Genetic Interactions Between GLC7, PPZ1 and PPZ2 in Saccharomyces cerevisiae

Corresponding author: Kelly Tatchell, Department of Biochemistry and Molecular Biology, Louisiana State University Medical Ctr., 1501 Kings Hwy., Shreveport, LA 71130., ktatch{at}mail.sh.lsumc.edu (E-mail)

Communicating editor: M. CARLSON

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

GLC7 encodes an essential serine/threonine protein type I phosphatase in Saccharomyces cerevisiae. Three other phosphatases (Ppz1p, Ppz2p, and Sal6p) share >59% identity in their catalytic region with Glc7p. ppz1 ppz2 null mutants have no apparent growth defect on rich media. However, null alleles of PPZ1 and PPZ2, in combination with mutant alleles of GLC7, confer a range of growth defects varying from slow growth to lethality. These results indicate that Glc7p, Ppz1p, and Ppz2p may have overlapping functions. To determine if this overlap extends to interaction with targeting subunits, Glc7p-binding proteins were tested for interaction in the two-hybrid system with the functional catalytic domain of Ppz1p. Ppz1p interacts strongly with a number of Glc7p regulatory subunits, including Glc8p, a protein that shares homology with mammalian PP1 inhibitor I2. Genetic data suggest that Glc8p positively affects both Glc7p and Ppz1p functions. Together our data suggest that Ppz1p and Ppz2p may have overlapping functions with Glc7p and that all three phosphatases may act through common regulatory proteins.

THE PPP family of serine/threonine protein phosphatases includes the well-studied enzymes PP1, PP2A, and calcineurin/PP2B (BARFORD 1996 ). The holoenzymes of these consist of a conserved catalytic subunit and one or more regulatory subunits. For PP1 (EGLOFF et al. 1997 ) and PP2A (COHEN 1989 ; FERRIGNO et al. 1993 ; DEPAOLI-ROACH et al. 1994 ), multiple unique regulatory subunits bind the catalytic subunit and either target the enzyme to its site of activity or otherwise regulate its activity. These regulatory subunits are thought to be necessary for the enzymes to carry out their wide range of physiological activities. In addition to PP1, PP2A, and PP2B, novel and generally less-well-studied phosphatases round out the gene family (COHEN 1997 ). Many of these novel phosphatases are very similar to either PP1 or PP2A within their catalytic domains but the extent to which targeting subunits contribute to their regulation is not well understood.

Saccharomyces cerevisiae contains 12 members of the PPP family of phosphatases, including 2 isoforms of PP2A, 2 isoforms of PP2B, 1 isoform of PP1, and 7 additional enzymes. These latter include three enzymes most similar to PP2A (Sit4p, Pph3p, and Ppg1p); three similar to PP1 (Ppq1p/Sal6p, Ppz1p, and Ppz2p); and Ppt1, a more distantly related member of the family (STARK 1996 ). The catalytic domains of the PP1 family members are >59% identical but Ppz1p, Ppz2p, and Sal6p have sequences upstream of the catalytic domain that are not found in Glc7p. The overall structures of these enzymes are diagrammed in Fig 1. Glc7p is the most extensively characterized member of this group. It is >70% identical to mammalian PP1 isoforms, for which crystal structures are known (EGLOFF et al. 1995 , EGLOFF et al. 1997 ; GOLDBERG et al. 1995 ), and participates in such diverse processes as cell cycle regulation (FRANCISCO et al. 1994 ; HISAMOTO et al. 1994 ; BLACK et al. 1995 ; MACKELVIE et al. 1995 ; BLOECHER and TATCHELL 1999 ; SASSOON et al. 1999 ), glycogen metabolism (FENG et al. 1991 ; FRANCOIS et al. 1992 ; CANNON et al. 1994 ; STUART et al. 1994 ), sporulation (CANNON et al. 1994 ; TU et al. 1996 ), and glucose repression (TU and CARLSON 1994 , TU and CARLSON 1995 ).

View larger version (9K): In this window In a new window Download PPT slide   Figure 1. Alignment of protein phosphatase sequences. Glc7p, Ppz1p, Ppz2p, and Sal6p share >60% identity at the amino acid sequence level in their catalytic domains. Sequences NH2-terminal to the catalytic domains are less conserved.

In contrast to the extensive investigations of Glc7p, much less is known about the substrates and physiological roles of Ppz1p, Ppz2p, and Sal6p/Ppq1p. PPZ1 was cloned by virtue of its sequence similarity to other phosphatases (DA CRUZ E SILVA et al. 1991 ; POSAS et al. 1992 ). PPZ2 was cloned both in the same manner as PPZ1 (DA CRUZ E SILVA et al. 1991 ) and as a dosage suppressor of recessive mutants in MPK1, a component of a MAP kinase pathway (LEE et al. 1993 ). Yeast strains lacking Ppz1p and Ppz2p are viable but exhibit an osmotic-remedial cell lysis defect in some genetic backgrounds (HUGHES et al. 1993 ; LEE et al. 1993 ; POSAS et al. 1993 ) and are resistant to high salt (POSAS et al. 1995B ). PPQ1 was cloned on the basis of its sequence similarity to related phosphatases (CHEN et al. 1993 ) and was identified independently by mutations that enhance translational suppressors (SAL6; VINCENT et al. 1994 ). Like PPZ1 and PPZ2, PPQ1/SAL6 is not essential, but ppq1 mutants have defects in protein synthesis (CHEN et al. 1993 ) and are sensitive to protein synthesis inhibitors (CHEN et al. 1993 ; VINCENT et al. 1994 ), suggesting that this enzyme functions in translational control.

The high sequence conservation within the catalytic domains of these four enzymes (Fig 1) and our finding that relatively few GLC7 alleles isolated by alanine-scanning mutagenesis conferred conditional growth phenotypes (BAKER et al. 1997 ) led us to ask if these enzymes share overlapping functions. The hypothesis that we have tested is that PPZ1, PPZ2, and SAL6 are at least partially redundant with GLC7 and that this redundancy is responsible for the relatively robust growth phenotype of many GLC7 mutants.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains, media, and growth conditions:
The strains used in this study are listed in Table 1 and are all congenic to KT1112 with the exception of PJ69-4A. The sal6::HIS3 disruption in strain KT1618 was constructed by digesting pAV194 (VINCENT et al. 1994 ) with BamHI and SalI and transforming into yeast. The sla1::URA3 disruption strain was kindly provided by D. Drubin. The gip1::HIS3 disruption in strain AB104 was constructed using plasmid pJT26 as described in TU et al. 1996 . The yol091w::kan^R and yal014c::kan^R disruption strains were obtained from EUROSCARF (Frankfurt, Germany) and Research Genetics (Huntsville, AL), respectively. The ppz1::TRP1, ppz1::URA3, and ppz2::LEU2 mutations in strains DL920, DL786, and DL789, respectively, were introduced into our genetic background by seven serial backcrosses. Yeast cells were grown either on YPD (1% yeast extract, 2% Bacto-peptone, 2% glucose) or on synthetic medium (0.67% yeast nitrogen base, 2% glucose) supplemented with the appropriate amino acids. Yeast genetic manipulations were carried out as described previously (SHERMAN and HICKS 1991 ). Sporulation was induced on medium containing 1% yeast extract, 2% peptone, and 2% potassium acetate. Glucose-repression defects were assessed by testing for growth on medium containing 1% yeast extract, 2% peptone, 200 µg/ml 2-DG (2-deoxyglucose; Sigma, St. Louis), and 2% sucrose under anaerobic conditions (GasPak; Difco, Detroit). Escherichia coli strains DH5F' and XL1-Blue were used for cloning and propagation of plasmids. Yeast transformations were carried out using the lithium acetate method (GIETZ et al. 1992 ).

GLC8 was disrupted by a modified method described by ERDENIZ et al. 1997 . The Kluyveromyces lactis URA3 gene was amplified by PCR using primers annealing to the ends of the URA3 gene and containing adaptamer sequences. Both 5' and 3' flanking sequences of GLC8 (~300 bp each) were amplified in separate PCR reactions. The primers for these reactions were designed such that two primers (BV-29 and BV-30) contained sequences complementary to the GLC8 gene followed by sequences complementary to adaptamer sequences of the primers used to amplify the K. lactis URA3 gene. The primers used to amplify the 5' end of GLC8 were BV-28 5'-CCCGAACACTACCAAATCAATAGCAG-3' and BV-29 5'-GATCCCCGGGAATTGCCATGCATTATTCTGCTGATGTGC-3'. The primers used to amplify the 3' end region of GLC8 were BV-30 5'-AATTCCAGCTGACCACCATGGAAGCATTACGACGTAAGG-3' and BV-31 5'-CGTTTCGCGTGATAACAAACAGTGC-3'. The PCR products from the three independent reactions were combined and the full-length product was amplified using primers BV-28 and BV-31. The resulting PCR product was transformed into KT1358, and Ura⁺ colonies containing the disrupted gene were selected.

Plasmid construction:
Plasmids used in this study are listed in Table 2. pUC19-PPZ1 (kindly provided by D. Levin; POSAS et al. 1992 ) contains the entire PPZ1 coding sequence. pUC19-PPZ1 was digested with PstI and religated to change the orientation of the insert relative to the pUC19 HindIII site, creating pBV207. The change in orientation allowed the entire PPZ1 gene to be released by digestion with HindIII. This fragment was ligated into the HindIII site of pNC160, yielding pBV216. A truncated PPZ1 gene lacking the first 360 codons was generated in several steps. The ClaI-BglII fragment of GLC7 in pHH1 [HindIII-SalI clone of HA-GLC7 from YCp50-HA-GLC7 (STUART et al. 1994 ) inserted into HindIII-XhoI in pBS-SK⁺] was replaced with a 423-bp ClaI-BglII fragment of PPZ1, corresponding to codons 361–501, yielding pBV210. A 2114-bp BglII fragment of pUC19-PPZ1, encoding the remainder of the C-terminal domain of PPZ1, codons 502–692, was inserted in the correct orientation into pBV210 at the BglII site, yielding pBV211. pBV211 therefore encodes the catalytic domain of Ppz1p (codons 361–692) fused in frame with a hemagglutinin (HA) tag (WILSON et al. 1984 ) and the first 11 amino acids (aa) of Glc7p, under the control of the native GLC7 promoter. The GLC7:PPZ1 fusion was released from pBV211 with HindIII and inserted into the HindIII site of pNC160 (pBV212).

For the two-hybrid screen, the catalytic domain of Ppz1p (aa 361–692) was fused to the DNA-binding domain of Gal4. In the first step of the construction, an EcoRI-SacI fragment of PPZ1 encoding the central catalytic portion of Ppz1p was inserted into pBluescript. Oligonucleotide primers were used to introduce a NcoI site at position +1081 in PPZ1. Primers used were BV-5 (5'-CAAAAAGCCCATGGATATTGATGAAACTATCC-3') and the complementary primer BV-6 (5'-GGATAGTTTCATCAATATCCATGGGCTTTTTG-3'), where the NcoI site is underlined. The two halves of the EcoRI-SacI fragment of PPZ1 were amplified independently by PCR using BV-5, BV-6, and primers complementary to the T7 and T3 promoter sequences in pBluescript, respectively. The two independent PCR products were then combined in a PCR reaction using the T7 and the T3 primers to recover the full-length EcoRI-SacI PPZ1 fragment containing the engineered NcoI site. The PCR product was digested with EcoRI-ClaI and inserted into pBV213 yielding pBV214. pBV213 is a plasmid containing a 2114-bp BglII fragment of PPZ1 inserted into the BamHI site of pUC18. The orientation of the PPZ1 BglII fragment in pBV213 is such that the EcoRI site of pUC18 is at the 5' end of PPZ1. pBV214 was cut with NcoI and SalI and the NcoI-SalI fragment containing the catalytic domain of PPZ1, corresponding to amino acids 361–692, was inserted into the two-hybrid vector pAS-CYHII (DURFEE et al. 1993 ), creating pBV222. The full-length Ppz1p was also fused to the Gal4-DNA-binding domain of the two-hybrid vector pAS-CYHII for use in our analysis. The N terminus of Ppz1p (aa 1–360) was amplified by PCR from pUC-19-PPZ1 using primers BV-46 (5'-CCTTCCTTTTCACCATGGCTAATTCAAGTTC-3'), which introduced the underlined NcoI site at the start codon and changed the Gly at the second amino acid position to Ala to abolish N myristoylation, and BV-47 (5'-TGTTCTCTTAGCAGCGTAGCCCGCATCCAG-3'), which annealed downstream of the ClaI site at position +1081. The PCR product was digested with NcoI and ClaI and cloned into pBV222. During this cloning step the 441-bp NcoI-ClaI fragment of pBV222 was replaced with the native 441-bp ClaI fragment of pBV216 in the correct orientation, creating pBV224.

To construct a GFP:PPZ1 gene fusion, the green fluorescent protein (GFP) variant GFP^F64L,S65T (CORMACK et al. 1996 ) was amplified from the pRSETB:GFP template by PCR using primers BV-9 (5'-GGGCGCTCGAGTCCCCCCGCTGAATTCATGAG-3') and BV-11 (5'-CGCGGCTCGAGTCTTTGTATAGTTCATCC-3') that introduced the underlined XhoI sites flanking the GFP coding region. The PCR product was cut with XhoI and cloned into a unique XhoI site present in pUC19-PPZ1 at nucleotide +54, creating pBV209. pBV209 was digested with HindIII to release the entire GFP:Ppz1p coding sequence, which was inserted into pNC160 and YEp351, yielding pBV215 and pBV223, respectively.

Genomic integration of glc7-109, glc7-127, glc7-129, glc7-132, and glc7-133:
The previously described alanine-scanning alleles of GLC7 were tested for function on the centromere vector pNC160 (BAKER et al. 1997 ). To facilitate genetic analysis of these mutants, we integrated glc7-109, glc7-127, glc7-129, glc7-132, and glc7-133 at the GLC7 locus using a two-step gene replacement technique (SCHERER and DAVIS 1979 ). The BglII-KpnI restriction fragments from plasmids that contain these alleles were cloned into the BamHI-KpnI sites of the integrating vector pRS306 (SIKORSKI and HIETER 1989 ). These plasmids, lacking the 5' end of the GLC7 gene, were linearized at the unique SalI site in GLC7 and transformed into haploid yeast strains KT1112 and KT1113. Ura⁺ transformants showed phenotypes previously reported for the plasmid-borne glc7 mutants. These transformants were grown on medium containing 5-fluoroorotic acid to select for recombination events that looped out the integrated plasmids. 5-FOA-resistant Ura^- segregants were screened for the phenotype of the glc7 mutant. Southern analysis and genomic PCR were used to confirm that only a single GLC7 gene was present in the GLC7 locus and that the wild-type GLC7 allele had looped out.

DNA sequencing:
The sequence of the NcoI-ClaI PCR fragment used in the construction of pBV222 was determined. Specific oligonucleotides were designed to allow sequence analysis of both strands using Sequenase (United States Biochemical, Cleveland) and the dideoxy-chain termination method. Primer BV-7 (5'-TCCGGAAACAAAAACGCTCC-3') was used at the 5' end and primer BV-8 (5'-GCTACGATAGCAGCTAATGG-3') was used at the 3' end. Comparison of the nucleotide sequence to wild type was performed using a BLAST search at the Stanford Genomic Resources Center. The sequence of the NcoI-ClaI fragment used in the construction of pBV224 was also determined. The primers used were GV-3 (5'-CCATTTGGATCATTGAAGGTG-3'), GV-4 (5'-TCAAACGTCCCTGATCCCTC-3'), and GV-5 (5'-AGACAACGACATCTCGCAC-3'). Primers and templates were sent to the DNA Sequencing Facility at Iowa State University and comparison of the nucleotide sequence to wild type was performed using a BLAST search at the Stanford Genomic Resources Center.

Immunoblot analysis:
Immunoblot analysis was performed on total cell extracts. Cell extracts were prepared by growing cells to mid-log phase, breaking the cells with glass beads in the presence of 5% trichloroacetic acid, and precipitating total protein (DAVIS et al. 1993 ). The pelleted protein samples were resuspended in a 6-M urea buffer, electrophoresed through an SDS-polyacrylamide gel, and immunoblotted with monoclonal HA antisera (12CA5; WILSON et al. 1984 ) at a dilution of 1:1000 and horseradish-peroxidase-conjugated secondary antibody at a dilution of 1:1000. The ECL detection system (Amersham, Arlington Heights, IL) was used to detect antibody binding.

Two-hybrid assay:
Yeast strain PJ69-4A (JAMES et al. 1996 ) was used for the two-hybrid assay. GLC7, PPZ1_{(aa 361–692)}, and PPZ1_{(aa 1–692)} were fused in frame to the Gal4DBD in pAS-CYHII. Glc7p-interacting proteins were fused in frame to the Gal4-AD in either pACTII or pGAD. Protein-protein interactions were measured by assaying ß-galactosidase enzyme activity (KAISER et al. 1994 ). Assays were performed in triplicates and the mean and simple standard deviation were calculated.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

To test the hypothesis that Glc7p, Ppz1p, Ppz2p, and Sal6p/Ppq1p have overlapping functions, we examined the phenotypes of yeast strains containing mutations in more than one of these phosphatase genes. A set of congenic strains was generated, each containing a gene deletion in PPZ1, PPZ2, and/or SAL6. These strains were crossed to GLC7 mutants. In our genetic background ppz1::URA3 and ppz2::LEU2 mutants, hereafter referred to as ppz1 and ppz2, respectively, have relatively mild growth phenotypes, growing at near wild-type rates at temperatures ranging from 24° to 37°. ppz1 ppz2 double mutants grow slowly at 11° and 15°, indicating that PPZ1 and PPZ2 likely have overlapping functions necessary for growth at low temperature. As reported, ppz1 mutants were resistant to high NaCl concentrations (POSAS et al. 1995B ) and were hypersensitive to caffeine (HUGHES et al. 1993 ; POSAS et al. 1993 ). ppz1 ppz2 strains also showed the salt resistance phenotype of ppz1 PPZ2 strains. sal6::HIS3 mutants, hereafter referred to as sal6, were also robust in our genetic background, exhibiting no obvious growth defect at temperatures ranging from 11° to 37° on rich and synthetic media. As reported, ppq1/sal6 mutants were mildly paromomycin sensitive (VINCENT et al. 1994 ), consistent with the role of PPQ1 in translational proofreading (CHEN et al. 1993 ; VINCENT et al. 1994 ).

To determine the effects of combining mutations in the phosphatase genes, tetrad analysis was carried out on diploid strains heterozygous for sal6, ppz1, ppz2, and glc7 mutations. The sal6, ppz1, and ppz2 mutations were scored by the associated auxotrophic markers and each glc7 mutation was scored by its associated phenotype. The glc7 alleles used in this analysis are listed in Table 3 with the accompanying phenotype of each mutant. The ppz1 ppz2 glc7 mutants were tested for their ability to grow using different carbon sources, at different temperatures, and in the presence of different salts and other growth inhibitors (e.g., 0.9 M NaCl and caffeine). The results of tetrad analysis of meiotic progeny of diploid strains heterozygous for glc7 and ppz1, ppz2, or sal6 are summarized in Table 4 and Fig 2. Whereas sal6 showed no genetic interaction with glc7 or ppz1 ppz2 mutations, multiple genetic interactions were observed between glc7 and ppz1 ppz2 mutations.

View larger version (88K): In this window In a new window Download PPT slide   Figure 2. Genetic interactions between GLC7, PPZ1, and PPZ2. (A) Ascosporal clones from crosses between a ppz1::URA3 ppz2::LEU2 (KT1492) strain and glc7-109 (KT1596) or glc7-132 (KT1599) strains. The four colonies in each row represent a tetrad. The smaller colonies are always those that contain ppz1::URA3, ppz2::LEU2, and either glc7-109 or glc7-132. Colonies of all other genotypes grow at wild-type rates. (B) Clones from the tetrads represented in A were patched onto YPD medium, allowed to grow overnight at 30°, and replica plated onto the media indicated on the left. The plates used for scoring growth at 30° and glycogen accumulation were incubated for 1 day. The plates containing 5 mM caffeine and 0.9 M NaCl were incubated for 2 days and the other plates were incubated for 3 days at their respective temperatures. Unless otherwise stated the plates were incubated at 30°. (C) The strains of indicated genotype were grown on media containing the indicated concentrationsof caffeine at indicated temperatures for 1 day. (D) Strains containing either ppz1 ppz2 (BV460), glc7-133 (BV459), ppz1 ppz2 glc7-133 (BV461), and ppz1 ppz2 glc7-133 (BV461) with pBV215 (GFP:PPZ1 in pNC160) were streaked onto either YPD media or YPD media containing 0.5 M sorbitol. The plates were incubated for 1 day at their respective temperatures.

glc7 mutants display complex genetic interactions with ppz1 and ppz2:
In crosses between each glc7 mutant and the ppz1 ppz2 strain, all spore clones of the triple mutant genotype (glc7 ppz1 ppz2) grew significantly more slowly than spore clones of the other genotypes. The glc7-127 and glc7-129 triple mutants germinated but never grew into macrocolonies. For glc7-109, glc7-132, and glc7-133, the triple mutants formed reproducibly smaller colonies than any single mutant. Representative tetrads are illustrated in Fig 2A. To illustrate in more detail the complex phenotypes displayed by these mutant combinations, representative strains from these crosses were grown on YPD medium and tested for growth in a variety of conditions. As shown in the second and third rows of Fig 2B, the triple mutants grow more slowly than strains of other genotypes on YPD at 37° and 11°, respectively. Microscopic examination and flow cytometry analysis of these cells grown at low temperature revealed no obvious cell cycle arrest (data not shown). The fourth row of Fig 2B illustrates a growth defect of the triple mutants on rich media containing nonfermentable carbon sources ethanol and glycerol (YPGE); the glc7-132 ppz1 ppz2 strain grows slowly while the glc7-109 ppz1 ppz2 strain fails to grow on YPGE.

ppz1 mutants are hypersensitive to caffeine (HUGHES et al. 1993 ; POSAS et al. 1993 ). In our strain background glc7-109 and glc7-132 strains grow more slowly than wild type on medium containing 10 mM caffeine but are less sensitive than ppz1 mutants (data not shown). As shown in the sixth row of Fig 2B, the ppz1 ppz2, ppz1 ppz2 glc7-109, and ppz1 ppz2 glc7-132 strains fail to grow in the presence of 5 mM caffeine. The triple mutants are consistently more sensitive to caffeine than ppz1 ppz2 mutants alone (Fig 2C).

ppz1 ppz2 double mutants exhibit an osmotic-remedial cell lysis defect at high temperatures (e.g., 37°) in some genetic backgrounds (LEE et al. 1993 ), suggesting a role for these phosphatases in some aspect of cell wall integrity. ppz1 ppz2 glc7-133 mutants grow slowly at all temperatures and are inviable at 37°. Microscopic examination of ppz1 ppz2 glc7-133 mutants revealed massive cell lysis (approaching 100%) at 37°. We found that 1 M sorbitol only partially suppressed the growth defect of the triple mutant at 37°. As shown in Fig 2D, growth at 37° was not restored to wild-type levels by sorbitol, and sorbitol actually repressed growth of the ppz1 ppz2 glc7-133 mutant at 30°. These results suggest that Glc7p, Ppz1p, and Ppz2p may all function in cell wall maintenance but further studies will be necessary to sort out possible roles.

PPZ1 and PPZ2 act redundantly in their genetic interactions with GLC7:
Several growth defects have been noted for strains containing disruptions of PPZ1. ppz1 mutants show increased salt tolerance (POSAS et al. 1995B ) and caffeine sensitivity (HUGHES et al. 1993 ; POSAS et al. 1993 ). Disruption of ppz2 has little influence on these defects. However, most of the genetic interactions observed between different glc7 alleles and ppz1 and ppz2 require that both ppz1 and ppz2 be disrupted. For example, glc7-127 and glc7-129 strains bearing disruptions of either ppz1 or ppz2 are viable, whereas the triple mutants are inviable (Table 4). Likewise, the slow-growth phenotype of ppz1 ppz2 strains containing glc7-133, glc7-109, or glc7-132 is not observed if either PPZ1 or PPZ2 is functional. One exception to this rule is glc7-109. The salt sensitivity conferred by glc7-109 is suppressed by ppz1 alone (data not shown).

Some phenotypic traits of glc7 mutants are independent of ppz1 and ppz2:
In contrast to growth rate, caffeine sensitivity, carbon source utilization, and salt sensitivity, other traits of glc7 mutants appear to be independent of the PPZ genotype. glc7-109 mutants hyperaccumulate glycogen, as shown in Fig 2B by the dark brown staining with iodine. This trait is not affected by deletion of either ppz1 or ppz2. In a similar manner both glc7-132 and glc7-132 ppz1 ppz2 strains accumulate low levels of glycogen (Fig 2B). Likewise, the sporulation competency and deficiency of glc7-109 and glc7-132 strains, respectively, are not altered by the PPZ genotype (data not shown). glc7 mutants that are defective in glucose repression, including glc7-133 and glc7-127, are resistant to 2-DG (BAKER et al. 1997 ; TU and CARLSON 1994 ). ppz1 ppz2 mutants are not 2-DG resistant, nor are any ppz1 ppz2 mutants in combination with any of our glc7 alleles, other than glc7-133 (glc7-127 ppz1 ppz2 strains are inviable). In a similar manner, the conditional mitotic arrest observed in other glc7 mutants (HISAMOTO et al. 1995 ; MACKELVIE et al. 1995 ), including glc7-129 (BLOECHER and TATCHELL 1999 ), is not observed in any ppz1 ppz2 glc7 mutants. These results indicate that glucose repression and mitotic regulation are two additional pathways regulated solely by GLC7.

reg1 and sla1 mutants display genetic interactions with ppz1 and ppz2:
The hypothesis that PP1/Glc7p is regulated by targeting subunits leads to the prediction that mutations in at least some of the genes encoding Glc7p regulatory subunits will result in synthetic phenotypes with ppz1 and ppz2 mutations. We tested this hypothesis by crossing reg1, reg2, sla1, gip1, gac1, pig1, yol091w, and yal014c null mutants, which encode bona fide and putative Glc7p-binding proteins, with ppz1 and ppz2 null mutants. As predicted from our earlier genetic analysis, ppz1 ppz2 null mutants displayed no obvious growth defects in combinations with mutants affecting either glycogen accumulation [e.g., gac1 (FRANCOIS et al. 1992 ; STUART et al. 1994 ) and pig1 (CHENG et al. 1997 )] or sporulation [e.g., gip1 (TU et al. 1996 ) and yol091w (TU et al. 1996 ; PEARSON et al. 1998 )]. ppz1 ppz2 null mutants also did not show any obvious growth defects in combination with either reg2 or yal014c null mutants. Reg2p is a Glc7p-binding protein involved in growth control (FREDERICK and TATCHELL 1996 ) and Yal014c is a putative Glc7p-binding protein with unknown function.

On the other hand, synthetic growth defects were observed in crosses between ppz1 ppz2 mutants and either reg1 or sla1 null mutants. Reg1p is required for glucose repression (TU and CARLSON 1995 ) and its activity requires the ability to bind Glc7p (ALMS et al. 1999 ; DOMBEK et al. 1999 ). reg1 mutants also have a slow-growth phenotype (MATSUMOTO et al. 1983 ; TUNG et al. 1992 ). As judged by the colony size of reg1::URA3 ppz1::TRP1 and reg1::URA3 ppz2 double mutants, ppz1 and ppz2 alone have no adverse effect on the growth rate of reg1::URA3 mutants. However, reg1::URA3 mutants are inviable in combination with both ppz1 and ppz2. From 11 tetrads analyzed in a cross between EG716-6B (MAT reg1::URA3 ppz2::LEU2) and BV352 (MATa ppz1::TRP1), the five spore clones judged to be reg1::URA3 ppz1 ppz2 failed to grow into macrocolonies. Four of the triple mutants arrested as microcolonies and one arrested as a single cell. All the remaining spore clones germinated and grew into macrocolonies. The fact that ppz1 ppz2 mutants are not defective in glucose repression yet are lethal in combination with reg1 suggests that the activities of Ppz1p and Reg1p that are responsible for the synthetic lethality of the triple mutant are not related to glucose repression.

Sla1p was originally identified in a synthetic lethal screen with Abp1p (actin-binding protein). Sla1p has been shown to be required for the proper formation of the cortical actin cytoskeleton (HOLTZMAN et al. 1993 ) and is required for the proper localization of Rho1p and Sla2p (AYSCOUGH et al. 1999 ). As determined by the colony size of sla1::URA3 ppz1 and sla1::URA3 ppz2 double mutants, ppz1 and ppz2 alone have no adverse effect on the growth rate of sla1::URA3 mutants. However, sla1::URA3 mutants are inviable in combination with both ppz1 and ppz2. From 10 tetrads analyzed from a cross between BV533 (MATa ppz1::URA3 ppz2::LEU2) and BV543 (MAT sla1::URA3), the four spore clones judged to be the triple mutants arrested as microcolonies and failed to grow into macrocolonies. Microscopic examination of the triple mutants revealed a cell lysis phenotype reminiscent of the cell lysis phenotype observed with the glc7-133 ppz1 ppz2 triple mutant. All but one of the remaining spore clones germinated and grew into macrocolonies. The synthetic phenotype of the reg1 ppz1 ppz2 and the sla1 ppz1 ppz2 mutants provides further evidence for possible overlapping roles for Glc7p and Ppz1/2p.

Ppz1p interacts with some Glc7p-binding proteins in the two-hybrid system:
The synthetic growth defects of ppz1 ppz2 glc7 mutants could be explained if a substrate(s) shared by all three phosphatases must be dephosphorylated to maintain normal cell viability or growth. If the substrate specificity of these phosphatases is regulated by targeting subunits, as is Glc7p, then one might predict that the three phosphatases would share at least a subset of Glc7p regulatory or targeting subunits. We tested two Ppz1p-containing fusion proteins in the two-hybrid assay with a panel of Glc7p-binding protein fusions. pAS-PPZ1_{(aa 1–692)} contains the entire PPZ1 open reading frame while pAS-PPZ1_{(aa 361–692)} contains only the catalytic domain of Ppz1p. In pAS-PPZ1_{(aa 1–692)} the codon encoding the N-myristoylated glycine residue was changed to alanine to avoid possible complications due to N myristoylation, a known modification of Ppz1p (CLOTET et al. 1996 ). For example, ANDRULIS et al. 1998 noted that membrane association of a GAL4-DNA-binding domain fusion protein to the nuclear periphery can alter gene expression. The Glc7p-binding proteins tested for Ppz1p interaction included Gac1p, a glycogen-specific regulatory protein (FRANCOIS et al. 1992 ; STUART et al. 1994 ); Gip1p, which is required for meiosis and sporulation (TU et al. 1996 ); and Reg2p, which is involved in growth control (FREDERICK and TATCHELL 1996 ). Each of these proteins has been shown to interact with Glc7p both in vivo and in vitro. We also tested proteins that were identified in the two-hybrid system as Glc7p-interacting proteins (TU et al. 1996 ) but which have not yet been confirmed by other methods to interact with Glc7p. These included Red1p, which is involved in synaptonemal complex formation (BAILIS and ROEDER 1998 ); Scd5p, which is involved in vesicular trafficking (NELSON et al. 1996 ); Sla1p, an actin-binding protein (HOLTZMAN et al. 1993 ); and the products of several uncharacterized open reading frames: YTA6, GIP2, YFR003c, YFL023w, YOL091w, and YAL014c. Glc8p, a PP1-inhibitor-2-related protein (CANNON et al. 1994 ; TUNG et al. 1995 ) that weakly interacts with Glc7p in the two-hybrid assay (RAMASWAMY et al. 1998 ), was also tested in our analysis.

pAS-PPZ1_{(aa 1–692)}, pAS-PPZ1_{(aa 361–692)}, and pAS-GLC7 were transformed with one of the Glc7p-binding pACT fusions into strain PJ69-4A and transformants were tested for protein-protein interactions by measuring ß-galactosidase (ß-gal) activity. As shown in Fig 3A, Glc7p interacted most strongly with Gac1p (ß-gal levels 60-fold above Snf4p), whereas Glc8p, Red1p, Sla1p, Yta6p, or Yfr003c with Glc7p resulted in 2- to 6-fold higher ß-gal activity than with Snf4p, which we used as the negative control (Fig 3A). All other Glc7p-binding proteins showed an intermediate level of interaction with Glc7p, exhibiting ß-gal levels 18- to 27-fold higher than Snf4p (Fig 3A).

View larger version (23K): In this window In a new window Download PPT slide   Figure 3. Two-hybrid interactions between Ppz1p(aa 1–692), Ppz1p(aa 361–692), and Glc7p-binding proteins. Yeast strain PJ69-4A, which contains the GAL7-LacZ reporter gene, was cotransformed with either pAS-GLC7 (A), pAS-PPZ1(aa 361–692) (B), or pAS-PPZ1(aa 1–692) (C) along with pACT plasmids containing one of the following clones: pGAD-GIP1, pGAD-GIP2, pGAD-RED1, pGAD-SCD5, pGAD-SLA1, pGAD-YTA6, pGAD-YFL023W, pGAD-YOL091W, pGAD-YAL014, or pGAD-YFR003C (TU et al. 1996 ); pSE1111 (pACT-SNF4), pDF112 (pACT-GLC8), or pDF116 (pACT-REG2). The pACT-SNF4 clone was used as a negative control. Protein-protein interactions were assayed by measuring ß-galactosidase activity. A single asterisk denotes statistically significant change vs. the negative control with P < 0.05 and a double asterisk denotes a statistically significant change vs. the negative control with P < 0.01. The levels of all the protein-protein interactions between Glc7p and the various subunits were significantly different (P < 0.01) than the negative control.

In contrast to Glc7p, which interacted with all Glc7p-binding proteins, full-length Ppz1p interacted with only a subset of Glc7p-binding proteins (Fig 3C). Glc8p and Yfr003c showed the strongest interaction, consistently giving ß-gal levels 2.5-fold higher than the negative control while Yol091w exhibited ß-gal levels 2-fold higher than the negative control. Eliminating the NH₂-terminal extension of Ppz1p resulted in a systematic increase in ß-galactosidase expression with all the Glc7p-binding proteins as well as the negative control. Ppz1p_{(aa 361–692)} displayed a significant interaction with Glc8p, at least 5-fold above the negative control (Fig 3B). Scd5p, Red1p, Sla1p, Gip2p, Yta6p, and Yfr003c also significantly interacted with the truncated Ppz1p (1.5- to 2-fold above the negative control; Fig 3B). DE NADAL et al. 1998 also noted that Hal3p, a negative regulator of Ppz1p, shows enhanced binding to the Ppz1p catalytic domain in the absence of the NH₂-terminal domain. The observed increase in ß-galactosidase activity between the truncated Ppz1p and the full-length Ppz1p is not simply due to increased protein expression, since immunoblot analysis demonstrated that the full-length Ppz1p is actually more abundant than the truncated version (data not shown). Together, these results support the possibility that Glc7p and Ppz1p share a subset of Glc7p regulatory subunits.

The catalytic domain of Ppz1p complements a ppz1 null mutant:
Our PPZ1 clone with the NH₂-terminal deletion interacted strongly with a number of Glc7p-binding proteins. However, CLOTET et al. 1996 reported that a similar deletion expressed from the PPZ1 promoter did not complement the caffeine sensitivity or LiCl resistance of a ppz1 null mutant, although overexpression of the catalytic domain did partially complement the ppz1 mutant. This raises the possibility that the NH₂-terminal domain may be critical for specificity and brings into question the significance of the two-hybrid interactions between the catalytic domain of Ppz1p_{(aa 361–692)} and the Glc7p regulatory proteins. As suggested by CLOTET et al. 1996 , the catalytic domain of Ppz1p may lack activity towards specific substrates but retain some nonspecific ser/thr phosphatase activity. Overexpression of this fragment could partially meet the demand for Ppz1p activity. Alternatively, the catalytic domain could contain the essential determinants of Ppz1p specificity, with the amino-terminal domain serving to control subcellular location, regulate stability, or modulate the binding of regulatory proteins. To address this question we constructed a GLC7:PPZ1 fusion clone comprised of the GLC7 promoter, a sequence encoding an HA epitope, and the first 10 codons of GLC7 fused in frame to the coding regions for the catalytic domain and the C terminus of Ppz1p_{(aa 361–692)}. We transformed CEN plasmids containing either the GLC7-PPZ1_{(aa 361–692)} fusion or the full-length PPZ1 into yeast to test expression and biological activity. Immunoblot analysis (Fig 4B) indicates that the Glc7-Ppz1p_{(aa 361–692)} fusion protein is stably expressed in yeast at steady-state levels only slightly lower than Glc7p. This is probably a higher level of expression than that of native Ppz1p. A direct comparison of GFP:Glc7p (A. BLOECHER and K. TATCHELL, unpublished results) and full-length GFP:Ppz1p (see below) protein levels expressed from CEN vectors under the control of their respective natural promoters revealed that GFP:Glc7p is at least 10-fold more abundant than GFP:Ppz1p (data not shown). As shown in Fig 4A, the Glc7-Ppz1p_{(aa 361–692)} protein retains the activity of full-length Ppz1p, as shown by its ability to complement not only the caffeine sensitivity of a ppz1 strain (Fig 4A), but also the lethality of a ppz1 ppz2 glc7-127 triple mutant and the slow-growth phenotype of ppz1 ppz2 glc7-109 and ppz1 ppz2 glc7-133 triple mutants (data not shown). In contrast, the Glc7-Ppz1p_{(aa 361–692)} fusion failed to complement any of the defects associated strictly with the loss of GLC7, including the lethality of a glc7 disruption, the 2-DG resistance of glc7-127 and glc7-133, and the LiCl sensitivity of glc7-109 (data not shown). In summary, the ability of the Glc7-Ppz1p_{(aa 361–692)} fusion to complement a ppz1 null mutation suggests that the catalytic domain of Ppz1p retains key determinants of Ppz1p function while its failure to complement any glc7-associated defects suggests that it has not gained a nonspecific ser/thr phosphatase activity. Likewise, it has also been noted that a SAL6 clone containing a partial deletion (130 amino acids) of the NH₂ terminus still complemented sal6 phenotypes (VINCENT et al. 1994 ). Together, these results indicate that the essential determinants of Ppz1p reside within the catalytic domain and strengthen the significance of the two-hybrid interactions. The discrepancy between these results and those of CLOTET et al. 1996 , where a truncated PPZ1 clone failed to complement a ppz1 null mutant, is likely due to differences in the levels of expression between the two clones. Ppz1p is expressed at a low level, relative to Glc7p, but is localized strictly to membranes. When an NH₂-terminal truncated Ppz1p is expressed from its native promoter, the local concentration of Ppz1p in the membrane is probably insufficient for function. Increased expression of the truncated Ppz1p from the more active GLC7 promoter may result in an increase in the local concentration of Ppz1p at membranes to the point where it is able to act on specific substrates. This model is compatible with the existing data.

View larger version (46K): In this window In a new window Download PPT slide   Figure 4. Complementation of ppz1::URA3 by an NH2-truncated form of Ppz1p. (A) A ppz1::URA3 (BV475) strain was transformed with pBV212 (HA-tagged catalytic domain of Ppz1p), pBV216 (wild-type PPZ1), and pNC160 (empty vector). Transformants were spotted on YPD medium, on YPD medium containing 6 mM caffeine, and on YPD medium containing 10 mM caffeine for 1, 3, and 5 days, respectively, at 30°. (B) Immunoblot analysis was performed on protein extracted from strain BV475 containing either pBV212 (HA-tagged NH2-truncated Ppz1p), pSB17 (HA-tagged Glc7p), or pNC160. The blot was probed with the 12CA5 anti-HA monoclonal antibody.

GLC8 regulates PPZ and GLC7 functions in vivo:
Our two-hybrid data suggest that previously identified Glc7p-interacting proteins can also associate with the catalytic domain of Ppz1p_{(aa 361–692)}. To test the significance of these data, we chose to investigate the interaction between Glc8p and Ppz1p in more detail because Glc8p has a well-documented role in Glc7p regulation and because Glc8p showed the strongest interaction with Ppz1p. Glc8p is a heat-stable protein most similar to inhibitor-2, a well-characterized PP1 regulatory protein (BOLLEN et al. 1994 ; KAKINOKI et al. 1997 ). The glc8 null mutation reduces glycogen levels and suppresses the chromosome-loss phenotype of ipl1-1 (CANNON et al. 1994 ; TUNG et al. 1995 ), two traits associated with reduced Glc7p activity. Overexpression of Glc8p also suppresses ipl1-1 (TUNG et al. 1995 ), suggesting that Glc8p can exert a negative role on Glc7p when overexpressed. Glc7p phosphatase activity is reduced in glc8 null mutants (TUNG et al. 1995 ), implying that Glc8p can directly modulate Glc7p activity. Surprisingly, Glc8p only weakly interacts with Glc7p in the two-hybrid assay (Fig 3A; RAMASWAMY et al. 1998 ) and has not been reported to associate with Glc7p in a more direct binding assay.

To test for genetic interactions between GLC8, PPZ1, and PPZ2 we first constructed a glc8::URA3 null mutant and examined it for defects reported for ppz1 mutants. As reported previously (CANNON et al. 1994 ), glc8::URA null mutants, hereafter referred to as glc8, have a mild glycogen accumulation defect. As shown in Fig 5, glc8 mutants are also tolerant to 0.1 M LiCl and sensitive to 10 mM caffeine. The LiCl tolerance and caffeine sensitivity of glc8 are intermediate between that of ppz1 and of wild type and are more pronounced at 37°. These results are consistent with Glc8p acting as a positive modulator of both Ppz1p and Glc7p activity.

View larger version (95K): In this window In a new window Download PPT slide   Figure 5. Phenotype of glc8::URA3. (Top) Wild-type (KT-1357), ppz1::TRP1 (BV519), and glc8::URA3 (BV470) yeast strains were plated onto YPD medium or YPD medium containing 0.1 M LiCl and grown at 37° for 2 days. (Bottom) Wild-type (KT1112), ppz1::URA3 (BV475), and glc8::URA3 (BV472) yeast strains were plated onto YPD medium or YPD medium containing 10 mM caffeine and were grown at 30° for 2 days.

If Glc8p acts as a positive regulator of Ppz1p we would predict that glc8 would also suppress the salt sensitivity of glc7 mutants. To test this, we mated glc8 mutants to strains containing five different glc7 alleles and characterized the meiotic progeny (Table 5). As shown in Fig 6, glc8 rescues the NaCl salt sensitivity of glc7-109, glc7-132, and glc7-127. The temperature-sensitive growth defect of the glc7-133 glc8 strain (Fig 6, row 2) is partially rescued by the addition of 1 M sorbitol (Fig 6, row 3), reminiscent of the partially osmotic-remedial growth defect of glc7-133 ppz1 ppz2 (Fig 2D). We also characterized the meiotic progeny of a cross between a glc8 strain and a ppz1::TRP1 ppz2::LEU2 strain (Table 5). The triple mutants were viable and relatively robust in growth, but were slightly more sensitive to caffeine than the ppz1::TRP1 ppz2::LEU2 double mutants (data not shown). Since glc7 and ppz1 both cause caffeine sensitivity, the enhanced caffeine sensitivity of glc8 ppz1 ppz2 mutants could be due to an effect of the glc8 null mutation on Glc7p function. Together, these results lend support to the hypothesis that Glc8p positively regulates both Ppzp and Glc7p.

View larger version (48K): In this window In a new window Download PPT slide   Figure 6. Genetic interactions between glc7 and glc8 mutations. Yeast strains with the relevant genotype designated were grown on YPD medium at 24° for 2 days and then replica plated onto the media listed on the left. The 0.9 M NaCl plates were incubated at 24° for 3 days. The 1 M sorbitol plates were incubated at 37° for 2 days and the YPD plates were incubated either at 24° for 3 days or at 37° for 1 day. Strains used were as follows: wild-type (KT1358), glc8::URA3 (BV470), glc7-109 (BV449), glc7-109 glc8::URA3 (BV479), glc7-133 (BV448), glc7-133 glc8::URA3 (BV481), glc7-132 (BV450), glc7-132 glc8::URA3 (BV483), glc7-127 (BV451), and glc7-127 glc8::URA3 (BV477).

Overexpression of Ppz1p has been shown to reduce the growth rate of wild-type cells (CLOTET et al. 1996 ). If Glc8p is indeed a positive modulator of Ppz1p function, we reasoned that mutations in GLC8 might alleviate this slow-growth defect. To test this hypothesis, we constructed a chimeric GFP:PPZ1 gene fusion (Fig 7A) that contained the GFP open reading frame between the 18th and 19th codon of PPZ1 and assayed it for biological activity. The GFP:Ppz1p on a low-copy vector was able to rescue the lethality of a ppz1 ppz2 glc7-127 triple-mutant strain (Fig 7B) as well as the caffeine sensitivity of a ppz1 strain (data not shown). Fluorescence microscopy revealed that GFP:Ppz1p localized primarily with internal and plasma membranes of yeast cells (data not shown). Expression of the GFP:PPZ1 fusion from a high-copy plasmid reduced the growth rate of wild-type but not of glc8 mutant strains (Fig 7C), providing further evidence that Glc8p positively affects Ppz1p function. To test whether Glc8p might affect the stability of GFP:Ppz1p, we performed immunoblot analysis for GFP on strains containing the GFP:PPZ1 fusion integrated at the URA3 locus either in a ppz1 background or a ppz1 glc8 double-mutant background. The levels of GFP:Ppz1p are equivalent in either background (data not shown), strongly suggesting that Glc8p affects GFP:Ppz1p activity. Together, these results suggest that Glc8p acts as a modulator of both Glc7p and Ppz1/Ppz2p.

View larger version (59K): In this window In a new window Download PPT slide   Figure 7. Mutations in glc8 suppress the growth defect conferred by GFP:Ppz1p overexpression. (A) Schematic diagram of GFP:PPZ1 construction. (B) Complementation of the glc7-127 ppz1::URA3 ppz2::LEU2 lethal phenotype by GFP:PPZ1 (pBV215) on a low-copy vector. Strains glc7-127 ppz1::URA3 ppz2::LEU2 kept alive with pBV215 (BV427) and glc7-127 ppz2::LEU2 (BV428) were grown on YPD medium for 3 days at 24°. (C) Strains KT1357 and BV470 were transformed with either YEp351 or pBV223 (YEp351-GFP:PPZ1). Transformants were grown on synthetic medium lacking leucine for 2 days at 30°.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Our tests for genetic interactions between GLC7, PPZ1, PPZ2, and SAL6 were prompted after our limited success at isolating conditional alleles of glc7 by alanine-scanning mutagenesis (BAKER et al. 1997 ). Our model for this approach was the isolation and characterization of alanine-scanning alleles of the actin gene ACT1 (WERTMAN et al. 1992 ). We felt that actin would be a good model for our effort because actin and protein phosphatase type 1 are both products of single essential genes, both exhibit a high degree of evolutionary conservation, and both have a large number of binding partners. However, only 2 out of 22 glc7 mutants had conditional growth phenotypes, in contrast to the 16 out of 36 conditional mutants identified for act1 (WERTMAN et al. 1992 ). Although many of our mutants contained missense mutations in highly conserved amino acid residues, few of these displayed strong or conditional growth phenotypes. One explanation for these results is that other phosphatases can partially compensate for the lack of Glc7p function. The most likely candidates for phosphatases with overlapping function are Ppz1p, Ppz2p, and Sal6p. These are most similar in sequence to Glc7p in their catalytic domains but none is essential for viability (VINCENT et al. 1994 ; POSAS et al. 1992 , POSAS et al. 1993 ; CHEN et al. 1993 ). The results of genetic analysis presented here are largely consistent with this hypothesis. All glc7 alanine-scanning alleles tested conferred either a more severe growth defect or lethality in a ppz1 ppz2 null background. Some of the changes observed were qualitatively similar but quantitatively more severe. For example, ppz1 ppz2 mutants and some glc7 mutants are sensitive to caffeine but the triple mutants are more sensitive. In other cases qualitatively new traits are observed in the triple mutant. For example, glc7-109 ppz1 ppz2 mutants are unable to utilize ethanol efficiently as a carbon source but this defect is not observed for any glc7 mutant nor for ppz1 ppz2 mutants.

The activity of type 1 protein phosphatase is regulated by targeting and/or regulatory subunits (HUBBARD and COHEN 1993 ). We predicted that if Glc7p and Ppzp have overlapping activities then some Glc7p-binding proteins would also interact with Ppzp. Precedence for this is provided by the association of Tap42p with both the PP2A and the Sit4p phosphatases (DI COMO and ARNDT 1996 ). The results of our two-hybrid assay confirmed this prediction for Ppz1p and Glc7p. Most of the Glc7-binding proteins that also associate with Ppz1p have not been characterized, but Glc8p, which has been shown to regulate Glc7p activity (TUNG et al. 1995 ), interacts more strongly with Ppz1p in the two-hybrid system than with Glc7p. The results of our genetic analysis indicate that Glc8p also regulates the activity of Ppz1p, as shown by the caffeine sensitivity and partial salt resistance of glc8 mutants and the ability of glc8 to suppress the slow growth associated with Ppz1p overexpression.

The complex genetic interactions observed between GLC7, PPZ1, and PPZ2 could be explained by a "nested overlap" hypothesis. In this model each phosphatase would have substrates that it is uniquely capable of dephosphorylating, whereas other substrates are shared by all three. Thus, each phosphatase would have unique as well as overlapping cellular functions. For GLC7, unique roles include glycogen synthesis, glucose repression, regulation of mitosis, and sporulation. For PPZ1, possible unique roles are regulation of the response to salt stress by the regulation of ENA1 (POSAS et al. 1995B ) and the control of the G1/S phase transition of the cell cycle (CLOTET et al. 1999 ). We do not yet know the physiological pathway(s) in which Glc7p, Ppz1p, and Ppz2p act redundantly. The cell lysis defect of glc7-133 ppz1 ppz2 mutants suggests that cell wall synthesis or maintainance may be such a pathway. The failure of glc7-109 ppz1 ppz2 mutants to utilize ethanol or glycerol as a carbon source suggests that mitochondrial function may require either activity. This type of functional redundancy has been observed for other protein phosphatases. PPH21 and PPH22, which encode isoforms of the catalytic subunit of PP2A, exhibit genetic interactions with PPH3, another PP2A-like phosphatase gene (RONNE et al. 1991 ).

Not all data are consistent with the nested overlap hypothesis. If Glc7p and Ppzp have overlapping activities we would predict that they would exhibit similar enzymatic activities. However, POSAS et al. 1995A found that the in vitro specificity of Ppz1p is qualitatively different from that of Glc7p. Recombinant Ppz1p was found to effectively dephosphorylate histone H1, myelin basic protein, and casein but was ineffective toward rabbit glycogen phosphorylase, a common substrate for PP1. The caveat for this experiment is that regulatory subunits may influence enzyme specificity, thus differential binding could result in differential specificity. It has also been noted that recombinant PP1 has enzymatic properties different from PP1 synthesized in vivo (ALESSI et al. 1993 ; MACKINTOSH et al. 1996 ).

The apparent functional overlap between Glc7p and Ppzp proteins could also reside at the level of the substrate, such that the function of a pathway would require the dephosphorylation of only one of two or more functionally redundant substrate proteins. One of these proteins could be a substrate of Glc7p while the other could be a substrate of Ppz1p/Ppz2p. The two phosphatase substrates could even be part of the same protein. One site(s) on such a protein would be a substrate of Glc7p while another site(s) would only be a substrate for Ppz1p. As long as one of the two sites is dephosphorylated the pathway can function. One appealing feature of this model is that the two phosphatases need not have similar substrate specificities. Distinguishing among these possibilities will require better understanding of the pathways regulated redundantly by Glc7p and the Ppzp enzymes.

	ACKNOWLEDGMENTS

We thank David Levin, Marian Carlson, Kelly Tindall, Philip James, Sue Liebman, Debra Frederick, Steve Elledge, Donald Pappas, and John Cannon for strains and plasmids used in this study. We also thank Lucy Robinson and Eric Aamodt for critically reading this manuscript. This work was supported by National Institutes of Health grant GM-477899.

Manuscript received December 2, 1999; Accepted for publication January 24, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALESSI, C. R., A. J. STREET, P. COHEN, and P. T. W. COHEN, 1993  Inhibitor-2 functions like a chaperone to fold three expressed isoforms of mammalian protein phosphatase-1 into a conformation with the specificity and regulatory properties of the native enzyme. Eur. J. Biochem. 213:1055-1066[Medline].

ALMS, G. R., P. SANZ, M. CARLSON, and T. A. J. HAYSTEAD, 1999  Reglp targets protein phosphatase 1 to dephosphorylate hexokinase II in Saccharomyces cerevisiae: characterizing the effect of a phosphatase subunit on the yeast proteome. EMBO J. 18:4157-4168[Medline].

ANDRULIS, E. D., A. M. NEIMAN, D. C. ZAPPULLA, and R. STERNGLANZ, 1998  Perinuclear localization of chromatin facilitates transcriptional silencing. Nature 394:592-595. (erratum: Nature 395: 525).[Medline].

AYSCOUGH, K. R., J. J. EBY, T. LILA, H. DEWAR, and K. G. KOZMINSKI et al., 1999  Sla1p is a functionally modular component of the yeast cortical actin cytoskeleton required for correct localization of both Rho1p-GTPase and Sla2p, a protein with talin homology. Mol. Biol. Cell 10:1061-1075[Abstract/Free Full Text].

BAILIS, J. M. and G. S. ROEDER, 1998  Synaptonemal complex morphogenesis and sister-chromatid cohesion require Mek1-dependent phosphorylation of a meiotic chromosomal protein. Genes Dev. 12:3551-3563[Abstract/Free Full Text].

BAKER, S. H., D. L. FREDERICK, A. BLOECHER, and K. TATCHELL, 1997  Alanine scanning mutagenesis of protein phosphatase type 1 in the yeast Saccharomyces cerevisiae.. Genetics 145:615-626[Abstract].

BARFORD, D., 1996  Molecular mechanisms of the protein serine/threonine phosphatases. Trends Biochem. Sci. 21:407-412[Medline].

BLACK, S., P. D. ANDREWS, A. A. SNEDDON, and M. J. R. STARK, 1995  A regulated MET3-GLC7 gene fusion provides evidence of a mitotic role for Saccharomyces cerevisiae.. Yeast 11:747-759[Medline].

BLOECHER, A. and K. TATCHELL, 1999  Defects in Saccharomyces cerevisiae protein phosphatase type I activate the spindle/kinetochore checkpoint. Genes Dev. 13:517-522[Abstract/Free Full Text].

BOLLEN, M., A. A. DEPAOLI-ROACH, and W. STALMANS, 1994  Native cytosolic protein phosphatase-1 (PP-1S) containing modulator (inhibitor-2) is an active enzyme. FEBS Lett. 344:196-200[Medline].

CANNON, J. F., J. R. PRINGLE, A. FIECHTER, and M. KHALIL, 1994  Characterization of glycogen-deficient glc mutants of Saccharomyces cerevisiae.. Genetics 136:485-503[Abstract].

CHEN, M. X., Y. H. CHEN, and P. T. W. COHEN, 1993  PPQ, a novel protein phosphatase containing a Ser+Asn-rich amino-terminal domain, is involved in the regulation of protein synthesis. Eur. J. Biochem. 218:689-699[Medline].

CHENG, C., D. HUANG, and P. J. ROACH, 1997  Yeast PIG genes: PIG1 encodes a putative type 1 phosphatase subunit that interacts with the yeast glycogen synthase Gsy2p. Yeast 13:1-8[Medline].

CLOTET, J., F. POSAS, E. DE NADAL, and J. ARIÑO, 1996  The NH₂-terminal extension of protein phoshatase PPZ1 has an essential functional role. J. Biol. Chem. 271:26349-26355[Abstract/Free Full Text].

CLOTET, J., E. GARI, M. ALDEA, and J. ARIÑO, 1999  The yeast ser/thr phosphatases sit4 and ppz1 play opposite roles in regulation of the cell cycle. Mol. Cell. Biol. 19:2408-2415[Abstract/Free Full Text].

COHEN, P., 1989  The structure and regulation of protein phosphatases. Annu. Rev. Biochem. 58:453-508[Medline].

COHEN, P. T., 1997  Novel protein serine/threonine phosphatases: variety is the spice of life. Trends Biochem. Sci. 22:245-251[Medline].

CORMACK, B. P., R. H. VALDIVIA, and S. FALKOW, 1996  FACS-optimized mutants of the green fluorescent protein (GFP). Gene 173:33-38[Medline].

DA CRUZ E SILVA, E. F., V. HUGHES, P. MCDONALD, M. J. STARK, and P. T. COHEN, 1991  Protein phosphatase 2Bw and protein phosphatase Z are Saccharomyces cerevisiae enzymes. Biochim. Biophys. Acta 1089:269-272[Medline].

DAVIS, N. G., J. L. HORECKA, and G. F. SPRAGUE, JR., 1993  Cis- and trans-acting functions required for endocytosis of the yeast pheromone receptors. J. Cell Biol. 122:53-65[Abstract/Free Full Text].

DE NADAL, E., J. CLOTET, F. POSAS, R. SERRANO, and N. GOMEZ et al., 1998  The yeast halotolerance determinant Hal3p is an inhibitory subunit of the Ppz1p Ser/Thr protein phosphatase. Proc. Natl. Acad. Sci. USA 95:7357-7362[Abstract/Free Full Text].

DEPAOLI-ROACH, A. A., I. K. PARK, V. CEROVSKY, C. CSORTOS, and S. D. DURBIN et al., 1994  Serine/threonine protein phosphatases in the control of cell function. Adv. Enzyme Regul. 34:199-224[Medline].

DI COMO, C. J. and K. T. ARNDT, 1996  Nutrients, via the Tor proteins, stimulate the association of Tap42 with type 2A phosph