Glc7p Protein Phosphatase Inhibits Expression of Glutamine-Fructose-6-phosphate Transaminase from GFA1 *

Inhibitor-1 (I-1) is a specific inhibitor of protein phos-phatase-1 (PP1). We assayed the ability of I-1 to inhibit Saccharomyces cerevisiae PP1, Glc7p, in vivo . Glc7p like other PP1 catalytic subunits associates with a variety of noncatalytic subunits, and Glc7p holoenzymes perform distinct physiological roles. Our results show that I-1 inhibits Glc7p holoenzymes that regulate transcription and mitosis, but holoenzymes responsible for meiosis and glycogen metabolism were unaffected. Ad-ditionally, we exploited a genetic screen for mutants that were dependent on I-1 to grow. This scheme can identify processes that are negatively regulated by Glc7p-catalyzed dephosphorylation. In this paper I-1-de-pendent gfa1 mutations were analyzed in detail. GFA1 encodes glutamine-fructose-6-phosphate transaminase. One or more phosphorylated proteins activate GFA1 transcription because the pheromone response and Pkc1p/mitogen-activated protein kinase pathways positively regulate GFA1 transcription. Our findings show that an I-1-sensitive Glc7p holoenzyme reduces GFA1 transcription. Therefore, GFA1 is a member of a growing list of genes that are negatively regulated by Glc7p dephosphorylation. analyses surveyed the ability of I-1 to inhibit a spectrum of Glc7p holoenzymes. Our data indicate that I-1 can inhibit some but not all Glc7p holoenzymes. Finally, we isolated mutants that required I-1 expression to grow. We found that some gfa1 mutants required I-1 or some other means of Glc7p inhibition to grow. A thorough analysis revealed that Glc7p activity inhibits transcription of GFA1 . By inhibiting Glc7p activity, GFA1 transcription increased sufficiently in the isolated gfa1 mutants to overcome their glucosamine auxotrophy trait.

subunit, Glc7p, with alternative noncatalytic subunits. Each Glc7p holoenzyme is likely to have distinctive functions in vivo. PP1 noncatalytic subunits dictate substrate dephosphorylation in part by controlling the subcellular localization of the holoenzyme. The traits of noncatalytic subunit mutations in S. cerevisiae have provided preliminary evidence about the physiological processes regulated by each Glc7p holoenzyme. For example, the Glc7p/Gac1p holoenzyme regulates glycogen metabolism (12), Glc7p/Sds22p governs mitosis (13,14), Glc7p/ Gip1p is involved sporulation (11), and Glc7p/Reg1p modulates glucose repression (8). All Glc7p noncatalytic subunits studied to date act positively because mutations in noncatalytic subunit genes mimic the traits of glc7 loss of function alleles.
To date, no negative regulators for Glc7p have been identified in S. cerevisiae. The yeast genome does not encode recognizable homologs to I-1, DARP-32, or NIPP-1. Moreover, Glc8p, which has sequence similarity to inhibitor-2, acts more like a positive regulator of Glc7p in S. cerevisiae (24,25). However, Glc7p activity must be tightly regulated in yeast because great alterations are detrimental to cell viability. Depletion of Glc7p leads to arrest in the G 1 or M-phase of the cell cycle (7,25,26). In contrast, overexpression of GLC7 causes hyperpolarized growth and cell death (25)(26)(27).
Phosphorylated I-1 is a potent Glc7p inhibitor in vitro (28). Indeed I-1 inhibition is a defining criteria used to classify protein phosphatases as type 1 protein phosphatases (1)(2)(3). Furthermore, I-1 interacts with Glc7p in vivo (29). In this study, we tested the ability of human I-1 to inhibit yeast Glc7p holoenzymes in vivo. Since there are a number of glc7 traits, some of which can be attributed to specific holoenzymes, these analyses surveyed the ability of I-1 to inhibit a spectrum of Glc7p holoenzymes. Our data indicate that I-1 can inhibit some but not all Glc7p holoenzymes. Finally, we isolated mutants that required I-1 expression to grow. We found that some gfa1 mutants required I-1 or some other means of Glc7p inhibition to grow. A thorough analysis revealed that Glc7p activity inhibits transcription of GFA1. By inhibiting Glc7p activity, GFA1 transcription increased sufficiently in the isolated gfa1 mutants to overcome their glucosamine auxotrophy trait.
Plasmid pJZ501 was isolated by complementation of the gfa1-97 mutation in JC1007-97. Derivatives of pJZ501 were made to localize the complementing region of DNA. GFA1 is the only known open reading frame on the gfa1-97 complementing plasmid, pJZ504, that has a 3.7kilobase pair EcoRI fragment in pRS316. Integrative plasmids, p2368 and p2397, contain the YSC1-GFA1 intergenic region (SalI-EcoRI or SalI-BamHI fragments) cloned into YIp356 or YIp366, respectively (85), to make GFA1-lacZ fusions.
Isolation of Mutants That Require I-1 for Growth-Strain CH1305 was transformed with pJZ203 (Table I) and used as the parent strain for the mutant hunt. These transformants were grown to mid-exponential phase (2 ϫ 10 7 cells/ml) in leucine-deficient medium. Cell suspensions were diluted in water and plated on YEP-glucose plates at approximately 2,000 cells per plate. These cells were mutagenized with UV irradiation to approximately 10% survival and then incubated at 30 o C for 5-7 days to allow the development of red color (see "Results"). As expected, most colonies were white or sectored. Of 2 ϫ 10 5 colonies screened, 330 solid red colonies were identified. Each individual red colony was streaked for single colonies on YEP-glucose to ensure that it was uniformly non-sectoring. The 330 non-sectoring strains were further transformed with pJZ205, a pRS316 plasmid that expresses I-1 but lacks ADE3 (Table I). Transformants were selected on uracil-deficient, low adenine medium. We identified four mutants (JC1007-78,-97,-319, and -323) that formed sectoring pJZ205 transformant colonies.
Isolation of Genes That Complement or Suppress I-1 Dependence-Plasmid pJZ501 was isolated from a YCp50 wild-type yeast library (33). It complemented the I-1-dependent mutation in strain JC1007-97 based on the colony-sectoring trait. It was isolated twice from approximately 40,000 Ura ϩ transformants. Additionally, YEp13 and YEp24 yeast genomic libraries (34,35) were used to transform JC1007-97 strain to isolate high copy gfa1-97 suppressors.
Construction of Gene Disruptions-A DNA fragment containing gfa1::URA3 was generated by a polymerase chain reaction using oligonucleotide primers GFA1a (5Ј-AGG AAT AAC TGT ATT TCT TTT CTT  ATA TAG TTA TCA GGG CCT GTG CGG TAT TTC ACA CCG) and GFA1b (5Ј-ATG GAA GTT CAA AAA TTA AAA GCG AAG GAG AAG TGA TTG TAG ATT GTA CTG AGA GTG CAC) and pRS316 template DNA. Each primer contains 20 3Ј bases that flank the pRS316 URA3 gene. The 40 5Ј bases of these primers have sequences proximal to the start and end of the GFA1 open reading frame. The polymerase chain reaction fragment was transformed into JC746-9D by selecting transformants on uracil-deficient medium supplemented with 2 mg of Dglucosamine/ml. Desired transformants were recognized by their glucosamine auxotrophy and confirmed by Southern analysis. The hxk2::URA3 mutation in p2341 was constructed by insertion of a 1-kilobase pair URA3 HindIII fragment into a HindIII site within HXK2 in a subclone of pBW112 (36). Correct integration of hxk2::URA3 in yeast transformants was confirmed by polymerase chain reaction analysis of yeast genomic DNA. Disruptions reg1::LEU2, reg2::URA3, and mig1::URA3 were introduced into yeast by transformation with fragments from pBM1966, BS-reg2::URA3, or pJN41, respectively, as described (37)(38)(39).
Protein and Glycogen Assays-Protein concentrations were determined by dye binding using a bovine serum albumin standard (Bio-Rad). Iodine staining was used for qualitative assessment of glycogen levels (5,40). Dry weights and glycogen were assayed from three cultures derived from independent transformants as described (40).
Enzyme Assays-To assay glutamine-fructose-6-phosphate transaminase, cell extracts in 60 mM KH 2 PO 4 , pH 7.0, 1 mM EDTA, 1 mM dithiothreitol were prepared by vortexing with glass beads at 4 o C. The assay mixture contained 15 mM D-fructose 6-phosphate (Sigma) and 15 mM L-glutamine (Sigma) in extract buffer with 1-2.5 mg/ml protein extract. Incubation was started by addition of extract and terminated after 1 h at 37 o C by heating at 100 o C for 2 min. To measure the background of D-glucosamine 6-phosphate and other interfering material, the reaction was stopped immediately after adding the extract. After cooling and centrifugation, D-glucosamine 6-phosphate was determined in the supernatant by a modified Morgan-Elson procedure (41) using a D-glucosamine standard. D-Glucosamine 6-phosphate yields 85% of the absorbance obtained with D-glucosamine (41), and this correction factor has been applied.
To assay Glc7p protein phosphatase activity, yeast cell extracts were made in 50 mM imidazole, pH 7.5, at 4 o C and used immediately. Phosphorylase phosphatase activity was assayed using 32 P-phosphorylase in the presence of 3 nM okadaic acid as described (27,28).
Permeabilized cells were used to assay ␤-galactosidase as described (42). All enzyme assays were done on cultures derived from three or more independent transformants. Two-tailed t tests were performed on data to determine statistical significance.
LEU2 lacZ 85 Western Immunoblotting-Cells were grown in synthetic media with or without glucosamine. To determine the effect of growth phase on the level of Glc7p, samples from middle logarithmic to stationary phase were harvested, and crude protein extracts were prepared by vortexing with glass beads. Typically, 50 g protein per lane was loaded and separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (10% w/v acrylamide). The resolved proteins were transferred to nitrocellulose membranes and blocked with 5% dehydrated skim milk in phosphate-buffered saline (8 mg of NaCl, 0.2 mg of KCl, 1.44 mg of Na 2 HPO 4 , and 0.24 mg of KH 2 PO 4 per ml, pH 7.4). Overnight incubation at 4 o C with anti-HA (1 g/ml; 12CA5 Roche Molecular Biochemicals), anti-yeast phosphoglycerate kinase (0.5 g/ml, PGK, Molecular Probes), or anti-I-1 (22) primary antibodies was followed by anti-mouse IgG-conjugated peroxidase (Cappel) secondary antibodies. Bands were finally visualized using luminol chemiluminescence reagents (Pierce) and several exposures to XAR5 (Eastman Kodak Co.) film.
Northern Analysis-Total yeast RNA was isolated, fractionated on formaldehyde-agarose gels, and hybridized with 32 P-labeled probes as described (31). The GLC7 probe was the 486-base pair EcoRI-SalI fragment from pKC886. The ACT1 probe was from the plasmid pYact1 (43).

I-1 Inhibits a Subset of Glc7p Holoenzymes in Vivo-Human
I-1 was expressed in yeast using the strong, constitutive GPD1 promoter in various low and high copy vectors. I-1 inhibits yeast PP1 in vivo (28). Surprisingly, we failed to find any growth impairments, which would be expected if I-1 expression severely inhibited Glc7p activity. Anti-I-1 immunoblot analysis of I-1 yeast transformant proteins confirmed expression of the human I-1 protein (data not shown). Therefore, we compared the traits of strains transformed with I-1 expressing and control plasmids to learn whether any Glc7p holoenzymes were modulated by I-1 in vivo.
The Gac1p/Glc7p holoenzyme promotes glycogen synthesis in yeast (12). Null mutations in GAC1 or missense mutations in GLC7 greatly reduce glycogen accumulation (5, 44 -46). Therefore, glycogen accumulation provides an assay of the Gac1p/ Glc7p holoenzyme in vivo activity. Yeast transformed with plasmids that express I-1 accumulated 20 -26% less glycogen than control transformants (Fig. 1A). However, this reduction was small compared with the greater than 95% reduction caused by some glc7 or gac1 mutations (12,46) and was not statistically significant (p Ն 0.18). These data indicate that the Gac1p/Glc7p holoenzyme activity is not significantly inhibited by I-1 in vivo.
Glc7p regulates two essential functions in the mitotic cell cycle as follows: the G 1 -to S-phase transition and chromosome segregation (7,24,26). The former function is poorly understood and has only been observed in pedigree analysis experiments (24). However, chromosome segregation appears to require Glc7p activity to dephosphorylate the kinetochore protein Ndc10p (47). Ndc10p phosphorylation by protein kinase, Ipl1p, inhibits mitotic spindle attachment (48). Temperature-sensitive lethal mutations in ipl1 cause a ploidy increase that can be suppressed by inhibition of Glc7p activity (6). We found that I-1 expression suppressed the temperature-sensitive trait of ipl1-1 (Fig. 1C). The Sds22p/Glc7p is most likely the holoenzyme that modulates chromosome segregation because sds22 mutants arrest in the cell cycle in preanaphase (13,14). Therefore, the ipl1-1 suppression by I-1 suggests that the Sds22p/Glc7p holoenzyme is inhibited by I-1 in vivo.
Glc7p holoenzymes participate in at least three distinct steps during sporulation (46). The first two steps exploit unknown Glc7p holoenzymes. First, Glc7p promotes transcription of IME1, encoding the main transcriptional inducer of early meiotic genes. Next, an undefined step after IME1 expression, but before premeiotic S-phase, requires Glc7p activity (46). 2 Finally, packaging of meiotic spores appears to require the Gip1p/ Glc7p holoenzyme (11,46). Diploid JC746 transformed with I-1-expressing or control plasmids was transferred to solid sporulation medium, and the frequency of sporulated cells was counted after 3 days. Remarkably, I-1 expression failed to alter significantly the sporulation frequency ( Fig. 1B). High copy plasmids that express truncated GLC7 genes behave like dominant-negatives because they can suppress ipl1 and gcn2 protein kinase mutations (6,9). However, as far as the sporulation functions of Glc7p are concerned, high copy truncated GLC7 had no detectable effect (Fig. 1B). Additionally, sporulation frequency was unaffected by high copy GLC8. Sporulation of yeast diploids normally yields four haploid spores within a tetrad. We previously found that some glc7 alleles yielded a high frequency of dyads (46). Dyads are produced instead of tetrads when one meiotic division fails to occur or spore packaging is incomplete. None of the transformants in Fig. 1B produced more than 2% dyads. Together these data show that neither truncated GLC7, I-1, nor high copy GLC8 inhibit Glc7p sporulation functions.
Some GFA1 Mutants Require I-1 Expression to Grow-Glc7p activity must be tightly controlled because it is essential for growth yet can inhibit growth if overexpressed (6,24,26,49). Since the results of the previous section showed that I-1 inhibits a subset of Glc7p holoenzymes, we took advantage of ectopic I-1 expression to isolate mutants that require I-1 expression for viability. Such mutants would illustrate novel aspects of Glc7p

FIG. 1. I-1 expression inhibits some Glc7p holoenzymes in vivo.
A, CH1305 transformants were grown for 48 h in uracil-deficient minimal medium. Glycogen was quantitated as described (40). A two-tailed t test showed that pJZ205 and pJZ206 transformant glycogen levels differed insignificantly from pRS426 control transformants (p ϭ 0.3 and 0.18, respectively). B, three independent JC746 transformants were sporulated for 3 days at 30°C on SPOR plates. For each of the three sporulated patches, at least 200 cells were counted, and sporulation frequency was calculated from the numbers of asci. C, the ipl1-1 strains JC1085-18B (top and left) and JC1117-7D (bottom and right) transformed with pJZ204 (2-m TRP1 I-1) or its parent vector, pG-3 (2-m TRP1), were streaked on tryptophan-deficient minimal medium and incubated at the indicated temperatures for 3 days.
function. We exploited a synthetic lethal screen in which mutants that require I-1 expression to grow exhibit a solid red colony color, whereas wild-type cells yield colonies with red and white sectors (see "Experimental Procedures"). This scheme identified four mutants that were dependent on I-1 for viability. Diploids heterozygous for these mutations did not require I-1 to grow. Therefore, each I-1-dependent mutation was recessive. Additionally, tetrad analysis of diploids that were heterozygous for these mutations displayed 2:2 segregation for growth rate, which indicates that the mutants harbor single, nuclear mutations that confer I-1 dependence for growth. Moreover, complementation analysis indicated three complementation groups. The remainder of this paper concerns the characterization of the two allelic mutations in strains JC1007-97 and JC1007-319. Because traits of the JC1007-97 strain were easier to score, our further analyses used JC1007-97 exclusively.
The I-1-dependent mutation in JC1007-97 is allelic to GFA1. We isolated plasmids from a wild-type YCp50 library that complement the I-1-dependent mutation in JC1007-97/pJZ203. Transformants with complementing plasmids were recognized by their sectoring trait. Two independent plasmids that complement the JC1007-97 I-1-dependent mutation contained overlapping DNA from chromosome VII. Further analysis localized the complementing gene to GFA1, which encodes glutamine-fructose-6-phosphate transaminase (EC 2.6.1.16) (50). A wild-type URA3 gene was integrated at the GFA1 locus in JC1053-3D by integrative transformation using plasmid p2368. When this transformant was crossed to JC1007-97/ pJZ203, tetrad analysis of the sporulated diploid revealed no recombination between the integrated URA3 and the I-1-dependent mutation. Therefore, by these complementation and linkage criteria, the I-1-dependent mutations in JC1007-97 and JC1007-319 are alleles of GFA1 and will be called gfa1-97 and gfa1-319, respectively.
The Gfa1-97 Mutation Reduces but Does Not Eliminate Glutamine-Fructose-6-phosphate Transaminase Activity-Glutamine-fructose-6-phosphate transaminase catalyzes the formation of glucosamine 6-phosphate, which is the first and ratelimiting step for protein glycosylation and chitin synthesis (51). We confirmed that disruption of the GFA1 gene generates a strain that is auxotrophic for glucosamine (50) by construction of a gfa1::URA3 mutant. We assayed the glutamine-fructose-6-phosphate transaminase enzyme activity of crude extracts from wild-type, gfa1-97, and gfa1::URA3 mutant strains. As shown in Table II, the gfa1-97 mutant had approximately 35% of the wild-type specific activity, whereas activity in a gfa1::URA3 null mutant was undetectable. Despite the reduction in enzyme activity in gfa1-97 (JC1007-97) cells, when transformed with an I-1-expressing plasmid (JC1007-97/ pJZ203), they grew at the same rate as wild-type on glucosamine-free minimal medium. This paradox was resolved by experimental results described later. Additionally, the GFA1 genotype significantly influenced the intracellular glucosamine (Table II); however, the mechanism is unknown.
Since glucosamine synthesis in the gfa1-97 mutant was reduced, we tested if extracellular glucosamine could suppress the I-1-dependent trait of gfa1-97. To test this possibility, JC1007-97/pJZ203 cells were streaked on plates that contained 5 mg of glucosamine/ml. On this medium, JC1007-97/pJZ203 mutant cells readily lost the plasmid pJZ203 and became white or sectored (Fig. 2A). Such gfa1-97 cells without a source of I-1 were dependent on glucosamine to grow. This result indicates that the reduced glucosamine synthesis in JC1007-97 is responsible for its I-1 dependence, and extracellular glucosamine could suppress the I-1 dependence of the gfa1-97 mutation.
Because gfa1::URA3 null mutants had an undetectable glutamine-fructose-6-phosphate transaminase activity, we reasoned that they might also be I-1-dependent. However, since gfa1::URA3 strain can only grow in the presence of glucosamine, the I-1 dependence could not be tested. Expressing I-1 in gfa1::URA3 mutant cells did not relieve the glucosamine requirement for growth.
Phenotypic Characterization of Gfa1-97-Suppression of the I-1 dependence by growth in medium containing 5 mg of glucosamine/ml provided an easy way to obtain JC1007-97 plasmid-free cells to study traits of gfa1-97 without I-1 present. These plasmid-free JC1007-97 cells produced completely white colonies of Leu Ϫ cells. To distinguish the plasmid-bearing and plasmid-free cells, they will be labeled explicitly as "JC1007-97/pJZ203" and "JC1007-97," respectively.
The gfa1-97 mutation caused temperature-sensitive growth on media without glucosamine. JC1007-97 cells grew as well as wild-type CH1305 cells at temperatures up to 25 o C. However, at 30 o C and higher temperatures, only 0.1% of the cells formed visible colonies after 2 days, the majority of cells either died at the single-cell stage or divided several times to form microcolonies (Fig. 3A). No specific cell cycle arrest was observed for these cells. Cells stopped growing and produced buds of differing sizes. At 37 o C, the growth defect trait was exacerbated; no colonies appeared after 5 days of incubation. The majority of cells did not divide and a large population lysed or had long, slender buds when examined microscopically (Fig. 3B).
Growth of JC1007-97 in liquid cultures showed three impairments compared with wild-type cells (Fig. 3C). First, the JC1007-97 mutant cells were significantly delayed at entering the exponential growth phase. Second, JC1007-97 cells divided at 80% the growth rate of wild-type cells during exponential phase. Third, the maximum cell density of the JC1007-97 mutant cells was 70% of the wild-type. Note that the rich YEPglucose medium used for liquid culture growth initially contains some glucosamine from the yeast extract ingredient. Growth on fructose was similar to that observed on glucose. However, growth on galactose was unimpaired. In fact, galactose suppressed the I-1 dependence trait of gfa1-97 and partially suppressed the temperature-sensitive trait (Fig. 2).
Glucosamine is important for chitin synthesis and protein glycosylation, which are required for cell wall integrity (51). Mutants with defects in cell wall integrity lyse at restrictive temperatures but can be rescued in media with greater osmolarity. Therefore, we tested if the addition of sorbitol or NaCl to the medium could rescue the growth defect of the JC1007-97 strain. JC1007-97 grew well at 30 o C in glucosamine-free media supplemented with 0.8 M sorbitol or 0.25 M NaCl (data not shown). Sorbitol suppressed the I-1 dependence and temperature-sensitivity traits of gfa1-97 (Fig. 2). NaCl suppressed a Cells with the indicated genotypes (congenic with CH1305) were grown in YEP-glucose media containing 2 mg of glucosamine/ml. Glutamine-fructose-6-phosphate transaminase assays were done as described under "Experimental Procedures." The glutamine-fructose-6phosphate transaminase-specific activity was determined as nanomoles of glucosamine synthesized per h/mg of crude extract protein. The data shown are the average of three independent cultures. b Intracellular glucosamine concentration was expressed as nanomoles of glucosamine per mg of crude protein in the extract. these traits identically to sorbitol (data not shown). This suppression was consistent with gfa1-97 causing a defect in cell wall integrity.
Inhibition of Glc7p Activity Suppresses Traits of Gfa1-97-In addition to the known function of I-1 to inhibit PP1 activity, it is possible that gfa1-97 results in a reliance on a different I-1 function. Therefore, we tested whether inhibition of Glc7p by some other means could suppress gfa1-97. High copy, truncated GLC7 genes (GLC7⌬) behave as dominant-negatives (6,9). Mutations in gcn2 and ipl1 protein kinase genes are suppressed by GLC7⌬ because these protein kinases phosphorylate known or suspected Glc7p substrates. Therefore, we tested whether the inhibition of Glc7p activity via GLC7⌬ could substitute for I-1. When JC1007-97 was transformed with p2168, which carries GLC7⌬ at high copy, the I-1 dependence was relieved (Fig. 4). Additionally, the slow growth of JC1007-97 at 30 o C was suppressed by GLC7⌬ (Fig. 3A).
I-1 inhibits PP1 activity only when threonine 35 of I-1 is phosphorylated (20,22). A mutant form of I-1 in which alanine is substituted for threonine 35 cannot inhibit PP1 activity in vitro because it cannot be phosphorylated (20,22). Wild-type I-1 interacts with Glc7p in the two-hybrid assay, whereas the I-1-T35A mutant does not (29). In addition, I-1 suppresses the ipl1 temperature-sensitive trait, whereas I-1-T35A does not ( Fig. 1 and data not shown). We tested whether I-1-T35A could suppress traits of gfa1-97. We found that expression of I-1-T35A could neither suppress the I-1 dependence of gfa1-97 as scored by the colony sectoring trait (Fig. 4) nor the temperature-sensitive growth at 30 and 37 o C on glucosamine-free media (Fig. 3A). Taken together, these genetic data show that gfa1-97 was isolated as a I-1-dependent mutation because Glc7p inhibition was required for glucosamine-free growth.
Glucosamine Does Not Alter Glc7p Activity or Protein Levels-Why do gfa1-97 cells require Glc7p inhibition for growth? The first possibility we considered was that reduced glucosamine biosynthesis in some way increased Glc7p activity, which is toxic (24,26,27). This explanation would predict that extracellular glucosamine reduces Glc7p activity because glucosa- FIG. 2. Suppression of gfa1-97 traits by growth media. A, JC1007-97 was streaked on low adenine synthetic complete medium with the indicated carbon sources, glucosamine (5 mg/ml), or sorbitol (0.8 M). Plates were incubated at 30°C 4 days before photography. Great retention of pJZ203 produces solid red colonies, which are dark in this figure. In contrast, white colonies or sectors derive from cells that lost pJZ203. B, glucosamine or sorbitol suppressed the gfa1-97 temperature sensitivity. Ten-fold serial dilutions of CH1305 (GFA1) and JC1007-97 (gfa1-97) were spotted on YEP-glucose, YEP-glucose with 5 mg glucosamine/ml, YEP-glucose with 0.8 M sorbitol, YEP-galactose, or YEP-fructose. Plates were incubated at 37°C for 4 days before photography. FIG. 3. Demonstration of gfa1-97 traits without I-1. A, temperature sensitivity caused by gfa1-97 was apparent from growth of CH1305 (GFA1) and JC1007-97 (gfa1-97) transformants at different temperatures. The genotypes of strains and plasmids are shown in the legend. Plasmids in JC1007-97 transformants are (starting at top, going clockwise) as follows: pJZ504, pJZ208, p2168, pJZ206, pJZ205, and pRS426, respectively. Transformants were streaked on YEP-glucose plates and incubated for three days at the indicated temperatures. B, the gfa1-97 mutation caused arrest at all points in the cell cycle on glucosaminefree medium. CH1305 or JC1007-97 cells grown at 37°C for 24 h in YEP-glucose were fixed and photomicrographed. C, the gfa1-97 cells exhibited a pre-exponential phase growth lag, retarded exponential growth rate, and reduced growth yield compared with wild-type cells in liquid cultures. Growth of liquid 30°C YEP-glucose (without added glucosamine) cultures of CH1305 (triangles, dashed lines) and JC1007-97 (circles, solid lines) were monitored by 600 nm light scattering. Cell counts by hemocytometer showed similar results. mine relieves the I-1-dependent trait of gfa1-97. Since glycogen synthase is a Glc7p substrate and its activity is dependent on phosphorylation (5), we analyzed glycogen accumulation in wild-type and gfa1-97 mutant cells. A slightly darker iodine staining was observed with gfa1-97 cells, indicating an increase in glycogen accumulation and thus potentially elevated Glc7p activity. Quantitative glycogen assays confirmed that gfa1-97 cells accumulated 20% more glycogen than wild type (data not shown). Although increased glycogen accumulation can be caused by increases in Glc7p activity, many other factors can influence glycogen levels (5). Glc7p activity was specifically assayed in yeast crude extracts by [ 32 P]phosphorylase dephosphorylation in the presence of okadaic acid (11,27,28). By this assay we did not find significant differences of Glc7p-specific activity between isogenic gfa1-97, gfa1::URA3, and wild-type cells (data not shown). Therefore, despite increases of glycogen that might be caused by increases of Glc7p activity, we were unable to document a significant change in Glc7p-specific activity by in vitro assays.
We also examined Glc7p protein levels using immunoblot analysis. Glc7p protein levels were monitored using an HA-GLC7 fusion gene on a centromeric plasmid. This gene complements all GLC7 traits and allowed the relative Glc7p protein levels to be assayed using anti-HA antibody. No consistent differences of Glc7p levels were found between wild-type and gfa1-97 cells. Furthermore, extracellular glucosamine had no effect on the levels of Glc7p in wild-type or gfa1-97 cells (data not shown). Moreover, we found only minor differences of GLC7 mRNA levels in wild-type and gfa1-97 cells harvested at equivalent time points (data not shown). The GLC7 mRNA abundance increased in latter growth phase samples for both cultures. This induction in late exponential phase has been noted before (4). These experimental results demonstrate that gfa1-97 cells do not have increased Glc7p activity or protein levels.
High Copy Suppressors of Gfa1-97-To help elucidate the reason gfa1-97 cells require Glc7p inhibition for growth, we sought high copy suppressors of gfa1-97. High copy libraries were screened for plasmids that suppressed the I-1 dependence of JC1007-97/pJZ203 by the colony-sectoring assay. Because Glc7p inhibition would increase the phosphorylation of Glc7p substrates, potential high copy suppressors were expected to be protein kinase genes. Therefore, 24 protein kinase genes in 2-m vectors were also specifically tested. By these routes, we found that RHO1, PKC1, BCK1, MKK1, MKK2, MPK1, SNF1, and PBS2 could suppress gfa1-97. Except for PBS2 and SNF1, these genes encode components of the Rho1/Pkc1 MAP kinase pathway (52)(53)(54)(55). This pathway maintains cell wall integrity (56). Except for RHO1, all these suppressing genes encode protein kinases. Glucosamine is used in the synthesis of glycoproteins and chitin, which are both components of the cell wall. The isolation of these high copy suppressors suggests that the stress in cell wall biosynthesis caused by gfa1-97 can be relieved by hyperactivity of the Rho1/Pkc1 pathway. Moreover, these suppression results implicate Glc7p in antagonizing the activity of this pathway.
The PPZ protein phosphatases encoded by PPZ1 and PPZ2 antagonize some aspects of the Rho1/Pkc1 pathway. Because the catalytic domains of PPZ proteins are very homologous to PP1 enzymes, we considered I-1 might inhibit activity of these PP1 homologues. Therefore, we tested whether I-1 expression recapitulated the increased salt or caffeine resistance traits of ppz1 (57). We found that wild-type yeast transformed with plasmids that express I-1 or I-1-T35A were equally sensitive to caffeine and lithium (data not shown). Therefore, we have no evidence that I-1 inhibits Ppz1p or Ppz2p in vivo.
Glc7p and the Pkc1p Pathway Regulate GFA1 Transcription-The pheromone-responsive transcription factor Ste12p and its coactivator Mcm1p have binding sites in the GFA1 promoter. Both Ste12p and Mcm1p are multiply phosphorylated proteins (58,59). Because the transcriptional activity of Ste12p and Mcm1p is regulated by their phosphorylation, we explored whether GFA1 transcription was altered in strains with inhibited Glc7p activity. The E. coli lacZ gene was fused to GFA1 at codon 73 and integrated at the GFA1 locus so that GFA1 transcription could be quantitated by ␤-galactosidase assays (see "Experimental Procedures"). Consistent with pheromone induction of GFA1 mRNA (50), ␤-galactosidase activity from GFA1-lacZ was induced ϳ3-fold by pheromone (Fig. 5A). High copy I-1 and GLC7⌬ increased GFA1-lacZ ␤-galactosidase activity 29 and 67%, respectively (Fig. 5B). Although these increases of GFA1 transcription were small, both were significant (p Ͻ 0.005 and p Ͻ 0.0005, respectively). Enzyme assays showed that I-1 increased glutamine-fructose-6-phosphate transaminase activity from gfa1-97 but not wild-type cells (data not shown). Moreover, elimination of the Glc7p/Reg1p holoenzyme by reg1::LEU2 increased GFA1-lacZ ␤-galactosidase activity 78% (Fig. 5B). These data reveal that Glc7p activity inhibits GFA1 transcription and that increased GFA1 transcription can overcome the gfa1-97 glucosamine auxotrophy.
To learn if the high copy suppressors we isolated also suppress gfa1-97 by means of increasing GFA1 transcription, GFA1-lacZ ␤-galactosidase activity was assayed for cells transformed with suppressor gene plasmids. These assays revealed that all high copy suppressor genes increased GFA1 transcription to various degrees ( Fig. 5B and data not shown). Induction of GFA1-lacZ by MPK1/SLT2, the MAP kinase of the cell wall integrity pathway, reveals that suppression by RHO1, PKC1, BCK1, MKK1, and MKK2 can be explained by their stimulation of Mpk1p.
Snf1p and Reg1p regulate the transcription of glucose-repressed genes (60). Our results showed that overexpression of SNF1 suppressed gfa1-97 by increasing GFA1 transcription. Inactivation of REG1 also increased GFA1-lacZ expression JC1007-97 transformed with various URA3 plasmids was grown for 6 days on low adenine uracil-deficient medium before photography. White or sectored colonies indicate loss of pJZ203 and suppression of the I-1 dependence trait. The smaller, white colonies visible in the panels are respiratory-deficient clones, which do not turn red, despite the retention of pJZ203. The panels are labeled with the plasmid genotypes. The plasmids used were pRS316, p2168, pJZ205, pJZ208, pJZ206, and pJZ209. (Fig. 5B). Therefore, we were interested in whether inhibition of Glc7p by I-1 or GLC7⌬ could increase transcription of other genes, such as HXT4, regulated by Snf1p and Reg1p (37). We found that either I-1 or GLC7⌬ induced HXT4-lacZ expression (Fig. 5C). Regulation of HXT4 also uses signals derived from HXK2-encoded hexokinase, and dephosphorylated Mig1p represses HXT4 (37). Gene disruptions of HXK2, MIG1, REG1, or REG2 were made in a diploid heterozygous for gfa1-97. These strains were sporulated, tetrads dissected, and spores analyzed to explore whether null mutations in these genes could suppress traits of gfa1-97. The reg1::LEU2 gfa1-97 double mutant spores grew poorly on all media, so potential gfa1-97 suppression by reg1 could not be scored. However, all of the other null mutations yielded two slow growing gfa1-97 spores per tetrad, showing that they could not suppress the gfa1-97 slow growth trait. DISCUSSION PP1 holoenzymes achieve diversity by the variety of noncatalytic subunits that can associate with catalytic subunits. The single PP1 catalytic subunit, encoded by GLC7 in S. cerevisiae, is known to regulate a variety of physiological processes. Glc7p binds to at least nine different noncatalytic subunits (10). Mammalian PP1 enzymes are also regulated by inhibitor proteins. Inhibitor I-1 was known to inhibit Glc7p activity in vitro (28). In this work we show I-1 also inhibits Glc7p in vivo, although not all holoenzymes appear sensitive to I-1.
Separate domains of I-1 participate in PP1 binding and inhibition (61,62). The amino acid sequence KIQF in I-1 is similar to a sequence found in many PP1 noncatalytic subunits, and it is required for I-1 inhibition function (23,61,63). This sequence is thought to be an interface for PP1 binding. Because Reg1p and Gac1p noncatalytic subunits contain a sequence similar to KIQF (63), I-1 is unlikely to bind and inhibit Glc7p while Glc7p is bound to either Reg1p or Gac1p. However, depending on relative affinities and protein levels, I-1 could potentially displace Reg1p or Gac1p from Glc7p, thereby reducing the catalytic activity of those holoenzymes. Our data are consistent with in vivo inhibition of the Reg1p/Glc7p holoenzyme by I-1 because expression of I-1 in yeast mimics the reg1 induction of GFA1 and HXT4 transcription (Fig. 5, and see Ref. 37). In contrast, Gac1p/Glc7p holoenzyme inhibition by I-1 is, at best, minor ( Fig. 1). Because Sds22p lacks a KIQF-homologous domain (63), the Sds22p/Glc7p holoenzyme could potentially bind I-1 without dissociation. Regardless of the mechanism, I-1 suppression of ipl1 temperature sensitivity argues that I-1 inhibits the Sds22p/Glc7p holoenzyme in vivo (Fig. 1).
I-1 residues around Thr-35 are thought to occupy the PP1active site and competitively inhibit PP1 substrate access (64). PP1 inhibition by I-1 requires I-1 Thr-35 phosphorylation (20). In mammalian cells cAMP-dependent protein kinase phosphorylates I-1 in response to extracellular hormones (1-3). However, yeast cAMP-dependent protein kinase is responsive to nutritional signals via the RAS GTPase (5,65). Nitrogen starvation and the absence of glucose promote sporulation of diploid yeast cells. These conditions apparently reduce cAMP-dependent protein kinase activity because hyperactive cAMPdependent protein kinase inhibits sporulation (66). We have previously found that Glc7p activity is required for at least three points in the meiotic sporulation program (46). In this work we failed to detect I-1 inhibition of Glc7p during sporulation (Fig. 1). Although this could be a consequence of I-1 hypophosphorylation, the observation that high copy GLC7⌬ or GLC8 transformants also sporulated normally indicates that the unknown Glc7p holoenzymes that control meiosis and sporulation are not subject to the same inhibitors as other holoenzymes.
FIG. 5. Attenuation of Glc7p activity increases GFA1 and HXT4 transcription. A, CH1305/p2368 (GFA1-lacZ) cells were grown to midexponential phase (A 600 ϭ 0.7-1.0) in YEP-glucose. The culture was split, and 10 g of ␤-factor per ml was added to one culture (triangles) at time 0, and cell samples were assayed for ␤-galactosidase. The control culture (circles) had no ␤-factor added. B, the GFA1-lacZ ␤-galactosidase activity was assayed for exponential phase cultures. The left three data are from CH1305/p2368 cells transformed with pTVS30, pJZ203, or p2155 grown in leucine-deficient medium. The center two bars show data from CH1305/p2368 or CH1305 reg1::LEU2/p2368 cells grown in YEP-glucose. The right three data are from CH1305/p2397 cells transformed with YEp352, YEp352-PKC1, or YEp352-MPK1 grown in uracil-deficient medium. C, CH1305/pBM2800 transformed with pTSV30, pJZ203, or p2155 were grown in leucine-, uracil-deficient medium and assayed for HXT4-lacZ ␤-galactosidase activity. B and C, the relevant genotype of the plasmids is shown within the bar. Assays were performed on three or more cultures derived from independent transformants.
We hypothesize that Glc7p is naturally subject to inhibition. Three observations prompt this hypothesis. First, intermediate Glc7p activity is required for normal cellular growth since substantial increases or decreases of the wild-type activity kills yeast cells. Second, several PP1 inhibitors exist in mammalian cells, and if they were evolutionarily ancient would also be found in the unicellular eukaryote, S. cerevisiae. However, only the I-2 homologue Glc8p has been found in yeast, and it appears to predominantly activate Glc7p activity (24,25). Our data demonstrate no Glc8p influence on Glc7p holoenzymes that regulate sporulation or GFA1 transcription, respectively. Finally, assays of yeast PP1 activity are exquisitely sensitive to crude extract dilution. 3 This behavior is similar to mammalian PP1 assays (1-3) and is evidence of PP1 inhibitors present in the crude extract. The prospect of natural Glc7p inhibitors spurred our hunt for mutants that depend on I-1 expression for viability.
Two independent mutants, which required I-1 expression for viability, were isolated and discovered to contain mutations in GFA1, which encodes glutamine-fructose-6-phosphate transaminase, the glucosamine biosynthetic enzyme. These mutants could not grow without glucosamine if I-1-T35A, which cannot inhibit PP1, was expressed. Moreover, inhibition of in vivo Glc7p activity by GLC7⌬ revealed that the gfa1 mutants need Glc7p inhibition to grow without glucosamine. Neither gfa1-97 nor extracellular glucosamine influenced Glc7p activity, Glc7p protein, or GLC7 RNA levels (data not shown). Therefore, gfa1-97 mutant cells have a condition where normal Glc7p activity forbids glucosamine-free growth. We considered that Glc7p regulated glutamine-fructose-6-phosphate transaminase by dephosphorylation because this enzyme from other fungal species is regulated by phosphorylation (51,67). However, changes in Glc7p activity did not significantly affect enzyme activity in wild-type cells (data not shown). We demonstrated that Glc7p inhibition induces GFA1 transcription (Fig. 5) and that increased GFA1 transcription restores gfa1-97 glutaminefructose-6-phosphate transaminase to wild-type levels (data not shown). Therefore, gfa1 mutants were isolated as I-1-dependent mutants because they required GFA1 transcription induction that results from Glc7p inhibition.
The mating pheromone response pathway induces GFA1 transcription (50). Pheromone exposure induces haploid cells to initiate polarized cell growth toward the pheromone source. Since glucosamine and its products are required for cell wall growth, GFA1 induction under these conditions is understandable. Ste12p and Mcm1p activate transcription of genes in response to pheromones in both haploid cell types (68), and DNA 5Ј to GFA1 contains binding sites for these two proteins (50,59). However, glucosamine biosynthesis is needed at all stages of growth, not merely in response to pheromones. In fact, Mcm1p coactivates transcription of many genes essential for cell cycle progression, cell wall, or cell membrane biosynthesis, and it may participate in the M/G 1 expression of GFA1 (59, 69 -71).
We showed that high copy RHO1, PKC1, and other protein kinase genes in the Pkc1p MAP kinase cascade could suppress the glucosamine auxotrophy of gfa1-97. This cascade activates cell wall biosynthetic genes during periods of polarized cell growth (56,72). The Pkc1p MAP kinase cascade suppression of gfa1-97 appears to be via induction of GFA1 transcription because even overexpressing Mpk1p, the terminal kinase in the cascade could induce GFA1-lacZ (Fig. 5B). Mpk1p phosphorylates the transcriptional factor, Rlm1p, which partially mediates Mpk1p function (73). However, the region 5Ј to GFA1 lacks recognizable Rlm1p-binding sites. Therefore, the activation of GFA1 may use an activation mechanism similar to FKS2, which is also responsive to Mpk1p activation via a cis-DNA region that lacks Rlm1p-binding sites (72). Surprisingly, high copy PBS2 also suppressed gfa1-97. Pbs2p is a protein kinase in the osmotic stress responsive Hog MAP kinase cascade, which is distinct from the Pkc1p MAP kinase cascade (56). PBS2 may suppress gfa1-97 by its documented activation of Mcm1p (74,75).
Glucosamine levels should be regulated in response to growth rate and nutrient levels to ensure an appropriate supply of this sugar for cell wall and glycoprotein biosynthesis. Evidence of a mechanism to modulate intracellular glucosamine levels was revealed by the influence that gfa1 genotype had on these levels (Table II). Since I-1 and Glc7p activity regulate the glucose transporter encoded by HXT4 (Fig. 5C), we hypothesize that they also regulate glucosamine transporters in a manner that senses glucosamine biosynthetic capacity. Growth on galactose allows derepression of glucose-repressed genes and can suppress the I-1-dependent trait of gfa1-97 (Fig.  2). Similarly, reg1 mutations or SNF1 overexpression also suppress gfa1-97 (this work) and derepress glucose-repressible genes (60). Glucose repression of GFA1 could explain these observations; however, this possibility has not been tested. If GFA1 transcription were glucose-repressible, it would be novel because mig1 mutations do not increase GFA1 transcription nor suppress gfa1-97 I-1 dependence. Mig1p is responsible for glucose repression of other genes (60). Furthermore, gfa1-97 and GFA1 ϩ cells grown on rich medium showed a greater difference in glutamine-fructose-6-phosphate transaminase activity than cells grown in minimal medium (data not shown). This finding vividly demonstrates the influence that extracellular nutrient levels and growth rate have on Gfa1p regulation.
Glc7p regulates transcription of several genes in S. cerevisiae. Evidence has implicated the Glc7p/Reg1p holoenzyme in regulating transcription of SUC2, INO1, ADH2, HXT2, HXT4, and IME1 (8, 37, 46, 76 -78) and Glc7p/Gac1p in regulating CUP1 (79). In some cases a candidate Glc7p phosphoprotein substrate is genetically implicated, but none have been proven by rigorous biochemical analyses. The results in this paper add GFA1 to the list of genes regulated by Glc7p/Reg1p. Similar to INO1 regulation, GFA1 transcription was increased by elevated Snf1p activity but not by mig1 or hxk2 mutations (76). Curiously, only Mcm1p and Ste12p are known to regulate GFA1 transcription, and these regulators are distinct from the proteins known to regulate INO1 (76,77). Therefore, to explain the role of Glc7p/Reg1p holoenzyme in transcriptional regulation requires postulating that it dephosphorylates a number of different proteins or that it dephosphorylates proteins that function at many promoters such as TATA-binding proteins or components of RNA polymerase II holoenzyme. One of the other I-1-dependent mutations we isolated was in SPT6, which encodes a protein that influences chromatin structure and RNA polymerase elongation (80,81). Further work will be required to define the Glc7p substrates responsible for these transcriptional effects.