Identification of Functional Connections Between Calmodulin and the Yeast Actin Cytoskeleton

Corresponding author: Yoshikazu Ohya, Department of Biological Sciences, Graduate School of Science, University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan., ohya{at}biol.s.u-tokyo.ac.jp (E-mail).

Communicating editor: M. D. ROSE

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

One of four intragenic complementing groups of temperature-sensitive yeast calmodulin mutations, cmd1A, results in a characteristic functional defect in actin organization. We report here that among the complementing mutations, a representative cmd1A mutation (cmd1-226: F92A) is synthetically lethal with a mutation in MYO2 that encodes a class V unconventional myosin with calmodulin-binding domains. Gel overlay assay shows that a mutant calmodulin with the F92A alteration has severely reduced binding affinity to a GST-Myo2p fusion protein. Random replacement and site-directed mutagenesis at position 92 of calmodulin indicate that hydrophobic and aromatic residues are allowed at this position, suggesting an importance of hydrophobic interaction between calmodulin and Myo2p. To analyze other components involved in actin organization through calmodulin, we isolated and characterized mutations that show synthetic lethal interaction with cmd1-226; these "cax" mutants fell into five complementation groups. Interestingly, all the mutations themselves affect actin organization. Unlike cax2, cax3, cax4, and cax5 mutations, cax1 shows allele-specific synthetic lethality with the cmd1A allele. CAX1 is identical to ANP1/GEM3/MCD2, which is involved in protein glycosylation. CAX4 is identical to the ORF YGR036c, and CAX5 is identical to MNN10/SLC2/BED1. We discuss possible roles for Cax proteins in the regulation of the actin cytoskeleton.

CALMODULIN, a highly conserved calcium-binding protein, has been implicated in Ca²⁺-mediated signaling cascades, including muscle contraction and neurotransmitter release (COHEN and KLEE 1988 ). Essential functions of calmodulin for cell proliferation have been studied in diverse eukaryotic cells, including Saccharomyces cerevisiae (DAVIS 1992 ; OHYA and ANRAKU 1992 ), Schizosaccharomyces pombe (TAKEDA and YAMAMOTO 1987 ), Aspergillus nidulans (RASMUSSEN et al. 1990 ), and Aspergillus oryzae (YASUI et al. 1995 ). It seems likely that calmodulin performs diverse functions by interacting with many different target proteins. Indeed, a large number of calmodulin-binding proteins possessing diverse activities in vitro have been identified so far (COHEN and KLEE 1988 ). Many of these proteins have been well characterized biochemically, but it has generally remained unclear which calmodulin targets have functional significance and how these functions are regulated in the cell.

Functions of calmodulin in cell proliferation have been extensively studied in S. cerevisiae. The yeast has a single essential calmodulin gene (DAVIS et al. 1986 ). Because the vertebrate calmodulin is able to functionally replace the endogenous yeast calmodulin (DAVIS and THORNER 1989 ; OHYA and ANRAKU 1989 ), essential functions of yeast calmodulin can be regarded as the archetype of vertebrates. We previously succeeded in systematic isolation of 14 temperature-sensitive calmodulin mutations by phenylalanine-to-alanine mutagenesis (OHYA and BOTSTEIN 1994B ). The seven well-conserved phenylalanine residues we mutagenized are likely to be important in interactions with target peptides, as judged by NMR and X-ray structural analysis of the complex between calmodulin and calmodulin-binding peptides (BABU et al. 1988 ; IKURA et al. 1991 , IKURA et al. 1992 ). The most striking finding from systematic mutagenesis of calmodulin was that the mutations formed four intragenic complementation groups (OHYA and BOTSTEIN 1994A ). Each group has a characteristic functional defect in actin organization (cmd1A), calmodulin localization (cmd1B), nuclear division (cmd1C), or bud emergence (cmd1D). Thus, each complementation group of calmodulin mutations can be thought of as producing a functional defect by loss of interaction with different essential target ligands. Complementing yeast calmodulin mutations have recently proved helpful in dissecting different steps of receptor-mediated endocytosis (GELI et al. 1998 ). Several calmodulin-dependent enzymes, including protein kinases and protein phosphatases, have been studied in yeast. Two essential calmodulin targets have thus far been identified in yeast. One is Myo2p (JOHNSTON et al. 1991 ; BROCKERHOFF et al. 1994 ) and the other is Nuf1p (GEISER et al. 1993 ; KILMARTIN et al. 1993 ; STIRLING et al. 1994 ). Myo2p is a class V myosin that is involved in polarized growth and functionally implicated in a post-Golgi stage of the secretory pathway (JOHNSTON et al. 1991 ; GOVINDAN et al. 1995 ), although the molecular mechanism is not well understood. Nuf1p plays a role in proper assembly of the spindle pole body, the primary microtubule organizing center in yeast.

To shed more light on the mechanism(s) by which calmodulin affects actin organization, a group of conditional-lethal calmodulin mutants (complementation group A) that displays a defect in intracellular actin organization has been characterized further. A temperature-sensitive calmodulin mutation that falls in this group, cmd1-226 (F92A), results in loss of localized actin cortical patches in the bud and disappearance of the actin cables at the restrictive temperature (OHYA and BOTSTEIN 1994A ). Direct binding between calmodulin and actin had not been detected in a gel overlay assay or by sedimenting F-actin in the presence of calmodulin (PIAZZA and WALLACE 1985 ), suggesting that the phenotypes of cmd1-226 must be an indirect effect mediated by an as yet unidentified actin-binding protein. Association of calmodulin with actin-binding proteins in other eukaryotes has often been reported. Another example is caldesmon, a calmodulin-binding protein found in smooth muscle that binds F-actin (SOBUE et al. 1981 ).

In the budding yeast S. cerevisiae, Myo2p appeared to be the only essential calmodulin target known to bind actin. Thus, Myo2p was an obvious candidate to be the mediator of the essential function of calmodulin upon the actin cytoskeleton. Myo2p contains a myosin-like head domain, a series of IQ motifs associated with calmodulin binding, and a C-terminal coiled-coil region, which are well conserved among the members of the class V myosin family (JOHNSTON et al. 1991 ). A temperature-sensitive myo2-66 mutation encodes a Glu-to-Lys change at position 511, which lies at the actin-binding face in the head domain (LILLIE and BROWN 1994 ). The myo2-66 cells stop growing at the restrictive temperature as large, unbudded cells with mislocalized actin patches (JOHNSTON et al. 1991 ). Immunofluorescence localization of Myo2p at the growing sites of cells further implicates the role of Myo2p in polarized growth (BROCKERHOFF et al. 1994 ; LILLIE and BROWN 1994 ). Physical interaction between Cmd1p and Myo2p was demonstrated both by immunofluorescence microscopy and gel overlay assay (BROCKERHOFF et al. 1994 ).

We present several lines of genetic and biochemical evidence, suggesting that Myo2p is a downstream target of calmodulin that is essential for actin organization. Analyses of replacement mutations at position 92 of calmodulin, which is altered in cmd1-226, indicate that hydrophobic and aromatic interactions are important between Cmd1p and Myo2p. We have also undertaken a genetic screen to identify mutations that show synthetic lethal interaction with cmd1-226 ("cax" mutations). Phenotypic and genetic analyses of cax mutants suggest that all the Cax proteins are involved in the regulation of the actin network.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Yeast strains, media and genetic methods:
The strains used in this paper are listed in Table 1. Rich medium (YPD) contains 1% Bacto-yeast Extract (Difco, Detroit, MI), 2% polypeptone (Nihon Seiyaku, Tokyo), and 2% glucose (Wako Chemicals, Osaka, Japan). Synthetic growth medium (SD) for selective growth was described previously, as were methods for tetrad analysis (KAISER et al. 1994 ). Lithium acetate (Wako Chemicals) was used for yeast transformation with a modification (SCHIESTL and GIETZ 1989 ) of the original method (ITO et al. 1983 ). FOA agar plates were made by adding 1 µg/ml 5-fluoroorotic acid (Sigma, St. Louis, MO) to SD agar.

DNA manipulations and plasmids:
Standard procedures were used for DNA manipulations and Escherichia coli transformation (SAMBROOK et al. 1989 ). Strains DH5F' or SCS1 were used to propagate plasmids. DNA sequencing was carried out with an automated DNA sequencer (model 373A; Applied Biosystems, Foster City, CA) equipped with a fluorescence detector for electrophoretically separated DNA fragments (CONNEL et al. 1987 ), using both universal primers and primers synthesized specifically as needed. Southern blotting analysis was performed with the ECL gene detection system (Amersham Pharmacia Biotech) to confirm sites of chromosomal integration.

pRB1612 (YCpU-CMD1) containing CMD1, CEN, and URA3 was described previously (OHYA and BOTSTEIN 1994B ).

pRB1612L (YCpL-CMD1) containing a 2-kb SalI-BamHI fragment of CMD1 was constructed from pRB1612 and pRS315.

pRB1616 (OHYA and BOTSTEIN 1994B ) was used for subcloning of the BstBI-SphI Phe92 to any amino acid cmd1 fragment amplified by PCR.

pRB1617 (YIpHade3) (OHYA and BOTSTEIN 1994B ) was used for chromosomal integration of the Phe92–X cmd1 mutations. To create YOC376, YOC377, YOC378, YOC379, YOC380, YOC381, and YOC382, an SphI-BamHI fragment containing the Phe92–X cmd1 mutation from pRB1616-derived plasmid was inserted into pRB1617.

pRB1617L (YIpLade3) was made by replacing the BamHI-XhoI HIS3 fragment of pRB1617 with the 2-kb BamHI-SalI LEU2 fragment from pJJ282 (JONES and PRAKASH 1990 ). To create YOC1170, YOC1172, YOC1174, YOC1175, YOC1176, YOC1178, YOC1180, and YOC1181, the SphI-BamHI fragment containing the allele of the wild-type calmodulin, cmd1-228, cmd1-233, or cmd1-239 from the pRB1616-derived plasmid (OHYA and BOTSTEIN 1994B ), was inserted into pRB1617L.

pYO693 (pYSLU1-CMD1) contains partially defective CEN3 (KOSHLAND et al. 1985 ), ADE3, URA3, and CMD1 (Y. OHYA, unpublished results). The plasmid was used for synthetic lethal screening in the ade2 ade3 strain.

pYO1148 (YCpL-MYO2) was made by insertion of a 5.5-kb ClaI-EcoRI fragment of p10-2B (JOHNSTON et al. 1991 ) containing MYO2 into the CEN, LEU2 plasmid, pRS315 (SIKORSKI and HIETER 1989 ).

Random replacement mutagenesis:
Four codons at positions 89–92 of CMD1 were replaced with random sequence using random replacement mutagenesis (PALZKILL and BOTSTEIN 1992A , PALZKILL and BOTSTEIN 1992B ). First, the nucleotide sequence 5'-GCTTTTAAAGTATTC-3' at positions 88–92 (covering codons 262–276) of CMD1 was changed to 5'-GCCGGCCTCGAGGGTCTCC-3' by a two-step PCR method (HO et al. 1989 ), with the result that the sequence at the region of mutagenesis was replaced with a sequence containing NaeI and BsaI recognition sites (underlined). Second, this new oligonucleotide sequence was released by digestion with NaeI and BsaI, which created a 12-bp deletion. To replace the deleted nucleotides with random sequence, a second linker (made from oligonucleotides 5'-NNNNNNGGAGACCGTCGACTTGTGAGCGGATAACAGGTCTCG-3' and 5'-NNNNNNCGAGACCTGTTATCCGCTCACAAGTCGACGGTCTCC-3' by annealing following blunting) was inserted: it contains 6 bp of random sequence at each end, along with embedded BsaI recognition sites. Libraries of independent linker insertions were constructed in E. coli, and the plasmid DNA was extracted and purified. The DNA was digested with BsaI again and religated, leaving an insertion of 12 random nucleotides.

Three independent libraries containing CMD1 random substitution mutations (i.e., altered sequences still capable of producing functional calmodulin) at positions 89–92 were used in this study. Plasmids containing these CMD1 mutations were recovered from yeast, amplified in E. coli, and transformed back into yeast to ensure that the phenotype was conferred by the plasmid rather than by any spontaneous genomic mutations. DNA sequencing of random replacement at positions 89–92 revealed that every plasmid had a different amino acid sequence.

PCR-based mutagenesis:
Introduction of mutations at position 92 was performed with a two-step PCR-based mutagenesis (HO et al. 1989 ), as described before (OHYA and BOTSTEIN 1994B ), using 16 oligonucleotides (Table 2). CMD1N+, corresponding to the translation start sequence of calmodulin, contained an artificial BstBI site without changing any amino acid sequence, and CMD1C-, corresponding to the 3'-noncoding sequence, contained the SphI site at the end. Long primers with Phe to Val, Ile, Leu, Trp, Met, Cys, and Tyr mutations were made in both sense and antisense strands.

In the first PCR reaction, both N- and C-terminal fragments were amplified separately and purified after electrophoresis on an agarose gel. These two fragments shared an overlapping region of at least 18 bp containing the mutations at position 92, so that in a second PCR reaction both fragments were mixed and the whole fragment was amplified with CMD1N+ and CMD1C- primers. The fragment containing the mutation was digested with BstBI and SphI, and was subcloned into the BstBI-SphI gap of pRB1616. All mutations were verified by DNA sequencing. DNA sequencing with two sequencing primers (5'-TGACCGGAAACTACTGAAC-3' and 5'-GATGAACGAAATAGATGTTGATGG-3') sufficed to cover the entire coding sequence of calmodulin.

Construction of yeast strains:
To integrate cmd1 mutations into the genome, we used the pRB1617-derived plasmids (using HIS3 as a selectable marker) or pRB1617L-derived plasmids (using LEU2 as a selectable marker). After digestion of these plasmids with SacII and AlwNI, the ade3-cmd1-HIS3-ade3 and ade3-cmd1-LEU2-ade3 fragments were used to transform YOC101 (MATa ade2 his3 leu2 lys2 trp1 ura3 cmd1-1::TRP1 [pRB1612]) or YOC102 (same as YOC101 except its mating type). Correct integrants were recognized as white transformants (ade2 ade3). Finally, strains that had lost pRB1612 were selected on FOA plates. The resulting calmodulin mutant strains are listed in Table 1. The cmd1 mutations used in this study are listed in Table 3.

To construct cmd1 myo2-66 double mutants carrying the wild-type calmodulin on a URA3 plasmid, crosses were made between YOC1105 (MAT ade2 his3 leu2 lys2 trp1 ura3 cmd1-1::TRP1 myo2-66 [pRB1612]) and representative calmodulin mutants (MATa ade2 his3 leu2 lys2 trp1 ura3 ade3::HIS3::cmd1-xxx cmd1-1::TRP1). The calmodulin strains used were YOC200, YOC226, YOC228, YOC231, YOC233, YOC239, YOC376, YOC377, YOC378, YOC379, YOC380, YOC381, and YOC382. After mating, the resulting diploids were sporulated. Segregants from the tetrads that simultaneously showed His⁺, Ura⁺, and Ts phenotypes were taken to be myo2-66 cmd1 double mutants harboring the wild-type calmodulin gene on a URA3 plasmid.

Purification and biotin labeling of calmodulins:
E. coli strains carrying the appropriate plasmids were incubated for 2 hr at 37° in TB medium containing 100 µg/ml ampicillin. Calmodulin production was induced by adding isopropyl l-ß-D-thiogalactopyranoside to a final concentration of 1 mM. After a 2-hr incubation, cells were collected by centrifugation, washed twice in 50 mM Tris-HCl, pH 7.5, containing 1 mM phenylmethylsulfonyl chloride and 1 mM MgCl₂, and freeze thawed. Cellular debris was removed by ultracentrifugation for 30 min at 200,000 x g. Calmodulin was purified essentially as described (OHYA et al. 1987 ; TAKAHASHI et al. 1996 ). The 2-ml supernatant was applied to the phenyl-sepharose column chromatography, and then the calmodulin was eluted with 50 mM Tris-HCl, pH 7.5, containing 5 mM EGTA and 0.1 M ammonium sulfate. Purified wild-type and mutant calmodulins were biotin labeled using ImmunoPure Sulfo-NHS-LC-Biotin (Pierce Chemical Co., Rockford, IL).

Calmodulin-Myo2-binding assay:
The procedures were based on those of BROCKERHOFF et al. 1994 .

Myo2 fusion protein was made using the GST Gene Fusion System (Pharmacia, Piscataway, NJ). The fragment of MYO2 that encodes residues 908–1086 of Myo2p was cloned into the pGEX plasmid. Myo2 fusion protein was expressed in an E. coli strain. Proteins (50 µg) were separated on a 10% SDS polyacrylamide gel and then electrophoretically transferred to a nitrocellulose membrane. The transfer buffer was 48 mM Tris and 39 mM glycine. After transfer, the membrane was blocked for 1 hr in blocking buffer (50 mM Tris-HCl, 150 mM NaCl, 0.05% Tween 20, 2% nonfat dry milk) and then probed at room temperature for 1.5 hr with 2.4 µM biotin-labeled wild-type or mutant calmodulin in Ca²⁺ buffer (10 mM Tris, 0.2% BSA, 1 mM CaCl₂) or EGTA buffer (10 mM Tris, 0.2% BSA, 10 mM EGTA). The membrane was then washed three times each for 10 min and air dried. Biotinylated horseradish peroxidase–avidin complex was formed using the Vectastain Elite standard kit (Vector Laboratories, Burlingame, CA) and stained using the POD immunostain set (Wako Chemicals).

Immunofluorescence microscopy:
Procedures for phalloidin staining and immunofluorescence microscopy were based on those of PRINGLE et al. 1989 . Actin staining was carried out using rhodamine-conjugated phalloidin, as a 3.3 µM stock in methanol from Molecular Probes (Eugene, OR). For staining of Myo2p, rabbit anti-Myo2p antibody (LILLIE and BROWN 1994 ) was used as a first antibody, biotinylated anti-rabbit antibody as a second antibody, and fluorophore-conjugated avidin, streptoavidin (Boehringer Mannheim, GmbH, Germany). For staining of Cmd1p, rabbit anti-Cmd1p antibody (OHYA et al. 1987 ) was used as a first antibody, and FITC-labeled anti-rabbit antibody was used as a second antibody. Stained cells were examined with an Olympus BX-FLA epifluorescence microscope (Olympus, Tokyo) and photographed on T-MAX 400 film (Eastman Kodak Co., Rochester, NY).

Isolation of cax mutants:
A colony-sectoring assay was used to screen mutations that create a requirement for CMD1 in cmd1-226 cells. The ade2 ade3 cmd1-226 strain carrying the wild-type calmodulin gene on a pYSLU1-CMD1 (YOC1119) was used as a parent strain. pYSLU1-CMD1 contains a mutation in the CEN3 region that makes the plasmid unstable. The principle of the colony-sectoring assay is that ade2 strains form red colonies, that ade2 ade3 strains form white colonies, while ade2 ade3 mutants with an ADE3 plasmid often lose the plasmid and produce abundant white sectors in an otherwise red colony. Therefore, after mutagenesis of YOC1119, one can get the mutants that require a calmodulin by screening for nonsectoring colonies. YOC1119 was mutagenized with EMS to ~70% survival (SHERMAN et al. 1974 ). The mutagenized cells were plated on YPD plates and incubated at 23° for 7 days. Red colonies were picked and rechecked for nonsectoring phenotype and for plasmid loss phenotype on plates containing 1 µg/ml FOA. The dependency of the strains on CMD1 was tested by transformation of a second copy of CMD1 on the LEU2/CEN plasmid (YCpL-CMD1) into the mutants. Only mutants whose sectoring phenotype and FOA sensitivity were restored by the second copy of CMD1 were analyzed further. Tetrad analysis was then carried out to pick up strains whose FOA sensitivity and nonsectoring phenotype were caused by a single mutation. Complementation tests were performed by mating mutant segregants and testing FOA sensitivity and sectoring phenotype on YPD plates.

Construction of cax mutants with an integrated CMD1 background:
The cax strains with an integrated CMD1 background were constructed by crossing the cax strains originally isolated by colony-sectoring assay (cax ade3::cmd1-226::HIS3 [pYSLU1-CMD1]) with the wild-type strain YOC1170 (MATa CAX ade3::CMD1::LEU2) or YOC1176 (same as YOC1170 except its mating type). The resulting diploids were sporulated. To isolate the cax CMD1::LEU2 strains, sensitivity to FOA, LEU, and HIS markers were checked. If two His⁺ segregants (with ade3::cmd1-226::HIS3) of a tetrad showed an FOA-insensitive phenotype, then the two other Leu⁺ segregants were assigned to harbor cax mutations. The possibility of gene conversion at the CAX locus was discarded by confirming that the phenotypes of the two possible cax segregants were the same.

Cell lysis assay:
Yeast cells were incubated on a YPD plate at 23° for 3 days. The plate was then overlaid with an alkaline phosphatase assay solution as described by PARAVICINI et al. 1992 . Colonies containing lysed cells turned blue within 1 hr, while control colonies remained unstained, even after 2 hr.

Cloning and genetic mapping of CAX1:
CAX1 was cloned by complementation of the FOA-sensitive growth phenotype of the corresponding mutant, using a YEp-based genomic library made by YOSHIHISA and ANRAKU 1989 . Plasmids that yielded FOA-insensitive transformants of the strain YOC1120 (cmd1-226::HIS3 cax1-1 [pYSLU1-CMD1]) were recovered. The plasmids were then checked to see whether they contained a CMD1 gene, both by restriction mapping and by PCR with primers that amplified the ORF of CMD1. Two plasmids that conferred FOA-insensitive phenotype in cax1 cells were recovered. One turned out to contain CMD1, and the other plasmid, pMS102, was studied further.

One end of the cax1-rescuing fragment was sequenced using the primer suited for direct sequencing of the DNA insert of YEp13. Comparison with the sequence database revealed that the insert was within a region of chromosome V. Subcloning analyses indicated that the 3-kb SalI-SpeI fragment has complementation activity on a single-copy plasmid and that NcoI is an essential site. The open reading frame defined by these tests is YEL036C, which encodes a previously characterized gene, ANP1 (also known as GEM3 and MCD2).

To show that the plasmid that confers the FOA-insensitive phenotype actually contains the CAX1 gene, integration mapping was carried out. The EcoRV-SpeI fragment located ~300 bp downstream from the ANP1 ORF was replaced with a LEU2 fragment from pJJ283 in YOC326 (MAT cmd1-226::HIS3 leu2). The integrant was mated to YOC1120. Tetrad analysis revealed that the LEU2 marker was tightly linked to the CAX1 locus (31 tetrads analyzed). Thus, we concluded that pMS102 contained the CAX1 gene itself.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The myo2-66 mutation shows synthetic lethal interaction with cmd1-226:
The yeast calmodulin mutant cmd1-226 has a characteristic functional defect in actin organization (OHYA and BOTSTEIN 1994A ). At the permissive temperature, the temperature-sensitive cmd1-226 cells grow slower than wild-type cells and frequently contain delocalized actin. At the restrictive temperature, >95% cells have lost localized actin cortical patches in the bud. In an attempt to genetically identify the essential calmodulin target that functions in the organization of the actin network, we looked for synthetic lethal interaction between the cmd1-226 mutation and a mutation in the MYO2 gene that encodes the only essential target of calmodulin known to bind actin. In the myo2-66 mutant, aberrant actin morphology had been observed, suggesting that Myo2p plays some role in actin organization (JOHNSTON et al. 1991 ).

By tetrad dissection, we constructed several cmd1 myo2-66 double-mutant strains harboring the wild-type CMD1 on a URA3-marked plasmid. The calmodulin alleles used were cmd1-226 (F92A) with a defect in actin organization, cmd1-228 (F12A F16A F19A) with a defect in calmodulin localization at sites of cell surface growth, cmd1-231 (F12A F89A) with a defect in bud emergence, cmd1-233 (F12A F140A) with a defect in bud emergence, and cmd1-239 (F65A F68A) with a defect in nuclear division. The double-mutant strains harboring the URA3-CMD1 plasmid were then tested for sensitivity to FOA. Only strains that can lose the URA3-CMD1 plasmid are able to grow on FOA medium.

We found that myo2-66 is synthetically lethal with cmd1-226. The cmd1-226 myo2-66 double mutant failed to grow at 23°, although each of the single mutants grew at this temperature (Figure 1B). The strains with either a cmd1-228 myo2-66, cmd1-231 myo2-66, or cmd1-239 myo2-66 mutation grew well on FOA medium at 23° (Figure 1C, Figure 1D, and Figure 1F). Instead, the cmd1-233 myo2-66 double mutant grew slowly, but eventually formed colonies (Figure 1E). We randomly picked 10–15 colonies of cmd1-233 myo2-66 from FOA medium and confirmed that they all had lost the URA3-CMD1 plasmid, indicating that the cmd1-233 myo2-66 double mutant is viable. We examined the doubling time and restrictive temperature of the cmd1 myo2-66 double mutants (Table 4). Among the viable double mutants, cmd1-233 myo2-66 exhibited the slowest growth rate. However, the restrictive temperature of the cmd1-233 myo2-66 mutant was the same as that of myo2-66. The other double mutants showed no obvious synthetic effect on doubling time or on restrictive temperature. Thus, a synthetic growth defect of cmd1-233 myo2-66 was observed, but it was not as strong as that of cmd1-226 myo2-66.

View larger version (50K): In this window In a new window Download PPT slide   Figure 1. cmd1-226 shows synthetic lethal interaction with myo2-66. Cells of cmd1 myo2-66 double mutants carrying the wild-type calmodulin gene on a URA3 plasmid were fully grown in YPD from single colonies at 23°. The cell suspensions were diluted with 1 µg/ml FOA medium and incubated at 23° for 2 days to allow Ura- cells to grow. The equal volumes of the liquid culture were streaked on an FOA-containing plate (A–F) or on a synthetic complete plate (G–L). The following strains were tested: (A and G) YOC1106 (CMD1 myo2-66 [YCpU-CMD1]), (B and H) YOC1107 (cmd1-226 myo2-66 [YCpU-CMD1]), (C and I) YOC1108 (cmd1-228 myo2-66 [YCpU-CMD1]), (D and J) YOC1109 (cmd1-231 myo2-66 [YCpU-CMD1]), (E and K) YOC1110 (cmd1-233 myo2-66 [YCpU-CMD1]), (F and L) YOC1111 (cmd1-239 myo2-66 [YCpU-CMD1]).

Cmd1-226p has decreased binding activity to Myo2p:
To examine directly the possible impairment of the ability of the mutant calmodulin to bind Myo2p, we analyzed the binding of a Myo2p fusion protein with wild-type and several mutant calmodulins by a gel overlay assay. The wild-type and mutant calmodulins used in this experiment exhibited a single band after running on an SDS-PAGE gel (Figure 2A). A DNA segment of MYO2 containing the coding sequence of the sixth IQ motif was fused to the glutathione S-transferase gene to produce a Myo2p fusion in E. coli (Figure 2B, lane 2). Figure 2B compares the binding ability of each mutant calmodulin with that of the wild-type calmodulin in the presence of Ca²⁺. Binding of Cmd1-226p to the Myo2p fusion was severely decreased. We detected a definite binding signal of Cmd1-228p, Cmd1-231p, Cmd1-233p, and Cmd1-239p to Myo2p fusion protein, although the binding of some mutant calmodulins may be slightly weaker compared with that of a wild-type calmodulin. When we performed the same experiment in the absence of Ca²⁺ (using 10 mM EGTA buffer), no binding was observed for each calmodulin (data not shown). It was reported that binding of calmodulin to the last four IQ motifs is Ca²⁺ independent, while binding to the first two IQ motifs was inhibited by Ca²⁺ (BROCKERHOFF et al. 1994 ). We used the last (the sixth) IQ motif in this study, and Ca²⁺ seems to have different effects on the interactions of the different last four IQ motifs with calmodulin.

View larger version (42K): In this window In a new window Download PPT slide   Figure 2. Purification of calmodulins and Cmd1p-Myo2p-binding assay. (A) Purification of wild-type and mutant calmodulins. Purified wild-type and various mutant calmodulins were subjected to SDS-PAGE. Lane 1, Cmd1-226p; lane 2, Cmd1-228p; lane 3, Cmd1-231p; lane 4, Cmd1-232p; lane 5, Cmd1-233p; lane 6, Cmd1-234p; lane 7, Cmd1-239p; lane 8, Cmd1-240p; lane 9, Cmd1-250p; lane 10, wild-type calmodulin; lane 11, markers (mol. wt. = 200, 97.4, 69, 46, 30, 21.5, and 14.3 kD). (B) Binding assay of Myo2p and each mutant calmodulin. Lane 1, markers (mol. wt. = 97.4, 69, 46, 30, and 21.5 kD). Myo2-GST fusion protein was expressed and subjected to SDS-PAGE gel (lane 2). The fusion was blotted with wild-type calmodulin (lane 3), Cmd1-228p (lane 4), Cmd1-239p (lane 5), Cmd1-231p (lane 6), Cmd1-226p (lane 7), or Cmd1-233p (lane 8). An arrowhead indicates the band of the Myo2-GST fusion protein. Binding results are expressed as the density of bands: +, binding; -, decreased binding. The experiment was performed more than twice, and we obtained the same results.

The result indicated that the cmd1-226 mutation results in a severe defect in binding to the Myo2p fusion protein. Because the cmd1-226 mutation contains only the phenylalanine-to-alanine change at position 92, Phe92 is likely to be important for the proper interaction with the IQ motif of Myo2p. Together with the genetic result of synthetic lethality between cmd1-226 and myo2-66, we suggest that decreased binding of Cmd1-226p to Myo2p, in combination with the weak interaction between Myo2p and actin, results in lethality. Because Cmd1-233p showed binding to Myo2p, our interpretation is that cmd1-233 does not mainly affect Myo2p function (see DISCUSSION).

The cmd1-226 mutation does not affect Myo2p localization, but affects Cmd1p localization:
We immunolocalized Myo2p and Cmd1p in the cmd1-226 cells. As reported previously (BROCKERHOFF et al. 1994 ; LILLIE and BROWN 1994 ), a Myo2p cap was seen in small and unbudded cells of the wild-type strain (Figure 3A). We found that apparent Myo2p staining remained at the bud tip in >90% of the cmd1-226 cells, and that the staining was seen clearly in cells that had lost actin spot polarization (Figure 3C and Figure 3D).

View larger version (106K): In this window In a new window Download PPT slide   Figure 3. Myo2p and Cmd1p localization in wild-type and cmd1-226 cells. Cells were shifted from 27° to 33° for 170 min. The cells were then double labeled with anti-Myo2 antibody (A and C) and rhodamine phalloidin (B and D), or cells were labeled with anticalmodulin antibody (E and F). Strains: (A, B, and E) YOC200 (wild type) and (C, D, and F) YOC226 (cmd1-226).

In the cmd1-226 cells, polarized calmodulin localization was lost (Figure 3F). The loss of calmodulin localization is a property shared between cmd1-226 (cmd1A) and cmd1-228 (cmd1B). Under the same conditions, however, cmd1-226 has delocalized actin organization, while cmd1-228 has wild-type actin organization (OHYA and BOTSTEIN 1994A ). Because cmd1A and cmd1B complement each other, it seems likely that the primary defects of these mutants are different. Based on the fact that localization of calmodulin is dependent on actin (BROCKERHOFF and DAVIS 1992 ), our current hypothesis is that the loss of calmodulin localization in cmd1A cells is caused by the indirect effect of delocalized actin organization.

Random replacement mutagenesis of calmodulin at positions 89–92:
The phenylalanine-to-alanine alteration at position 92 impairs the essential function of calmodulin. To analyze the requirement of amino acid residues at position 92 and to compare requirements at different phenylalanine residues, we analyzed >50 calmodulin mutations replaced with random sequences at positions 89–92 (see MATERIALS AND METHODS). This region contains two phenylalanine residues (Phe89 and Phe92). Table 5 shows the properties of the replacement mutants, including growth phenotypes and the predicted amino acid residues. We found a strong preference at position 92. Phenylalanine, valine, isoleucine, and leucine at position 92 were recovered in the sequence, resulting in robust growth at any temperature. Alanine and tryptophan were recovered only in the temperature-sensitive mutations. No charged amino acid was allowed (the probability that this occurs by chance is P < 7 x 10^-4). Amino acids with neither the hydroxyl nor amide group were allowed (the probability that this occurs by chance is P < 7 x 10^-4). These results suggested that hydrophobic and aromatic amino acid residues are the only ones consistent with function at position 92. In contrast, we found no strong preference at position 89. Charged and polar amino acids were allowed at this position, although they appeared only in "Ts" and "partial Ts" mutants (Table 5).

Site-directed mutagenesis at the position Phe92:
Random replacement mutagenesis uses a functional selection approach suited for determining which are the most critical amino acid residues in the randomizing region. This technique, however, does not always give a perfect solution for the amino acid requirement, partly because only a statistical interpretation is available for the amino acids that are not recovered in the complementing sequence, and partly because the individual replacements usually contain additional amino acid changes. To obtain more accurate information for the requirement at position 92, we used site-directed mutagenesis at this position.

We constructed seven replacement mutations at position 92, resulting in an alteration to valine, isoleucine, leucine, tryptophan, methionine, cysteine, or tyrosine, based on the results of random replacement mutagenesis. Valine, isoleucine, and leucine were expected to be "good" residues at position 92. Tryptophan was expected to be an "allowed but temperature-sensitive" residue. Not even statistical data were available for methionine, cysteine, and tyrosine. All the mutants were constructed by insertion of the calmodulin mutations into the genome after verifying the mutations. We found that all of the seven mutants constructed grew at 38°, as well as at 25° (Table 6). The phenotype of the strain with the tryptophan replacement mutation (cmd1-379) was particularly unexpected because all the random replacement mutants with Trp at position 92 exhibited a temperature-sensitive phenotype. On the basis of these results, we assume that additional mutations at positions 89–91 in cmd1-307, cmd1-308, cmd1-309, cmd1-310, cmd1-325, and cmd1-326 (Table 5) were involved in the temperature-sensitive phenotypes observed for these random replacement mutants.

View this table: In this window In a new window   Table 6. Properties of Phe92X cmd1 mutations

The actin morphology of the site-directed mutants was examined for comparison with that of cmd1-226 (F92A). The mutant cells were shifted from 27° to 33°, incubated for 170 min, and then stained with rhodamine-phalloidin. We found that none of the mutants, other than cmd1-226 (F92A) cells, have defects in actin organization. While <10% small-budded cmd1-226 cells contained a polarized actin network (Table 6; Figure 4B), >85% small-budded cells of other site-directed mutants had polarized actin organization (Table 6; Figure 4, C–E).

View larger version (92K): In this window In a new window Download PPT slide   Figure 4. Actin morphologies of the Phe92X calmodulin replacement mutants. Cells of the replacement mutants were shifted from 27° to 33° for 170 min and then stained with rhodamine phalloidin. Strains: (A) YOC200 (wild type), (B) YOC226 (cmd1-226: F92A), (C) YOC376 (cmd1-376: F92V), (D) YOC379 (cmd1-379: F92W), and (E) YOC380 (cmd1-380: F92M).

Of eight replacement mutations, cmd1-226 (F92A) and cmd1-380 (F92M) mutations showed synthetic lethal interaction with myo2-66 (Table 6). However, the F92M calmodulin can bind to the Myo2p fusion and to wild-type calmodulin, based on the gel overlay assay (data not shown).

Isolation of mutations that show synthetic lethal interaction with cmd1-226:
We sought to identify additional components involved in the regulation of actin organization by calmodulin by screening for synthetic lethal interactions with cmd1-226, as found for myo2-66. As described in MATERIALS AND METHODS, we began with a cmd1-226 also carrying an intact plasmid-borne CMD1 gene, and we isolated eight recessive mutations that display recessive lethality with cmd1-226. We called this phenotype Cax (calmodulin-dependent in cmd one two twenty-six). The cax mutants harboring the wild-type CMD1 gene on a URA3-marked plasmid failed to grow on FOA plates (Figure 5A); using this phenotype, we carried out complementation analysis, which revealed that the cax mutations could be divided into five complementation groups. All the mutants became able to grow on FOA plates after transformation with the second plasmid carrying the wild-type CMD1 gene (Figure 5B).

View larger version (61K): In this window In a new window Download PPT slide   Figure 5. Growth of cax mutants on FOA plates. (A) cax mutants fail to lose the wild-type CMD1 gene on a URA3 plasmid. cax mutant cells were streaked on an FOA plate and incubated at 23° for 5 days. Growth on a permissive medium (SD-Ura) was tested as a control. The strains used are as follows: YOC1119 (CAX), YOC1120 (cax1-1), YOC1126 (cax2-1), YOC1130 (cax3-1), YOC1132 (cax4-1), and YOC1134 (cax5-1). (B) Growth of cax mutants does not depend on URA3. cax cells were transformed with YCpL-CMD1, streaked on an FOA-Leu plate, and incubated for 5 days. The cax strains used are the same as described in A.

Allelism of each cax group was tested with myo2-66 because myo2-66 also showed synthetic lethal interaction with cmd1-226. Cax strains were transformed with a single-copy plasmid pYO1148 containing MYO2. The plasmid failed to complement the FOA sensitivity of the cax mutations, indicating that none of the recessive cax mutations are allelic to myo2. We also confirmed that cax mutations recombine freely with the CMD1 locus (data not shown).

Growth characteristics and morphological properties of the cax mutants:
To further analyze the cax mutations, we constructed cax strains harboring the wild-type CMD1 gene integrated in the chromosome in place of cmd1-226. We found that cax2-2 and cax3-1 grew poorly at 37° (Table 7). The same mutants simultaneously exhibited a calcium-sensitive phenotype. cax4-1 hardly grew on plates containing 1 µg/ml FK506, a drug that inhibits the activity of calmodulin-dependent phosphoprotein phosphatase, calcineurin (KUNZ and HALL 1993 ). Because cax2-2, cax3-1, and cax5-1 cells looked fragile when examined by phase-contrast microscopy, we assayed cell lysis in these cells with a simple plate overlay assay. We found that alkaline phosphatase was easily leaked from cells of cax2-2, cax3-1, and cax5-1 at 23° (Table 7). Among the lytic mutants, the temperature-sensitive phenotype of cax2-2 and cax3-1 was suppressed by addition of 1 M sorbitol (data not shown). These results suggest that Cax2p, Cax3p, and Cax5p are involved in the maintenance of osmotic integrity. Several cells of cax2, cax3, and cax5 appeared larger than wild-type cells in the presence (Figure 6D, Figure 6E, and Figure 6G) or absence of 1 M sorbitol (data not shown).

View larger version (12K): In this window In a new window Download PPT slide   Figure 6. All cax strains have defects in actin organization at 25°. Log-phase cells were fixed and stained with rhodamine phalloidin. Wild-type, myo2-66, cax1-1, and cax4-1 cells were cultured in YPD. cax2-2, cax3-1, and cax5-1 cells were cultured in YPD + 1 M sorbitol. The strains used are as follows: (A) YOC1170 (wild type), (B) YOC1102 (myo2-66), (C) YOC1139 (cax1-1), (D) YOC1140 (cax2-2), (E) YOC1141 (cax3-1), (F) YOC1142 (cax4-1), and (G) YOC1143 (cax5-1). Bar, 10 µm.

We examined the morphology of the actin cytoskeleton in wild-type, myo2, and cax cells. To avoid the secondary effect caused by cell lysis, we used medium containing osmotic stabilizer (1 M sorbitol) for cax2, cax3, and cax5 cells. In a small-budded, wild-type cell, actin patches were mostly concentrated at the bud (only one or two actin patches were seen in most of the mother cells, Figure 7A), and actin cables were aligned along the axis of the formation of the bud (Figure 6A). Decreased polarity of actin patches and faint actin cables were observed in myo2 and all cax mutants. The percentage of cax and myo2 cells with three or more actin patches in mother cells is 3–4.5-fold larger than in wild-type cells (Figure 7A). The percentage of budded cells that lost actin spot polarization was much larger in myo2 mutants, and slightly larger in cax1, cax2, cax3, cax4, and cax5 mutants than in a wild-type strain (Figure 7B).

View larger version (12K): In this window In a new window Download PPT slide   Figure 7. Morphological properties of cax cells. More than 150 small-budded cells were scored for each mutant at 25°. The strains and culture conditions used were the same as described in Figure 6. (A) Percentage of small-budded cells that have three or more actin patches in mother cells. (B) Percentage of budded cells that have lost actin spot polarization.

Synthetic lethal interaction between cax and cmd1 mutations:
The cax mutations showed synthetic lethal interaction with cmd1-226. To test for allele specificity, we examined synthetic lethality with other calmodulin mutations. For this experiment, we used cmd1-228 (defective in calmodulin localization at the bud tip), cmd1-233 (defective in bud emergence), and cmd1-239 (defective in nuclear division). We revealed (Table 8) that cax1-1 exhibited synthetic lethal interaction only with cmd1A (cmd1-226). cax4-1 showed synthetic lethal interaction with cmd1A (cmd1-226) and cmd1B (cmd1-228). cax5-1 showed synthetic lethal interaction with cmd1A and cmd1D (cmd1-233). cax2-2 and cax3-1 showed synthetic lethal interaction with cmd1A, cmd1B, and cmd1D. None of the cax mutations showed synthetic lethal interaction with cmd1C (cmd1-239), which affects nuclear division. These results suggest that Cax1p plays some crucial roles in the function of Myo2p. Because cmd1A, cmd1B, and cmd1D seem to be defective in cell polarity, Cax2p, Cax3p, Cax4p, and Cax5p may have overlapped or diverse functions in the establishment or maintenance of cell polarity.

CAX1 was identical to ANP1/GEM3/MCD2:
The CAX1 was cloned, which complemented the FOA sensitivity of the cax1 mutant. Deletion analysis and subcloning were used to identify the minimum complementing region. Linkage of the genomic copy of the cloned DNA to CAX1 was demonstrated (see MATERIALS AND METHODS).

The CAX1 gene was found to be identical to ANP1/GEM3/MCD2 (amino nitrophenol propandiol resistance gene/Golgi enzyme maintenance gene/multinucleated cells with morphogenetic defects). Deletion of the gene confers sensitivity to amino nitrophenol propandiol (MCKNIGHT et al. 1981 ), improper Golgi function, defects in secretion, altered protein glycosylation (CHAPMAN and MUNRO 1994 ; SIPOS et al. 1995 ), and resistance to sodium orthovanadate (KANIK-ENNULAT et al. 1995 ). MONDESERT et al. 1997 isolated ANP1 in a screen for mutants defective specifically in polarized growth, but with no drastic impairment in actin cytoskeleton. In contrast, we did detect obvious impairment in actin spot polarization in cax1 (Figure 6C and Figure 7), although this impairment is not as strong as that observed in mutants of actin-binding proteins. The other phenotypes of anp1 deletion cells include killer toxin resistance and flocculation, suggestive of defective cell walls (CHAPMAN and MUNRO 1994 ). Our cax1-1 mutant exhibited consistent resistance to sodium orthovanadate and flocculation (data not shown).

To examine whether deletion of ANP1/CAX1 results in synthetic lethal interaction with cmd1-226, we created a strain harboring a deletion mutation of CAX1, cmd1-226, and the wild-type CMD1 gene on a URA3-marked plasmid. We found that the strain was unable to grow on FOA medium, indicating that the CAX1 deletion combined with cmd1-226 results in inviability. Next, synthetic lethal interaction of the deletion was examined with myo2-66. We created a strain harboring a CAX1 deletion, myo2-66, and the wild-type MYO2 gene on a URA3-marked plasmid and found that the strain was unable to grow on FOA medium. Furthermore, we confirmed by tetrad analysis that most of the cax1 myo2-66 cells were inviable. We conclude that deletion of ANP1/CAX1 shows a synthetic growth defect with myo2-66.

We found that CAX4 encodes a novel protein with no significant homology to any other known proteins (the ORF name is YGR036C). Deletion of CAX4 showed FK506 sensitivity and defects in actin organization, which are phenotypes similar to the cax4-1 strain (M. SEKIYA and Y. OHYA, unpublished results). The CAX5 gene was found to be identical to MNN10/SLC2/BED1 (mannan defective/synthetic lethality with cap2/bud emergence delay). The gene shows significant homology to galactosyl transferase. The slc2 mutant has defects in actin cytoskeleton (KARPOVA et al. 1993 ). cax5-1, as well as mnn10 (BALLOU et al. 1991 ), showed resistance to orthovanadate (data not shown).

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Synthetic lethality with the complementing calmodulin mutations:
Synthetic lethal interaction is often observed between mutations in genes that functionally interact. Based on the assumption that each intragenic complementation group of calmodulin mutations results in a single functional defect, allele-specific genetic interaction with the complementing calmodulin mutations seemed likely to serve as a good tool for the dissection of the individual cellular processes regulated by calmodulin.

We found here that among the complementing mutations, only cmd1-226 (F92A) exhibits synthetic lethal interaction with a myo2 mutation. A plausible explanation for this result is that among the mutant calmodulins, only Cmd1-226p loses binding activity to Myo2p. Biochemical analysis supported this model, as we found that Cmd1-226p is severely defective in binding to Myo2p, compared to the other calmodulins. The overall structure of Cmd1-226p seems not to be altered by biotinylation because Cmd1-226p was able to bind to another CaM-binding protein, Cna2p (Ca²⁺/calmodulin-dependent protein phosphatase), as strongly as wild-type calmodulin (H. OKANO and Y. OHYA, unpublished results).

The cmd1-233 (cmd1D) mutation does not enhance the temperature-sensitive phenotype of myo2-66, but cmd1-233 and myo2-66 show a synthetic growth defect at 23° (Table 4). Examination by gel overlay assay indicates that Cmd1-233p physically interacts with Myo2p (Figure 2B). Our explanation is that an unidentified calmodulin target(s) uncoupled by cmd1-233 exacerbates the myo2 phenotype. Another possibility is that Cmd1-233p, as yet less severely than Cmd1-226p, causes defects of the Myo2p function. Because we found no synthetic growth defect between myo2-66 and cmd1-231 that belongs to the same complementation group of cmd1-233(cmd1D), we think that another essential target uncoupled by cmd1D is involved in bud emergence.

Amino acid requirement at position Phe92 of calmodulin:
In this study, we demonstrate that Phe92 of yeast calmodulin is essential for the Myo2p–Cmd1p interaction. As reported previously, the target recognition by calmodulin is mainly stabilized by hydrophobic interaction (IKURA et al. 1992 ). Phe92, which is perfectly conserved among eukaryote calmodulins, is one of the residues involved in the interaction. Although the amino acid sequence of yeast calmodulin is only 60% identical to the sequences of other eukaryote calmodulins, functional importance of Phe92 in target recognition seems to be well conserved during evolution. Only hydrophobic and aromatic residues are consistently allowed at Phe92 (Table 5).

Among the several replacement mutants at position 92, only the phenylalanine-to-alanine change results in a temperature-sensitive phenotype. This was an unexpected result, because all tryptophan replacements constructed by random replacement mutagenesis showed a temperature-sensitive phenotype. We assume that additional mutations at positions 89–91 affect the temperature-sensitive phenotypes. This result illustrates, in a striking way, that random replacement methods and single-residue replacements do not give the same information, just as was found previously in the case of bacterial ß-lactamase (PALZKILL and BOTSTEIN 1992A , PALZKILL and BOTSTEIN 1992B ).

Synthetic lethal interaction between cmd1-380 (F92M) and myo2-66 suggests that the F92M substitution of calmodulin alters the Myo2p-Cmd1p interaction. Because the actin morphology of cmd1-380 cells was indistinguishable from that of wild-type cells (Figure 4E), it is likely that the cmd1-380 mutation causes a slight defect of the Cmd1p-Myo2p interaction. In the gel overlay assay experiment, however, we observed no obvious impairment in the binding of Cmd1-380p to the Myo2p fusion. Because the gel overlay assay was performed under the condition where the concentration of calmodulin molecule is much higher than in vivo, one possibility is that F92M calmodulin in vivo might have a defect in binding to Myo2p. Another possibility is that the binding of F92M calmodulin to different IQ motifs might be decreased. Methionine contains a relatively long hydrophobic side chain compared to other nonaromatic hydrophobic residues. Introduction of methionine at position 92 possibly alters the conformation of the hydrophobic surface of calmodulin that is involved in target recognition.

Function of Myo2p:
Myo2p has recently been identified as a member of the class V myosins (JOHNSTON et al. 1991 ). Other members include p190 from vertebrate brain (ESPINDOLA et al. 1992 ), dilute from mouse (MERCER et al. 1991 ), and Myo4p from yeast (HAARER et al. 1994 ). Several observations implicated myosin V in polarized vesicle transport. dilute is involved in polarized transport of pigment-containing melanosomes (MERCER et al. 1991 ). dilute is also involved in the transport of smooth endoplasmic reticulum membranes in neuronal dendrites (TAKAGISHI et al. 1996 ).

In yeast, the myo2-66 mutant arrests as an unbudded cell with a large number of accumulated intracellular vesicles (JOHNSTON et al. 1991 ; GOVINDAN et al. 1995 ). Because actin is thought to be required for polarized secretion in yeast (NOVICK and BOTSTEIN 1985 ), it is possible that Myo2p serves as an actin-based vesicle motor. However, nobody has yet observed accumulation of secretory product within myo2-66 cells.

Here we present both genetic and biochemical evidence indicating that the cmd1-226 mutation results in loss of binding with Myo2p. This is consistent with the observation that myo2-66 shows defects in actin organization, even at the permissive temperature (JOHNSTON et al. 1991 ). Because Myo2p is localized correctly in a cmd1-226 mutant (Figure 3C), binding of calmodulin to Myo2p is not necessary for intracellular localization of Myo2p, but is probably necessary for the Myo2p activity in organizing the actin cytoskeleton.

Several possibilities can be envisioned for the Myo2p function in actin organization. Myo2p itself may be involved in localizing or moving the actin cytoskeleton toward the growing tip. Alternatively, cargoes in secretory vesicles driven by Myo2p may anchor or stabilize the actin cytoskeleton. Myo2p might be capable of cross-linking actin because of the predicted and demonstrated ability of class V myosins to dimerize (CHENEY et al. 1993 ). There is growing evidence suggesting that myosins regulate the actin network in yeast and other organisms. For example, loss of myosin I function in yeast results in defective actin organization (GELI and RIEZMAN 1996 ; GOODSON et al. 1996 ). It was suggested that 95F myosin (a class VI unconventional myosin) in Drosophila may be involved in the formation of actin furrows through the transport of cytoplasmic components (MERMALL and MILLER 1995 ).

Isolation and characterization of cax mutations:
To search for other components that function together with calmodulin and Myo2p, we used a synthetic lethal screen beginning with cmd1-226. One property common to myo2-66 and all the cax mutants is altered morphology of actin cytoskeleton, suggesting that all the CAX genes somehow function in the regulation of actin organization.

From the similarity of the phenotype between cax1-1 and myo2-66 strains, we think that Anp1p/Cax1p is likely to play a crucial role in the regulation of the actin network through calmodulin and Myo2p. Because deletion of ANP1/CAX1 shows a synthetic growth defect with myo2-66, Anp1p/Cax1p may share common functions with Myo2p. Anp1p is known to be involved in protein glycosylation. Given that Myo2p is involved in polarized secretion and that protein glycosylation occurs in the secretory process, one possibility is that the performance of Anp1p function depends on the secretory process regulated by Myo2p. We have found evidence that many late secretory mutations cause, as part of their secretion phenotypes, disorganization of the actin cytoskeleton (MULHOLLAND et al. 1997 ).

Another possibility is that Anp1p might regulate the components of the actin cytoskeleton simply because proper glycosylation of proteins might be required for actin organization. We are attracted to this idea because (1) invertase was not accumulated in the cmd1-226 (data not shown) and myo2-66 cells (GOVINDAN et al. 1995 ), and (2) cmd1-226, myo2-66, and anp1/cax1 share a common feature of defects in actin organization.

In conclusion, examination of genetic interaction with complementing calmodulin mutations appears to be an effective approach to identify the target that is severely uncoupled by cmd1A. By using this approach, we found evidence suggesting that novel components, including Myo2p and Anp1p, are involved in the process of actin organization regulated by calmodulin.

	ACKNOWLEDGMENTS

We wish to thank G. C. JOHNSTON for providing us with the myo2-66 strain. We also thank S. S. BROWN for providing us with the anti-Myo2 antibody. This work was supported by a grant for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan (Y.O.), Takeda Science Foundation (Y.O.), Kowa Life Science Foundation (Y.O.), National Institutes of Health grant GM-46406 (D.B.), and a grant from the Japan Society for the Promotion of Science for Young Scientists (M.S.-K.).

Manuscript received February 3, 1998; Accepted for publication May 27, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

BABU, Y. S., C. E. BUGG, and W. J. COOK, 1988  Structure of calmodulin refined at 2.2 Å resolution. J. Mol. Biol. 204:191-204[Medline].

BALLOU, L., R. A. HITZEMAN, M. S. LEWIS, and C. E. BALLOU, 1991  Vanadate-resistant yeast mutants are defective in protein glycosylation. Proc. Natl. Acad. Sci. USA 88:3209-3212[Abstract/Free Full Text].

BROCKERHOFF, S. E. and T. N. DAVIS, 1992  Calmodulin concentrates at regions of cell growth in Saccharomyces cerevisiae. J. Cell Biol. 118:619-629[Abstract/Free Full Text].

BROCKERHOFF, S. E., R. C. STEVENS, and T. N. DAVIS, 1994  The unconventional myosin, Myo2p, is a calmodulin target at sites of cell growth in Saccharomyces cerevisiae. J. Cell Biol. 124:315-323[Abstract/Free Full Text].

CHAPMAN, R. E. and S. MUNRO, 1994  The functioning of the yeast Golgi apparatus requires an ER protein encoded by ANP1, a member of a new family of genes affecting the secretory pathway. EMBO J. 13:4896-4907[Medline].

CHENEY, R. E., M. K. O'SHEA, J. E. HEUSER, M. V. COELHO, and J. S. WOLENSKI et al., 1993  Brain myosin-V is a two-headed unconventional myosin with motor activity. Cell 75:13-23[Medline].

COHEN, P., and C. B. KLEE (Editors), 1988 Calmodulin. Molecular Aspects of Cellular Regulation. Elsevier, New York.

CONNEL, C., S. FUNG, C. HEINER, J. BRIDGHAM, and V. CHEKERRIAN et al., 1987  Automated DNA sequence analysis. Bio Techniques. 5:342-348.

DAVIS, T. N., 1992  What's new with calcium? Cell 71:557-564[Medline].

DAVIS, T. N. and J. THORNER, 1989  Vertebrate and yeast calmodulin, despite significant sequence divergence, are functionally interchangeable. Proc. Natl. Acad. Sci. USA 86:7909-7913[Abstract/Free Full Text].

DAVIS, T. N., M. S. URDEA, F. R. MASIARZ, and J. THORNER, 1986  Isolation of the yeast calmodulin gene: calmodulin is an essential protein. Cell 47:423-431[Medline].

ESPINDOLA, F. S., E. M. ESPREAFICO, M. V. COELHO, A. R. MARTINS, and F. R. C. COSTA et al., 1992  Biochemical and immunological characterization of p190-calmodulin complex from vertebrate brain: a novel calmodulin-binding myosin. J. Cell Biol. 118:359-368[Abstract/Free Full Text].

GELI, M. I. and H. RIEZMAN, 1996  Role of type I myosins in receptor-mediated endocytosis in yeast. Science 272:533-535[Abstract].

GELI, M. I., A. WESP, and H. RIEZMAN, 1998  Distinct functions of calmodulin are required for the uptake step of receptor-mediated endocytosis in yeast: the type I myosin Myo5p is one of the calmodulin targets. EMBO J. 17:635-647[Medline].

GEISER, J. R., H. A. SUNDBERG, B. H. CHANG, E. G. D. MULLER, and T. N. DAVIS, 1993  The essential mitotic target of calmodulin is the 110-kilodalton component of the spindle pole body in Saccharomyces cerevisiae. Mol. Cell. Biol. 13:7913-7924[Abstract/Free Full Text].

GOODSON, H. V., B. L. ANDERSON, H. M. WARRICK, L. A. PON, and J. A. SPUDICH, 1996  Synthetic lethality screen identifies a novel yeast myosin I gene (MYO5): myosin I proteins are required for polarization of the actin cytoskeleton. J. Cell Biol. 133:1277-1291[Abstract/Free Full Text].

GOVINDAN, B., R. BOWSER, and P. NOVICK, 1995  The role of Myo2, a yeast class V myosin, in vesicular transport. J. Cell Biol. 128:1055-1068[Abstract/Free Full Text].

HAARER, B. K., A. PETZOLD, S. H. LILLIE, and S. S. BROWN, 1994  Identification of MYO4, a second class V myosin gene in yeast. J. Cell Sci. 107:1055-1064[Abstract].

HO, S. N., H. D. HUNT, R. M. HORTON, J. K. PULLEN, and L. R. PEASE, 1989  Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene 77:51-59[Medline].

IKURA, M., S. SPERA, G. BARBATO, L. E. KAY, and M. KRINKS et al., 1991  Secondary structure and side-chain 1H and 13C resonance assignments of calmodulin in solution by heteronuclear multidimensional NMR spectroscopy. Biochemistry 30:9216-9228