Two Yeast Genes Encoding Calmodulin-dependent Protein Kinases ISOLATION, SEQUENCING, AND BACTERIAL EXPRESSIONS OF CMKl AND CMKZ*

We have isolated two genes from Saccharomyces cerevisiae that both encode a calmodulin-dependent protein kinase (CaM kinase). The CMKl gene has been cloned by hybridization using an oligonucleotide probe synthesized on the basis of the peptide sequence of purified yeast CaM kinase (Londesborough, J. (1989) J. Gen. Microbiol. 135, 3373-3383). The other gene, CMK2, which is homologous to C M K l , has been isolated by screening at low stringency with a CMKl fragment as a probe. The CMK2 product expressed in bacteria shows Ca2+- and CaM-dependent protein kinase activity, indicating that CMK2 also encodes a CaM kinase. The CMKl and CMK2 products expressed in bacteria were found to have different biochemical properties in terms of autoregulatory activity and preference for yeast CaM or bovine CaM for maximal activity. Antibody raised against a peptide fragment of the CMKl protein cross-reacts with the CMK2 product. Immunoblotting with this antibody indicated that the CMKl and CMK2 products have apparent molecular masses of 56 and 50 kDa, respectively, in

step in the regulatory cascade involved in control of a diverse range of physiological processes, such as carbon metabolism, muscle contraction, and nerve signal transmission (1)(2)(3). An intracellular Ca2+ signal evoked by extracellular stimuli is thought to promote protein phosphorylation through a signaltransducing system involving CaM and CaM kinases. Several lines of evidence also suggest that a CaM kinase plays fundamental roles during cell cycle progression, especially in mitosis (4,5).
CaM kinases are classified into two groups on the basis of their substrate specificities (6): one with narrow substrate specificity and the other with broad substrate specificity. The former group includes phosphorylase kinase (7), myosin light chain kinase (S), CaM kinase I (9), CaM kinase I11 (lo), and CaM kinase Gr (11). Each CaM kinase in this group phosphorylates a target protein specific for the kinase, and thereby regulates the activity of its own discrete process. The latter group includes CaM kinase I1 isozymes (12, 13), which phosphorylate more than 10 different target proteins, modulating their functions simultaneously (14). These findings, together with the fact that isozymes of the CaM kinase I1 family (50)(51)(52)(53)(54)(55)(56)(57)(58)(59)(60) are distributed widely in many tissues and organisms (15, 16), suggest that this kinase is responsible for the regulations of a wide variety of proteins in uiuo and plays central roles in Ca2'/CaM signal transduction pathways.
The primary structure of CaM kinase I1 has been studied extensively in the last 4 years. Five homologous rat brain cDNAs encoding subunits of CaM kinase I1 isozymes have thus far been identified. The similarities in the entire amino acid sequences of the cy (478 amino acid residues) (17, 18), @ (542 amino acid residues) (19), p' (527 amino acid residues) (19), y (527 amino acid residues) (20), and 6 (533 amino acid residues) (21) subunits are more than 80%. Analyses of the deduced amino acid sequences have suggested that each subunit is composed of three functional domains: a catalytic domain within the N-terminal half, a multifunctional regulatory domain mapped in the central portion, and a C-terminal portion, which is thought to play a role in subunit assembly and/or subcellular localization (22, 23). The multifunctional regulatory domain is composed of an autoinhibitory region (residues 281-302 of the cy subunit) (24) and a CaMbinding region (residues 296-309) (25). This regulatory domain shows the unique features that autophosphorylation of Thr'"' in the autoinhibitory region greatly diminishes the binding of this region to the catalytic domain and converts the kinase of a Ca"/CaM-independent form (26-32).
The primary structures of other specific CaM kinases, such as phosphorylase b kinase (33) and myosin light chain kinase (34) have also been determined. A comparison of their catalytic domains shows that these three CaM kinases are closely related (18) and are members of the same protein kinase subfamily (35). In contrast, their other domains have less or no similarity.
Two previous papers reported the existence of CaM kinase activity in cell extracts of the yeast Saccharomyces cereuisiae (36, 37), and recently, Londesborough (38) reported purification of yeast CaM kinase to near homogeneity. He found that the purified fraction contains an autophosphorylatable 56-kDa CaM kinase with broad substrate specificity. These biochemical findings suggest that the properties of the yeast enzyme are similar to those of mammalian CaM kinase 11.
Analysis of conditional lethal mutants showed that yeast CaM is involved in nuclear division (39, 40), but little is known about the essential target of CaM in mitosis, and there is no information available on the role of yeast CaM kinase in this process. We attempted to determine the functional role of the CaM kinase in cell proliferation and growth control of yeast by molecular genetic studies. This paper reports the nucleotide sequences of two yeast CMK genes both encoding a CaM kinase. The first CaM kinase gene (CMKI ) encodes the CaM kinase purified previously (38). The second CaM kinase, the product of the CMKZ gene, is very similar in primary sequence to the CMKl product, but is distinct from it.

RESULTS
Peptide Sequence Analysis of Yeast CaM Kinase-The partial amino acid sequence of yeast CaM kinase was determined from peptide fragments of the purified protein. Yeast CaM kinase was purified as described in Ref. 38. Traces of two yeast polypeptides contaminating the final preparation (38) were removed by one-dimensional polyacrylamide gel electrophoresis. Proteins were detected by staining with Coomassie Brilliant Blue, and regions containing the CaM kinase were cut out and treated with lysylendopeptidase (41). Peptide fragments released from the gel were recovered in the supernatant, separated by reverse-phase high performance liquid chromatography (42), and sequenced with an automated gasphase peptide sequenator. Six discrete sequences (TN18,  TN19, TN28A, TN28B, TN33, and TN40) recovered from five peaks of material separated by high performance liquid chromatography were determined (Fig. lA).
Isolation of the CMKl Gene-On the basis of the TN33 peptide sequence, two types of 44-mer oligonucleotide probe were synthesized (Fig. 1B). The TN33N probe was synthesized according to the most frequent codon usage in S. cereuisiae (43). The TN33C probe was designed to take advantage of the weak G-T mismatch in base pairing (44). For characterization of these two probes, we hybridized them to yeast genomic DNA digested with BamHI (Fig. 2). Under a low stringency condition at 49 "C, TN33C hybridized to only one fragment of 1.5 kb, while TN33N hybridized to many fragments. In a more stringent condition a t 54 "C, TN33C still hybridized to the same fragment, but no fragment was detected with TN33N. From these results, the TN33C probe was chosen for cloning the yeast CaM kinase gene. We used a yeast genomic bank, which was divided into 48 pools, each containing 196 independent clones (45). On screening DNA from the 48 pools by dot hybridization, 5 pools gave positive Portions of this paper (including "Materials and Methods" and Figs. 1-3 and 8) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. signals. By a second screening of these 5 pools, 5 independent positive clones (pJOY101-pJOY105) were obtained. These five plasmids carried a 1.5-kb BamHI fragment that hybridized to the TN33C probe. Moreover, restriction enzyme mapping indicated that they all contained the same DNA region (data not shown). Subcloning of the pJOYlOl revealed that it contained a 2.8-kb KpnI-ClaI fragment that hybridized to the TN33C probe.
Nucleotide sequencing of the 2.8-kb KpnI-ClaI fragment from pJOYlOl revealed the presence of a single open reading frame (ORF) that encodes a polypeptide of 446 amino acids with a calculated molecular mass of 50,295 daltons (Figs. 3 and 4). The ATG (position l), a start codon of the ORF, is immediately preceded by translation termination codons in all three reading frames. No conserved splicing signal sequences are present near the coding region. Of particular significance is the fact that nucleotide sequences that could encode the six peptide fragments obtained from the purified CaM kinase (38) were all found to be included in the ORF. Thus we named the gene CMKl (for CaM-dependent multifunctional protein kinase).
Isolation of the CMK2 Gene-From structural analysis of rat CaM kinase I1 cDNA, it has been established that the a, p, y, and d subunits of rat CaM kinase I1 represent a gene family. Therefore, we next examined whether yeast contains any gene that is homologous to CMKl by Southern blotting (Fig. 5). We used a 1.1-kb SplI-Hind111 fragment within the CMKl coding region as a probe, and digested yeast genomic DNA with BamHI. Under both high (65 "C) and low (57.5 and 50 "C) stringent conditions, the CMKI probe hybridized strongly to CMKl (1.5-and 2.0-kb fragments). In addition, it hybridized weakly to another homologous fragment (>lo-kb fragment) under the low stringent condition. Digestions of yeast genomic DNA with three other restriction enzymes (EcoRI, HindIII, and PstI) gave essentially similar results, suggesting the existence of one homologous gene in S. cereuisiae. For cloning this gene, we used dot hybridization with the yeast genomic library described above (45). Positive signals were obtained under the low stringent condition (50 "C) from 9 to 48 pools, including 5 pools which contained the CMKl clone. Positive clone (pJOY201, pJOY202, pJOY203, and pJOY301) were obtained independently from the other 4 pools by colony hybridization and analyzed with a restriction enzyme. Three of the four clones (pJOY201, pJOY202, and pJOY203) were shown to contain a 2.8-kb PstI fragment that hybridized to the CMKl probe only at low stringency. We, therefore, characterized this fragment further. Analysis of the restriction enzyme map and partial DNA sequence showed that the pJOY3Ol plasmid contained a chimeric DNA derived from CMKl and another genomic region (data not shown), and so was possibly an artifact generated during construction of the library.
Nucleotide sequencing of the 2.8-kb PstI fragment from pJOY201 revealed the presence of a single ORF encoding a protein of 447 amino acids (Mr = 50,447) (Figs. 3 and 6) that is highly homologous to the predicted CMKl product. The deduced amino acid sequence is 63% identical to that of the CMKl product. Moreover, when conserved amino acid replacements are included, the similarity is more than 90%. The identity at the nucleotide level is more than 60%. From these facts, the cloned CMKl-related gene was designated as CMK2. Genomic Southern blot hybridization using the 1.2kb BglII-BglII fragment from the CMKZ gene as a probe was performed under the low stringency condition. However, this mosomes on orthogonal field-agarose gel electrophoresis gels (46). The probes used were a 1.1-kb SplI-Hind111 fragment from pJOYlOl and a 1.3-kb BglII-BglII fragment from pJOY201. Under the stringent condition, each probe hybridized to a single fragment: CMKl was assigned to chromosome VI, and CMK2 to chromosome XV. Tetrad analysis provided additional information about the map positions of CMKl and CMK2. A diploid strain YOJ211 heterozygous for both CMKl and CMK2 loci (CMKllAcmk1::TRPl CMK2IAcmk2: LEU2) was constructed, and subjected to meiosis and spore formation. More than 90% of the tetrads produced 4 viable spores and phenotypic analysis showed that 50 of 54 tetrads were either the parental ditype or the nonparental ditype (PD:NPD:T = 23:27:4). These observations indicate that the CMKl and CMK2 genes are both tightly linked to the centromere. Linkage between CMK2 and pho80 (47) was also confirmed by tetrad analysis (PD:NPD:T = 43:0:11, Fig. 7). Recently, a plasmid-based DNA library of chromosome VI was constructed completely: Restriction enzyme analysis as well as Southern blot analysis with this DNA library revealed that CMKl was located on the right arm of chromosome VI, adjacent to the SUP11 gene (48): the order is CENG-SUP11-CMKl (Fig. 7).
Expressions of CMKl and CMK2 Proteins in Escherichia coli-We think that CMKI encodes the 56-kDa yeast CaM ' N. Ogasawara, personal communication.   rying PET-CMK1 produced a polypeptide of 56 kDa and those carrying PET-CMK2 produced two polypeptides of 50 and 46 kDa (Fig. 8); these proteins were not present in uninduced bacteria, or in induced bacteria containing a control plasmid. Densitometric analysis showed that these PET-CMK1-and PET-CMK2-dependent polypeptides constituted 8 and 17%, respectively, of the total cellular proteins. Fractionation analysis showed that the CMKl protein was mainly present in the soluble fraction, whereas about 80% of the CMK2 proteins were present in the particulate fraction (Fig. 8).
The CMK proteins expressed in E. coli were partially purified on CaM-conjugated Sepharose CL-4B (see "Materials and Methods"). The CMKl and CMK2 proteins constituted 85 and 93%, respectively, of the proteins in the purified samples (Fig. 9A, lanes 2 and 3). Fig. 9B shows the autophosphorylation activities of these proteins. After incubation with [y3'P]ATP, samples were subjected to electrophoresis on SDS-polyacrylamide gel, and analyzed by autoradiography. In each case, only one band corresponding to the CMKl and CMK2 protein was detected (lanes 2 and 3 ) . The levels of autophosphorylation reach to the plateau after incubation for 15 min (data not shown).
The protein kinase activities of the partially purified samples were next assayed with and without Ca'+ and CaM, using synthetic peptide as substrate. Fig. 10 shows that the CMKl or CMK2 protein has Ca2+-and CaM-dependent phosphorylation activities. The protein kinase activity was increased about 2.5-4.2-fold in the presence of Ca'+ and CaM. For maximal activity, CMKl kinase preferred bovine CaM to yeast CaM, while CMK2 kinase preferred yeast CaM. Purified yeast CaM kinase has previously been shown to prefer bovine CaM for maximal activity (Fig. 10; Ref. 38), like the CMKl gene product. Next, we determined the CaZ'/CaM-independent activities after the CMK kinases had been subjected to autophosphorylation with cold ATP. CMKl kinase was not activated by preincubation with ATP, whereas CMK2 kinase was activated 1.8-fold by this treatment (Fig. 8). Preincubation in the absence of ATP showed no stimulation. This finding indicates that CMK2 kinase expressed in E. coli shows autoregulatory activity, like mammalian CaM kinase 11. Table I shows a comparison of substrate specificities of the CMKl and CMK2 kinases for six conventional substrates, assayed as described in Ref. 50. All the substrates were phosphorylated by purified yeast CaM kinase, the CMKl and CMK2 kinases, showing that the yeast CaM kinases have broad substrate specificity. It was found that myelin basic protein and myosin light chain were markedly phopshorylated by yeast CaM kinases even in the absence of CaM, whereas, casein was phosphorylated strictly in a Ca2+/CaM-dependent manner. Purified yeast CaM kinase show a 50-fold stimuiation by Ca2+ and CaM. Since mammalian CaM kinase shows a Ca'+ dependence for each substrate tested here (50), this Ca2'/CaM-independent property may be unique for the yeast CaM kinase. The yeast CaM kinases phosphorylated synthetic peptides containing gizzard myosin light chain (Kemptamide) more than that containing phosphorylation site 2 of glycogen synthase (Syntide 2). This is another characteristic of the yeast CaM kinases, since brain CaM kinase I1 was reported to phosphorylate Syntide 2 5.5-fold more than Kemptamide (50).
Detection of CMKl and CMK2 Products in Yeast-To identify the authentic CMK products in yeast, we prepared specific rabbit antiserum. The antiserum was raised against a polypeptide fragment (residues 69-341) of CMKl (see "Materials and Methods"), and found to cross-react with both the CMKl and CMK2 proteins expressed in bacteria (data not shown). Affinity purified antibody was used in later experiments.
The CMKl product was identified by immunoblot analysis of total yeast proteins as follows. A 56-kDa protein was detected in wild-type yeast cells and was present in greater abundance in cells carrying the CMKl gene on a multicopy plasmid, YEP-CMK1 (Fig. 11). This protein was not present in a mutant carrying the deletion Acmkl ::TRPl. Furthermore, the immunoreactive 56-kDa protein detected in a yeast lysate migrated with the purified 56-kDa yeast CaM kinase (38) on SDS-PAGE. Besides the 56-kDa protein, a 90-kDa protein was detected, but its origin is unknown. Cells carrying the CMK2 gene on a multicopy plasmid, YEP-CMK2, contained a 50-kDa protein. This protein is not a degradation product of the CMKl protein, as it was also detected in a Acmkl deletion mutant carrying the YEP-CMK2 plasmid. The 50-kDa protein was not detected by immunoblot analysis in wildtype cells, indicating that it was detected only in cells with multicopies of the CMK2 gene (Fig. 11). From these results, we concluded that the CMKl and CMK2 products in yeast have apparent molecular masses of 56 and 50 kDa, respectively.
The expression of chromosomal CMKl and CMK2 genes was examined by immunoprecipitation (Fig. 12). Both the 56and 50-kDa proteins were detected in wild-type cells (CMKl CMK2), and the former was the CMKl product as described before. The 50-kDa protein was proved to be the CMK2 gene product, since it was detected even in the Acmkl CMK2 cells and not in Acmk2 cells.    protein kinases whose sequences are now available, the CMK kinases show the greatest homology with mammalian CaM kinase I1 (Fig. 13). CMKl kinase exhibits overall similarity with mammalian CaM kinase 11, and the N-terminal catalytic domain shows remarkable conservation. On the basis of the alignment in Fig. 13 (34), and the phosphorylase b kinase y chain (33), are closely related to one another and known to be members of the same protein kinase subfamily (35). CMKl kinase also shows similarity to myosin light chain kinase and phosphorylase b kinase, but their sequence identities with the CMKl kinase in the catalytic domain are lower than that of CaM kinase 11. From an extensive study using synthetic peptide analog, the calmodulin-binding domain of the a subunit of rat CaM kinase I1 has been mapped at residues 296-309 on the Cterminal side of the catalytic domain (25). The similarity between the CMKl kinase and CaM kinase I1 within this region is not so high: only 3 of 14 residues are identical. However, homologous positions in the two yeast CMK kinase sequences seem to have the ability to form a basic amphipathic helical structure, which is a common structural feature of calmodulin-binding domains. The hydrophilic C-terminal region of rat CaM kinase I1 is proposed to be a domain for assembly. Similarity within this region is also not high, although hydrophilic characteristics are conserved in both CMK kinases.
The CMKl and CMK2 kinases in yeast appear to belong to the CaM kinase I1 family, but they differ markedly with respect to their autophosphorylation sites (26, 27). CMK2 kinase contains a putative autophosphorylation site at position 316, but CMKl kinase does not have one in a homologous region. In the CMKl sequence, the Thr3% residue on the amino-terminal side of the CaM-binding site is conserved, but the Arg located 3 residues upstream of ThraW is absent. This difference is ,of interest because the homologous position of rat CaM kinase I1 plays a role in autonomous activation of kinase activity.
During the homology search, we found another CMK-related gene, the yeast SNFl gene essential for the regulation of carbon catabolite repression (53). In the catalytic domain, 92 of 252 amino acids are identical (Fig. 13), and if conserved amino acid replacements are taken into account, the similarity is 74.6%. This gene is of further interest in that the similarity extends beyond the catalytic domain. Optimal positional similarity is achieved by deletion of the CaM-binding domain in the S N F l sequence. In other words, the CaM-binding domain is obviously absent in the S N F l product, but the domain required for assembly is partially conserved. It remains to be seen whether the sequence similarity between CMKl and S N F l reflects any underlying structural or functional similarity.
Disruption of CMKl and CMK2 Gene-To obtain information on the physiological functions of yeast CaM kinases, we analyzed the phenotypes of a Acmkl mutant, a Acmk2 mutant, and a Acmkl Acmk2 double mutant. Deletion mutants of both genes with long deletions within their coding regions were constructed (Fig. 14). Diploid strains heterozygous for CMKl and CMK2 were constructed and subjected to tetrad analysis. Tetrad analysis showed that more than 90% of the asci generated 4 viable colonies and some spores showed the Trp+ Leu+ phenotype. Southern blot hybridization with the CMKl and CMK2 probes showed the chromosomal gene disruption (Fig. 14). In addition, no immunoprecipitable CMK proteins were detected with antibody against the CMK products in the Acmkl Acmk2 double disrupted mutant (Fig. 12). Therefore, cells lacking CMKl, CMK2, or both genes are viable. In addition, cells carrying double disruptions appeared to grow normally under a variety of conditions: they grew at 17,23,30, and 37 "C in rich medium and in the presence of a nonfermentable carbon source, and in synthetic medium. They could conjugate well to cells with an a or a mating type, and showed no defect in meiosis or sporulation." Thus, even the double disruption of the CMK genes has no effect on any cellular activities thus far examined.

DISCUSSION
CMKl and CMK2, Two Yeast CaM Kinase Genes-We have isolated two CaM kinase genes, named CMKl and was also concluded to encode a CaM kinase on the basis of biochemical analysis of the CMK2 product expressed in E. coli and the striking similarity of the CMK2 product with the CMKl product. Moreover, the CMK2 protein was detected in a yeast cell lysate, indicating that this gene is also expressed in yeast cells.

CMK Kinase as a Mammalian CaM Kinase 11 Homologue-
Previous biochemical characterization has suggested the functional similarity of yeast CaM kinase to mammalian CaM kinase I1 (38). The sequence comparison in this study confirmed that the yeast CMK kinases are homologues of mammalian CaM kinase 11, both CMK kinases showing greater overall similarity to the CaM kinase I1 than to any other protein kinases so far identified. In particular, the conservation of the catalytic domains of these kinases strongly suggests a functional homology in their catalytic actions. The CMK kinases also show similarity to myosin light chain kinase and phosphorylase b kinase, both of which are members of the same CaM-dependent protein kinase subfamily. But their similarities to the CMK kinases are restricted to the catalytic domain, indicating that the yeast CMK kinases are not so closely related to these protein kinases as to mammalian CaM kinase 11. The similarity between the two yeast CMK kinases is much higher than that between the yeast and mammalian CaM kinases, suggesting that these two yeast kinases separated after evolution of yeast. One major difference between the yeast and mammalian CaM kinases is that the amino acid sequences of yeast kinases ( C M K l , 446 amino acid residues; CMK2,447 amino acid residues) are shorter than that of CaM kinase I1 (a subunit, 478 amino acid residues; /3 subunit, 542 amino acid residues). There are long deletions (approximately 80 amino acid residues) in the C-terminal portion of the yeast kinases. The C-terminal portion is thought to be involved in subunit assembly (22), so this deletion could lead to differences in the degree of oligomerization. In fact, Londesborough (38) reported that native yeast CaM kinase is a dimer, whereas the CaM kinase I1 isozymes in rat forebrain are composed of 6 to 12 subunits (54). contained the CMKI product (Fig. 9). Each of the CMKl and CMK2 products expressed in bacteria showed a protein kinase activity by itself (Fig. 10). These observations imply that yeast CMKl and CMK2 kinases can function independently in the cells. However, we have not ruled out the possibility that an isozyme species composed of both the CMKl and CMK2 products may exist in yeast. Differences between CMKl and CMK2"Biochemical studies on the proteins expressed in bacteria showed that the CMKl product has similar properties to the CMK2 product, but is a distinct protein. One difference is in preference for CaM species for maximal phosphorylation activity: the CMKl kinase is activated more by bovine CaM than by yeast CaM, whereas the CMK2 kinase is activated more by yeast CaM (Fig. 10). Yeast CaM shows 60% sequence identity with bovine CaM (5.9, and the extent of maximal activation of some CaMdependent enzymes by bovine CaM is higher than that by yeast CaM (56, 57). Only 7 of 14 amino acid residues within the putative CaM-binding domains of the two enzymes are identical. This sequence difference may account for the difference in biochemical properties. Another difference noted is the appearance of Ca'+/CaM-independent activity after autophosphorylation: the CMKl kinase is not activated by autophosphorylation (38; Fig. lo), whereas the CMK2 kinase is. In the CY subunit of rat CaM kinase 11, the ThrZS6 residue is known to be the autophosphorylation site required for autoactivation. This phosphorylation site is conserved in CMK2 but not in CMKl (Fig. 13). Further analysis is needed to identify the residue autophosphorylated in the CMK2 se-quence, but probably a homologous Thr residue is phosphorylated, thereby converting the kinase to a Ca"/CaM-independent form. Miyakawa et al. (37) reported that a partially purified yeast CaM kinase showed CaZ+/CaM-independent activity after autophosphorylation, so their preparation probably contained the CMK2 product. Autoregulatory activity has been thought to be important after the Ca'+ signal has been evoked and decays. The role of the autoregulatory activity of yeast CaM kinases in cellular function is still unknown, but our data indicate that the autoregulatory activities of the two enzymes are different.
Physiological Functions of the CMKl and CMK2 Products-Purified yeast CaM kinase (CMKl kinase) phosphorylates endogenous substrates of 50-and 500-kDa polypeptide in vitro (38). The identities of these substrates are not known, but the Ca'+-dependent phosphorylation of these proteins suggests the possible function of CaM kinase in Ca'+-dependent cellular processes. The different properties of the CMKl and CMK2 kinases also suggest that these two CMK kinases both participate in cellular processes. However, elimination of both kinase genes is not lethal. Our present working hypothesis is that some other gene(s) may substitute for the CMKl and CMK2 functions in double mutants. Judging from the observation that no additional members of this gene family were detected with either the CMKl or CMK2 probe, yeast cells may contain a distantly related gene product. For instance, yeast contains three cyclin genes, two of which were isolated by a multicopy suppressor of cdc28 mutation (58), while the third one is a distantly related gene. For control of cell proliferation, the three genes can be replaced by one another: all combinations of double disruptions result in cells with minimal growth defects (59). Homology search with CMK sequences revealed the S N F l gene, which has been shown to be essential for the regulation of carbon catabolite repression (53), as a candidate for the third related gene in yeast. Its similarity extends beyond the catalytic site to other regions. Although it is not known whether the substrate specificity of S N F l kinase is similar to those of the CMK kinases, it is clear that S N F l contains no homologous CaM-binding site and may not be regulated by CaM. Genetical analysis of CMK double disruption in combination with snfl mutation may be helpful in determining the function of CMK kinase in yeast cells. Another genetic approach is isolation of a conditional lethal mutant in which the third gene is mutated, using a synthetic lethal phenotype (60). Studies along these lines are in progress. 14.