Regulation of Yeast Glycogen Metabolism and Sporulation by Glc7p Protein Phosphatase
Nadja T. Ramaswamy, Li Li, Miriam Khalil, John F. Cannon

Abstract

Glc7p is an essential serine/threonine type 1 protein phosphatase (PP1) from the yeast Saccharomyces cerevisiae, which has a role in many processes including cell cycle progression, sporulation, glycogen accumulation, translation initiation, and glucose repression. Two hallmarks of PP1 enzymes are very high amino acid sequence conservation and association of the catalytic subunit with a variety of noncatalytic, regulatory subunits. We tested the hypothesis that PP1 sequence conservation was the result of each PP1 residue playing a role in multiple intermolecular interactions. Analysis of 24 glc7 mutants, isolated primarily by their glycogen accumulation traits, revealed that every mutated Glc7p residue altered many noncatalytic subunit affinities and conferred unselected sporulation traits to various degrees. Furthermore, quantitative analysis showed that Glc7p affinity for the glycogen-binding noncatalytic subunit Gac1p was not the only parameter that determines the glycogen accumulation by a glc7 mutant. Sds22p is one Glc7p noncatalytic subunit that is essential for mitotic growth. Surprisingly, several mutant Glc7p proteins had undetectable affinity for Sds22p, yet grew apparently normally. The characterization of glc7 diploid sporulation revealed that Glc7p has at least two meiotic roles. Premeiotic DNA synthesis was undetectable in glc7 mutants with the poorest sporulation. In the glc7 diploids examined, expression of the meiotic inducer IME1 was proportional to the glc7 diploid sporulation frequency. Moreover, IME1 hyperexpression could not suppress glc7 sporulation traits. The Glc7p/Gip1p holoenzyme may participate in completion of meiotic divisions or spore packaging because meiotic dyads predominate when some glc7 diploids sporulate.

REVERSIBLE protein phosphorylation controls a variety of cellular activities by altering the function of substrate proteins. Studies of a variety of organisms indicate that type 1 serine/threonine protein phosphatases (PP1) regulate many processes (Cohen 1989; Shenolikar and Nairn 1991). The catalytic subunit of these enzymes interacts with a collection of noncatalytic subunits that modulate subcellular localization, substrate specificity, and enzyme activity (Hubbard and Cohen 1993). Saccharomyces cerevisiae has a single gene that encodes PP1, GLC7. Amino acid sequences of PP1 proteins are highly conserved (Bartonet al. 1994). The Glc7p sequence is 81% identical to that of rabbit PP1, and the structure of the latter has recently been solved by X-ray crystallography (Egloffet al. 1995; Goldberget al. 1995). In S. cerevisiae, there are at least nine Glc7p-interacting proteins (reviewed in Stark 1996).

Reduced glycogen accumulation was the first of several traits recognized for glc7 mutations (Cannonet al. 1994). At least three Glc7p noncatalytic subunits (Gac1p, Glc8p, and Reg1p) assist in wild-type glycogen synthesis levels. Dephosphorylation of glycogen synthase by Glc7p is required for glycogen synthesis. The Glc7p/Gac1p holoenzyme dephosphorylates glycogen particle-bound glycogen synthase in vivo by exploiting the Gac1p affinity for glycogen (Françoiset al. 1992). GLC8 encodes a homolog of mammalian inhibitor-2, which inhibits PP1 activity in vitro (Tunget al. 1995). GLC8 is not essential for cell growth and at normal gene dosage Glc8p appears to enhance Glc7p function in vivo (Morcos 1993; Tunget al. 1995). Gene disruption of glc8 reduces glycogen accumulation by only 20% compared to the ~90% reduction caused by glc7 or gac1 mutations (Françoiset al. 1992; Cannonet al. 1994). Mutations in REG1 have affects on several physiological processes, and deletion of REG1 causes greater than wild-type glycogen accumulation (Tu and Carlson 1995; Huanget al. 1996).

In addition to the nonessential functions of glycogen synthesis, Glc7p performs essential functions in mitotic cell cycle progression. Temperature-dependent glc7 mutations cause arrest in the G2/M phase (Hisamotoet al. 1994; MacKelvieet al. 1995), whereas Glc7p depletion by plasmid loss results in G1 arrest (Cannonet al. 1995). Sds22p is an essential PP1 noncatalytic subunit that mediates G2/M phase PP1 activity (Hisamotoet al. 1995; MacKelvieet al. 1995). Sds22p from Schizosaccharomyces pombe controls the substrate specificity of the PP1 catalytic subunit (Stoneet al. 1993).

Sporulation is also dependent on Glc7p activity. Diploid S. cerevisiae sporulate when grown on nonfermentable carbon sources and starved for nitrogen (reviewed in Esposito and Klapholz 1981; Kupiecet al. 1997). These environmental conditions induce cells to stop mitotic growth and initiate meiosis. Ume6p, which represses meiotic genes during mitosis, is converted into an activator of meiotic genes by binding to Ime1p (Bowdish and Mitchell 1993; Bowdishet al. 1995; Steber and Esposito 1995; Rubin-Bejeranoet al. 1996). Therefore, expression of Ime1p and its regulation of meiotic functions are indispensable for meiosis and sporulation. Gip1p and Red1p are two Glc7p-binding proteins with meiotic functions (Tuet al. 1996). GIP1 is transcribed during the middle of meiosis and is required for some late meiotic gene expression. Meiosis I chromosome segregation and synaptonemal complex formation require Red1p (Smith and Roeder 1997).

The great conservation of PP1 amino acid sequences between species can be rationalized in terms of the number of proteins with which these enzymes must interact. Hypothetically, each residue in Glc7p may have roles in interacting with multiple noncatalytic subunits or in recognizing several substrates. This hypothesis was tested through the isolation of a collection of glc7 mutants that reduced glycogen accumulation, yet grew normally. Subsequently, mutant Glc7p proteins were assayed for their noncatalytic subunit affinity and for their affects on the unselected sporulation trait. Consistent with the hypothesis, all mutant Glc7p proteins exhibited altered noncatalytic subunit affinities and displayed sporulation traits. Because sporulation was reduced by every glycogen-deficient glc7 mutant, the connection between these two traits was examined quantitatively, and the meiotic processes affected by sporulation-defective glc7 mutants were characterized. These studies have revealed at least two meiotic roles for Glc7p.

MATERIALS AND METHODS

Yeast strains and media: Yeast strains used in this study are listed in Table 1. Yeast strains JC981, JC993, and JC746 have been previously described (Cannonet al. 1994; Dalley and Cannon 1996). Other yeast strains were either derived from JC746 by transformation (JC821A and JC821D) or have been serially backcrossed, at least five times, to JC746-derived haploids (all other strains). When JC821A cells were transformed with pKC886 or pKC886-x plasmids, the p1855 plasmid initially present was always eliminated before further analysis. JC821D/pKC886-x transformants were derived from JC821A transformants by homothalism (HO)-induced diploidization using YCp50-HO followed by YCp50-HO plasmid loss (Herskowitz and Jensen 1991). Plasmids p1855 and YCp50-HO were eliminated from strains by screening colonies growing on YEPD plates for spontaneous Ura clones. Rich (YEPD), omission (synthetic complete, SC), and sporulation (SPOR) media have been previously described (Shermanet al. 1986). The composition of yeast extract-peptone-acetate (YEPA) differed from that of YEPD in that 2% (w/v) potassium acetate was used as the carbon source. Liquid sporulation medium (SPM) contained 2% (w/v) potassium acetate and required amino acids at 25% the normal concentrations (Shermanet al. 1986).

Plasmids: Plasmids used in this study are listed in Table 1. Plasmids with mutant GLC7 genes are named with a suffix that indicates the glc7 allele (i.e., pKC886-R73C contains an Arg-73 to Cys mutation). Most mutant GLC7 genes were originally isolated on pKC886 (see below). The two exceptions were glc7-R73C and glc7-R121K, which were retrieved from the yeast genome as described using plasmid p1855 (Cannonet al. 1994). Swapping restriction fragments made derivatives of pKC886 harboring these alleles. pAS1-GLC7 was made by inserting the 1.5-kb GLC7 BamHI fragment from pCV5 (Stuartet al. 1994) into the BamHI site of pAS1. Mutant GLC7 genes were transferred to pAS1-GLC7 by a series of restriction fragment swaps. A NdeI site containing the first methionine codon of GLC8 was introduced into p1945 (Cannonet al. 1994) by oligonucleotide mutagenesis (Morcos 1993). The resulting NdeI-HindIII GLC8 fragment was cloned into pET-16b(Novagen, Madison, WI). Subsequently, a NcoI-EcoRI GLC8 fragment from this plasmid was ligated to similarly digested pACTII to produce pACTII-GLC8. SDS22 DNA was amplified from yeast genomic DNA by PCR with primers 229 (5′-GCTCTATTATATTGTATCGTGTC-3′) and 230 (5′-TTCATG CTCGAGGTTTTTAAGGG-3′), and the amplicon was cloned into pGEX-5X-1 (Pharmacia, Uppsala, Sweden)after digestion with EcoRI and XhoI. The recovered SDS22DNA was completely sequenced and was identical to wild type (R. Gitan, unpublished results). A BamHI-XhoI SDS22 fragment from this plasmid was cloned into pACTII. The resulting plasmid was then digested with BamHI, filled in using Klenow DNA polymerase, and ligated so that the reading frame of SDS22 would be in frame with Gal4p(768-881), yielding pACTII-SDS22. Plasmid pRS425-IME1 was constructed in several steps. It has a 3.5-kb KpnI-SalI fragment encoding IME1 from pAM504 (Smith and Mitchell 1989). All mutations and reading frames were verified by DNA sequencing using an ABI model 373A DNA sequencer and Taq DyeDeoxy Terminator Cycle Sequencing (ABI, Foster City, CA).

Isolation of glc7 mutants: The plasmid shuffle method was used to isolate glc7 mutations on pKC886 (Sikorski and Boeke 1991). Plasmid pKC886 (CEN TRP1 GLC7) was mutagenized with hydroxylamine (Rose and Fink 1987) and used to transform JC821A (Itoet al. 1983). Transformants were selected on tryptophan-deficient minimal medium (SC-Trp) at 30° and then replica-plated to SC-Ura and two YEPD plates. The YEPD plates were incubated at either 30° or 37° and were used to score glycogen accumulation and temperature sensitivity, respectively. If JC821A transformants contained a pKC886 plasmid that provided Glc7p essential functions, then plasmid p1855 was readily lost. Plasmid p1855 was particularly unstable in these transformants because it is a high-copy plasmid and elevated GLC7 expression inhibits growth (Blacket al. 1995; Cannonet al. 1995; Zhanget al. 1995). Only transformants that could lose p1855 were further characterized. This constraint guaranteed that mutant glc7 gene products could perform all essential Glc7p functions and eliminated potential plasmid replication mutants. DNA was prepared from JC821A Ura (i.e., p1855-free) derivatives that exhibited increased or decreased glycogen accumulation or temperature sensitivity. Putative mutant pKC886 plasmids were recovered and amplified via Escherichia coli DH5α transformation. Only plasmids that reproducibly conferred mutant traits in JC821A transformants were further analyzed. DNA sequences of these mutant GLC7 genes were determined as described above.

Sporulation and viability experiments: For liquid sporulation, cells were grown to exponential phase in YEPD or omission media at 30° and then diluted into YEPA. After growth in YEPA for 10to 24 hr, cells were centrifuged and resuspended in 2 vol of SPM. Sporulation on SPOR plates used ~5 × 105 cells/mm2. Total and sporulated cells were counted using a hemacytometer after the cells had been sonicated briefly in water. At least 200 cells were counted in two or more experiments for all data reported. Viable cells were determined by counting colonies of dilutions plated on YEPD after 2 to 3 days of growth at 30°. When diploids were transformed with 2μ plasmids, the colonies used for viable counts were scored for plasmid retention by replica-plating to diagnostic omission media.

View this table:
TABLE 1

Plasmids and strains

Glycogen and two-hybrid assays: Yeast strains JC981 and JC993 (Dalley and Cannon 1996) were transformed with Gal4p DNA-binding hybrids (pASI-GLC7-x) or Gal4p transcription activation hybrids, respectively. Individual transformants were purified on SC-Trp or SC-Leu plates and allowed to grow at 30°. Transformants containing the two hybrids were mated, and diploids were selected on SC-Leu-Trp. Quantitative β-galactosidase assays were performed on diploids grown in selective minimal medium as described using o-nitrophenol-β-galactopyranoside (Yocumet al. 1984). Units ofβ-galactosidase activity were calculated (Miller 1972) from triplicate independent cultures and varied <15%. For glycogen assays, cells were harvested from growth medium and washed with water. Dry weights and glycogen were assayed as described (Lillie and Pringle 1980).

Propidium iodide staining and flow cytometry: Yeast cells (106) fixed in 70% (v/v) ethanol were incubated with 1 mg RNase A/ml for 1 hr at 37° in 300 μl of 50 mm sodium citrate, pH 7.4, then stained with 50 μg propidium iodide/ml for 15 min at room temperature. Following staining, cells were suspended in 1 ml of 50 mm sodium citrate, pH 7.4, and stored in the dark at 4°. DNA content was measured by a Coulter EPICS (Miami Lakes, FL) 753 flow cytometer. Stationary phase diploids were used as the standard for the 2N fluorescence level. Stained, exponentially growing haploid cells had two levels of fluorescence, one equal to the 2N and another at one-half the fluorescence, 1N. Exponentially growing diploids also showed two levels of fluorescence, one equal to 2N, the other double the 2N fluorescence, which we used as the 4N standard.

RNA analysis: Total yeast RNA was isolated, fractionated on formaldehyde agarose gels, and hybridized with 32P-labeled probes as described (Shermanet al. 1986). IME1 and ACT1 probes were prepared from fragments of pHS101 and pYact1, respectively (Ng and Abelson 1980; Smith and Mitchell 1989).

RESULTS

Isolation of mutant glc7 alleles that affect glycogen or growth traits: Mutant glc7 alleles that exhibited glycogen deficiency, hyperaccumulation, or growth thermosensitivity were isolated using the plasmid shuffle method (materials and methods). Mutations were identified that exhibited one or more of these traits, yet retained sufficient activity to permit viability when the mutant glc7 allele supplied the only PP1. Yeast strain JC821A has a glc7::HIS3 mutation that inactivates the chromosomal PP1-encoding gene. A high-copy, unstable plasmid (p1855) initially provides wild-type Glc7p to keep this strain alive. The glc7 mutations were isolated on the low-copy, CEN plasmid, pKC886. Effects on glycogen accumulation were evident in JC821A/pKC886 transformants during mutant screening because of their iodine-staining trait (Cannonet al. 1994). The majority of transformants stained the same wild-type brown color, but yellow or dark-staining colonies were also found (<1%). Comparison of 37° and 30° plates revealed rare temperature-sensitive candidates (<0.1%). Sixty apparently mutant derivatives of pKC886 were isolated, which could allow the loss of p1855 from JC821A, yet conferred either a glycogen accumulation or temperature-sensitive trait. A subset of these mutations was characterized in this work.

The DNA sequences of GLC7 in 24 mutant pKC886 derivatives were determined, and in each case a single C to T transition was discovered, consistent with the mutagenic specificity of hydroxylamine. Residue alterations caused by these mutant alleles are reported in Table 2. This table also includes the original glc7-1 (R73C) allele (Cannonet al. 1994) and a cold-sensitive allele, glc7-R245Q, which was constructed by site-directed mutagenesis from information provided by cold-sensitive fission yeast protein phosphatase alleles (Kinoshitaet al. 1990). On centromere vectors glc7-R245Q did not complement the glycogen deficiency of glc7-R73C nor restore viability of glc7::HIS3 mutants. Furthermore, on high-copy vectors, glc7-R245Q confers a reduced glycogen accumulation and cold-sensitive phenotype (Cannonet al. 1995). All mutations were found to change the amino acid sequence of the encoded Glc7p protein except for the temperature-sensitive mutation found in the intron. The glc7 intron mutation converted the consensus TACTAAC sequence to TATTAAC. A similar mutation within the ACT1 intron still allowed splicing to occur at 98% of wild-type levels (Vijayraghavanet al. 1986). Isolation of the Glc8p-dependent glc7-R121K allele will be reported elsewhere (J. F. Cannon, unpublished results). The 18 Glc7p residues altered in this collection of glc7 mutants were each conserved in PP1 sequences from other organisms, and these mutations were distributed throughout the molecule (Table 2). Remarkably, even at residues that exhibit variations between PP1 enzymes, the hydroxylamine-derived mutation frequently changed the Glc7p residue to a naturally occurring one found in another PP1 enzyme (e.g., R42K and A278V).

Glycogen accumulation and sporulation functions of Glc7p are separable: GLC7 mutations may affect glycogen and sporulation traits differently, depending on how the mutant residue influences Glc7p interactions with substrates and noncatalytic subunits. Indeed, analysis of the cid1 allele (glc7-T152K) showed that glc7 alleles could reduce sporulation but not affect glycogen accumulation (Tu and Carlson 1994). Therefore, we wanted to quantitatively assay these two traits for the collection of novel glc7 mutations to learn if sporulation and glycogen accumulation was affected to the same degree for these novel glc7 mutations.

Glycogen accumulation was assayed for JC821A transformed with wild-type or mutant pKC886 plasmids. Cells were grown until early stationary phase in liquid YEPD medium for these assays to mimic the plate growth conditions initially used to recognize these mutants, even though greater glycogen accumulation occurs if the cultures grow longer. As predicted by their iodine-staining trait, mutants that stained yellow accumulated less glycogen than wild type and mutants with dark staining accumulated more (Figure 1). In contrast, mutants in which iodine staining was not the selected trait (the Glc8p-dependent R121K and the intron temperature-sensitive) accumulated glycogen to nearly wild-type levels. Of the mutants that were isolated on the basis of low glycogen, some accumulated very low levels (<2% of wild type), whereas others accumulated 50–70% that of wild type. This exquisite sensitivity of the iodine staining method to small changes in glycogen levels has been noted previously (Cannonet al. 1994; Dalley and Cannon 1996).

View this table:
TABLE 2

Summary of glc7 mutations

The sporulation function of mutant Glc7p proteins was assayed in transformants of two diploids, JC821D and JC835 (Table 1). These are closely related, but not isogenic, strains. Their maximum sporulation frequencies were 60 and 41%, respectively, when transformed with pKC886 (CEN GLC7+). This difference could be because these strains are not isogenic or from idiosyncrasies of genomic vs. plasmid GLC7 expression. The pKC886-x plasmids provided the only functional glc7 allele in JC821D, which is homozygous for glc7::HIS3. Therefore, this strain reports the intrinsic sporulation trait of the glc7 alleles. All glc7 alleles except for the intron mutation showed a sporulation defect in JC821D (Figure 2A). Moreover, the majority of the glc7 alleles promoted undetectable (<0.5%) sporulation in this diploid. Sporulation of glc7-R73C in JC821D was similar to its sporulation when homozygous in the genome (Cannonet al. 1994). Allele glc7-A268T was unique among the novel, glycogen-deficient alleles because its intrinsic sporulation was the highest, reaching about one-third of wild-type GLC7 sporulation frequency (Figure 2A). Therefore, the unselected sporulation trait was always found for glc7 mutants isolated because of a glycogen deficiency.

Figure 1.

Glycogen accumulation of JC821A strains transformed with pKC886-x derivatives. Strain JC821A is glc7::HIS3, and pKC886-x plasmids provide wild-type or mutant GLC7. Triplicate glycogen assays were performed on cells grown in YEPD for 24 hr. Average glycogen content, reported as a percentage of dry weight, varied <15% between triplicates.

We also examined whether the novel glc7 alleles could enhance the sporulation of JC835, which is homozygous for the previously characterized glc7-R73C mutation. Sporulation of JC835 reports the ability of glc7 alleles to complement the glc7-R73C sporulation trait. This strain sporulated 7% without pKC886-x plasmids, and this frequency was increased for most glc7 alleles (Figure 2B). These findings reveal that most glc7 alleles, incapable of promoting sporulation alone, can complement the sporulation trait of glc7-R73C. Even glc7-R73C on a centromeric pKC886 plasmid promoted significant sporulation in JC835 (58% that of wild type). Therefore, apparently subtle increases in glc7-R73C gene dosage can increase sporulation. Such phenotypic effects of small increases in wild-type GLC7 gene dosage have been noted previously (Giotet al. 1995). Three glc7 alleles did not complement the glc7-R73C sporulation trait. JC835 transformed with alleles R142H, D202N, and G214D sporulated close to the same frequency as untransformed JC835. We interpret these results to indicate that mutant Glc7p proteins encoded by these alleles cannot dephosphorylate the substrates essential for sporulation.

Sporulated yeast cells produce two-spored dyads instead of tetrads if one of the meiotic divisions does not occur or if spores are incompletely packaged (Esposito and Klapholz 1981; Kupiecet al. 1997). Sporulation of some JC835 transformants produced many dyads (Figure 2B). Alleles R142H and G214D produced predominantly or exclusively dyads in the few sporulated cells found, and the greatest dyad yield came from alleles Q48K and A278V, which complemented the R73C sporulation trait well. Because the percentage or yield of dyads was independent from the sporulation frequency, dyad production can be considered a separate glc7 trait potentially mediated by Glc7p substrates different from those that allow sporulation initiation. Sporulation data from JC835 transformants allow some of the glc7 alleles to be classified by these two traits. For example, alleles R142H and G214D can be considered defective for both traits because not only did they sporulate poorly, but they also predominantly yielded dyads. In contrast, alleles D202N and P177S also sporulated poorly, but because they generated tetrads, they do not appear defective in the second sporulation function. The high yield of dyads for the Q48K and A278V alleles suggests that they are only defective in the second sporulation function.

Figure 2.

Sporulation of glc7 diploids. (A) JC821D (without p1855) transformed with pKC886-x derivatives sporulated on SPOR plates for 2 days. GLC7 alleles on the pKC886-x plasmids are indicated. Asterisks indicate transformants with undetectable (<0.5%) sporulation. GLC7-R245Q cannot complement glc7::HIS3 and was not assayed in this experiment. (B) JC835 (homozygous for glc7-R73C) transformed with pKC886-x derivatives grown 24 hr in YEPA, then sporulated in liquid SPM for 2 days (materials and methods). The vertical dashed line indicates sporulation of untransformed JC835. Sporulation close to or below this level indicates a failure of the pKC886-x glc7 allele to complement the sporulation trait of glc7-R73C. The unfilled portion of each bar indicates the contribution of dyads to the total sporulated cells.

View this table:
TABLE 3

Sporulation and viability of JC821D transformants

Starvation intolerance does not explain the glc7 sporulation trait: The reduction in sporulation for glc7 mutants could be because of a loss of viability in the sporulation media. For example, bcy1 mutations, which elevate cAMP-dependent protein kinase activity, reduce sporulation because the mutant cells are intolerant of the starvation conditions required for sporulation (Cannonet al. 1990). To test the viability of glc7 diploids during sporulation, JC821D diploids transformed with sporulation-deficient glc7 alleles on pKC886-x plasmids were sporulated 3 days on SPOR plates, and the percentage of viable cells assayed (Table 3). The alleles chosen for this experiment exhibit a variety of sporulation efficiencies (Figure 2). Cell viability decreased for all diploids with mutant glc7 alleles compared to those transformed with wild-type GLC7. Although poor sporulation was associated with greater inviability, the viability does not quantitatively account for the sporulation reduction. For example, glc7-G24A shows only a threefold reduction in viability compared to wild type, but over a 50-fold reduction in sporulation.

We analyzed sporulation kinetics to learn if glc7 diploids sporulated slower than wild type. For this experiment two isogenic diploids that differed in their chromosomal glc7 alleles were used. Using these strains circumvented potential inviability caused by pKC886 plasmid loss that might occur in the JC821D diploids above. Diploid JC1277 (GLC7/glc7-R73C) began to sporulate after 10 hr in liquid sporulation medium, eventually reaching 60% sporulation after 48 hr. Under identical conditions the diploid JC1276 (glc7-R73C/glc7-R73C) sporulated only 2% after 24 hr, and this frequency did not increase (Figure 3A). Therefore, the failure to detect JC1276 sporulation was not explained by slower sporulation kinetics of the glc7-R73C mutant cells. Both JC1276 and JC1277 cells arrested as unbudded cells in sporulation medium, indicating that both sensed the environmental conditions that trigger sporulation. Viability of both mutant and wild-type diploids decreased during the growth in YEPA (Figure 3B). Nitrogen sources in this medium typically repress sporulation. We have evidence that early meiotic events occur during the YEPA preincubation (see below). The viability did not continue to change during the time interval in which the wild-type cells sporulated. This result shows that these diploids lose viability soon after transfer to sporulation-promoting medium and that glc7-R73C diploids are more susceptible than wild-type diploids to this viability loss.

Figure 3.

Sporulation and viability of wild-type and glc7 diploids. JC1277 (GLC7/glc7-R73C) (circles) and JC1276 (glc7-R73C/glc7-R73C) (triangles) were grown in YEPA 24 hr and then transferred to SPM. Percent sporulation (A) and viability (B) were quantitated as described in materials and methods at the times indicated in SPM.

GLC7 mutations hinder sporulation before premeiotic DNA synthesis: Premeiotic DNA synthesis is one of the earliest events after the onset of meiosis-inducing signals (Esposito and Klapholz 1981; Mitchell 1994). We examined premeiotic DNA synthesis in JC821D diploids with glc7 alleles R142H, D202N, and G214D. These alleles were chosen because they exhibited the greatest sporulation defects. We assayed premeiotic DNA synthesis by monitoring the cellular DNA content using flow cytometry. An experiment comparing the DNA levels for wild-type GLC7 and glc7-R142H is shown in Figure 4. Diploid JC821D transformants grown to early stationary phase in YEPD were transferred to YEPA for 10 hr and then transferred to SPM medium. After the 10 hr of YEPA growth, the cultures remained in stationary phase as evidenced by the few G2 cells with a 4N DNA content. After 4 hr in SPM, premeiotic DNA synthesis initiated in wild-type transformants, producing a peak of 4N DNA. Diploid quantities of DNA disappeared by 20 hr, and only 4N DNA content was detected. Ultimately, 60% of the wild-type JC821D transformants sporulated, and they contained exclusively 4N DNA. In contrast, glc7-R142H diploids showed no evidence of premeiotic DNA synthesis even after 30 hr in sporulation medium. This failure to initiate premeiotic DNA synthesis can explain the poor sporulation of glc7-R142H diploids. The glc7 alleles D202N and G214D were similarly tested with identical results to R142H (data not shown).

Figure 4.

DNA content of JC821D diploids during sporulation. JC821D transformed with pKC886 or pKC886-R142H plasmids were grown in YEPA for 10 hr and then transferred to SPM medium at zero time. Samples of cells after 0, 4, 10, 20, or 30 hr in SPM were analyzed by propidium iodide staining by flow cytometry (materials and methods). Cell numbers (y-axis) are plotted versus fluorescence (x-axis).

Regulation of early meiotic genes IME1 and IME2 by Glc7p: Ime1p and Ime2p are positive inducers of meiosis (Mitchellet al. 1990). Expression of IME1 is essential for meiosis, and it promotes IME2 transcription. Because sporulation-defective glc7 diploids were blocked prior to meiotic prophase (see above results), the time during which early meiotic gene expression occurs (Mitchell 1994), we suspected that these mutants could be defective in expression of one or more early meiotic genes. To test this hypothesis, IME1 mRNA expression was assayed for the wild type and glc7-R142H diploids used above for DNA analysis. Wild-type diploids had the greatest IME1 mRNA levels after growing in YEPA for 10 hr (Figure 5A). This mRNA diminished substantially by 20 hr after transfer to SPM medium. In contrast, no IME1 transcript was detectable at any time in R142H diploids cultured identically.

The wild-type IME1 expression in YEPA suggests that sporulation initiated in this medium. However, premeiotic DNA synthesis was undetectable until this strain was transferred to SPM medium (Figure 4). If left in YEPA, this wild-type diploid eventually sporulated to ~10% after 5 days. Sporulation in the nitrogen-rich YEPA medium is not common for many yeast strains and may be related to the lower cAMP-dependent protein kinase activity of our strains (Cannonet al. 1994).

Poor IME1 mRNA expression in sporulation-defective glc7 diploids was verified using strains transformed with pKB100, a plasmid containing IME1 responsive, IME2 upstream regulatory sequences fused to lacZ (Bowdish and Mitchell 1993). JC821D transformed with wild-type GLC7 induced IME2-lacZ β-galactosidase 9 hr after transfer to SPM (Figure 5B; open symbols). Transformants with pKC886-120 (containing the temperature-sensitive glc7 intron mutation) behaved similarly (data not shown). In comparison, glc7-R73C diploids reached a maximum one-quarter that of wild type. Moreover, no IME2-lacZ induction was detected for R142H (Figure 5B), G24A, P82S, R142H, D202N, or P269L diploids (data not shown). Note that several of these glc7 alleles allow some sporulation in JC821D (Figure 2A; Table 2). JC821D transformed with pKC886-A268T sporulated about one-half the frequency of wild-type GLC7 transformants (Figure 2A). Diploids with glc7-A268T induced one-half the wild-type levels of IME2-lacZ (data not shown). Additionally, many JC835/pKC886-x transformants were also tested for IME2-lacZ induction. In every case the maximum level of β-galactosidase activity from IME2-lacZ was proportional to the sporulation efficiency of the diploid (data not shown). Therefore, we conclude that absence of early meiotic inducers Ime1p and Ime2p is responsible in part for the glc7 meiotic defect and that Glc7p could control sporulation via Ime1p and Ime2p expression.

As Ime1p positively regulates expression of IME2 and a number of other meiotic genes (Mitchell 1994), we asked whether overexpression of IME1 could override the sporulation defect of glc7 diploids. In all glc7 diploids tested, a high-copy IME1 plasmid enhanced IME2-lacZ induction. The results for glc7-R73C and glc7-R142H are shown in Figure 5B (filled symbols). The high-copy IME1 plasmid increased induction of IME2-lacZ in the glc7 mutant diploids to levels that were comparable to wild type. Additionally, the induction maximum occurred earlier than that seen in wild-type diploids; it occurred after only 5 hr in SPM medium. Despite these Ime1p and Ime2p increases, there was no sporulation enhancement for these and other glc7 diploids (Table 4). DNA content analysis of JC821D glc7-R142H diploids transformed with high-copy IME1 showed no evidence of premeiotic DNA synthesis (data not shown).

Figure 5.

Expression of IME1 and IME2 during sporulation of wild-type and glc7 diploid cells. (A) Northern analysis of RNA isolated from JC821D transformants from Figure 4 under sporulation conditions. (B) Expression of IME2-lacZ in wild-type and glc7 diploid cells during sporulation. JC821D/pKC886 (circles), JC821D/pKC886-R73C (squares), and JC-821D/pKC886-R142H (triangles) transformed with pRS425-IME1 (filled symbols) or pRS425 (open symbols) were grown 8 hr in YEPA before shifting to SPM medium. All diploids also harbored pKB100 (IME2-lacZ). β-Galactosidase activity (in Miller units) was determined for cells harvested at the indicated times after the shift.

View this table:
TABLE 4

Suppression of glc7 sporulation trait by IME1

Glc7p two-hybrid interaction assays: PP1 catalytic subunits associate with a variety of noncatalytic subunits in vivo (Stark 1996). Formation of these alternate holoenzymes is essential for the PP1 catalytic subunit to dephosphorylate substrate phosphoproteins. Traits of glc7 mutations could be caused by altered affinities for one or more of these noncatalytic subunits. In fact, reduced affinity for the glycogen-binding subunit Gac1p, by the glc7-R73C gene product, has accounted for the glycogen deficiency of this mutant (Stuartet al. 1994). Therefore, we assayed the interaction of novel, mutant Glc7p proteins with the noncatalytic subunits Gac1p, Sds22p, Glc8p, and Gip1p by the two-hybrid assay (Fields and Song 1989) to learn if affinities to these proteins could explain the glycogen accumulation and sporulation traits of these mutants.

Our two-hybrid assays used wild-type or mutant Glc7p proteins expressed as fusions to the Gal4p DNA-binding domain (residues 1-147 of Gal4p) and noncatalytic subunits as Gal4p activation domain fusions. Interaction of the two hybrids was assayed by β-galactosidase activity from a GAL1 promoter driven lacZ reporter gene in diploid JC981 × JC993 cells (materials and methods). The Gal4p-Glc7p fusions in pAS1-GLC7-x plasmids included the hemagglutinin (HA) epitope (Durfeeet al. 1993), which we used to assay expression levels of the mutant fusions. Equivalent expression levels were found for all Glc7p fusions (data not shown). Therefore, these fusions had similar stabilities in vivo and faithfully reported the relative affinities for the noncatalytic subunits.

The majority of the glc7 alleles reduced apparent Glc7p affinity for Gac1p (Figure 6A). Besides glc7-R73C, which was previously shown to be defective in Gac1p binding (Stuartet al. 1994), we found that alleles P195F, R245Q, and P269L encode proteins with undetectable Gac1p binding by this assay. Other alleles (P23S, P82S, R142H, and A278V) were also substantially reduced. Poor Gac1p binding can account for the reduced glycogen accumulation by these glc7 alleles because Gac1p localizes Glc7p to glycogen particles where it dephosphorylates glycogen synthase (Françoiset al. 1992; Stuartet al. 1994). Reduced Gac1p binding was found for all glycogen-deficient glc7 mutants, but there was great variation between them. Furthermore, enhanced affinity was found for glc7-G279S, which exhibits greater than wild-type glycogen accumulation. However, Gac1p apparent affinity does not fully account for glycogen accumulation levels because these two parameters were not strictly correlated. For example, alleles R73C and G198D promote similar glycogen levels (Figure 1) but differ substantially in Gac1p affinity (Figure 6A). The glc7-D202N gene product bound reasonably well to Gac1p, yet it resulted in very low glycogen levels. Wild-type glycogen levels accumulated in the glc7-R121K mutant even though the mutant gene product had attenuated Gac1p affinity.

Figure 6.

Two-hybrid assay of Glc7p mutant protein interactions with (A) Gac1p, (B) Sds22p, (C) Glc8p, (D) Gip1p(32-573), and (E) Gip1p(161-573) proteins. Three independent transformants of JC981 × JC993 diploids transformed with pAS1-GLC7 and pACTII derivative plasmids were assayed for β-galactosidase activity (materials and methods); these levels differed by <20% for the replicates. Background activity of pAS1-SNF1 and pACTII-GLC8 was <0.004 Miller units. Asterisks denote hybrid pairs that yielded undetectable β-galactosidase activity (less than three times background). Note that different x-axis scales are used in each.

Figure 7.

Inverse correlation of Ga1p and Sds22p affinities by two-hybrid assays. Miller units of β-galactosidase activity for each protein interacting with Glpc7p mutant proteins are plotted. The asterisk indicates wild-type Glc7p. The curve shows a reciprocal relationship with an r2 correlation of 0.8.

The Glc7p noncatalytic subunit Sds22p is essential for viability (Hisamotoet al. 1995; MacKelvieet al. 1995). Because our selection ofglc7 mutations constrained them to encode a Glc7p protein that could perform essential functions, we anticipated all mutant proteins would exhibit Sds22p affinity. Amazingly, the majority of mutant Glc7p proteins isolated in this work exhibited reduced Sds22p affinity. In fact, five affinities (intron, G24A, A268T, A278V, and G279S) were either very close to background or undetectable (Figure 6B). On the other hand, four glc7 alleles (R73C, P195F, G198D, and R245Q), which reduced glycogen accumulation, encode mutant Glc7p proteins that apparently bound Sds22p better than wild-type Glc7p. Mutant Glc7p proteins with the poorest Gac1p binding (R73C and R245Q) were the best Sds22p binders. Conversely, the best Gac1p binder (G279S) was a poor Sds22p binder. Plotting the Gac1p and Sdss22p affinities shows a convincing inverse relationship (Figure 7). Only two Glc7p proteins, including wild type, are obvious outliers in this correlation. Such a relationship would be found if Gac1p and Sds22p binding to Glc7p were mutually exclusive.

Two-hybrid interactions with Glc8p were the weakest of any Glc7p-interacting proteins assayed in this work (Figure 6C). Interaction between Glc7p and Glc8p, although subtle, may be important for enhancing Glc7p function in vivo (Tunget al. 1995). Most mutations throughout Glc7p reduced binding to undetectable levels (Figure 6C). In contrast, mutations altering residues near the carboxy terminus substantially increased Glc8p binding.

Gip1p is a Glc7p noncatalytic subunit required for sporulation (Tuet al. 1996). Failure of Glc7p mutant proteins to interact with Gip1p might explain the sporulation trait of some glc7 mutants. Interaction of Glc7p mutant proteins with two Gip1p fusion proteins was assayed (Figure 6, D and E). Although affinities were about fourfold greater for the larger Gip1p(32-573) fusion, the pattern of affinities for the two fusions was very similar. We assumed that any mutant Glc7p protein that could promote some sporulation when it was the sole source of Glc7p would bind to Gip1p. Six out of eight alleles that allowed some JC821D sporulation (Figure 2A) also showed detectable Gip1p affinity (Figure 6, D and E). Conversely, if Gip1p affinity were the attribute that defined the sporulation frequency promoted by a mutant Glc7p protein, then glc7 alleles that encoded proteins with significant Gip1p affinity would also sporulate to some degree. Five out of six mutant Glc7p proteins with greater than one-quarter of the wild-type Gip1p affinity promoted sporulation to some degree when they were the sole source of Glc7p. Moreover, the sole exception, G279S, complemented the glc7-R73C sporulation trait well (Figure 2B). Surprisingly, two out of the three alleles with the poorest sporulation showed detectable Gip1p affinity. Affinities of the R142H and G214D mutant proteins for Gip1p were much greater than mutants that sporulated better. These alleles illustrate that Gip1p binding to a mutant Glc7p protein is necessary but not sufficient for the Glc7p to promote sporulation.

Two-hybrid experiments have demonstrated an interaction between Red1p and Glc7p (Tuet al. 1996). Diploids homozygous for red1 sporulate well but yield only 1% viable spores because of nondisjunction during the first meiotic division (Rockmill and Roeder 1988). If Red1p activity was reduced in any glc7 mutant, meiotic spore viability should be reduced. Spore viability for JC821D diploids transformed with pKC886-x plasmids containing wild-type, G24A, R73C, P82S, A268T, and P269L glc7 alleles was tested by tetrad dissection. Some spores were anticipated to be inviable because meiotic segregation of the centromeric pKC886-x plasmids would produce plasmid-free glc7::HIS3 spores. Despite this source of spore inviability, these diploids produced 37–80% viable spores. This high level of spore viability is inconsistent with a complete loss of Red1p function.

DISCUSSION

The conservation of PP1 amino acid sequences from diverse species is unusually high (Bartonet al. 1994). Some homology is anticipated for the residues that define the enzyme active site, and PP1 enzymes show their greatest homology with related serine/threonine protein phosphatases in such active-site residues (Egloffet al. 1995; Goldberget al. 1995). However, conservation of residues outside the active site suggests that other factors constrain amino acid variability in this set of proteins. The results in this article show that mutation of residues in the Glc7p PP1 enzyme can lead to multiple effects on in vivo Glc7p function.

Association with numerous alternative noncatalytic subunits distinguishes PP1 from related protein phosphatases, and maintaining potential interaction with these subunits may hypothetically limit potential PP1 sequence variation. At least nine PP1 noncatalytic subunits exist in yeast (Stark 1996); presumably more will be found in other species. Interaction of PP1 with a segment of one noncatalytic subunit has recently been delineated by X-ray crystallography (Egloffet al. 1997). The contiguous noncatalytic subunit residues that participate in binding to PP1 define a short motif found in many, but not all, known noncatalytic subunits. For example, Gac1p and Gip1p contain one and two of these proposed PP1 interaction regions, respectively (Egloffet al. 1997), whereas Sds22p and Glc8p (and its homolog inhibitor-2) do not have recognizable homologies to those regions. If noncatalytic subunits bound to PP1 equivalently, we would have anticipated identical patterns of two-hybrid results for each noncatalytic subunit tested. However, it is clear that the pattern of affinities for each noncatalytic subunit is different (Figure 6). We conclude, therefore, that noncatalytic subunit binding to PP1 may be overlapping, but not equivalent.

The concept of a common PP1 site that binds alternative noncatalytic subunits suggests that such binding would be competitive. Indeed, two-hybrid assays of Gac1p and Sds22p binding display an apparent reciprocal binding relationship consistent with mutually exclusive binding (Figure 7). No other pair of Glc7p noncatalytic subunits shows this relationship (Figure 6 and data not shown); perhaps because Gac1p and Sds22p are the major, high-affinity subunits, they compete for the same binding site. Examination of such a relationship would only be valid if there are no biases in the mutant collection used for analysis. We believe that such biases are absent because affinities to Gac1p and Sds22p were both unconstrained. Although these glc7 mutants were primarily deficient in glycogen accumulation, Glc7p/Gac1p interaction is necessary but not sufficient to promote glycogen synthesis. Consequently, Gac1p affinity can be considered an independent variable. Our demand that mutant Glc7p proteins perform essential functions could potentially limit mutants to those that bind to Sds22p. Nevertheless, Glc7p mutant proteins exhibited Sds22p affinities that ranged from undetectable to greater than wild type. Therefore, we conclude that the reciprocal relationship of Gac1p and Sds22p two-hybrid results suggests that the binding of these noncatalytic subunits is mutually exclusive.

If Glc7p substrates and noncatalytic subunits bound to Glc7p equivalently, it would be difficult to isolate glc7 mutations that resulted in hyperphosphorylation of a subset of Glc7p substrates. This contention was tested here by characterizing a collection of glycogen deficient glc7 mutants. Consistent with this hypothesis, almost all glycogen accumulation mutants were defective in sporulation. Moreover, quantitative comparisons of sporulation frequencies and glycogen accumulation rule out the possibility of cause and effect. This was initially a suspicion because sporulation occurs in diploid cells under environmental conditions similar to those that promote glycogen accumulation. Furthermore, these two traits are separable by glc7 mutations. In particular, the two mutations that accumulated more glycogen than wild type, D9K and G279S, sporulated as poorly as alleles like R142H, P195F, and A278V, which accumulated minor amounts of glycogen (Figures 1 and 2A). Sporulation of A268T was reasonably high despite the very low glycogen accumulation in this mutant. The Glc8-dependent mutation R121K had wild-type glycogen levels, yet sporulated poorly. In summary, from comparison of the sporulation and glycogen accumulation traits of several glc7 alleles it appears that Glc7p substrates and noncatalytic subunits bind to Glc7p differently.

Glc7p controls glycogen accumulation by dephosphorylating glycogen synthase, which renders its catalytic activity glucose-6-phosphate independent (Fenget al. 1991). The Glc7p/Gac1p holoenzyme seems to be responsible for this dephosphorylation in vivo (Françoiset al. 1992; Stuartet al. 1994). Hence, mutant Glc7p proteins with reduced Gac1p affinity would necessarily be glycogen deficient because they would mirror gac1 mutations. This can account for the low glycogen accumulation of R73C, R142H, P195F, P269L, and A278V mutants. However, G198D, D202N, and G214D glycogen-deficient alleles do not greatly reduce Gac1p affinity (Figure 6A). Furthermore, these alleles do not reduce affinity to Glc8p (Figure 6C), which is also required for wild-type glycogen levels (Cannonet al. 1994). Therefore, these alleles require another explanation for their low glycogen accumulation. The residues mutated in these alleles could affect recognition of the Glc7p substrate, glycogen synthase. Residues D202 and G214 in the carboxy-terminal PP1 subdomain influence structure of the “acidic groove,” which is thought to bind to substrates (Goldberget al. 1995). Mutations of these residues also could not complement the R73C sporulation trait (Figure 2B), which is consistent with a failure to dephosphorylate sporulation specific substrate(s). Not all Glc7p substrate recognition was affected by these mutations, however, because these mutant Glc7p proteins sufficiently dephosphorylated mitotic substrates, which would have prevented growth in the phosphorylated state.

Glycogen hyperaccumulation was unexpected for recessive, hypomorphic glc7 alleles. Such glycogen hyperaccumulation alleles were found among alanine-scanning glc7 mutations that altered residues D7 and D9 (glc7-121) or K259 and R260 (glc7-109) (Bakeret al. 1997). Likewise, we found that D9K individually as well as G279S conferred this trait. Glycogen hyperaccumulation traits for these glc7 alleles could result from enhanced Gac1p binding and/or increased glycogen synthase dephosphorylation activity. In fact, Gac1p affinity to the GLC7-G279S gene product was significantly greater than any other mutant Glc7p protein tested (Figure 6A). However, Glc7p does not control glycogen metabolism exclusively through Gac1p. REG1 encodes a Glc7p-binding protein, mutations in which increase glycogen accumulation (Tu and Carlson 1995; Huanget al. 1996). GLC7 glycogen hyperaccumulation alleles are proposed to reduce Reg1p affinity. Consistent with this hypothesis, residues K259, R260, and G279 are found between or within three proximal, carboxy-terminal beta sheets (Egloffet al. 1995; Goldberget al. 1995). Therefore, they influence the surface structure of PP1 that may provide the Reg1p interface. Glc7p D9, however, is in a distal, amino-terminal subdomain, and if it participates in this potential Reg1p binding, then it illustrates that noncatalytic PP1 subunits can bind to multiple and distant PP1 residues.

Temperature- and cold-sensitive glc7 mutants arrest late in mitosis. An identical arrest for sds22 mutants suggests that activity of the Glc7p/Sds22p holoenzyme performs an essential function late in mitosis (Hisamotoet al. 1995; MacKelvieet al. 1995). The majority of mutant glc7 alleles examined in this work could perform essential Glc7p functions. Therefore, it was surprising that many mutant Glc7p proteins exhibited reduced Sds22p affinity, and interaction with three mutant proteins was undetectable (Figure 6B). With the exception of the temperature-sensitive intron mutant, none of these poor Sds22p binders have any obvious growth impairments. Because elevated GLC7 expression suppresses the sds22 null lethality (Hisamotoet al. 1995), we propose that glc7 alleles that bind Sds22p poorly (G24A, A268T, and A278V) might compensate for weak affinities by increasing Glc7p activity or by adopting a substrate specificity normally induced by Sds22p association.

The temperature-sensitive glc7 intron mutation (glc7-120) did not affect glycogen accumulation or sporulation traits (Figures 1 and 2). Affinity to Gac1p or Gip1p was unaffected, but Sds22p and Glc8p affinities were both reduced. Because the mutation in this allele is in the intron, its traits must result from expression alterations of an otherwise wild-type protein. Splicing of mRNA has been shown to be particularly sensitive to heat-shock (Vogelet al. 1995), and this provides the most economical explanation of the temperature-sensitive trait of this intron mutation. However, glc7-120 effects on Sds22p and Glc8p affinities are more difficult to explain. Potentially, the strong, constitutive ADH1 promoter used to express the Gal4p-Glc7p fusion containing the mutant intron could produce a significant amount of unspliced fusion protein, which could bind well to Gac1p and Gip1p, but not to Sds22p or Glc8p. Much lower expression from the native GLC7 promoter was used in experiments where glycogen accumulation, sporulation, and temperature-sensitive traits were scored.

The glc7-R245Q mutation is a dominant-negative and cold-sensitive mutation (Cannonet al. 1995). Cold sensitivity and increased phenotypic severity resulting from increased glc7-R245Q gene dosage are consistent with this mutant protein binding to a Glc7p positive factor in a manner that excludes wild-type Glc7p from interaction. Greater than wild-type Sds22p affinity at the permissive temperature for the GLC7-R245Q mutant gene product suggests that Sds22p is the crucial positive factor. Indeed, S. pombe sds22+ was originally isolated on the basis of its ability to suppress an analogous cold-sensitive PP1 mutation in dis2 (Kinoshitaet al. 1990; Ohkura and Yanagida 1991).

The majority of Glc8p interaction/negative alleles affected amino acids in the amino-terminal subdomain (residues 42-142) of Glc7p (Figure 6C). Although none of these amino acids are found on the protein surface, three arginines (residues 42, 121, and 142) participate in intramolecular salt bridges. These mutations could change the distance between lobes composed of A- and F-helices in the N-terminal subdomain (Egloffet al. 1995; Goldberget al. 1995). Perhaps Glc8p recognizes Glc7p via these lobes. The distance of these lobes from the active site suggests that Glc8p binding to PP1 would not necessarily block substrate access, but it could depending on the Glc8p-binding orientation. This might be the binding mechanism of inhibitor-2, a Glc8p homologue, during in vitro PP1 inhibition.

The other major findings in this article concern the role of Glc7p during meiosis. Meiosis is an alternate cell cycle pathway available to diploid yeast under suitable environmental conditions. Protein phosphorylation controls some steps of meiosis because several protein kinases have been identified that regulate this developmental process (Mitchell 1994; Kupiecet al. 1997). The majority of these enzymes inhibit sporulation when they have reduced activity via recessive mutations. Therefore, they are poor candidates for phosphorylating Glc7p substrates because Glc7p substrates inhibit sporulation when overphosphorylated. The cAMP-dependent protein kinase (PKA) is the only known kinase that inhibits sporulation when hyperactive (Matsuuraet al. 1990). However, hyperactivity of PKA causes loss of viability during starvation in sporulation medium (Cannonet al. 1990), which contrasts with the viability retention of glc7 diploids (Table 3). These data suggest that PKA phosphorylates substrates not dephosphorylated by Glc7p or that another kinase phosphorylates Glpc7p substrates that inhibit sporulation. In fact, PKA regulates trehalose metabolism, which mediates stress tolerance, whereas Glc7p does not regulate this pathway (Cannonet al. 1994).

Ime1p activity is pivotal for early meiotic events (Mandelet al. 1994; Mitchell 1994). Its diminished expression in sporulation-defective glc7 diploids is sufficient to explain the failure of premeiotic DNA synthesis, a prelude to sporulation. We propose that one or more dephosphorylated Glc7p substrates normally activate IME1 transcription. Currently, three independent pathways are known to enhance IME1 expression. These pathways employ the gene products of IME4, MCK1, and the combination of RIM1, RIM8, RIM9, and RIM13 (Su and Mitchell 1993). We are currently exploring which pathway of IME1 induction is defective in glc7 diploids. The observation that high-copy IME1 induces IME2-lacZ transcription in glc7-R73C and glc7-R142H diploids (Figure 5B) suggests that IME1 is repressed in these mutant diploids and that an IME1 repressor is limiting.

Poor IME1 expression is not the only reason that glc7 diploids fail to sporulate. A high-copy IME1 plasmid can induce IME2-lacZ transcription in glc7-R73C and glc7-R142H diploids to levels that are comparable to wild type (Figure 5B). Nevertheless, sporulation of glc7 diploids was not improved in such transformants (Table 4). The failure to suppress the glc7 sporulation trait by high-copy IME1 shows that reduced IME1 expression is not the only block to glc7 spore formation. Finding two Glc7p-binding proteins, Gip1p and Red1p, expressed late in meiosis (Thompson and Roeder 1989; Tuet al. 1996) presaged a late meiotic function for Glc7p holoenzyme(s).

Mutant Glc7p proteins can promote sporulation despite great reductions in Gip1p affinity. GLC7 alleles R73C and P269L had undetectable Gip1p affinity (Figure 6D), yet sporulated to some degree when they provided the sole source of Glc7p (Figure 2A). Although the poorest sporulation alleles (R142H, P177S, D202N, and G214D) had weakened Gip1p affinities, they were not as low as other glc7 alleles, which sporulated well. Therefore, function of the Glc7p/Gip1p holoenzyme does not control the sporulation frequency.

Sporulation of diploid yeast cells normally results in two meiotic divisions followed by packaging the four progeny nuclei into spores. However, dyads result if either of the meiotic divisions does not occur or if all progeny nuclei are not packaged (Esposito and Klapholz 1981; Kupiecet al. 1997). The glc7 alleles Q48K, R142H, G214D, and A278V produced a great fraction of dyads in sporulated cells (Figure 2B). As a group, these four alleles have very low Gip1p affinities (Figure 6D). Moreover, no mutant Glc7p proteins with intermediate Gip1p affinities (above one-third of wild-type Glc7p) produced a large fraction of dyads. Therefore, these data suggest that the Glc7p/Gip1p holoenzyme dephosphorylates substrate proteins that encourage completion of meiotic divisions or spore packaging during the later stages of sporulation.

Two-hybrid assays indicated an association of Red1p and Glc7p (Tuet al. 1996). Mutations in red1 cause poor meiotic spore viability (Rockmill and Roeder 1988). Consequently, if the sole function of Red1p were to localize Glc7p, mutant forms of Glc7p that do not bind to Red1p would also exhibit poor meiotic spore viability. In fact, all glc7 mutants tested have high meiotic spore viability. We found that many glc7 alleles reduced Red1p affinities (L. Li and J. F. Cannon, unpublished results). Hence, Red1p function is normal in the glc7 mutants tested and does not appear to be influenced by Glc7p binding.

In summary, we found that virtually all glc7 mutations studied affected the interaction of Glc7p with regulatory noncatalytic subunits. Thus, glc7 mutants isolated because of traits caused by malfunction of one Glc7p holoenzyme, invariably affected other Glc7p holoenzymes and therefore other traits. These data are consistent with the hypothesis that high PP1 sequence conservation results from each residue influencing multiple protein interactions with PP1. Furthermore, scrutinizing the Gac1p affinities and effects on sporulation by glc7 mutations shows that Glc7p has at least two roles in regulating glycogen metabolism and sporulation.

Acknowledgments

We thank David Eide, Shirish Shenolikar, Arnold Smith, and Judy Wall for valuable discussions, suggestions, and comments on this manuscript, Marian Carlson, Karen Clemens, Raad Gitan, Aaron Mitchell, Paul Morcos, and Kelly Tatchell for sharing plasmids, and Brian Dalley for generating strains used in this study. This work was supported by National Institutes of Health grant GM-40326 to J.F.C. and by fellowships from the University of Missouri Molecular Biology Program to N.T.R. and L.L.

Footnotes

  • Communicating editor: A. P. Mitchell

  • Received June 5, 1997.
  • Accepted February 10, 1998.

LITERATURE CITED

View Abstract