yielding plasmid pMM25. To place expression of full-length CLK1 under GAL1 promoter control, the HindIII fragment in pMM25 was replaced by the HindIII fragment from pMMA3, generating plasmid pMM51.
To express a catalytically inactive mutant version of full-length CLK1 from the GAL1 promoter, the HindIII fragment in pMM25 was replaced with the HindIII fragment from plasmid pMM45 (described below), which contains the clk1(K201R) allele that was generated by in vitro site-directed mutagenesis (see "In Vitro Site-directed Mutagenesis"), yielding plasmid pMM52. To express a catalytically inactive mutant version of the C-terminally truncated derivative of CLK1 from the GAL1 promoter, several steps were required. First, the BamHI-SalI fragment of pMM42 (described below) was inserted into pBluescript SK that had been digested with BamHI and SalI. Second, a ClaI-EcoRI fragment of the resulting plasmid was then replaced by the ClaI-EcoRI fragment of pMM45, generating pMM55. Finally, the BamHI-SalI fragment of pMM25 was replaced with the BamHI-SalI fragment of pMM55, to yield the desired plasmid pMM56.
To facilitate purification, chimeras containing various portions of Clk1 fused to the C terminus of Schistosoma japonicum glutathione S-transferase (GST) were constructed. A chimera containing codons 135-610 of CLK1 joined in frame to the GST coding sequence was generated by ligating the EcoRV-KpnI fragment of pMMA3 into the E. coli expression vector, pGEX-3x (Pharmacia), that had been digested with SmaI and KpnI, yielding pMM23. A chimera containing all 610 codons of the CLK1 gene joined in frame to the GST coding sequence was generated in two steps. First, the BamHI-EcoRI fragment of pMM25 was ligated into pGEX-3x that had been cleaved with BamHI and EcoRI, yielding plasmid pMM36. Second, the EcoRI fragment of pMMA3 was ligated into the EcoRI site of pMM36, yielding plasmid pMM37, which expresses the desired fusion. A chimera containing codons 1-486 of the CLK1 gene joined in frame to the GST coding sequence was generated by replacing the HindIII fragment in pMM36 with the HindIII fragment from pMM25, yielding plasmid pMM42, which expresses the GST::Clk1(⌬487-610) fusion. A chimera containing the clk1(K201R) allele fused in frame to the GST coding sequence was constructed in two steps. First, the HindIII fragment in pMM37 was replaced with the HindIII fragment of pMM21 (see "In Vitro Sitedirected Mutagenesis"), yielding plasmid pMM44. Second, the EcoRI fragment of pMM44 was replaced with the EcoRI fragment of pMMA3, to yield pMM45.
In Vitro Site-directed Mutagenesis-To generate a template for con-3 B. Baum, personal communication.
structing the Lys-to-Arg mutation at codon 201 of the CLK1 coding sequence, the PstI-EcoRI fragment from pMMR2 was ligated into the double-stranded replicative form of the phage vector, M13mp19 (New England Biolabs), and single-stranded phage DNA was prepared from the resulting construct using standard procedures (26). Using a commercial in vitro mutagenesis kit (Amersham Corp.) and the mutagenic primer, 5Ј-GCCTTTTTAATAACTCTAATGGCAACAGCTT-3Ј (mismatched base underlined), mutant DNA was generated and purified, according to the manufacturer's recommendations. Introduction of the desired mutation was confirmed by direct nucleotide sequence analysis (31). The corresponding fragment was excised from the replicative form of the mutant phage by digestion with EcoRI and PstI and ligated into pMM15 to generate plasmid pMM21. The PstI-KpnI fragment of pMM21 was inserted into the corresponding sites in YEp352 to yield plasmid pMM22.
DNA Sequence Determination, Physical Mapping, and Transcript Analysis of the CLK1 Locus-Plasmid pMMR2 was purified using a Magic Mini-Prep kit (Promega), and the nucleotide sequence of most of the insert was determined on both strands by the dideoxynucleotide method and a double-stranded DNA sequencing procedure (32) using the Sequenase (U.S. Biochemical Corp.) form of T7 DNA polymerase, [␣-35 S]dATP (Amersham) (33) and, where necessary, custom-synthesized oligonucleotide primers. The GenBank TM accession number for this sequence is U23464.
To confirm the map location of the CLK1 locus, an internal (EcoRI-NarI) fragment of the CLK1 coding sequence was excised from plasmid pMMR2, purified by gel electrophoresis, radiolabeled by a random primer method (34), and hybridized to a blot of intact yeast chromosomes (gift of G. Anderson, Dept. of Plant Biology, University of California, Berkeley) separated by orthogonal field gel electrophoresis (35). The same probe also was hybridized to a set of filters containing an ordered array of fragments of the yeast genome in a bacteriophage vector (36), generously provided by L. Riles and M. Olson (Dept. of Genetics, Washington University School of Medicine, St. Louis, MO).
Total RNA and poly(A) ϩ RNA species were isolated as described (23). Poly(A) ϩ RNA was fractionated in a formaldehyde-containing agarose gel, transferred to nitrocellulose filter, and hybridized against the same EcoRI-NarI probe derived from pMMR2 as described (26).
Construction of clk1 Null Mutations-The EcoRI-NarI fragment in pMMR2 was replaced with an EcoRI-NarI fragment from pJJ217 (37), which contains the entire HIS3 gene, to yield pMM11. The ApaI fragment of pMM11 was excised and used for DNA-mediated transformation of diploid strain YPH501. His ϩ colonies were analyzed by restriction endonuclease digestion and DNA hybridization (38) using a 2.5-kb EcoRI-ScaI fragment excised from YCpMM1.6 as the probe to identify transformants in which the CLK1 locus on one homolog had been replaced with the clk1-⌬1::HIS3 allele by homologous recombination. Such a clk1-⌬1::HIS3/CLK1 heterozygous diploid (YMM3) was sporulated on potassium acetate plates, and the resulting tetrads were dissected and analyzed for the segregation of markers. To generate a cmk1⌬ cmk2⌬ clk1⌬ triple mutant, a MATa clk1-⌬1::HIS3 spore derived from YMM3 was mated with strain MPY62 (MAT␣ cmk1-⌬1::HIS3 cmk2-⌬1::TRP1) (17), to produce diploid strain YMM6. Haploids containing all three mutant loci were recovered from those tetrads in which His ϩ :His Ϫ segregated 2:2 and in which the Trp ϩ marker also co-segregated into the His ϩ spores.
Production of Rabbit Polyclonal Anti-Clk1 Antibodies-As determined by immunoblotting (39) prior to immunization, two rabbits (1188 and 1189) displayed negligible cross-reaction against a cell-free extract of strain YPH499 and were injected, as the primary inoculation, with 50% complete Freund's adjuvant (Sigma) containing 1 mg of a GST-Clk1(⌬487-610) fusion protein that had been purified by adsorption and elution from glutathione-agarose beads from lysates of E. coli cells induced to express plasmid pMM42 by treatment with 0.5 mM isopropyl 1-thio-␤-D-galactopyranoside. Four weeks and 8 weeks later, each rabbit was injected with 200 g of the same antigen in 50% incomplete Freund's adjuvant (Sigma). One rabbit (1188) received an additional injection 19 weeks after the primary inoculation. Bleeds (4 -20 ml) were drawn at 6, 10, and 14 weeks after the initial injection for rabbit 1189 and at 6, 10, and 21 weeks for rabbit 1188 and stored in 0.02% sodium azide at Ϫ70°C.
Immunoblot Analysis of Proteins-Equal amounts of protein from cell extracts or in immune complexes were subjected to electrophoresis on polyacrylamide gels containing sodium dodecyl sulfate (SDS-PAGE), electrophoretically transferred to nitrocellulose using a dry blotting system (Multiphor II, Pharmacia), and incubated with a 1:10,000 dilution of either preimmune serum or an appropriate antiserum. For detection of Clk1, antiserum from rabbit 1188 was used. For detection of Clk1 derivatives tagged with the c-Myc epitope, a monoclonal antibody (mAb), designated 9E10 (40), was used. To detect immobilized primary antibodies, an appropriate secondary antibody (either horse anti-mouse or goat anti-rabbit immunoglobulins) coupled to horseradish peroxidase was used and visualized using chemiluminescence (ECL TM , Amersham) and x-ray film (Biomax or XAR, Eastman Kodak Co.).
Assays of Clk1 Protein Kinase Activity-Autophosphorylation was measured as the incorporation of radioactivity from [␥-32 P]ATP into a species that co-migrated with authentic Clk1 protein and was immunoprecipitable with anti-Clk1 antibodies. To measure the phosphotransferase activity of Clk1, immune complexes generated as described immediately below were incubated under the same conditions but in the presence of various exogenously added protein substrates. For these purposes, a culture (40 ml) of yeast strain YMM3.1a (MATa clk1-⌬1::HIS3) containing an appropriate plasmid was grown in a selective medium (SGlc or SGal) to late exponential phase (a value of 100 on a Klett-Summerson photoelectric colorimeter equipped with a number 66 red filter). The cells were harvested by centrifugation, washed by resuspension and recentrifugation in sterile distilled deionized water, resuspended in 400 l of lysis buffer (25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride), and lysed by five 1-min bursts of vigorous vortex mixing with an equal volume of glass beads (0.45-0.5-mm diameter) alternating with 1-min periods of cooling on ice. The resulting crude extract was separated from the beads by placing the bead-lysate slurry in a conical Eppendorf tube punctured at the bottom with 25-gauge syringe needle and recovered into another Eppendorf tube by brief sedimentation in a clinical centrifuge at 3,000 rpm. After clarification by centrifugation in a microcentrifuge at maximum rpm for 5 min, the resulting supernatant solution was subjected to sedimentation at 100,000 ϫ g in a tabletop ultracentrifuge (TL100, Beckman). A sample (ϳ20 mg of total protein) of the resulting supernatant fraction was incubated overnight with a 1:100 dilution of ascites fluid containing anti-c-Myc mAb 9E10. The immune complexes so generated were collected by adsorption to 20 mg of Protein A-Sepharose beads (Pharmacia), washed three times by resuspension and recentrifugation in lysis buffer, and resuspended in kinase assay buffer (50 mM Tris-HCl, pH 7.5, 0.1 mM EGTA, 0.1 mM EDTA, 5% glycerol). For each reaction, a sample of the resuspended immune complexes (typically 5 l, representing 1 ⁄10 of the total volume) was incubated for 10 -20 min at either 25 or 30°C in a final volume of 40 l in kinase assay buffer containing 10 Ci of [␥-32 P]ATP (3 Ci/mmol, Amersham) in the presence and absence of various divalent cations (Mg 2ϩ , Mn 2ϩ , and/or Ca 2ϩ ), with or without purified yeast Cmd1 protein (0.5 g) (gift of M. Pausch, this laboratory). Reactions were terminated by the addition of 40 l of SDS-PAGE sample-loading buffer (41) followed immediately by boiling for 2 min. After removal of the beads by centrifugation in a microcentrifuge, the quenched reaction mixtures were subjected to SDS-PAGE (41). To fix the gel and remove any unincorporated label, the gel was soaked for at least 2 h in 10% acetic acid, 40% methanol and then in dH 2 O for at least 2 h to further reduce the residual radioactivity. After drying the gel onto a filter paper backing, the radioactive species present were visualized by autoradiography using x-ray film (XAR, Kodak). To normalize for the amount of protein present and to determine unambiguously the migration position of the Clk1 polypeptide, another portion of the same immune complexes was blotted to nitrocellulose and probed with the 9E10 antibody, as described under "Immunoblot Analysis of Proteins".
As an alternative, and to permit rapid purification and minimize perturbation of the C-terminal domain, full-length and other forms of Clk1 were expressed in E. coli as fusions to the C terminus of S. japonicum GST. The GST fusions were purified from bacterial cell lysates by adsorption to glutathione-agarose beads (Pharmacia), washing the beads thoroughly with Tris-buffered saline (TBS; 50 mM Tris-HCl, pH 7.5, 100 mM NaCl), and eluting the bound protein with 5 mM glutathione in TBS. To measure either autophosphorylation activity or phosphotransferase activity, the purified fusion proteins (typically, ϳ1 g of protein) were incubated in solution under conditions essentially identical to those used for the immune complex protein kinase assay described immediately above.
Phosphoamino acid analysis of the radioactivity incorporated into Clk1, its derivatives, or other phosphoacceptor substrates either after incubation with [␥-32 P]ATP in vitro or after labeling of yeast cells in vivo by growth of cultures (5 ml) in LPSM medium (42) with 0.2 mCi of [ 32 P]PO 4 3Ϫ /ml for 2 h was performed using standard methods for protein hydrolysis and two-dimensional electrophoresis (43).
Assay of CaM Binding-A culture (200 ml) of E. coli strain BL21(DE3) (44) carrying plasmid pMM23 was grown at 37°C to an A 600 of 0.5, and then expression of the GST-Clk1(135-610) fusion protein encoded in the plasmid was induced by treatment of the cells with isopropyl 1-thio-␤-D-galactopyranoside at a final concentration of 0.5 mM for 2 h. The cells were collected by centrifugation, resuspended in TBS containing 0.1% Triton X-100 and 1 mM phenylmethylsulfonyl fluoride, and lysed by four 3-s bursts of sonic irradiation at the optimum setting of a probe sonicator (model W1850, Heat Systems-Ultrasonics, Inc.). The crude lysate was clarified by centrifugation for 20 min at 15,000 rpm in the SS-34 rotor of a Sorvall refrigerated centrifuge. The resulting supernatant solution contained nearly all of the fusion protein and was loaded onto a bed (0.5 ml) of glutathione-Sepharose 4B (Pharmacia) in a minicolumn. After loading, the column was washed with 2 ml of TBS containing 0.1% Triton X-100 and then with 2 ml of TBS containing 2 mM CaCl 2 . A solution (10 l) containing 2 mM CaCl 2 and 10 g of purified Cmd1 protein (gift of M. Pausch, this laboratory) was applied to the column. The flow-through fraction and six subsequent washes (0.3 ml each) with TBS containing 2 mM CaCl 2 were all collected. The GST-Clk1(135-610) fusion was then eluted with four washes (0.3 ml each) of TBS containing 5 mM glutathione. All of the fractions were subjected to SDS-PAGE and transferred to nitrocellulose. The content of Cmd1 in the various fractions was determined by immunoblotting of the nitrocellulose using rabbit polyclonal anti-Cmd1 antibodies (45). The content of fusion protein in each fraction was determined by resolving samples on an identical gel and staining with Coomassie Brilliant Blue dye.
Immunofluorescence Detection of Clk1-Subcellular distribution of Clk1 was assessed using minor modifications of standard procedures for fixation, permeabilization, and labeling of yeast cells by indirect immunofluorescence (46). Permeabilized cells were incubated with a 1:300 dilution of primary antibody for 1.5 h at room temperature. Prior to use, rabbit polyclonal anti-Clk1 antiserum 1188 was depleted for nonspecific immunoreactivity by incubation with nitrocellulose filter strips on which a large excess of proteins from an extract of a clk1⌬ strain had been immobilized. To remove excess primary antibody, wells were washed 10 times (20 l each) with the sorbitol-containing phosphate buffer. The slides were then incubated for 1.5 h at room temperature with either goat anti-rabbit immunoglobulin antibodies (when the primary antibody was serum 1188) or sheep anti-mouse immunoglobulin antibodies (when the primary antibody was 9E10 mAb), which were conjugated to fluorescein isothiocyanate (Boehringer Mannheim). To remove excess secondary antibody, wells were washed again 10 times in the same way. Before viewing, slides were dipped briefly in 1 mg/ml 4,6-diamino-2-phenylindole; then a drop of Citifluor (Citifluor Ltd.) was placed in each well, and a coverslip was secured on top using nail polish. Slides were examined using the ϫ 100 objective of a Nikon Optiphot epifluorescence microscope equipped with an ultraviolet filter and photographed using Kodak Ektachrome Elite 400 slide film.
Isolation of the RCK1 Gene and Construction of rck1 Null Mutations-Based on the published sequence of RCK1 (47), an oligonucleotide (5Ј-GGTGTCGGCGGAGGCGGTATTGGTT-3Ј) was synthesized, end-labeled, purified, and used to probe bacterial colonies containing a yeast genomic DNA library as described above. In this way, a plasmid (pMM46) was isolated that contained the entire RCK1 gene, as judged by restriction endonuclease site mapping and direct DNA sequence analysis of the insert. A ClaI fragment containing the RCK1 gene was excised from pMM46 and inserted into the corresponding site of Bluescript SK, yielding plasmid pMM48. pMM48 was cleaved within the RCK1 coding sequence by digestion with NdeI, and the resulting 5Јoverhangs were converted to flush ends by incubation with the Klenow fragment of E. coli DNA polymerase I and all four deoxyribonucleoside 5Ј-triphosphates (200 M). Linearized plasmid was then ligated to a SalI fragment containing the entire LEU2 gene, which had been excised from plasmid pJJ283 (37) and converted to blunt ends by the same procedure, yielding plasmid pMM49. A XbaI-SalI fragment of pMM49 containing the rck1-1::LEU2 disruption allele was introduced by DNAmediated transformation into haploid strains YPH499, YMM3.1a (clk1⌬), and YMM6.1a (cmk1⌬ cmk2⌬ clk1⌬) and into diploid strain YMM6 (cmk1⌬/CMK1 cmk2⌬/CMK2 clk1⌬/CLK1). Leu ϩ transformants containing authentic transplacements at the RCK1 locus were identified by restriction endonuclease digestion and Southern blot hybridization analysis, as described above, using the radiolabeled ClaI fragment of plasmid pMM48 as the probe. The resulting strains were designated, respectively, YMM15, YMM16, YMM20, and YMM21.

RESULTS
The CLK1 Locus Encodes an Open Reading Frame Homologous to Known Ca 2ϩ /CaM-dependent Protein Kinases-To de-termine if the genomic region upstream of the deposited sequence of the YEF3 gene (27), which encodes a yeast translation elongation factor, contains a bona fide open reading frame, we used a synthetic oligonucleotide probe based on the deposited sequence to isolate chromosomal fragments from a yeast genomic DNA library. We then determined the complete nucleotide sequence of this entire region, which revealed a continuous open reading frame with the capacity to encode a 610-residue polypeptide (Fig. 1). Using the BLAST algorithm (47) to compare the deduced protein against all available data bases, the greatest degree of amino acid sequence identity was shared with known and biochemically well characterized CaM kinases from both mammals and yeast (Fig. 2). Hence, the gene was designated CLK1, for CaM kinase-like protein kinase. Hybridization of a probe internal to the CLK1 coding region to a yeast chromosome blot and to a yeast contig set confirmed that CLK1 resides on the right arm of chromosome 12 (data not shown), as reported for the immediately juxtaposed YEF3 gene (27), just centromere-distal to the CDC42 gene. 4 While our work was in progress, another laboratory isolated a putative protein kinase gene, RCK1, from S. cerevisiae and also noted that a portion of the data base entry for YEF3 had homology to it (48). This group also reisolated and resequenced the DNA upstream of the YEF3 gene and reported a larger and continuous open reading frame, which they designated RCK2 (48). RCK2 is the same locus as CLK1; however, the open reading frame reported for RCK2 (48) is prematurely terminated. Likewise, sequence determined for this same region of chromosome 12 and deposited in GenBank TM (accession number U20865) as part of the international effort to determine the complete nucleotide sequence of the entire S. cerevisiae genome, also contains errors that alter the CLK1 open reading frame. The corrected and full-length CLK1 gene product shares significantly greater identity to the predicted RCK1 gene product than it does to CaM kinases (Fig. 2); yet, Rck1 is 98 codons shorter than Clk1, primarily because it lacks the prominent C-terminal extension present in Clk1. To ensure that the reported RCK1 sequence was not incomplete at its 3Ј-end, we reisolated and resequenced this gene. Our sequence agrees completely with that previously reported (48).
CLK1 Is an Expressed Gene and Encodes the Predicted Polypeptide-Poly(A) ϩ RNA was isolated from MATa cells and from MATa cells treated with ␣-factor, separated by agarose gel electrophoresis, and hybridized to an internal portion (1.1-kb EcoRI-ApaI fragment) of the CLK1 gene. Two transcripts were detected (data not shown): a major one (ϳ2,600 nucleotides) and a more minor one (ϳ2,100 nucleotides). Neither transcript appeared to be induced after a 1-h exposure to mating pheromone. Both of these mRNAs are of more than sufficient length to encode the entire CLK1 open reading frame (1830 nucleotides) as well as 5Ј-and 3Ј-untranslated regions and a poly(A) tail.
To confirm that the CLK1 sequence represents one continuous open reading frame and that the predicted protein is actually produced in cells, an epitope tag was placed immediately downstream of and in frame with the last amino acid (residue 610) at the C terminus of the deduced CLK1 coding sequence. As judged by immunoblotting with the 9E10 mAb directed against the epitope tag, a single polypeptide was readily detected, even when expressed from the CLK1 promoter (Fig. 3A). The Clk1::myc protein also cross-reacted with a rabbit polyclonal antibody raised against a GST-Clk1(⌬487-160) fusion protein that lacks the deduced C-terminal extension (data not shown). The apparent molecular mass of the epitope-tagged polypeptide (as estimated from its mobility upon SDS-PAGE relative to standards of known molecular mass) was ϳ89 kDa. Even if the 2 kDa contributed by the C-terminal epitope tag is subtracted, this value is significantly higher than the molecular mass calculated for the Clk1 polypeptide (68.7 kDa) based on its predicted amino acid sequence. However, in extracts of normal yeast cells, the rabbit polyclonal anti-Clk1 antibodies recognized a single species with an apparent molecular mass of 87 kDa (see Fig. 5A), confirming that the authentic protein also migrates with an anomalously slow mobility. Both the N-and C-terminal portions of Clk1 are rich in charged amino acids, and both segments contain several uninterrupted tracts of acidic residues. Such highly negatively charged regions in other proteins confer a markedly slower electrophoretic mobility upon SDS-PAGE than that predicted by the amino acid sequence of the corresponding polypeptide (29,49).
The CLK1 Gene Product Possesses Ser/Thr-specific Protein Kinase Activity-The kinase domain of the deduced Clk1 polypeptide contains all of the most highly conserved residues and hallmark motifs found in known Ser/Thr-specific protein kinases (2,50). Except for Rck1, Clk1 is significantly more similar to the CaM kinase superfamily than it is to other classes of protein kinases, such as the yeast cyclic 3Ј,5Ј-AMPdependent protein kinase catalytic subunit (TPK1 gene product) (51) (Fig. 2). In fact, Clk1 is as similar to mammalian CaM kinases as it is to yeast Cmk1 and Cmk2 (Fig. 2). When compared with mammalian CaM kinases, however, Clk1 has a significantly longer N-terminal extension upstream of the kinase domain and a 31-residue Gly-rich insert situated between conserved domains VI (HRDXKPENLL) and VII (LADFGL). Most strikingly, residues 493-513 of Clk1 (but not the corresponding region of Rck1) possess the highly basic character found in other CaM-binding sites (8). Furthermore, when a 17-residue segment of this same region (residues 504 -520) is displayed as an ␣-helix, positively charged residues are arrayed on the side of the helix opposite from hydrophobic residues, reminiscent of the amphipathic nature of other known CaMbinding sequences (52).
For these reasons, it was of interest to determine whether Clk1 is an active protein kinase and whether its catalytic activity is affected by yeast CaM (CMD1 gene product). The Clk1 derivative tagged with the C-terminal c-Myc epitope allowed the use of an immune complex kinase assay to assess, first, the autophosphorylation activity of Clk1. The Clk1::myc polypeptide was collected from extracts of yeast cells expressing this protein using the mAb 9E10, adsorbed to Protein A-agarose beads, and the resulting immune complexes were incubated in a buffer containing 2 mM Mg 2ϩ and [␥-32 P]ATP. No Cmd1 was detectable in these immunoprecipitates by immunoblotting (as discussed further below). Reaction products were resolved by SDS-PAGE and examined by autoradiography. As a control, a mutant form of the Clk1::myc protein containing a Lys-to-Arg substitution in conserved domain II (VAIK) was constructed; conversion of the Lys residue at the equivalent position in all other protein kinases examined to date eliminates their catalytic activity (2,50). Radioactivity was incorporated into a single species (Fig. 3A) that co-migrated exactly with the Clk1::myc protein, as judged by immunoblotting (Fig. 3B); in contrast, no incorporation whatsoever was observed in the immune complexes containing the Clk1(K201R)::myc polypeptide (Fig. 3A), although this mutant protein was present at a level identical to that of Clk1::myc (Fig. 3B). These results demonstrated that Clk1 is capable of autophosphorylation.
The effect of other divalent metal ions and purified yeast Cmd1 was then examined using the same assay method (Fig.  3C). Either Mg 2ϩ or Mn 2ϩ (but not Ca 2ϩ ) supported autophosphorylation, but Mg 2ϩ was significantly more efficacious. With Mg 2ϩ present, the addition of Ca 2ϩ and/or Mn 2ϩ had no marked effect (either stimulatory or inhibitory). Under all conditions tested, the addition of Cmd1 produced no significant enhancement of incorporation (Fig. 3C).
To determine the nature of the residues autophosphorylated in vitro, GST-Clk1 was incubated with [␥-32 P]ATP and 2 mM Mg 2ϩ , resolved by SDS-PAGE, transferred electrophoretically to a polyvinylidene difluoride filter (Immobilon-P, Millipore), and located by autoradiography. The corresponding region of the filter was excised, and bound protein was subjected to acid hydrolysis (43). Hydrolysis products were mixed with nonradioactive phosphoserine, phosphothreonine, and phosphotyrosine (as carriers and internal standards), resolved by twodimensional electrophoresis, and visualized by both ninhydrin staining and autoradiography. Radioactivity present in each spot was quantitated using a PhosphorImager (Molecular Dynamic, Inc.). Of the total radioactivity, 90% was present as phosphoserine and only 10% as phosphothreonine (data not shown). There was no detectable phosphotyrosine.
We were concerned that binding of the 9E10 mAb to the epitope tag on the C terminus of Clk1::myc might interfere with binding of yeast Cmd1, might mimic the effect of Cmd1 binding, or might otherwise disrupt proper regulation. Therefore, we also expressed Clk1 and various derivatives in bacterial cells as GST-Clk1 fusions. The GST fusions were purified by adsorption to and elution from glutathione beads. In solution, both GST-Clk1 and a C-terminally truncated derivative, GST-Clk1(⌬487-610), displayed autophosphorylation activity (Fig.  4); in contrast, a catalytically inactive mutant derivative, GST-Clk1(K201R), failed to incorporate any detectable radioactivity. During production and purification, some of the GST-Clk1 suffered proteolytic degradation and was converted to a size quite similar to that of the purposefully engineered GST-Clk1(⌬487-610) truncation. On a per molecule basis, both this proteolytic product and the truncated protein showed a significantly greater extent of autophosphorylation than the fulllength protein (data not shown). The presence of Ca 2ϩ and purified Cmd1 stimulated autophosphorylation of the GST-Clk1 protein only slightly, if at all, and had no detectable effect on autophosphorylation of either the GST-Clk1 degradation product or the GST-Clk1(⌬487-610) truncation (Fig. 4).
Phosphotransferase Activity of Clk1-Several exogenously added proteins were tested as potential phosphoacceptor substrates for Clk both in the immune complex kinase assay and using the soluble GST-Clk enzyme. Neither ␤-casein nor autocamtide-2 (53) (a synthetic peptide based on the autophosphorylation site sequence in mammalian CaM kinases) was detectably phosphorylated by either bead-bound or soluble Clk1. In contrast, a modest degree of incorporation into myelin basic protein was observed under both assay conditions (data not shown).
During the course of these studies, our attention was drawn to the possibility that yeast translation elongation factor 2 (EF-2), product of the EFT1 and EFT2 genes (54), might serve as an effective substrate for Clk1 for several reasons. First, a novel type of Ca 2ϩ /CaM-dependent protein kinase activity (dubbed Ca 2ϩ /CaM-dependent protein kinase III) has been implicated in the phosphorylation and regulation of mammalian EF-2 by in vitro biochemical studies (55). Second, the purified mammalian enzyme is able to phosphorylate yeast EF-2 and has a reported molecular mass (ϳ95 kDa) that is in the same size range as the apparent molecular mass of Clk1 (56). Third, a similar activity has been detected and partially purified from yeast (57); however, the gene encoding this Eft1/Eft2 kinase has not yet been identified. Finally, as dosage suppressors of the growth-inhibitory effect of overproduced C-terminally truncated Clk1 (see "Discussion"), we isolated genes implicated in translation and RNA binding (58).
Partially purified yeast EF-2 (generous gift of A. C. Nairn, Rockefeller University, New York) was readily phosphorylated by both GST-Clk1 and GST-Clk1(⌬487-610), but not at all by the catalytically inactive GST-Clk1(K201R) mutant (Fig. 4). The presence of Ca 2ϩ and purified Cmd1 did not significantly stimulate the ability of GST-Clk1 to phosphorylate yeast EF-2. In contrast, the autophosphorylation activity of purified yeast Cmk1, a CaM kinase with a typically broad substrate specificity (17,18), was greatly stimulated by the presence of Ca 2ϩ and Cmd1; yet, this enzyme was unable to phosphorylate yeast EF-2 under any conditions. Thus, the ability of GST-Clk1 to phosphorylate yeast EF-2 was remarkably specific.
To determine the nature of the residues in yeast EF-2 phosphorylated by Clk1 in vitro, the yeast EF-2 preparation was incubated in solution with purified GST-Clk1, [␥-32 P]ATP, and Mg 2ϩ and then subjected to SDS-PAGE. The resulting gel was dried onto a filter paper backing (Whatman, 3MM), and the portion containing the radiolabeled EF-2 was excised and then subjected to both phosphoamino acid analysis and phosphopeptide mapping (kindly performed in the laboratory of A. C. Nairn). Phosphoserine was found in the labeled EF-2, but no phosphothreonine or phosphotyrosine was detected (data not shown).
Clk1 Does Not Have a High Affinity for CaM-Because Clk1 has a sequence (residues 493-520) resembling known CaMbinding sites, yet neither its autophosphorylation nor its phosphotransferase activity responded to the presence of Ca 2ϩ and Cmd1 in vitro, other independent methods were applied to FIG. 3. Immune complex assay of Clk1 kinase activity isolated from yeast. Extracts were prepared, as described under "Experimental Procedures," from a clk1⌬ yeast strain (YMM3.1a) carrying vector alone (YEp352), the same plasmid overexpressing Clk1::myc (pMM16), or the same plasmid overexpressing the catalytically inactive Clk1(K201R)::myc mutant (pMM22). Immune complexes were captured on Protein A-Sepharose beads after incubation of the extracts with an anti-c-Myc mAb (9E10), washed with TBS, and incubated with 2 mM Mg 2ϩ and [␥-32 P]ATP, as described in detail under "Experimental Procedures." The products of the reaction were subjected to SDS-PAGE in a 10% gel. The gel was dried onto filter paper and analyzed by autoradiography (A). Another gel containing portions of the same immune complexes was transferred to a nitrocellulose filter, and the blot was incubated with the 9E10 antibody and visualized using enzyme-linked chemiluminescence (B). Reactions were conducted, as in A, with the immune complexes prepared from cells expressing Clk1::myc in the presence and absence of the various divalent cations indicated, with and without the addition of purified Cmd1 protein (yeast CaM) (C).
FIG. 4. Solution assay of Clk1 kinase activity using bacterially expressed enzyme. Extracts were prepared, as described under "Experimental Procedures," from an E. coli strain (DH5␣) expressing either GST-Clk1 (from pMM37), GST-Clk1(K201R) (from pMM45), and pGST-Clk1(⌬487-610) (from pMM42). These fusion proteins (ϳ0.5 g each; purified by adsorption and elution from glutathione beads, as described under "Experimental Procedures") or a sample of highly purified Cmk1 enzyme (ϳ500 ng) (17) was incubated with [␥-32 P]ATP and 2 mM Mg 2ϩ in the presence of 2 mM Ca 2ϩ and 0.5 g of Cmd1 (ϩ) or in the presence of 1 mM EGTA (Ϫ), either alone (Ϫ) or in the presence (ϩ) of a partially purified preparation (ϳ0.5 g of total protein) of yeast translation elongation factor 2 (YEF-2) (generous gift of A. Nairn). The products of the reaction were resolved using SDS-PAGE. The gel was dried onto filter paper and analyzed by autoradiography. The migration positions of the proteins indicated were confirmed (data not shown) by immunoblotting a gel containing otherwise identical samples with rabbit polyclonal anti-Clk1 antibodies and anti-Eft1/Eft2 antibodies (provided by A. Nairn).
determine if Clk1 can associate with Cmd1. First, a GST-Clk1(135-610) fusion protein was produced and purified from E. coli cells. Glutathione-agarose beads were loaded with an excess of this fusion protein, washed exhaustively, and then loaded with purified yeast Cmd1 (10 g) in the presence of 2 mM Ca 2ϩ . An essentially identical experiment was performed on another column in which the Cmd1 was loaded in buffer containing EGTA. The flow-through of each column was collected. The columns then were washed twice with the appropriate buffer, and the bound GST-Clk1(135-610) was eluted with glutathione. The resulting fractions were resolved by SDS-PAGE, and the distribution of Cmd1 was analyzed by immunoblotting with rabbit anti-Cmd1 antibodies. Either with or without Ca 2ϩ , all of the Cmd1 protein was in the flowthrough and in the initial wash fraction; none was detected in the eluted fraction, which contained all of the GST-Clk1(135-610) (data not shown). Thus, Cmd1 was unable to bind tightly to the C-terminal segment of immobilized Clk1.
As a second means to attempt to detect association of Clk1 with Cmd1, both proteins were overproduced in yeast cells from multicopy plasmids. CLK1 and CMD1 were expressed from the GAL1 promoter, and CLK1::myc was expressed from its own promoter. Clk1 (or Clk1::myc) was then immunoprecipitated from extracts of such cells using either the rabbit polyclonal antibodies (or the 9E10 mAb). The immunoprecipitates were resolved by SDS-PAGE and examined by immunoblotting. No Cmd1 was detectable, although copious amounts of Clk1 (or Clk1::myc) were present (data not shown).
As a third means to assess the ability of Clk1 to bind Cmd1, we used a CaM overlay method, as described previously for Cmk1 and Cmk2 (17). Extracts of E. coli cells expressing GST-Clk1 fusions were resolved by SDS-PAGE, transferred to filters, and incubated in the presence and absence of Ca 2ϩ with purified Cmd1 that had been freshly labeled by reaction with 125 I-labeled Bolton-Hunter reagent. Under no condition tested was a detectable amount of radiolabeled Cmd1 protein retained by the GST-Clk1 bound to the filter, whereas a readily detectable amount of label was retained by an equivalent amount of purified Cmk1 protein (data not shown).
Overexpression of CLK1 Inhibits Yeast Cell Growth-Removal of the C-terminal CaM-binding sequences downstream of the catalytic domain in CaM kinases generates a CaMindependent and constitutively active enzyme (15,16,59,60). To examine the role of the C-terminal domain of Clk1, high copy plasmids were constructed that express from the galactose-inducible GAL1 promoter full-length CLK1, the C-terminally truncated version, CLK1(⌬487-610), or derivatives of each of these genes containing the K201R mutation to ablate catalytic activity. On glucose-containing medium, which represses expression from the GAL1 promoter, cells carrying all four plasmids grew equivalently to cells carrying the vector alone (Fig. 5B). As expected, when shifted to galactose medium, yeast cells carrying each of the four plasmids overproduced the corresponding protein product, as judged by immunoblotting with polyclonal anti-Clk1 antibodies (Fig. 5A). When streaked to single colonies on galactose-containing plates, however, it was revealed that cells overproducing Clk1 grew much more slowly than control cells carrying the vector alone, as judged by relative colony size (Fig. 5B). More strikingly, overproduction of Clk1(⌬487-610) prevented cell growth almost completely (Fig. 5B). These growth inhibitory effects required the kinase activity of Clk1 because overproduction of catalytically inactive derivatives, Clk1(K201R) and Clk1(K201R ⌬487-610) had no detectable effect on growth (Fig. 5B). Growth inhibition caused by elevated expression of either Clk1 or Clk1(⌬487-610) was not lethal, however, because even after 10 -12 h in galactose medium nearly 100% viability was retained when cells were returned to glucose medium (data not shown). These genetic findings indicate that the C-terminal segment of Clk1 is a negative regulatory domain.
Clk1 Is Localized in the Cytoplasm-The basic region (residues 493-513) in the C-terminal portion of Clk1 does not resemble a typical nuclear localization signal (61). To shed additional light on the potential physiological function of Clk1, it was of interest to determine in which subcellular compartment Clk1 resides. Using indirect immunofluorescence and rabbit polyclonal antibodies, it was not possible to detect the endogenous level of Clk1. However, when Clk1 was modestly overproduced by expression of CLK1 from its own promoter on a multicopy plasmid, it was obvious that Clk1 was abundantly present in the cytosol and specifically excluded from the nucleus (Fig. 6). Cells carrying the clk1-⌬1::HIS3 mutation (see "CLK1 Is Not an Essential Gene") probed with the polyclonal anti-Clk1 antibodies (Fig. 6) or normal cells probed with preimmune serum from the same rabbit (data not shown) displayed only weak background staining. Cells expressing Clk1::myc and incubated with the 9E10 mAb also showed the same cytosolic staining and exclusion from the nucleus (data Samples (ϳ20 g) of extracts of a yeast strain (AS306 GAL ϩ ) harboring vector alone (YEp351GAL) (29), and the same plasmid expressing from the GAL1 promoter full-length Clk1 (pMM51), the catalytically inactive full-length Clk1(K201R) mutant (pMM52), the C-terminally truncated Clk1(⌬487-610) (pMM25), or a catalytically inactive version (K201R) of the C-terminally truncated Clk1(⌬487-610) (pMM56) were resolved by SDS-PAGE, transferred to a nitrocellulose filter, and incubated with rabbit polyclonal anti-Clk1 antibodies raised against GST-Clk1(⌬487-610) as the antigen, as described under "Experimental Procedures." B, overproduction of Clk1 and C-terminally truncated Clk1 is growthinhibitory. The same five strains described in A were restreaked to single colonies from glucose-containing plates onto selective medium containing either 2% glucose (SGlc-Leu; Glucose) or 2% galactose (SGal-Leu; Galactose). The plates were incubated at 25°C for 4 days and then photographed. not shown), whereas wild-type cells examined with the same antibody showed no detectable staining, as has also been observed by others who have used this same anti-epitope antibody for examining yeast cells by indirect immunofluorescence (62).
CLK1 Is Not an Essential Gene-To determine if the CLK1 locus is an essential gene, a deletion-insertion allele was constructed. This mutation removed 40% of the CLK1 coding region (including more than a third of the kinase domain) and substituted the HIS3 gene (Fig. 7A). Diploid His ϩ transformants heterozygous for the mutation were readily obtained (Fig. 7B), and sporulation of such a diploid strain (YMM3) yielded many tetrads containing four viable spores and in which His ϩ :His Ϫ segregated 2:2 (Fig. 7B). This result indicated that CLK1 function is not required for normal vegetative growth on rich or minimal medium.
Because of the similarity of CLK1 in certain respects to the CMK1 and CMK2 loci, it was also of interest to determine the phenotype of cells lacking all three genes. For this purpose, a spore carrying the clk1-1⌬::HIS3 allele was mated with a haploid of opposite mating type that carried the cmk1-1⌬::HIS3 and cmk2-1⌬::TRP1 mutations (17). The resulting diploid (YMM6) was sporulated and dissected. In 3 tetrads out of the 40 analyzed, His ϩ :His Ϫ segregated 2:2 and both of the His ϩ spores were Trp ϩ . Restriction endonuclease digestion and Southern blot hybridization analysis with appropriate DNA probes confirmed that the His ϩ Trp ϩ spores from such tetrads did indeed contain all three mutant alleles. All of these clk1⌬ cmk1⌬ cmk2⌬ triple mutant haploids grew essentially like wild-type cells on rich medium containing glucose, galactose, or glycerol as the carbon source or on minimal selective medium (data not shown).
Because the predicted product of the RCK1 gene shares an even greater degree of amino acid sequence identity to Clk1 than do Cmk1 and Cmk2, it was of obvious interest to determine the phenotype of a clk1⌬ rck1 double mutant. For this purpose, a rck1::LEU2 disruption allele was constructed and introduced into appropriate haploid and diploid strains, as described under "Experimental Procedures." It was possible to obtain clk1⌬ rck1 double mutant haploids at the expected frequency either from sporulation of diploids heterozygous at both loci and via direct transformation of clk1⌬ haploids with rck1 mutant DNA. Thus, RCK1 and CLK1/RCK2 do not constitute an essential gene pair, as previously reported (48).
With the strains we generated, it was also possible by appropriate crosses to construct strains to attempt to recover clk1⌬ rck1 cmk1⌬ cmk2⌬ quadruple mutant cells. Viable quadruple mutant spores were obtained at the expected frequency and were shown to possess all four mutant loci by restriction endonuclease digestion and Southern blot hybridization analysis using appropriate DNA probes. The growth of these clk1⌬ rck1 cmk1⌬ cmk2⌬ quadruple mutant cells was essentially indistinguishable from wild-type cells on both rich medium and minimal selective medium (data not shown), suggesting that CLK1, RCK1, CMK1, and CMK2 do not constitute an extended and essential gene family. DISCUSSION CaM kinases are thought to mediate many important regulatory signals in mammalian cells (16,60). Therefore, it was a somewhat surprising observation that yeast cells lacking both of the known CaM kinase homologs, Cmk1 and Cmk2, displayed no detectably deleterious phenotype (17,18). Our initial motivation in examining the CLK1 locus in detail was the possibility that the protein kinase encoded by this gene might have properties similar to other Ca 2ϩ /CaM-dependent protein kinases and might overlap in function with Cmk1 and Cmk2. In this way, the existence of the CLK1 gene might account for the observed viability of cmk1⌬ cmk2⌬ double mutants. However, the biochemical findings presented here demonstrate that, although the CLK1 gene product shares certain sequence features with other known and well characterized Ca 2ϩ /CaMdependent protein kinases, Clk1 is not a Ca 2ϩ /CaM-dependent protein kinase in vitro and does not appear to interact to any detectable extent with yeast CaM. Furthermore, the genetic results presented here indicate that neither CLK1 nor its homolog, RCK1, are essential genes and that CLK1, RCK1, CMK1, and CMK2 do not constitute a gene family with overlapping functions.
On the other hand, the properties of Clk1 are interesting and novel. Clk1 was able to phosphorylate the yeast translation elongation factor 2, which is thought to be the substrate of a FIG. 6. Clk1 is located in the cytosol and excluded from the nucleus. Wildtype cells (YPH499) harboring a plasmid (pMM12), which expresses Clk1 from its own promoter on a multicopy plasmid (YEp351), and an otherwise isogenic strain (YMM3.1a), carrying the clk1-1⌬:: HIS3 mutation and harboring the vector alone, were grown to mid-exponential phase, fixed, affixed to slides, permeabilized, and stained with 4,6-diamino-2phenylindole (right side) and then with primary and secondary antibodies, as described in detail under "Experimental Procedures." The primary antibody was a 1:300 dilution of a rabbit polyclonal anti-Clk1 antiserum (1188) that had been depleted of nonspecific cross-reacting activity by preadsorption to nitrocellulose filter replicas of extracts of clk1-1⌬::HIS3 cells separated by SDS-PAGE. The secondary antibodies were a 1:200 dilution of fluorescein isothiocyanate-conjugated goat anti-rabbit immunoglobulin antiserum (left side).
Ca 2ϩ /CaM-dependent protein kinase in both mammalian cells and yeast (56,57). The substrate selectivity of Clk1 is highlighted by the fact that Cmk1 was unable to phosphorylate yeast EF-2, even though Cmk1 has a rather broad substrate specificity in vitro (17,18). Furthermore, overexpression of a C-terminally truncated, and presumably hyperactive, form of Clk1 led to severe growth inhibition, and this toxicity required the catalytic activity of the kinase. Thus, the C-terminal domain of Clk1 is a negative regulatory element, as has also been demonstrated for the mammalian CaM kinases (16,59,60). Indeed, overexpression of C-terminally truncated derivatives of Cmk1 and Cmk2 also markedly inhibit yeast cell growth (58).
Our data do not support the conclusion that Clk1 is a Ca 2ϩ / CaM-responsive enzyme. We were unable to detect any effect of Cmd1 on Clk1 activity or to observe any stable association between Clk1 and Cmd1. On the other hand, it is possible that, in the immune complex assay, binding of the mAb to the Cterminal epitope altered the conformation of the C-terminal domain so as to alleviate its negative regulatory effect and, thereby, cause CaM independence. Although less likely, it could also be argued that, even in the GST-Clk1 fusion, the structure of the C-terminal domain is perturbed, leading to constitutive activation of the enzyme and precluding binding of Cmd1. Furthermore, even for mammalian CaM kinases, the degree of CaM-dependence varies for different phosphoacceptor substrates (16,60). Thus, it is also possible that Clk1 might have displayed a greater responsiveness to CaM activation if the "correct" substrate had been used.
The molecular mass of Clk1 is larger than that of Cmk1 and Cmk2 (and mammalian CaM kinases) but in the same size range as that reported for partially purified activities from mammalian cells (56) and yeast (57) that phosphorylate yeast EF-2 in a Ca 2ϩ /CaM-dependent manner. Despite its apparent lack of CaM responsiveness, Clk1 was able to phosphorylate EF-2. Studies by others have indicated that hyperphosphorylation of EF-2 inhibits translation (63,64). Two other observations are at least consistent with a potential role for Clk1 in regulation of protein synthesis. First, Clk1 is confined to the cytosol and, thus, is most abundant in the same subcellular compartment where translation occurs. Second, as a genetic approach to examine the physiological function of the Clk1, dosage suppressors and chromosomal mutations were isolated that ameliorate the growth inhibitory effect of overproduction of C-terminally truncated Clk1. All of the dosage suppressors analyzed to date are genes that encode RNA-binding proteins and other factors known to be involved in translation (58).
The RCK1 locus is the gene most closely related to CLK1 in the S. cerevisiae genome. RCK1 was identified by another group via its ability, when expressed in a Schizosaccharomyces pombe rad1 mutant, to suppress the UV sensitivity of these cells (48). These workers isolated CLK1/RCK2 because of its homology to RCK1. Their subsequent work, which has continued to involve the expression of these S. cerevisiae genes in S. pombe, suggests that the presence of Clk1 and Rck1 extends the G 2 phase of the cell cycle in S. pombe (65). In this regard, it has also been reported that a constitutively active form of mouse CaM kinase also inhibits the G 2 /M transition when expressed in S. pombe (66). However, we find that S. cerevisiae clk1⌬ and rck1 single mutants and clk1⌬ rck1 double mutants are no more UV-sensitive than otherwise isogenic wild-type cells (58), suggesting that the normal role of Clk1 (and Rck1) is not in repair of radiation damage (48) or in checkpoint control (65). At least as judged by budding pattern, cultures of cells growth-inhibited by overexpression of C-terminally truncated Clk1 do not accumulate cells with large buds, as would be diagnostic of a G 2 -specific arrest in S. cerevisiae (67). If Clk1 is a EF-2 kinase in vivo, one explanation for the ability of hyperactive Clk1 to inhibit growth in S. cerevisiae (and to retard the G 2 /M transition in S. pombe) is that the resulting hyperphosphorylation of EF-2 inhibits the synthesis of proteins, like cyclins, that are especially critical for cell cycle progression.
We attempted to determine if the state of EF-2 phosphorylation in S. cerevisiae is affected by the status of the CLK1 gene. However, anti-Eft1/Eft2 antibodies available to us (gift of A. C. Nairn) were not effective in immunoprecipitating detectable amounts of radioactive EF-2 protein from extracts of yeast cells labeled by growth in minimal medium containing [ 32 P]PO 4 3Ϫ . 5 In contrast, immunoprecipitation of Clk1 from cells labeled in this fashion demonstrated that Clk1 is a phosphoprotein in vivo, that the label is almost exclusively present in phosphoserine, and that neither the C-terminal truncation (⌬487-610) nor the catalytically inactive mutant (K201R) shows any significant diminution in incorporation (58). Thus, major phosphorylation sites in vivo do not reside in the C-terminal domain and can arise from the action of other cellular protein kinase(s), 5 M. Melcher, unpublished observations. FIG. 7. Structure and analysis of a clk1 null mutation. A, partial restriction endonuclease cleavage site map of the 3-kb ApaI-ApaI genomic DNA fragment that contains the CLK1 locus. To generate the clk1-⌬1::HIS3 allele, an EcoRI-NarI fragment specifying codons 327-564 of the CLK1 coding sequence was removed and replaced with the 2-kb EcoRI-Nar1 fragment containing the HIS3 gene and its promoter. The portion of the catalytic domain removed is shown schematically. B, DNA containing the clk1-1⌬::HIS3 mutation was used for DNA-mediated transformation of a haploid strain (YPH499) and an otherwise isogenic MATa/MAT␣ diploid strain (YPH501). DNA (ϳ5 g) from a resulting His ϩ haploid transformant (YMM1) and three His ϩ diploid transformants (YMM3, YMM3Ј, YMM3Љ) was examined for the presence of authentic gene transplacement by digestion with EcoRI and blotting with a probe (2.5-kb EcoRI-ScaI fragment) internal to the CLK1 coding sequence, as described under "Experimental Procedures." Sporulation and tetrad analysis of the YMM3 diploid yielded four viable spores (YMM3.1a, YMM3.1b, YMM3.1c, and YMM3.1d), in which the clk1-l⌬::HIS3 allele segregated 2:2. The diploid YMM6 was constructed by mating YMM3.1a with a cmk1⌬ cmk2⌬ MAT␣ strain (MPY62), and segregation of the clk1-1⌬::HIS3 mutation was followed in the resulting spores (YMM6.1a, YMM6.1b, YMM6.1c, and YMM6.1d) in the same way. and not obligatorily from autophosphorylation.
Despite the sequence relatedness of CLK1 to RCK1, CMK1, and CMK2, these genes do not constitute a family with shared and essential functions, since a clk1⌬ rck1 cmk1⌬ cmk2⌬ quadruple mutant was viable. No other closely related genes are present in the S. cerevisiae genome. Thus, these protein kinase activities do not appear to serve any function that is critical for vegetative growth but rather may be necessary for the cell to adjust efficiently to changing growth conditions. In this regard, the type II Ca 2ϩ /CaM-dependent protein kinases seem to contribute to the ability of yeast cells to adjust to heat stress (68). Thus, it is possible that Clk1 and its related protein kinases help adjust cellular metabolism to cope with environmental stress.