YPI1 and SDS22 Proteins Regulate the Nuclear Localization and Function of Yeast Type 1 Phosphatase Glc7*

We have recently characterized Ypi1 as an inhibitory subunit of yeast Glc7 PP1 protein phosphatase. In this work we demonstrate that Ypi1 forms a complex with Glc7 and Sds22, another Glc7 regulatory subunit that targets the phosphatase to substrates involved in cell cycle control. Interestingly, the combination of equimolar amounts of Ypi1 and Sds22 leads to an almost full inhibition of Glc7 activity. Because YPI1 is an essential gene, we have constructed conditional mutants that demonstrate that depletion of Ypi1 leads to alteration of nuclear localization of Glc7 and cell growth arrest in mid-mitosis with aberrant mitotic spindle. These phenotypes mimic those produced upon inactivation of Sds22. The fact that progressive depletion of either Ypi1 or Sds22 resulted in similar physiological phenotypes and that both proteins inhibit the phosphatase activity of Glc7 strongly suggest a common role of these two proteins in regulating Glc7 nuclear localization and function.

PP1 (protein Ser/Thr phosphatase-1) is a ubiquitous eukaryotic enzyme that regulates a variety of cellular processes, such as carbohydrate and lipid metabolism, protein synthesis, and cell cycle progression (for review see Refs. [1][2][3][4]. The PP1 catalytic subunit is highly conserved throughout evolution. In the yeast Saccharomyces cerevisiae, there is only one PP1, named Glc7, which is essential for cell viability (5,6). Similarly to its mammalian counterpart, Glc7 participates in the regulation of many different physiological processes, such as glycogen metabolism, glucose repression, ion homeostasis, cell cycle reg-ulation, sporulation, vacuole fusion, endocytosis, polyadenylation termination, and the maintenance of cell wall integrity (7)(8)(9)(10)(11)(12)(13). The functional versatility of PP1 is achieved by the existence of numerous regulatory subunits that target PP1 to different subcellular compartments and/or substrates, confer substrate specificity, and/or modulate enzymatic activity (1)(2)(3)14). These subunits are structurally diverse, but almost all of them contain a consensus binding motif (R/K)(V/I)X(F/W) necessary for PP1 regulation, which also accounts for the mutually exclusive binding of the different subunits to PP1 (1-3, 14 -17). PP1 activity is essential, but it must be tightly controlled as overexpression or hyperactivation of PP1 phosphatase is deleterious to the cell. Consequently, a large number of physiological inhibitors of PP1 have been identified in higher eukaryotes (1,3,14,18). We have recently identified the first inhibitory subunit of Glc7 in budding yeast. It is a small (155 amino acids), hydrophilic, heat-stable protein that we named Ypi1 (Yeast Phosphatase Inhibitor 1). This protein contains the typical consensus binding motif (R/K)(V/I)X(F/W) necessary to bind PP1. Deletion of YPI1 is lethal, suggesting a relevant role of the inhibitor in yeast physiology. On the other hand, overexpression of Ypi1 displays a number of phenotypes consistent with an inhibitory role of this protein on Glc7 activity. Structural homologues of Ypi1 can be found in yeast, plants, and animals, suggesting a strongly conserved function of this protein (19).
In this work, we provide evidence that Ypi1 interacts with Sds22, another regulatory subunit of Glc7 that targets the phosphatase to substrates involved in mitosis and chromosome segregation (20 -23). Sds22 lacks the consensus (R/K)(V/I)X(F/W) recognition motif found in other Glc7 regulatory subunits (15,17). However, interaction between the human orthologue of Sds22 and PP1 is mediated by the 11 leucine-rich repeats (LRR) 3 that Sds22 has in its central domain and occurs at a site in PP1 different from the one used to bind the (R/K)(V/ I)X(F/W) motif (24). In yeast, Sds22 is an essential protein of 40 kDa largely found in the nucleus, despite the absence of a clear nuclear localization sequence (NLS). In addition to its role in mitosis, Sds22 plays a role in maintaining the normal nuclear localization of Glc7 (25). In this work, we show that Ypi1 and Sds22 form a complex with Glc7, and present data suggesting that, similarly to Sds22, Ypi1 may function within the nucleus, regulating cell growth and also maintaining the nuclear localization of Glc7. The fact that Glc7, Ypi1, and Sds22 are proteins conserved among all eukaryotes suggests that the proposed model of regulation might be also conserved.

Strains and Culture Conditions
Escherichia coli DH5␣ was used as the recipient cell for all plasmids and constructs, whereas E. coli BL21 (DE3) codon plus-RIL (Stratagene) was used to produce recombinant proteins. S. cerevisiae strains used in this work are described in Table 1. Strain MMR09-4, in which the expression of YPI1 is under the control of the tetO 7 promoter, was constructed as follows. A PCR-amplified KanMX4-tetO 7 cassette was made with oligonucleotides 5ЈTETO-YPI1 and 3ЈTETO-YPI1 (Table  2); this cassette was inserted immediately upstream from the initiating ATG codon of the chromosomal YPI1 coding region by homologous recombination in the CML476 strain.
Strain MMR11-1 was made as follows. A 2.7-kbp DNA fragment containing the entire GLC7-yEmCitrine::SpHis5MX cassette, flanked by 206 nucleotides upstream from the GLC7 start codon and 209 nucleotides immediately downstream from the GLC7 stop codon, was PCR-amplified from genomic DNA from the KT2422 strain using Glc7-1256 and Glc7-1673 oligonucleotides (Table 2) as primers. This fragment was integrated in the genome of the MMR09-4 strain by homologous recombination to generate the MMR11-1 strain. The indicated GLC7-yEmCitrine cassette was also introduced in the strain SAY302 (sds22-5ts) to yield strain MMR13-4.
Standard methods for genetic analysis and transformation were used. Yeast cultures were grown in rich medium (YPD) or synthetic complete (SC) medium lacking appropriate supplements to maintain selection for plasmids (26), containing the indicated carbon sources.

Oligonucleotides
Oligonucleotides used in the present study are described in Table 2.

Reverse Transcription-PCR
Cells from 20 ml of yeast culture (MMR09-4 strain) were used for each time point after doxycycline or mock treatments. Yeast cells were collected at 4°C and washed in cold water, and the dried cell pellets kept at Ϫ80°C. Total RNA was extracted using the RiboPure-Yeast kit (Ambion) following the manufacturer's instructions. RNA quality was assessed by denaturing agarose gel electrophoresis, and the RNA quantification was carried out in a BioPhotometer (Eppendorf). 1 g of RNA was used to generate cDNA with the OneStep RT-PCR kit (Invitrogen) according to the manufacturer's instructions using the RT_YPI1_UP and RT_YPI1_DO oligonucleotides ( Table 2). 24 cycles of PCR were performed.

Co-immunoprecipitation Assays and Immunoblot Analysis
Preparation of yeast protein extracts for co-immunoprecipitation assays was essentially as described previously (35). Extraction buffer was 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% Triton X-100, 1 mM dithiothreitol, and 10% glycerol and contained 2 mM phenylmethylsulfonyl fluoride and a complete protease inhibitor mixture (Roche Applied Science). Yeast extracts (500 g) were incubated with either 1 l of anti-LexA polyclonal antibody (Invitrogen), 1 l of anti-HA monoclonal antibody (Sigma), or 1 l of anti-GST polyclonal antibody (Amersham Biosciences), and 50 l of protein A-Sepharose beads for 1 h at 4°C and then washed four times with extraction buffer. Proteins retained by the affinity system were detected by SDS-PAGE followed by immunoblot using anti-HA monoclonal, anti-LexA polyclonal, and anti-GST polyclonal antibodies and chemiluminescence reagents (ECL, Amersham Biosciences).

Purification of Recombinant Proteins in E. coli and Yeast
Purification of the fusion protein GST-Ypi1 expressed in E. coli was carried out as described previously (36). Transfor-mants were grown at 37°C until the absorbance at 600 nm reached a value of around 0.3. Isopropyl 1-thio-␤-D-galactopyranoside was then added to a concentration of 0.1 mM, and the cultures were grown overnight at 25°C. Cells were harvested and resuspended in 20 ml of sonication buffer (50 mM Tris-HCl, pH 7.6, 0.2 mM EGTA, 150 mM NaCl, 10% glycerol, 0.1% Triton X-100, 2 mM dithiothreitol, 2 mM phenylmethylsulfonyl fluoride, and complete protease inhibitor mixture (Roche Applied Science)). Cells were disrupted by sonication, and the fusion proteins were purified by passing the extracts through a 1-ml bed volume of glutathione-Sepharose columns (Amersham Biosciences). GST fusion proteins were eluted from the column with 10 mM glutathione. Samples were stored at Ϫ80°C. Bacterial expression and purification of GST-Glc7 fusion protein were described previously (19). GST-Sds22 fusion protein was expressed in yeast and purified as above.

Effect of Depletion of Ypi1 and Sds22 on Cell Growth
To evaluate the effect of depletion of Ypi1 in cell growth, wild type CML476 and the conditional tetO:YPI1 mutant (MMR09-4) were grown overnight in YPD, diluted at A 600 of 0.01, and treated with 100 g/ml doxycycline (or vehicle), and growth was resumed for 12 h at 30°C. The culture was diluted again until an A 600 of 0.01 and again received 100 g/ml doxycycline before growth was resumed. Samples were taken at the indicated intervals, and the A 600 of the culture measured. The doxycycline treatment was renewed every 12 h to account for possible degradation of the drug.
To check which step of the cell cycle the Ypi1-depleted cells were arrested, cultures (15 ml) of the MMR09-4 strain in YPD at A 660 of 0.01 were treated with doxycycline (100 g/ml), and growth was resumed at 30°C for 12 h. The ␣-factor (10 g/ml) and fresh doxycycline (100 g/ml) were then added, and after 2 h, cells were washed and resuspended in 15 ml of fresh YPD containing doxycycline (100 g/ml) and treated with 0.2 M hydroxyurea for 30 min. Cells were washed again, resuspended in 15 ml of fresh YPD with doxycycline, and growth resumed for 4 h. The same protocol was used for cells untreated with doxycycline and for wild type cells (CML476). Strain SAY306 (wild type) and SAY302 (sds22-5ts) were synchronized as above, except that growth temperature was 24°C. After blockage with hydroxyurea, cells were shifted to the nonpermissive temperature (37°C) and growth resumed for 4 h. In all cases, at different times after the release from the blockage, aliquots of cells were fixed with 3.7% formaldehyde for 60 min at room temperature and stained with DAPI to visualize the nuclei. 120 min after the release, cells were also collected for tubulin staining (see below).

Tubulin Staining
Cells were prepared for indirect immunofluorescence as in Ref. 37 and incubated with 1:250 diluted rat monoclonal antibody YL 1/2 raised against yeast ␣-tubulin (Serotech) (a generous gift of Dr. Jesús Avila, CBM, Madrid, Spain) and subsequently with 1:100 diluted fluorescein isothiocyanate-conjugated goat anti-rat IgG (Alexa Fluor 488; Molecular Probes), to label microtubules. A Color View 12 CCD camera coupled to a Nikon Eclipse E800 microscope was used in combination with the Analyze 3.0 software (Soft Imaging System) to capture the images.

Subcellular Localization of Ypi1-GFP and Glc7-yEmCitrine
Exponentially growing yeast cells containing plasmid pADH1-Ypi1-GFP were used to visualize Ypi1-GFP fusion. Aliquots (2 l) of the cultures were placed on microscope slides and covered with 18 ϫ 18-mm coverslips. The images of Ypi1-GFP and DAPI-stained nuclei were directly captured by fluorescence microscopy, using a Leica DMRXA2 microscope and the Leica FW4000 software.
MMR11-1 transformants (expressing Glc7-yEmCitrine) were used to localize Glc7 in the absence of a functional Ypi1 protein in the following way: yeast cells were inoculated into YPD, grown to saturation, diluted at an A 600 of 0.01 in YPD, and growth resumed for 1 h. Doxycycline was then added from a 5 mg/ml stock solution (made in 50% ethanol) to achieve a final concentration of 100 g/ml. Control cells received the same volume of vehicle. Growth was resumed, and after 12 h, the same amount of doxycycline or vehicle was added to account for degradation of the antibiotic. Cells were collected after 24 h and fixed by resuspension in phosphate-buffered saline/formaldehyde (2%) for 5 min at room temperature. Cells were washed several times with phosphate-buffered saline and kept at 4°C until further use. Similarly, strain MMR13-4 was used to monitor localization of Glc7 in the absence of a functional Sds22 protein. Cells were grown at 26°C in YPD until an A 600 of 0.5 and then maintained at the same temperature or shifted to 37°C for 1 h. Samples were taken and fixed as described above. Fluorescently labeled Glc7 and DAPI-stained nuclei were observed under the microscope (Nikon Eclipse E-800) by mixing on a slide 2 l of the samples with 2 l of a DAPI-containing mounting solution as described previously (38).

Other Techniques
Protein Phosphatase Assays-Protein phosphatase activity using p-nitrophenyl phosphate as substrate was determined essentially as described previously (39). The reaction buffer was 50 mM Tris-HCl, pH 7.5, 0.1 mM EGTA, 2 mM MnCl 2 , and 1 mM dithiothreitol. Samples were incubated 10 min at 30°C, and the reaction was then stopped by adding 1% Tris (final concentration). For phosphatase inhibition assays, different amounts of the purified inhibitors were incubated with the purified phosphatases during 5 min at 30°C, prior to the addition of p-nitrophenyl phosphate.
␤-Galactosidase Assay-␤-Galactosidase activity was assayed in permeabilized cells and expressed in Miller units as described in Ref. 40.

Ypi1
Interacts with Sds22 and Glc7-We have recently characterized Ypi1 as an inhibitory subunit of yeast Glc7 protein phosphatase and shown that it performs an essential function (19). Recent evidence obtained through large scale affinity coimmunoprecipitation approaches suggested that Ypi1 could interact physically with Sds22 (41,42). Because Sds22 is an essential protein of 40 kDa that interacts with Glc7 and targets the phosphatase to substrates involved in mitosis and chromosome segregation (see Introduction), we considered that the Ypi1-Sds22 interaction might have functional relevance. However, because it is known that a large number of interactions defined by high throughput methods turn out to be false positives (43), we decided to validate such interaction by different methods. We first confirmed the interaction between Sds22 and Ypi1 by a direct co-immunoprecipitation method using cell extracts from yeast expressing GST-Sds22, LexA-Ypi1, and HA-Glc7. Using anti-GST antibodies (to immunoprecipitate GST-Sds22), immunoblot analysis revealed the presence of both Ypi1 (LexA-Ypi1) and Glc7 (HA-Glc7) in the immunoprecipitates (Fig. 1A). Similarly, when the cell extracts were immunoprecipitated with anti-LexA antibodies (to immunoprecipitate LexA-Ypi1), we were able to recover Sds22 (GST-Sds22) and Glc7 (HA-Glc7) in the immunoprecipitates (Fig. 1B), and finally, when we immunoprecipitated the cell extracts with anti-HA antibodies (to immunoprecipitate HA-Glc7), we recovered Ypi1 (LexA-Ypi1) and Sds22 (GST-Sds22) in the immunoprecipitates (Fig. 1C). These results indicated that Sds22, Ypi1, and Glc7 were able to associate physically within the yeast cell. To analyze whether these three proteins formed a complex, we used a triple-hybrid system. This approach has been successfully applied by a number of groups to demonstrate that co-expression of an auxiliary bait is sufficient to strengthen ternary interactions (44,45). As shown in Table 3, overexpression of Ypi1 improved the two-hybrid interaction between Sds22 and Glc7 by 14-fold. Similarly, overexpression of Sds22 also improved the magnitude of the two-hybrid interaction between Ypi1 and Glc7 by 150-fold (Table 3). These results suggested that Sds22, Ypi1, and Glc7 may form a stable ternary complex. To confirm these results, we analyzed by gel filtration a crude extract of cells expressing LexA-Sds22, HA-Ypi1, and HA-Glc7. As shown in Fig. 2, most of the three proteins appeared in high molecular weight fractions, with an estimated molecular mass of around 130 kDa (fractions 39 and 41), very close to the expected molecular mass of a putative ternary complex (65 kDa (LexA-Sds22) ϩ 40 kDa (HA-Glc7) ϩ 33 kDa (HA-Ypi1)). We were also able to detect the three proteins in very high molecular weight fractions (fraction 23), very close to the void volume of the column (fraction 21), suggesting that the three proteins might form part of a supramolecular complex. The amount of HA-Glc7 in these very high molecular weight fractions was higher than HA-Ypi1, probably suggesting that HA-Glc7 may participate in different very high molecular weight complexes (46). We next mapped the regions involved in the interaction between Ypi1 and Sds22. Deletion of the Sds22 N-terminal region (from residues 1-40; Fig. 3A) only reduced slightly the interaction between Sds22 and Ypi1 (Fig. 3B), whereas removal of the last C-terminal 22 amino acids (from residues 316 -338) of Sds22, which largely consists of the LRR-cap domain, prevented the interaction of Sds22 with Ypi1 (Fig. 3B). Deletion of both N-terminal and C-terminal regions of Sds22 gave similar results to the deletion of only the C-terminal region (data not shown). To determine whether the LRR-cap domain was responsible for the interaction, we constructed a fusion protein containing only this domain (residues 316 -338), but we did not observe any interaction with Ypi1 (Fig. 3B). These results suggested that the LRR-cap domain was necessary but not sufficient for the interaction with Ypi1. We also tested the interac-  LRR, leucine rich repeats; LRR-cap, motif downstream of the last and incomplete LRR. B and C, two-hybrid interaction between GAD-Ypi1 (B) or GAD-Glc7 (C) and different truncated forms of Sds22. TAT7 yeast cells were transformed with pACT2-Ypi1 (B) or pACT-Glc7 (C) and the appropriate pBTM-Sds22 plasmids (pBTM-Sds22, pBTM-Sds22⌬41, pBTM-Sds22⌬316, and pBTM-Sds22LRRcap). Transformants were analyzed for two-hybrid interaction as described in the legend of Table 3. Values were normalized to the activity present in the interaction with the full-length forms (LexA-Sds22 ϩ GAD-Ypi1, 12.9 ␤-galactosidase units; LexA-Sds22 ϩ GAD-Glc7, 49.9 ␤-galactosidase units). Bars indicate standard deviation. Crude extracts were prepared from representative transformants expressing GAD-Ypi1 and analyzed by Western blotting using anti-LexA antibodies, to check the production of the different LexA-Sds22 derivatives. Similar results were obtained with transformants expressing GAD-Glc7 (not shown). D, diagram of motifs present in Ypi1; RVXW, consensus site for Glc7 binding; NLS?, putative bipartite nuclear localization sequence. E and F, two-hybrid interaction between GAD-Sds22 (E) and GAD-Glc7 (F) and different truncated forms of Ypi1. TAT7 yeast cells were transformed with pACT2-Sds22 (E) or pACT-Glc7 (F) and the appropriated pBTM-Ypi1 plasmids (pBTM-Ypi1, pBTM-Ypi1-Nterm, and pBTM-Ypi1-Cterm). Transformants were analyzed as above. Values were normalized to the activity present in the interaction with the full-length form (LexA-Ypi1 ϩ GAD-Sds22, 347.7 ␤-galactosidase units; LexA-Ypi1 ϩ GAD-Glc7, 5.8 ␤-galactosidase units). Bars indicate standard deviation. Crude extracts were prepared from representative transformants expressing GAD-Sds22 and analyzed by Western blotting using anti-LexA antibodies, to check the production of the different LexA-Ypi1 derivatives. Similar results were obtained with transformants expressing GAD-Glc7 (not shown).

Bait
Prey Additional protein ␤-Galactosidase activity

Ypi1 and Sds22 Regulate Glc7 Nuclear Function
tion between Glc7 and the different fragments of Sds22 and obtained similar results, although we observed a better interaction of Glc7 with the N-terminal truncated form of Sds22 in comparison with the full-length protein (Fig. 3C). Western blot analyses indicated that the truncated proteins were produced at similar levels in all the cases (Fig. 3B). Similarly, we constructed N-terminal and C-terminal deletions of Ypi1 (Fig. 3D) and found that Sds22 interacted mainly with the C-terminal part of Ypi1 (from residue 97-155), which contained a putative bipartite NLS (Fig. 3E), whereas Glc7 interacted mainly with the N-terminal region of Ypi1 (from residues 1-93), which contained the (R/K)(V/I)X(F/W) motif (Fig. 3F). The interaction with this fragment was much better than with full-length Ypi1, probably because of the higher level of expression of the truncated form of the protein (Fig. 3E) and/or to a better accessibility to the (R/K)(V/I)X(F/W) motif in the truncated form. Both domains of Ypi1 were necessary for activity because the expression of any of the truncated forms of Ypi1 could not rescue the lethal phenotype of a ypi1::KanMX mutant (strain MMR18; data not shown).
Ypi1 Is Largely Located Inside the Nucleus-Next, we analyzed the subcellular localization of Ypi1. As observed in Fig. 4, a Ypi1-GFP fusion protein was enriched in the nucleus. These results were consistent with the nuclear enrichment of Sds22 and Glc7 (25).
Inhibitory Capacity of Ypi1 Is Enhanced by Sds22-Because Ypi1 inhibits the phosphatase activity of Glc7 (Fig. 5) (19), we tested whether Sds22 had the same properties. As shown in Fig.  5, a GST-Sds22 fusion protein produced in yeast also inhibited (53% inhibition) the activity of Glc7 used in the assay. More interestingly, the combination of equimolar amounts of Ypi1 and Sds22 (0.224 M each) displayed a higher inhibitory capacity on Glc7, leading to an almost full inhibition (Fig. 5). Ypi1 and Sds22 displayed an additive inhibitory capacity as indicated by the dose-response curve of equimolar amounts of both proteins (Fig. 5).
Mutations in YPI1 and SDS22 Produce Similar Phenotypes-The results presented thus far suggested that the function of Ypi1 and Sds22 could be related. To study this possibility, we    Table 2). The expected 400-bp DNA fragment is shown. C, MMR09-4 (tetO 7 :YPI1) cells were transformed with a centromeric plasmid containing YPI1 regulated under its own promoter (pRS316-Ypi1) or with an empty plasmid (pRS316). Transformants were cultured in YPD at 28°C in the absence (open symbols) or presence (filled symbols) of doxycycline, as above. FEBRUARY 2, 2007 • VOLUME 282 • NUMBER 5 first tested whether the function of Sds22 could be replaced by Ypi1 overexpression and vice versa. However, this was not the case because overexpression of Sds22 could not rescue the lethal phenotype of a ypi1::KanMX mutant (strain MMR18), and overexpression of Ypi1 was not able to rescue the lethal phenotype of an sds22-5ts mutant at the nonpermissive temperature (strain SAY302; data not shown). Therefore, the functions of Ypi1 and Sds22 were not interchangeable.

Ypi1 and Sds22 Regulate Glc7 Nuclear Function
If the functions of Ypi1 and Sds22 were related, we reasoned that it should be possible to identify similar cellular phenotypes as a result of the inactivation of any of these two proteins. Because the deletion of YPI1 is lethal (19), we constructed conditional mutants. Among the different strategies tested, placing YPI1 under the control of a tetO 7 promoter at its own chromosomal location (strain MMR09-4; see "Experimental Procedures") gave the best results. In this strain, YPI1 expression could be switched off by addition of doxycycline. As shown in Fig. 6A, addition of doxycycline to the MMR09-4 strain resulted in a dramatic inhibition of cell growth. RT-PCR analysis clearly showed a marked decrease of YPI1 mRNA levels several hours after addition of doxycycline (Fig. 6B). The growth defect was specifically because of the lack of Ypi1 because introduction in strain MMR09-4 of a centromeric plasmid expressing YPI1 with its own promoter (pRS316-Ypi1) completely rescued the growth defect in doxycycline-containing media (Fig. 6C). These data confirmed a key role of Ypi1 in cell physiology.
It has been described that Sds22 is essential for the progression from metaphase to anaphase in the cell cycle and that sds22-deficient cells are arrested in mid-mitosis with condensed chromosomes and short mitotic spindle (23). Therefore, we decided to test whether the lack of Ypi1 could result in an equivalent phenotype. To this end, cells were synchronized in S-phase in the presence of doxycycline and the position and number of nuclei monitored after release from the blockage. As observed (Fig. 7A), the conditional tetO:YPI1 mutant (strain MMR09-4) suffers a severe blockage during anaphase, similarly to what it is observed in FIGURE 7. DAPI staining of tetO:YPI1 and sds2-5ts conditional mutants. A, wild type strain CML476 and its tetO:YPI1 derivative (MMR09-4) were synchronized with ␣-factor and hydroxyurea (see "Experimental Procedures"), and finally resuspended in YPD medium containing 100 g/ml doxycycline to resume growth. B, strains SAY306 (wild type) and SAY302 (sds22-5ts) were synchronized as above (except that growth temperature was 24°C). Cells were resuspended in fresh YPD and shifted from 24 to 37°C, and growth was resumed. In all cases culture samples were taken at different time intervals, fixed with 3.7% formaldehyde for 60 min, and stained with DAPI to visualize the nuclei. A total of 300 cells in each time were classified into four groups according to the position and number of nuclei (see schematic), and the results were plotted as a function of the time after release from the blockage, in the presence of doxycycline (A) or after the shift to 37°C (B). The experiment was carried out twice with similar results.
Micrographs are examples of cultures after 180 min of release, in all cases. SAY302 cells, carrying a thermosensitive sds22 allele (sds22-5ts), shifted to the nonpermissive temperature (Fig. 7B). Furthermore, when we checked the length of the mitotic spindle in ypi1-deficient cells using an anti-tubulin antibody (Fig. 8), we could observe that, whereas wild type cells presented long mitotic spindles, indicative of normal chromosome segregation, in Ypi1-depleted cells the mitotic spindles were short and similar to the ones observed in the sds22-5ts mutant grown at the nonpermissive temperature (Fig. 8). Wild type cells treated with doxycycline or incubated at the nonpermissive temperature presented long mitotic spindles (data not shown).
Because it has been suggested that Sds22 plays a role in the nuclear localization of Glc7 (25), we also evaluated the possible role of Ypi1 in regulating this process. With this aim, we introduced a GLC7-yEmCitrine cassette (a generous gift from Dr. K. Tatchell) into strain MMR09-4 and examined cells grown in the absence or presence of doxycycline. As shown in Fig. 9, fluorescently labeled Glc7 was mostly localized in the nucleus in cells expressing Ypi1. However, in Ypi1-depleted cells, Glc7-derived fluorescence was more diffuse, no longer nuclear, and occasionally presented a punctate pattern, suggesting that localization of Glc7 was altered in the absence of Ypi1. This phenotype resembled the one observed in strain MMR13-4, which expressed GLC7-yEmCitrine in an sds22-5ts background, grown at the nonpermissive temperature (Fig. 9), in agreement with a previous report (25). These data suggested a common role of Ypi1 and Sds22 in regulating Glc7 nuclear localization.
Sds22 has also been found as a dosage suppressor of the temperature-sensitive phenotype of the ipl1-321 mutant (47). Ipl1 is a protein kinase involved in the regulation of kinetochoremicrotubule interactions and microtubule function, whose activity is antagonized by Glc7 (9,(47)(48)(49). It was suggested that Sds22 could act as a Glc7 chaperone that could titrate Glc7 away from essential Ipl1 targets (47). To know whether Ypi1 had the same effect, we overexpressed Ypi1 in an ipl1-321 mutant and found that it suppressed the lethal phenotype of the mutant grown at the nonpermissive temperature, as in the case of overexpressing Sds22 (Fig. 10). These results suggested that overexpression of Ypi1 or Sds22 caused a decrease in the specific activity of Glc7 toward a particular substrate related to the Ipl1 protein kinase pathway.

DISCUSSION
It has been described that among other physiological processes, Glc7 PP1 phosphatase regulates cell cycle progression (2,9,10). Some of the functions of Glc7 in cell cycle progression are achieved by its binding to Sds22, a regulatory subunit that targets Glc7 to substrates involved in regulation of mitosis and chromosome segregation (20 -23). Sds22 is an atypical Glc7 regulatory subunit because it lacks the consensus (R/K)(V/ I)X(F/W) motif present in most Glc7 regulatory subunits. However, despite the absence of this motif, Sds22 still binds to Glc7, and it does so through its central domain composed of 11 leucine-rich repeats. Binding occurs at a site in Glc7 different from the one used to bind RVXF motif-containing interactors. In fact, Sds22 binds to a domain in PP1, composed of ␣4, ␣5, and ␣6 helices, that is located far away from other well known regulatory binding sites of PP1, such as the RVXF hydrophobic binding channel, the ␤12-␤13 loop, and the acidic groove, involved in binding of RVXF motif-containing interactors (24). The observation that the C-terminal half of PP1, including all residues that contribute to the RVXF-binding channel, is not required for the interaction with Sds22 (24, 50) raises the interesting possibility that in Sds22-associated PP1 holoenzymes this channel is free for interaction with a specific additional RVXF-containing subunit. In this way, Sds22-PP1 holoenzymes would resemble other trimeric PP1 holoenzymes known to contain an RVXF-containing and an RVXF-less regulator, such as the CPI-17-Mypt1-PP1 complex (51).
In this study, we provide strong evidence indicating that Ypi1, a recently identified inhibitory subunit of Glc7 phosphatase containing a typical consensus (R/K)(V/I)X(F/W)-binding motif (19), interacts physically with both Sds22 and Glc7. Interestingly, overexpression of Ypi1 enhances the interaction between Sds22 and Glc7, and overexpression of Sds22 enhances the interaction between Ypi1 and Glc7. A possible explanation for these results is that the overexpression of either Ypi1 or Sds22 could stabilize the interaction between the other two components (Glc7-Sds22 or Glc7-Ypi1, respectively) and/or displace other endogenous regulatory subunits from their binding to Glc7, making the phosphatase more available for interaction with the other component of a putative ternary FIGURE 8. Tubulin staining. A, growth of MMR09-4 (tetO:YPI1) and SAY302 (sds22-5ts) cells was synchronized as in the legend to Fig. 7. After 120 min of release from the blockage, aliquots were prepared for indirect immunofluorescence using anti-tubulin antibody. B, data indicate the mean of the length of the mitotic spindle in 300 cells at mid-mitosis for each condition. dox, doxycycline. FEBRUARY 2, 2007 • VOLUME 282 • NUMBER 5

Ypi1 and Sds22 Regulate Glc7 Nuclear Function
complex. This complex was confirmed by co-immunoprecipitation and gel filtration analyses. In the latter we found that most of the Sds22, Glc7, and Ypi1 proteins were present in fractions corresponding to an estimated molecular mass of 130 kDa. The existence of this complex would provide an explanation for the isolation in Schizosaccharomyces pombe, by gel filtration, of a trimeric complex of 105 kDa containing Sds22, Sds21 (PP1), and an unidentified phosphoprotein of 25 kDa (23). In this case, the 25-kDa unidentified protein could be the S. pombe orthologue of Ypi1 (GenBank TM accession number CAB11073).
We have also mapped the domains involved in the interaction among these three proteins, and our results suggest that the C-terminal domain of Ypi1 is responsible for the interaction with Sds22, whereas Glc7 binds to the N-terminal domain of Ypi1, which indeed harbors the PP1-binding (R/K)(V/ I)X(F/W) motif. We have also noted that binding of Sds22 to Ypi1 may be independent of the presence of Glc7 in the complex. This notion came out from our finding that a Ypi1-W35A mutant, which binds Glc7 very poorly because of an altered RVXW motif (19), interacted with Sds22 with the same strength as wild type (data not shown). We have also observed that the N-terminal part of Sds22 is dispensable for the interaction with Glc7 (in agreement with previous results (24)) and with Ypi1, and that the LRR-cap domain is necessary but not sufficient for the interaction with both Ypi1 and Glc7. The latter results may suggest that the LRR-cap domain is required for the proper folding of Sds22. One could envisage that the binding of the RVXF-containing interactors to Glc7, like Ypi1, would not compete with the binding of Sds22, because both interactors bind Glc7 at separate locations. However, binding of Sds22 and RVXF-containing interactors to Glc7 can also be mutually exclusive, as in the case of Gac1, an RVXF-containing regulator that targets Glc7 to substrates involved in glycogen metabolism (52).
We also present strong evidence indicating that the functions of Ypi1 and Sds22 are related, because the inactivation of any of these two proteins results in similar cellular phenotypes. We demonstrate that upon Ypi1 depletion, Glc7 shows an altered subcellular localization. This phenotype is similar to the one observed upon inactivation of Sds22, indicating that both Ypi1 and Sds22 are necessary for the nuclear localization of Glc7. Because Glc7 regulates cell cycle progression (2,9,10), the mislocalization of the phosphatase would lead to cell growth inhibition. In agreement with this suggestion, we have observed that depletion of Ypi1 caused cell growth arrest at mid-mitosis, and that this phenotype was similar to the one observed by inactivating Sds22 (25). Microscopic examination of Ypi1-depleted cells revealed that these cells contained short mitotic spindles, as in the case of sds22-5ts cells grown at the nonpermissive temperature. Therefore, the presence of both proteins is required for Glc7 to perform its nuclear essential functions. In addition, the function of Ypi1 and Sds22 is not redundant because overexpression of one of these regulatory FIGURE 9. Depletion of Ypi1 and Sds22 affect Glc7 localization. Strain MMR11-1 (tetO:YPI1), which contains an integrated GLC7-yEmCitrine construct, was grown in the absence or presence of 100 g/ml doxycycline (dox) as described under "Experimental Procedures." Cells were taken after 25 h of growth and fixed for microscopic observation. Similarly, strain MMR13-4 (sds22-5ts), containing the same GLC7-yEmCitrine construct, was grown at 26°C until an A 660 of 0.5 and maintained at 26°C or switched to 37°C for 1 additional h. Fluorescence of GLC7-yEmCitrine expression was detected by using a fluorescein isothiocyanate filter. Samples were also processed for DAPI staining. SBY214 (wild type (WT)) and SBY322 (ipl1-321) cells were transformed with plasmids pWS93 (empty), pWS-Ypi1 and pWS-Sds22. Transformants were grown at 30°C in selective SC-2% glucose until they reached the exponential phase (A 660 0.5). 10-Fold dilutions of the cultures were prepared, and 3 l of each were spotted on selective SC-2% glucose plates. Plates were then incubated at 30 or 37°C for 48 h.
subunits cannot suppress the lethal phenotype derived from the absence of the other.
In this study we also show that both Ypi1 and Sds22 have, in vitro, the ability to inhibit Glc7 and that the combination of equimolar amounts of both proteins leads to an almost complete inhibition of Glc7 activity. The inhibitory role of Sds22 on PP1 activity has already been documented in higher eukaryotes such as rat hepatocytes (53) and Schistosoma mansoni (54). In addition, in bovine spermatozoids, it is known that Sds22 inhibits PP1␥2 phosphatase activity. In these cells, the inhibitory role of Sds22 is controlled by binding to an unknown protein of 17 kDa, forming an inactive complex that is unable to bind PP1. Only when Sds22 is released from its binding to p17, is it able to bind and inhibit PP1 (55,56). We think that the situation in yeast is different, because our triple-hybrid analysis suggests that the expression of either Ypi1 or Sds22 improves the interaction between the other two components of a putative ternary complex. In addition, if the function of Ypi1 was to maintain Sds22 in an inactive state, in Ypi1-depleted cells Sds22 would be free to interact with PP1 and inhibit its function. However, our phenotype experiments indicate that depletion of Ypi1 gives similar phenotypes as impairing Sds22 function, which is not compatible with the proposed model. The inhibitory ability of Ypi1 and Sds22 is consistent with the fact that overexpression of any of these two proteins can suppress the lethal phenotype of ipl1-321 mutants grown at the nonpermissive temperature. A possible explanation for these results could be that overexpression of Ypi1 would inhibit Glc7 activity toward substrates involved in the Ipl1 protein kinase pathway, as already suggested for Sds22 (47).
The fact that all the proteins present in the Glc7-Ypi1-Sds22 complex have well conserved orthologues throughout evolution raises the interesting possibility that the function of each component and/or the function of the complex as a whole should be also conserved. We are currently investigating whether inhibitor 3 (PPP1R11; human orthologue of yeast Ypi1) and hSds22 (PPP1R7; human orthologue of yeast Sds22) form a complex with PP1 (human orthologue of yeast Glc7), similar to the one we have described in yeast.