GPI7 involved in glycosylphosphatidylinositol biosynthesis is essential for yeast cell separation.

GPI7 is involved in adding ethanolaminephosphate to the second mannose in the biosynthesis of glycosylphosphatidylinositol (GPI) in Saccharomyces cerevisiae. We isolated gpi7 mutants, which have defects in cell separation and a daughter cell-specific growth defect at the non-permissive temperature. WSC1, RHO2, ROM2, GFA1, and CDC5 genes were isolated as multicopy suppressors of gpi7-2 mutant. Multicopy suppressors could suppress the growth defect of gpi7 mutants but not the cell separation defect. Loss of function mutations of genes involved in the Cbk1p-Ace2p pathway, which activates the expression of daughter-specific genes for cell separation after cytokinesis, bypassed the temperature-sensitive growth defect of gpi7 mutants. Furthermore, deletion of EGT2, one of the genes controlled by Ace2p and encoding a GPI-anchored protein required for cell separation, ameliorated the temperature sensitivity of the gpi7 mutant. In this mutant, Egt2p was displaced from the septal region to the cell cortex, indicating that GPI7 plays an important role in cell separation via the GPI-based modification of daughter-specific proteins in S. cerevisiae.

Glycosylphosphatidylinositol (GPI) 1 anchoring is a posttranslational modification conserved in many cell surface proteins of all eukaryotes. GPI anchors attach to certain proteins in the endoplasmic reticulum (ER) and determine their final localization. In mammalian polarized cells, the GPI anchor has been implicated in the potential recruitment of proteins to the apical domain (1,2). In yeast, GPI-anchored proteins are one of the major components of mannoproteins at the cell wall and play important roles in cell wall biogenesis and cell wall assembly (3). The yeast Saccharomyces cerevisiae has a cell wall composed of mannoproteins, glucan, and chitin. The cell wall is a rigid structure that maintains cell morphology; however, it must be remodeled during the cell cycle, including bud emergence and the cell separation process.
The biosynthesis of the GPI anchor is essential for cell growth in yeast. Phosphatidylinositol is modified by the stepwise addition of sugars and ethanolaminephosphate (EtN-P), thus forming a complete precursor lipid in the ER. GPI transamidase complex recognizes a carboxyl-terminal signal in substrate proteins, cleaves a carboxyl-terminal peptide from the proprotein substrate, and attaches a GPI moiety to the nascent carboxyl terminus, referred to as the -site (4). GPI7/LAS21, a gene involved in the GPI biosynthetic pathway, is required for the addition of a side chain EtN-P to the second mannose portion of the GPI core glycan structure (5). GPI7 is not essential for cell viability, and GPI lacking a side chain EtN-P due to the loss of GPI7 function is still transferred to proteins (4 -6). Deletion of GPI7 causes a growth defect at high temperature, hypersensitivity to calcofluor white, and a decreased replacement of primary lipid moiety of GPI anchors by ceramide (5,7,8). Mutations in GPI7 also affect cell wall anchorage in S. cerevisiae and Candida albicans (9). GPI7 is involved in chlamydospore formation, budding patterns, and cell shape in C. albicans (10) and invasive growth in the dimorphic yeast Yarrowia lipolytica (11). However, how GPI7 is involved in these phenomena is still unknown.
Cell separation is the last step of the cell division cycle in budding yeast. It is completed by the degradation of the septum, a specialized structure of the cell wall (12). A chitinase and some glucanases, which are expressed only in daughter cells, are transported to the bud neck and degrade the septum (12,13). These enzymes are under the control of a transcription factor, Ace2p, and Cbk1 kinase (14,15). In this report, we provide several lines of evidence that GPI biosynthesis, especially the addition of EtN-P to the glycan portion of the GPI anchor by Gpi7p, is involved in the control of cell separation via modification of daughter-specific cell wall assembly proteins. Furthermore we propose a novel hypothesis that EtN-P of the GPI anchor is important for the correct targeting of the GPIanchored proteins in yeast cell separation.

EXPERIMENTAL PROCEDURES
Strains, Media, and Plasmids-Yeast strains used in this study are listed in Table I. YPAD and synthetic complete (SC) media were used as described previously (16). SDCA contains 0.67% yeast nitrogen base without amino acids (Difco), 2% glucose, 0.5% casamino acids, and 0.004% adenine sulfate. The disruption of genes in yeast was performed with a one-step gene disruption method as described previously (17). All the primer sets used for the gene disruption are listed in Table II. Three tandem copies of the HA epitope were placed just before the stop codon of GPI7 as follows. TOp1014, which carried the GPI7 gene in YCUp4 (CEN4, ARS1, URA3), was kindly provided by Dr. Akio Toh-e (University of Tokyo). Using TOp1014 as a template, an NheI restriction site was introduced by PCR just before the stop codon of GPI7, and the PCR product was subcloned into the centromere plasmid pRS316 (18) to generate pMF3 and sequenced for confirmation. The NheI-NheI cassette containing three copies of the HA epitope was prepared from pYT11 (19) and inserted into the NheI site of pMF3 to generate pMF7 (pRS316-GPI7-HA). To create gpi7 point mutations, we used the QuikChange site-directed mutagenesis kit (Stratagene). The centromere plasmid pRS315-HA-EGT2, in which 3xHA-tagged EGT2 is ex-pressed under its own promoter, was constructed as follows. A DNA fragment containing the EGT2 open reading frame, a 0.7-kb upstream region, and a 0.5-kb downstream region was amplified by PCR and subcloned in pRS315 (18) to generate pMF295. Using pMF295 as a template, the SmaI site was introduced 78 nucleotides downstream from the start codon of EGT2 by PCR. The PCR product was subcloned into pRS315 to generate pMF296 and sequenced for confirmation. The SmaI-SmaI cassette containing three copies of the HA epitope was inserted into the SmaI site of pMF296 to generate pMF297 (pRS315-HA-EGT2). The centromere plasmid pRS315-HA-ENG1, in which 3xHA-tagged ENG1 is expressed under its own promoter, was constructed as follows. A DNA fragment containing the ENG1 open reading frame, a 0.8-kb upstream region, and a 0.4-kb downstream region was amplified by PCR and subcloned in pRS315 to generate pMF306. Using pMF306 as a template, the SmaI site was introduced 600 nucleotides downstream from the start codon of ENG1 by PCR. The PCR product was subcloned into pRS315 to generate pMF309 and sequenced for confirmation. A SmaI-SmaI cassette containing three copies of the HA epitope was prepared and inserted into the SmaI site of pMF309 to generate pMF310 (pRS315-HA-ENG1). pRS315-HA-ENG1-EGT2, the 3xHA-tagged ENG1-EGT2 chimeric gene expressed under the ENG1 promoter, was constructed as follows. Using pMF310 as a template, a HindIII restriction site was introduced by PCR just before the stop codon of ENG1, and the PCR product was subcloned into the centro- mere plasmid pRS315 to generate pMF355 and sequenced for confirmation. Using pMF295 as a template, a HindIII site was introduced 150 nucleotides upstream and a XhoI site was introduced 0.5-kb downstream from the stop codon of EGT2 by PCR. The PCR product was subcloned into pMF355 to generate pMF357 (pRS315-HA-ENG1-EGT2) and sequenced for confirmation. Genetic Screens-To obtain multicopy suppressors of gpi7-2, we transformed gpi7⌬ cells carrying pRS316-gpi7-2-HA (gpi7-2, uracil marker) with a yeast genomic library (leucine marker). From about 15,000 transformants, 48 colonies that could grow at 35.7°C on SC-Leu-Ura plates were obtained. These colonies were grown on SC-Leu plates containing 0.1% 5-fluoroorotic acid to remove pRS316-gpi7-2-HA at 30°C. Then the multicopy plasmids recovered from these cells were introduced again into gpi7-2 cells. Of these colonies, seven could grow at 35.7°C on SDCA. Restriction analysis or DNA sequencing was performed for the seven plasmids. We obtained two kinds of plasmids. One plasmid contained GPI7 itself. The others were derived from a locus different from each other. After subcloning, the suppressor genes in these plasmids were determined to be WSC1, RHO2, ROM2, GFA1, and CDC5.
To screen for extragenic suppressors of gpi7⌬, a yeast genomic library mutagenized by an mTn-3xHA/GFP transposon kindly provided by Dr. Michael Snyder (Yale University) was used to transform MFY11 (gpi7⌬). A total of ϳ33,000 transformants was obtained on SCϪUra plates. The colonies were replica-plated on SCϪUra plates and cultured at 35.7°C to select mutants. The colonies that could survive at 35.7°C were streaked again on SCϪUra plates and cultured at 35.7°C to confirm the recovery of temperature sensitivity. Tagged transposon sites were identified by direct genomic sequencing as described previously (20).

myo-[2-3 H]Inositol
Labeling of GPI Intermediates-GPI intermediates were labeled with myo-[2-3 H]inositol (PerkinElmer Life Sciences) as described previously with some modifications (21). Cells were grown in SDCA medium, and 2.5 ϫ 10 7 cells were resuspended in 600 l of synthetic dextrose inositol-free medium containing 0.67% yeast nitrogen base without inositol and amino acids (Bio101), 5% glucose, and nutrient supplements. Cells were divided into two tubes and preincubated for 20 min at 25 or 35.7°C, and 0.25 MBq of myo-[2-3 H]inositol was added. The cells were then incubated for 90 min at 25 or 35.7°C, respectively. After the 90-min incubation, 10 mM (final concentration) NaF and NaN 3 were added to the reaction mixture to stop the reaction. The cells were washed with water. Lipids were extracted and desalted by butanol extraction as described previously (22). The lipid extracts were analyzed by ascending TLC using 0.2-mm-thick silica gel plates with a solvent system of chloroform/methanol/water (10:10:3). Radioactivity was detected using Molecular Imager FX (Bio-Rad).
Western Blotting-The samples were denatured with sample buffer for 1 h at 4°C for membrane proteins and run on 7.5% SDS-polyacrylamide gels as described previously (23). After SDS-PAGE, the proteins on the polyacrylamide gels were transferred to polyvinylidene difluoride membranes (Millipore). Gpi7-HAp was detected with the anti-HA monoclonal antibody 16B12 (Babco). Immunoreactive bands were visualized by staining with horseradish peroxidase-conjugated goat antimouse IgG (Cell Signaling Technology) and chemiluminescence with ECL-Plus (Amersham Biosciences).
Fluorescence Microscopy-All fluorescence images were observed using a fluorescence microscope, BX50 (Olympus). For calcofluor white staining, cells washed with water were sonicated for 10 s as described above, and calcofluor white solution (1 mg/ml calcofluor white (Sigma) in water) was added. The cells were then washed twice with water. For HA-Egt2p, HA-Eng1p, and HA-Eng1-Egt2p localization, cells grown in SC medium were sonicated, incubated at 35.7°C for 4 h, fixed with 3.7% formaldehyde for 1 h at room temperature, and stained as described previously with some modifications (24). Cells were washed with 250 l of phosphate-buffered saline with bovine serum albumin (1 mg/ml) and rat anti-HA antibody 3F10 (Roche Applied Science) and incubated at 4°C for 1 h. The cells were then washed twice in phosphate-buffered saline and incubated in phosphate-buffered saline buffer with bovine serum albumin and Alexa 488-conjugated goat anti-rat IgG at 4°C for 1 h.

Isolation and Characterization of gpi7-2 Mutants-
To understand the cellular functions of GPI7, we adopted two different genetic approaches: a multicopy suppressor analysis with a gpi7 missense mutant and an extragenic suppressor analysis with a gpi7 gene-disrupted (gpi7⌬) mutant. Although several multicopy suppressors were isolated from gpi7⌬ cells (7,8,25), no significant information was obtained to elucidate the biological function of GPI7. Our purpose in this study was to obtain new suppressors from gpi7 missense and gpi7⌬ mutant cells, which may help to elucidate the biological functions of Gpi7p. First, we tried to isolate missense mutants of GPI7 that show a deficiency in only a part of the several proposed functions of Gpi7p (5-9, 25) among site-directed mutagenic constructs. We replaced the positively or negatively charged amino acid(s) with alanine in a region that is conserved among the GPI7 open reading frames of S. cerevisiae (GenBank TM accession numbers in parentheses) (NP_012473), C. albicans (AAL83897), Y. lipolytica (AAK52677), Schizosaccharomyces pombe (CAA91096), Caenorhabditis elegans (NP_495820), and Homo sapiens (BAC11227). One mutant, gpi7-2, in which the 153rd and 154th aspartic acids are replaced by two alanines (Fig. 1A, shown as "AA"), showed temperature sensitivity at 35.7°C (Fig. 1B). This mutational region was conserved among GPI13 orthologues ( Fig. 1A) but not MCD4 orthologues (data not shown). Both Gpi13p and Gpi7p transfer EtN-P to position 6 of the mannose, whereas Mcd4p transfers it to position 2 (4). Therefore, the mutation site in gpi7-2 is conserved in the EtN-P transferases, which transfer EtN-P to position 6 of the mannose. The mutant cells recovered their cell growth on addition of an osmotic stabilizer, 1 M sorbitol (Fig. 1C), and were hypersensitive to calcofluor white (Fig. 1D), indicating that the mutant cells are defective in cell wall biogenesis or assembly. Western blotting analysis of HA-tagged Gpi7p revealed that the mutant Gpi7-2 proteins were of the same size as Gpi7-HA fusion protein (90 kDa) and were not significantly degraded in the cells by the point mutation (Fig. 1E). Therefore, the temperature sensitivity of the gpi7-2 strain might be due to the loss of the Gpi7p enzymatic activity and not due to the degradation of mutant Gpi7 proteins. Next we examined whether this mu- , which is specifically detected in gpi7⌬ cells, at permissive and non-permissive temperatures (5). The lipid fraction from gpi7-2 cells treated at the non-permissive temperature (35.7°C) showed M4 accumulation, while the M4 accumulation was decreased at the permissive temperature (25°C) (Fig. 1F). These results indicate that the gpi7-2 gene product partially functions in EtN-P transfer activity at the permissive temperature but loses its function at the non-permissive temperature. Therefore, we selected strain gpi7-2 for further characterization.
We investigated the morphology of the gpi7 mutant cells. Interestingly gpi7 mutants showed an abnormal cell arrangement after 4-h incubation at the restrictive temperature (35.7°C), while the morphology of the gpi7 cells was normal at permissive temperature (25°C) (data not shown). Approximately 30% of the gpi7-2 cells showed a unique arrangement in which the mother cell is attached to two daughter cells even after sonication (Fig. 2B). This abnormal cell arrangement was still observed after 10-h incubation at 35.7°C. Judging from methylene blue staining, most of the gpi7-2 cells were still viable after 10-h incubation at the non-permissive temperature (data not shown), suggesting that gpi7-2 cells might arrest at this stage. The DNA content of propidium iodide-stained cells was analyzed by flow cytometry (Fig. 2, A and B). An abnormal peak, 3C DNA content, was observed in gpi7-2 cells after 4-h incubation at 35.7°C (Fig. 2B), while only the 1C and 2C DNA peaks were observed in wild-type cells. The 3C peak was also observed in gpi7⌬ cells at the non-permissive temperature (data not shown). These 3C peaks tightly correlated to the abnormal cell arrangement observed in gpi7 mutants at the restrictive temperature (Fig. 2B). To confirm the three-cell arrangement observed in gpi7 mutant cells more precisely, we examined the DNA content and cell morphology during the cell cycle progression using G 1 synchronous cultures treated with ␣-factor. The synchronized wild-type and gpi7-2 cells were released into YPAD medium at the non-permissive temperature, 35.7°C. The ␣-factor-treated gpi7-2 cells were not completely uniform but showed 1C plus a few 2C peaks (Fig. 2E), whereas wild-type cells were uniformly 1C (Fig. 2C). Wild-type cells normally progressed through the cell cycle, regenerating G 1 phase cells with 1C DNA content at about 90 min after the release (Fig. 2, C and D). In contrast, gpi7 mutant cells failed to regenerate a distinct 1C peak, and 3C DNA cells accumulated at about 180 min after the release (Fig. 2E). The observation of cell morphology revealed that most of the gpi7 mutant cells (more than 80% of total cells) showed a three-cell arrangement (Fig. 2F).
On screening for multicopy suppressors of gpi7-2, we identified five genes that could recover the growth of gpi7-2 at 35.7°C. The suppressor genes were determined to be WSC1/ SLG1, RHO2, ROM2, GFA1, and CDC5 (Fig. 3). WSC1 encodes a plasma membrane protein required for maintenance of cell wall integrity and for stress response during vegetative growth (26). The RHO2 gene product is a GTPase, a member of the rho family in the ras superfamily (27). ROM2 encodes a GDP-GTP exchange factor for Rho1p and Rho2p that can be activated by cell wall defects (27). GFA1 encodes a glutamine-fructose-6phosphate aminotransferase, a key enzyme that catalyzes the first step in the pathways for chitin, GPI, and N-glycosylation biosynthesis (28). CDC5 encodes a serine/threonine protein kinase required for mitotic exit (29,30). The presence of these multicopy suppressor genes suggests that gpi7-2 may be defective in cell wall integrity and/or the cell cycle.
Microscopic Observation and Flow Cytometric Analysis of gpi7 Mutants Carrying Multicopy Suppressors-The gpi7 mutants showed an abnormal cell arrangement with two daughter cells being attached to a mother cell at restrictive temperature (Fig. 4B). We further examined whether multicopy suppressors might recover the cell separation defect of the gpi7-2 mutant. Interestingly loose peaks higher than 3C peaks were observed in gpi7-2 strains carrying multiple copies of WSC1 (Fig. 4C), RHO2 (data not shown), ROM2 (data not shown), and GFA1 (Fig. 4D)  a unique phenotype with more than three cells attached to each other without cell separation (Fig. 4, C and D).
Diffuse peaks containing more than 3C were also observed in gpi7-2 strains carrying multiple copies of the CDC5 gene (Fig.  4E). These unusual peaks resulted from the overexpression of CDC5 in gpi7 mutants because they were not observed in wild-type cells carrying multiple copies of CDC5, indicating the same profile as for the wild-type cells (data not shown). The gpi7-2 strain carrying multiple copies of CDC5 proliferated without cell separation (Fig. 4E), consistent with the appearance of loose peaks in the gpi7-2 strain carrying multiple copies of CDC5 in the flow cytometric analysis.
Genetic Interaction between GPI Biosynthesis and the Cbk1p Pathway-We also used a transposon insertion mutagenesis method (31) to identify mutations that can bypass the requirement of GPI7 for cell growth. From this screening, we obtained a mutant that could suppress the growth defect of gpi7 deletion mutant cells at 35.7°C. Sequencing the DNA region adjacent to the transposon revealed that it is inserted into the carboxyl  3. Multicopy suppressors of gpi7-2 mutant. The suppressor genes were identified as WSC1, RHO2, ROM2, GFA1, and CDC5. gpi7-2 cells that harbor multicopy plasmids carrying the genes indicated were grown in synthetic medium (SCϪLeu for gpi7Ϫ2 cells carrying multiple copies of WSC1, RHO2, GFA1, and CDC5 or SC-Ura for ROM2). Cells were washed and adjusted to a density of 10 7 cells/ml. Five-microliter aliquots of 10-fold serial dilutions were spotted on YPAD plates and incubated at 30 or 35.7°C for 3 days. terminus of CBK1 (Fig. 5A). The insertion site was located 2040 base pairs downstream of the start codon of the 2268-base pair open reading frame of CBK1. Since the predicted phosphorylation site Thr-743 in the conserved carboxyl-terminal region of Cbk1p is deleted (32), it is likely that this transposon insertion results in a loss of function allele of CBK1. Cbk1p plays important roles in polarized growth and cell separation in S. cerevisiae (33,34).
To investigate the genetic interaction between GPI7 and CBK1 more precisely, CBK1 was disrupted in gpi7⌬ mutants. The double mutation gpi7⌬ cbk1⌬ overcame the temperaturesensitive phenotype of gpi7⌬ at 35.7°C (Fig. 5B). It is known that CBK1 regulates cell separation via an Ace2p-dependent pathway and polarized growth via an Ace2p-independent pathway (33,35). To examine whether the suppression of gpi7⌬ cells by cbk1⌬ disruption depends on ACE2 or not, ACE2 was disrupted in the gpi7⌬ background. The gpi7⌬ ace2⌬ double mutant could grow as fast as the gpi7⌬ cbk1⌬ double mutant at 35.7°C (Fig. 5B), indicating that the disruption of the Cbk1p-Ace2p pathway bypasses the growth defect of the gpi7 mutant. Ace2p activates a set of genes that is essential for motherdaughter cell separation, such as CTS1 and SCW11, that encode proteins involved in degradation of the septum (14). It is reported that EGT2, which is also regulated by the Cbk1p-Ace2p pathway, encodes a GPI-anchored protein (15,36). EGT2 is specifically expressed at the early G 1 phase in daughter cells, and it has been proposed that Egt2p is a glucanase or a regulator of glucanase (13,37). The egt2⌬ mutants were shown to have defects in cell separation (37). These reports prompted us to investigate the relationship between GPI7 and EGT2. Deletion of EGT2 partially suppressed the growth defects of gpi7 single mutants at the non-permissive temperature (Fig. 5B). We also disrupted another glucanase encoded by BGL2 to check the specificity of suppression. Bgl2p is a non-GPI type endoglucanase, and the expression is not regulated by Ace2p (38). Mutation in BGL2 did not suppress the temperature sensitivity of gpi7 cells (Fig. 5B).
By microscopic observation, we found that the two daughter cells could not separate from the mother cell, and the cell cycle was arrested in gpi7 mutants (Figs. 4B and 5C). Since the Cbk1p-Ace2p pathway is involved in cell separation, the cell morphology was investigated in suppressor mutants. We found that gpi7⌬ cbk1 mTn cells grow as clumps of round cells joined in regions that stained brightly with calcofluor white, a chitinbinding dye (Fig. 5C). This phenotype looks identical to the phenotype of cbk1⌬ single mutant cells that show a severe cell separation defect (33). The gpi7⌬ cbk1⌬ and gpi7⌬ ace2⌬ double mutants also showed serious defects in cell separation (Fig.  5C). The gpi7⌬ egt2⌬ double mutant cells showed defects in cell separation, but the phenotype was less severe than that of the gpi7⌬ cbk1⌬ or gpi7⌬ ace2⌬ double mutant (Fig. 5C).
Localization of Egt2p in gpi7 Mutants-Daughter-specific proteins for cell separation are transported to the septum and degrade septum components from the daughter's side (12,15). To explain why a mutation in EGT2 overcomes the temperature sensitivity of gpi7 mutants, we investigated the localization of Egt2p. EGT2 tagged with a sequence encoding HA after a signal peptide sequence (pRS315-HA-EGT2, Fig. 6A) was introduced into egt2⌬ cells. The HA-tagged protein was functional because no cell separation defect was observed in egt2⌬ cells carrying pRS315-HA-EGT2, whereas egt2⌬ cells carrying a control vector had such a defect (data not shown). Fluores- cence microscopy revealed that HA-Egt2p specifically localized at the septum in wild-type cells (egt2⌬ carrying pRS315-HA-EGT2, Fig. 6B). However, in the gpi7⌬ background, HA-Egt2p dispersed in the cell wall, sometimes concentrating in punctate and cortical structures (gpi7⌬ egt2⌬ carrying pRS315-HA-EGT2, Fig. 6B). ENG1/DSE4, which is one of the daughterspecific genes regulated by Ace2p, encodes a secreted endoglucanase (13). We also analyzed the distribution of Eng1p. Eng1p was localized to the septum during cell separation (eng1⌬ carrying pRS315-HA-ENG1, Fig. 6C) (13). Even in the gpi7 mutant cells (gpi7⌬ eng1⌬ carrying pRS315-HA-ENG1), Eng1p was localized at the septum (Fig. 6C). These observations indicate that only the GPI-anchored proteins, such as Egt2p, are not transported to the septum correctly in gpi7 mutants, resulting in the defect in cell separation.
Finally we constructed a chimeric gene with ENG1 and EGT2 to examine the cellular localization of the product. The carboxylterminal region of Egt2p containing the GPI attachment signal sequence was fused to the carboxyl-terminal end of HA-Eng1p (named HA-Eng1-Egt2p). As a control, localization of HA-Eng1p was observed in cells with the same background. HA-Eng1p was localized to the septum in both wild-type and gpi7⌬ mutant cells (Fig. 7B), consistent with the result in Fig. 6C. In wild-type cells, HA-Eng1-Egt2p was localized at the septum (Fig. 7C, WT  pRS315-HA-ENG1-EGT2). In contrast, the chimeric protein was not only localized at the septum but also displaced to the cell cortex in the gpi7⌬ mutant (Fig. 7C, gpi7⌬ pRS315-HA-ENG1-EGT2). These results indicate that modification of the GPI anchor in the carboxyl-terminal region of Egt2p is responsible for its localization at the septum in wild-type cells and its displacement from the septum to the cell cortex in gpi7 mutant cells.

DISCUSSION
GPI7 encodes a transferase that adds a side chain EtN-P to the second mannose of the GPI core structure (5). However, the biological significance of this addition has remained unknown.
In this study, we demonstrated that the function of GPI7 is required for the completion of cell separation. First, we found that gpi7 mutants are abnormal in their cell morphology, showing defects in cell separation and daughter cell growth (Figs. 2B and 5C). Second, we isolated WSC1, RHO2, ROM2, GFA1, and CDC5 as multicopy suppressors of the gpi7 mutants (Fig. 3), suggesting that gpi7 mutant cells may have defects in cell wall integrity and/or the cell cycle. We also showed that multicopy suppressors overcome the growth defect of gpi7 mutants but not the cell separation defect (Fig. 4). Third, we found that mutations in the Cbk1p-Ace2p pathway bypass the temperature-sensitive growth of gpi7 mutants (Fig. 5, A and B). Finally, a GPI-anchored protein, Egt2p, that is required for cell separation was not transported to the septum correctly in gpi7 mutants (Fig. 6). In summary, these results indicate that GPI7 is involved in cell separation via targeting of daughter-specific GPI-anchored proteins.
In wild-type cells, cytokinesis occurs when cells exit mitosis together with actomyosin ring contraction and septum formation to accomplish a fission of mother and daughter cell cytoplasms (12,39,40). After cytokinesis, Cbk1p localizes at the bud neck and activates Ace2p. Ace2p up-regulates the expression of daughter-specific genes such as CTS1, SCW11, DSE1, DSE2, DSE3, ENG1, and EGT2. Notably Egt2p is modified by a GPI anchor in the ER after protein synthesis (36). Daughterspecific cell wall proteins are transported to the bud neck and FIG. 5. Disruption of the Cbk1p-Ace2p pathway overcomes the growth defect of gpi7⌬ cells. A, transposon-inserted CBK1 (cbk1 mTn ) suppresses the growth defect of gpi7⌬. The gpi7⌬ and gpi7⌬ cbk1 mTn cells were streaked on a YPAD plate and incubated at 35.7°C for 3 days. B, both cbk1⌬ and ace2⌬ suppress the growth defect of gpi7⌬, whereas egt2⌬ partially suppresses the growth defect of gpi7⌬. The gpi7⌬, gpi7⌬ cbk1⌬, gpi7⌬ ace2⌬, gpi7⌬ egt2⌬, and gpi7⌬ bgl2⌬ cells were spotted onto YPAD plates and incubated at the indicated temperatures for 2 days. C, cells were incubated in synthetic medium at 25°C for 12 h and then incubated at 35.7°C for 4 h. They were stained with calcofluor white (CFW) and observed with fluorescence microscopy (right panels) to highlight mother/daughter junctions and cell shape. Differential interference contrasts (DIC) are shown on the left. Bar, 10 m.
degrade the components of the septum such as ␤-1,3-glucan and chitin (Fig. 8). After that, wild-type cells are ready to proceed through phases G 1 and S. In the case of gpi7 mutants, it is likely that Egt2p is not transported or localized to the bud neck correctly, resulting in a defect of cell separation (Fig. 6B). Mislocalized Egt2p probably injures cell wall components because it has glucanase activity (Fig. 8). The weakened cell wall layer might give rise to cellular stress, arresting the cell cycle in the daughter cell at the G 1 phase. The mother cell could proceed to the next stage of the cell cycle despite the arrest of the daughter cell. This might be the reason why gpi7 mutants have defects in cell separation and daughter cell growth, showing a three-cell arrangement.

Roles of Multicopy Suppressors in gpi7
Mutants-Microscopic observation revealed that gpi7 mutants carrying multicopy suppressors grew as clumps of round cells without completion of cell separation (Fig. 4). Multicopy suppressors could suppress the growth defect of gpi7 mutants but not the cell separation defect (Fig. 8). We isolated two classes of suppressors. The first class of suppressors, such as WSC1, RHO2, ROM2, and GFA1, facilitates the production of cell wall components ␤-1,3-glucan and chitin. It is known that Wsc1p, Rho2p, and Rom2p act in the same pathway and are activated by cellular stress such as cell wall damage (41,42). Protein kinase C, Pkc1p, which was also isolated as a multicopy suppressor of gpi7⌬ (7), is the downstream component of this pathway. FKS1 and its homologue FKS2, MNN1, and CHS3, FIG. 6. Egt2p is displaced from the septal region to the cell cortex in gpi7⌬ mutant cells. A, construction of 3xHA-tagged EGT2 and ENG1 genes. The 3xHA tag was inserted after a sequence corresponding to the signal sequence (78 or 600 bp from the start codon in the EGT2 or ENG1 open reading frame, respectively). It has the promoter and terminator of EGT2 or ENG1, respectively. B, HA-Egt2p was visualized by indirect immunofluorescence assay in wild-type (egt2⌬ pRS315-HA-EGT2) and gpi7⌬ (gpi7⌬ egt2⌬ pRS315-HA-EGT2) cells. M, D1, and D2 indicate mother cell, first daughter cell, and second daughter cell, respectively. DIC, differential interference contrasts. Bar, 5 m. C, HA-Eng1p was visualized by indirect immunofluorescence assay in wild-type (eng1⌬ pRS315-HA-ENG1) and gpi7⌬ (gpi7⌬ eng1⌬ pRS315-HA-ENG1) cells. which encode subunits of the ␤-1,3-glucan synthase complex, an ␣-1,3-mannosyltransferase for mannan synthesis, and a chitin synthase, respectively, have been shown to depend on this pathway for their full expression (43). GFA1 is also involved in chitin synthesis because Gfa1p catalyzes the first step in the production of N-acetylglucosamine, which is a component of chitin. It has been reported that Gfa1p is the ratelimiting enzyme in chitin production (44). Since each suppressor gene, WSC1, RHO2, ROM2, and GFA1, promotes the production of cell wall components, the temperature-sensitive growth of gpi7 cells that might be due to a defect in cell wall integrity could be overcome by supplementation of cell wall components (Fig. 8). In contrast, CDC5, the second class of suppressor, seems to be different from the others. Cdc5p is a yeast orthologue of polo-like kinase, a conserved protein kinase family found in many eukaryotic organisms from yeast to mammals. It plays multiple key roles in the coordination of mitosis (30,45). CDC5 was identified as a multicopy suppressor of various mutants related to progression of the M phase, such as CDC15, CDC20, DBF2, and TEM1 (46,47). CDC5 was also obtained as a multicopy suppressor of an allele of DBF4, which arrests at the G 1 phase of the cell cycle (48). Based on these studies, it is likely that the overproduction of CDC5 overcomes the cell cycle arrest of the daughter cells in gpi7 mutants (Fig. 8).
Relationship between GPI7 and Cbk1p-Ace2p Pathway-We also found that the mutation in CBK1 overcomes the temperature-sensitive growth of gpi7 mutants through extragenic suppressor screening (Fig. 5A). The double mutation gpi7⌬ ace2⌬ suppressed the growth defect of gpi7 mutant cells as did the mutation gpi7⌬ cbk1⌬ (Fig. 5B). Cbk1p is a protein kinase belonging to the nuclear Dbf2-related family that is highly conserved from yeast to human (32). Nuclear Dbf2-related family kinases are important regulators of cell morphogenesis and cell proliferation. In S. cerevisiae, Cbk1p is important for polarized growth, mating projection formation, and cell separation (35). The cbk1⌬ cells fail to separate after mitosis and grow in clumps (33,35). Interestingly both gpi7⌬ cbk1⌬ and gpi7⌬ ace2⌬ double mutants showed a severe defect in cell separation, which is similar to the phenotype of cbk1⌬ and ace2⌬ single mutant cells that grow as clumps joined at the septum (Fig. 5C).
It is likely that the suppression of the gpi7 growth defect by the inactivation of Ace2p is due to the down-regulation of genes whose expression is regulated by Ace2p. Ace2p activates a set of daughter cell-specific genes that are essential for motherdaughter cell separation (14,15). These daughter-specific proteins are transported to the bud neck after cytokinesis and formation of the septum and degrade the septum from the side of the daughter cell (12)(13)(14). At least one gene required for cell separation, EGT2, encodes GPI-anchored proteins presumed to be glucanases (15,36). In gpi7 mutants, GPI-anchored proteins are modified by a GPI anchor lacking a second EtN-P, which is added by Gpi7p. Mutations in EGT2 (egt2⌬) partially suppressed the temperature-sensitive growth defect of gpi7 mutants, whereas mutation in BGL2 (bgl2⌬), which encodes a non-GPI glucanase and is not regulated by Ace2p, could not suppress the defect at all (Fig. 5B). Egt2p was not transported to the bud neck correctly in gpi7 mutant cells (Fig. 6B), suggesting that daughter-specific GPI-anchored proteins for cell separation cause cellular stress in gpi7 mutants and affect the cell cycle (Fig. 8). Although the gpi7 cbk1 and gpi7 ace2 double mutants have a cell separation defect, daughter cells can grow because daughter-specific proteins, which are required for cell separation and may injure the cell wall if mislocalized, are not expressed (Fig. 8).
Eng1p was transported to the septum in both wild-type and gpi7 mutant cells (Figs. 6 and 7). Eng1p, which was tagged with HA at the carboxyl terminus, was also reported to be transported correctly to the septum (13), indicating that the carboxyl-terminal tagging of Eng1p itself did not change its localization. We observed a displacement of Eng1-Egt2 chimeric protein to the cell cortex in gpi7⌬ cells but not in wildtype cells (Fig. 7). Therefore, this mislocalization depends on GPI anchoring of Eng1p. These results suggest that there exist at least two different mechanisms for the localization of proteins to the septum region, i.e. GPI-dependent and GPI-independent mechanisms.
Specificity of Cell Separation Control by GPI7-It is important to address whether the regulation of cell separation is FIG. 8. Summary of the regulation of cell separation by GPI7. Cbk1p and Ace2p activate the expression of daughter-specific genes. Egt2p, but not Eng1p, is modified by a GPI anchor in the ER. Eng1p and Egt2p are transported to the bud neck. In gpi7 mutants, Egt2p is transported to the cell surface but not to the bud neck. Multicopy suppressors overcome the temperature-sensitive growth defect of gpi7 mutants but not the cell separation defect. general to any GPI biosynthetic genes or specific to the GPI7 gene. Previous study indicated that gpi1 mutant cells, which are defective in the first step of GPI biosynthesis, were large, round, and budded with a separation defect at a semipermissive temperature (49). The gwt1-20 cells, which have a defect in inositol acylation of GPI, showed a swelling of whole cells, clumping of cells, and in some cases cell lysis (50). However, the defective phenotypes of gpi1 and gwt1-20 mutants are dissimilar to those of gpi7 mutant cells. In gpi7 mutants, GPI anchors are still transferred to proteins (5,6,9), whereas the transfer to proteins is defective in other mutants because the biosynthesis of GPI is stopped at the intermediate stage (49 -53). In gwt1-20 cells, Egt2p was not detected at the cell surface after incubation at the non-permissive temperature, consistent with the previous result that Egt2p is modified by the GPI anchor (36) (data not shown). Furthermore mutation in CBK1 or ACE2 could not suppress the temperature sensitivity of gwt1-20 (data not shown). Therefore, it is likely that the cell separation defect observed in gpi7 mutants is not general among GPI biosynthesis but that the addition of EtN-P to the side chain of the second mannose of the GPI glycan portion by Gpi7p is required for the completion of cell separation.
Reasons for Mislocalization of GPI-anchored Proteins Required for Cell Separation-The gpi7 mutants have defects in cell separation and daughter cell growth. It is likely that the mislocalization of daughter-specific GPI-anchored proteins causes a daughter-specific cell cycle arrest in gpi7 mutants (Figs. 5 and 6). Previous studies present several explanations for the cell separation defect and mislocalization of Egt2p in gpi7 mutants. First, GPI-anchored proteins are components of lipid rafts, which are sphingolipid-and sterol-rich microdomains of the membrane. Polarization of sterol-rich domains to the shmoo tip was observed in mating pheromone-treated cells in S. cerevisiae (54). The mating efficiency decreased to 1 ⁄100 of the wild-type level when both mating partners were gpi7⌬ strains (8). Lipid raft formation also contributes to the hyphal growth that is required for virulence in C. albicans (55). Interestingly gpi7⌬/gpi7⌬ cells were defective in pseudohyphal formation in C. albicans (10). Lipid rafts have been implicated in membrane trafficking and signaling in mammalian cells (56). In yeast, lipid raft formation is essential for the correct targeting of several proteins (54,57,58). Ceramide remodeling on GPI-anchored proteins in the Golgi and plasma membrane was significantly reduced in gpi7⌬ mutants (5). The lipid remodeling process might help GPI proteins to sort into membrane subdomains. It is possible that the addition of EtN-P to the GPI glycan portion by Gpi7p may be required for stability of lipid rafts, giving rise to the mislocalization of GPI proteins in gpi7 mutants. Second, GPI-anchored proteins are transported from the ER to the Golgi apparatus in vesicles distinct from those containing non-GPI proteins. Particular components are required for sorting GPI-anchored proteins from other secretory proteins upon exit from the ER (59,60). The p24 family is thought to be components of cargo receptors for GPI-anchored proteins. Emp24p, a member of this family, is necessary for efficient packaging of Gas1p into ER-derived vesicles and can be directly cross-linked to Gas1p (61). EMP24 was also identified as a BST2 gene that bypasses the requirement for Sec13p in coat protein complex II (COPII) vesicle formation (62). Recently it was reported that BST1, which also bypasses sec13 mutant, encodes a functional yeast homologue of mammalian GPI inositol deacylase (63). Therefore, another possibility is that particular cargo receptor(s), such as the members of the p24 family, might recognize the EtN-P side chain of the GPI either to concentrate GPI-anchored proteins or to form a vesicle for correct targeting at the cell separation stage. Although these possibilities should be addressed further in the future, to our knowledge this is the first report indicating that a complete GPI structure is essential for the correct targeting of GPIanchored proteins.