SAC1 Encodes a Regulated Lipid Phosphoinositide Phosphatase, Defects in Which Can Be Suppressed by the Homologous Inp52p and Inp53p Phosphatases*

The yeast protein Sac1p is involved in a range of cellular functions, including inositol metabolism, actin cytoskeletal organization, endoplasmic reticulum ATP transport, phosphatidylinositol-phosphatidylcholine transfer protein function, and multiple-drug sensitivity. The activity of Sac1p and its relationship to these phenotypes are unresolved. We show here that the regulation of lipid phosphoinositides in sac1 mutants is defective, resulting in altered levels of all lipid phos- phoinositides, particularly phosphatidylinositol 4-phosphate and phosphatidylinositol 4,5-bisphosphate. We have identified two proteins with homology to Sac1p that can suppress drug sensitivity and also restore the levels of the phosphoinositides in sac1mutants. Overexpression of truncated forms of these suppressor genes confirmed that suppression was due to phosphoinositide phosphatase activity within these proteins. We have now demonstrated this activity for Sac1p and have characterized its specificity. The in vitro phosphatase activity and specificity of Sac1p were not altered by some mutations. Indeed, in vivo mutant Sac1p phosphatase activity also appeared unchanged under conditions in which cells were drug-resistant. However, under different growth conditions, both drug sensitivity and the phosphatase defect were manifest. It is concluded that SAC1 encodes a novel lipid phosphoinositide phosphatase in which specific mutations can cause the sac1phenotypes by altering the in vivo regulation of the protein rather than by destroying phosphatase activity.

The Saccharomyces cerevisiae gene SAC1 was originally isolated as a mutant allele able to suppress the phenotypes seen in strains containing the temperature-sensitive act1-1 actin allele (1). The gene was subsequently cloned when SAC1 was also identified as a mutant suppressor of mutations in the yeast phosphatidylinositol-phosphatidylcholine transfer protein, sec14-1 (2). In addition to the ability to suppress the effects of act1-1 and sec14-1 mutations, a variety of perplexing additional phenotypes have been attributed to yeast strains containing an assortment of mutations in the SAC1 gene (sac1-1 to sac1-29, mds1-sac1) (1)(2)(3). These include disruption to the localization of the actin cytoskeleton and chitin deposition, reminiscent of the act1-1 phenotype (1). Some, but not all, mutant sac1 strains also show inositol auxotrophy (3,4); however, there are no apparent defects in de novo inositol biosynthesis or in utilization of inositol for phosphatidylinositol biosynthesis (4). Some, but not all, mutant sac1 strains also display multiple-drug sensitivity characterized by supersensitivity to a wide variety of drugs and detergents, including brefeldin A and novobiocin (3). Mutants also show secretory defects and deficiencies in ATP transport activity in the endoplasmic reticulum as well as an inability to translocate nascent prepro-␣-mating factor and preprocarboxypeptidase Y (5,6). In addition to these phenotypes, sac1 mutants also show cold sensitivity (1,2) and synthetic lethality when in combination with the trp1 allele (3). Despite this considerable body of work, a function for Sac1p that relates to phenotypic changes has remained to be described.
New insight as to the function of Sac1p became possible during the cloning of the presynaptic inositol-5-phosphatase synaptojanin-1 (7) when a region of considerable homology between this protein and Sac1p was identified. The synaptojanins are phosphatidylinositol-polyphosphate 5-phosphatases that remove phosphate from the D-5-hydroxyl position of phosphatidylinositol phosphates. They are members of a large group of 5-phosphatases (reviewed in Ref. 8) that also include the type I and II 5-phosphatases. All 5-phosphatases contain a 5-phosphatase domain; the type II 5-phosphatases additionally contain a type II domain; and the synaptojanin-like proteins, in addition to the type II domain, contain a region of homology to Sac1p, the Sac domain (see Fig. 1). Synaptojanin-1 represents one member of a family of Sac domain-containing inositide phosphatases (8) that also include the yeast proteins Inp51p, Inp52p, and Inp53p (also called Sjl1p Sjl2p, and Sjl3p) (9,10).
Despite the fact that Sac1p shows no homology to the domains in these proteins shown to catalyze phosphatase activity, we (3) and others (11,12) had speculated that phosphatidylinositol (PtdIns) 1 and its phosphorylated derivatives, the lipid phosphoinositides, could represent a medium by which Sac1p could administer its pleiotropic effects on actin function, secretion, inositol metabolism, and drug sensitivity. We describe here the direct demonstration that a function of Sac1p confers such control, albeit in a regulated manner.
Yeast Labeling and HPLC Analysis-Strains were grown in YPD or SD medium with appropriate amino acids at 30°C to A 600 ϳ 0.7 in medium containing 50 Ci/ml [ 3 H]inositol or [ 14 C]inositol (Amersham Pharmacia Biotech). Cells were pelleted and arrested by resuspension in 200 l of methanol. Lipid extraction, deacylation, and head group analysis by HPLC were carried out as described previously (18).
Characterization of Monophosphorylated Phosphatidylinositol-Mutant sac1-deleted cells were labeled for lipid extraction, deacylation, and separation with [ 14 C]inositol as described above. Fractions containing the monophosphorylated head groups were pooled and neutralized with triethylamine, diluted 10-fold in water before desalting (19). The GroPIns(3)P and GroPInsP standards were purified from 3 H-radiolabeled yeast (as described). The GroPIns(5)P standards were generated from PtdIns(3,5)P 2 isolated from 3 H-radiolabeled yeast. The [ 3 H]PtdIns (3, 5)P 2 was deacylated, purified, and desalted as described above and then incubated with 0.5 ml of erythrocyte ghosts as described (18) for 3-6 h at 37°C in 12 mM Hepes/KOH (pH 7.5), 5 mM EDTA, and 1 mM EGTA. The ghosts were then diluted to 1.5 ml, and perchloric acid was added to 1 M before incubation on ice to precipitate protein. The samples were neutralized with 2 M KOH and 0.5 M Hepes and again incubated on ice for 30 min before pelleting at 15,000 ϫ g for 10 min at 4°C to remove potassium perchlorate and protein. HPLC of GroPInsPs was performed on a Partisphere 5-m strong-anion exchange column (250 ϫ 40 mm) with a flow rate of 1 ml/min using the following gradient separation: 0 min, 0% solvent B; 5 min, 0% solvent B; 15 min, 2% solvent B; and 100 min, 2% solvent B (solvent A ϭ H 2 O and solvent B ϭ 1.25 M (NH 4 ) 2 HPO 4 (pH 3.8)).
Phosphatase Assays-The phosphatase activity of proteins was determined using colorimetric or radioactive assays. Activity for yeast 5-phosphatase (Inp52p) was determined by incubation for 5 min at 37°C in the presence of 0.125 mM lipid phosphoinositide, 0.25% (w/v) octyl glucoside, 4 mM MgCl 2 , and 0.1 M Tris (pH 7.4). The released phosphate was then assessed colorimetrically as described before (20). Briefly, the 20-l assay was stopped by adding 25 l of malachite green/molybdate reagent, followed by 80 l of water. The reaction was allowed to take place for 30 min before measurements were taken at 610 nm in an enzyme-linked immunosorbent assay reader (SpectraMax Plus, Molecular Devices). Basal phosphate contents were determined by adding the enzyme after termination of the reaction with the malachite green/molybdate reagent.
The phosphatase activity of the Sac1p phosphatases was assessed in the presence of 10 M lipid phosphoinositide, 0.25% (w/v) octyl glucoside, 2 mM MgCl 2 , and 100 mM Tris (pH 7.4) for 20 min at 37°C using radioactively labeled substrate that was prepared as described below. Phosphatidylinositol monophosphates were labeled with [␥-32 P]ATP using PtdIns 3-and 4-kinases with PtdIns as a substrate (21). PtdIns(5)P was incubated in the presence of a 4-fold excess of phosphatidylethanolamine with phosphoinositide 3-kinase to produce PtdIns(3,5)P 2 . The 32 P-labeled lipids were separated by TLC (16), purified, and supplemented with unlabeled lipids (Echelon) before assay (21). The assay (50 l) was stopped by adding 0.4 ml of methanol/CHCl 3 (1:1), followed by 0.2 ml of 1 N HCl and 10 l of Folch bovine brain extracts (Sigma) as carrier lipid (17). After phase separation, both phases were dried down under vacuum. The water phase was directly counted (scintillation), whereas the CHCl 3 phase was subjected to TLC (17). After development, PtdInsP and PtdInsP 2 spots were scrapped and counted in scintillant. In both types of assay, phosphoinositides (Echelon) and phosphatidylinositol (Sigma) were first repurified by chloroform extraction according as described (22) and subsequently quantified by ashing the lipids, followed by a colorimetric phosphate assay (23).

Inp52p and Inp53p Suppress sac1 Mutant Phenotypes-We
had previously demonstrated that the mds1-sac1 allele of the gene SAC1 and some other sac1 mutants display multiple-drug sensitivity, a phenotype characterized by sensitivity to drugs such as novobiocin and brefeldin A (3). Using high-and lowcopy wild-type yeast genomic libraries, we had screened for sequences able to suppress sac1 multiple-drug sensitivity and had identified a variety of sequences able to complement the mutant (3). However, the analysis was not comprehensive, and we had speculated that other open reading frames may be able to complement sac1 mutant phenotypes. Principally, we were interested in testing the genes that had been described as encoding proteins with homology to Sac1p, namely the S. cerevisiae genes INP51, INP52, INP53, and FIG4, and, in addition, human synaptojanin-1. To examine the ability of these genes to complement sac1 mutant phenotypes, we transformed the genes, cloned into an inducible expression vector, into wildtype and sac1-deleted strains. Transformed strains were then tested for growth on plates containing a variety of drugs and inhibitors under conditions that would suppress (glucose) or promote (galactose) induction of the plasmid-borne gene. Expression of the genes in wild-type yeast appeared to have no effects on growth (data not shown), whereas expression of the genes in the sac1 mutant showed significant differences. Yeast sac1 mutants grew slightly slower than wild-type cells; however, mutants transformed with wild-type SAC1, INP52, or INP53 and grown on galactose grew at wild-type rates (Fig. 1). When grown on medium containing inhibitory concentrations of drugs, the effect was more pronounced. Yeast sac1 disruptants were drug-sensitive and did not grow; however, mutants transformed with wild-type SAC1, INP52, or INP53 and incubated on medium containing galactose grew at wild-type rates ( Fig. 1). Therefore, the effects of sac1 mutations can be suppressed by some property contributed by overexpressed Inp52p and Inp53p phosphatases.
Altered Lipid Phosphoinositide Levels of sac1-deleted S. cerevisiae-The demonstration that sac1 phenotypes could be suppressed by phosphatidylinositol-phosphate 5-phosphatases strengthened our hypothesis that the lipid phosphoinositides may represent part of the mechanism by which Sac1p is able to mediate its pleiotropic effects (3). Other observations such as the effects of Sac1p on actin function (1), which can be regulated by PtdIns(4,5)P 2 ; the effects of Sac1p on Sec14p function, which regulates the ratio of PtdIns and phosphatidylcholine (24); and the fact that some SAC1 mutants show inositol auxotrophy (3) also suggested that the lipid phosphoinositides may be important for Sac1p function. Therefore, we sought to measure the levels of these lipids in wild-type and mutant sac1 strains. Wild-type and yeast strains containing a completely deleted SAC1 gene were incubated with [ 3 H]inositol before total cell phospholipids were extracted and deacylated to allow separation of the resulting head groups by HPLC. The elution characteristics of the inositol-containing head groups have been well characterized (17,18), allowing comparison of the lipid phosphoinositides. The data in Fig. 2A indicate that cells lacking the SAC1 gene were characterized by reduced levels of GroPIns(4,5)P 2 and an increase in GroPIns(3)P and GroPIns(3,5)P 2 . Most notably, the sac1-deleted strains showed a dramatic increase in a glycerophosphoinositide eluting at a time consistent with GroPIns(4)P. However, using the HPLC protocol described, the lipid deacylation products GroPIns(4)P and GroPIns(5)P cannot be efficiently separated (25). Having shown that Inp52p and Inp53p, both putative 5-phosphatases, could suppress the effects of sac1 mutants that appear to be characterized by a dramatic increase in a D-4-phosphorylated lipid, it remained a possibility that the increase in PtdIns(4)P was in fact a peak of PtdIns(5)P.
To establish the precise identity of the monophosphorylated glyceroinositide seen at the GroPIns(4)P position, we labeled cells containing a completely deleted sac1 gene with [ 14 C]inositol. As before, total cell phospholipids were then extracted and deacylated to allow separation of the glyceroinositides derived from lipid phosphoinositides by HPLC. Fractions containing the elevated levels of head group corresponding to the phosphatidylinositol monophosphate were collected and desalted. This purified [ 14 C]inositol compound was then mixed with 3 Hlabeled GroPIns(3)P, GroPIns(4)P, and GroPIns(5)P before sep- FIG. 1. SAC1, INP52, and INP53 suppress sac1 multiple-drug sensitivity. Sac1p and Fig4p contain Sac domains, whereas Inp51p, Inp52p, Inp53p, and human synaptojanin-1 also consist of type II phosphatase and 5-phosphatase domains in addition to a proline-rich region. FY833.sac1⌬URA3 yeast mutants were transformed with vectors (pYX243) with genes encoding these proteins regulated by a galactose (gal)-inducible and glucose (glu)-repressed promoter. Shown is the ability to grow as wild-type (ϩϩϩ) or mutant (ϩϩ) sac1 yeast or complete inhibition of growth (Ϫ) on SD plates containing glucose or galactose with or without a restrictive concentration of drugs (novobiocin, 1.8 mg/ml, or kanamycin, (100 g/ml). Mutant strains complemented with SAC1, INP52, or INP53 showed wild-type drug resistance, whereas INP51, FIG4, human synaptojanin-1 (SYN 1), and mutant mds1-sac1 (data not shown) continued to display drug sensitivity. arating the species by HPLC, confirming that the lipid species proposed to be PtdIns(4)P was indeed PtdIns(4)P (Fig. 2B). This confirmed that the 10-fold increase in lipid seen in sac1 mutants was indeed an increase in PtdIns(4)P.
We have consistently seen an ϳ7-fold increase in PtdIns(3,5)P 2 , a 1.5-fold increase in PtdIns(3)P, a 5-fold decrease in PtdIns(4,5)P 2 , and up to a 10-fold increase in PtdIns(4)P. Recently, similar analyses of lipid phosphoinositide levels in sac1 mutants also indicated that levels of PtdIns(4)P increase, by ϳ6 -10-fold (11,12). However, although there is agreement with respect to the elevation of PtdIns(4)P, PtdIns(3)P, and PtdIns(3,5)P 2 , we show here a 5-fold decrease in PtdIns(4,5)P 2 not reported before. Whether this distinction is due to strain differences or labeling protocols remains to be resolved; however, we have observed the decrease in PtdIns(4,5)P 2 levels in sac1 deletions in two different strain backgrounds (FY833 and YPH499; data not shown).
Restored Lipid Phosphoinositide Levels of Complemented sac1-deleted S. cerevisiae-Having shown that yeast sac1 mutants have dramatically altered levels of lipid phosphoinositides, yeast mutants transfected with vectors containing SAC1 and INP52 were grown in galactose to induce production of Sac1p and Inp52p and were labeled with [ 3 H]inositol to test the lipid phosphoinositide profiles of these yeast mutants. As shown in Fig. 3, in galactose, sac1 mutants continued to show defects in the levels of all of the lipid phosphoinositides compared with wild-type yeast. Transformation of the yeast mutants with a vector containing the wild-type SAC1 gene re-stored the level of lipids to almost wild-type levels. Similarly, the gene INP52 conferred a restoration of lipid levels to the mutant on induction, a feature not seen when grown on glucose (data not shown).
The Sac Domain Is Sufficient to Restore Partial Drug Resistance-Although the restoration of the level of lipid phosphoinositides by Inp52p and Inp53p could have been due to the carboxyl-terminal 5-phosphatase domain, we speculated that suppression was in fact due to the only region of homology between these proteins, the amino-terminal Sac domain (Fig.  1). To examine this, the Sac domains from Sac1p, Inp51p, Inp52p, Inp53p, Fig4p, and synaptojanin-1 were subcloned into a galactose-inducible vector. These constructs were transformed into wild-type and mutant sac1 yeast, and the growth characteristics of these strains were examined on medium containing glucose or galactose and various drugs. As shown in Table I, the construct containing the Sac domains from Sac1p, Inp52p, and Inp53p conferred a growth advantage on the mutant cells. The suppression seen with these domains was not as complete as that seen with full-length proteins, as colonies of Sac domain-containing yeast mutants appeared only several days after yeast mutants transformed with full-length genes. However, as mutant cells transformed with Sac domains from Inp52p and Inp53p both conferred the same growth advantage as the Sac domain from Sac1p, the results indicate that the Sac domains of the phosphatases are sufficient to restore wild-type phenotypes to sac1 mutants.
The Sac Domain of Sac1p Is a Lipid Phosphoinositide Phos-

TABLE I
Some Sac domains partially suppress sac1 deletion strain novobiocin sensitivity FY833.sac1⌬URA3 yeast mutants transformed with pYX243-Sac domain constructs were grown on SD plates containing glucose or galactose with or without a restrictive concentration of novobiocin (1.8 mg/ml) or kanamycin (100 g/ml). Cells transformed with constructs encoding the Sac domains from Sac1p, Inp52, and Inp53 had a slight growth advantage (ϩϩϩ) over those without (ϩϩ) when grown on galactose. These strains also showed growth (ϩ) on drug-containing gene induction plates, whereas all noninduced (glucose) and INP51, FIG4, synaptojanin-1, and mutant mds1-sac1 Sac domain-containing strains showed no growth (Ϫ). phatase-Recently, it has been demonstrated that the Sac domains of Inp52p and Inp53p, but not that of Inp51p, contain phosphatase activity (11). The differential effect of the Sac domains on the rescue of Sac1p-depleted cells questioned whether the properties ascribed to Inp52p and Inp53p Sac domains were also retained by Sac1p. This issue is emphasized by the finding that Fig4p was unable to rescue despite being the closest Sac1p-related sequence (data not shown). To determine if the Sac domain of Sac1p displayed phosphatase activity, we cloned SAC1 and a construct consisting of only the Sac domain of Sac1p into GST-tagged vectors for expression in yeast. Using a variety of radiolabeled lipid phosphoinositides, we tested GST-Sac1p for phosphatase activity. As shown in Fig.  4, Sac1p displayed activity principally against PtdIns(3)P and PtdIns(4)P. The protein showed only very slight activity for PtdIns(3,5)P 2 and no detectable activity for PtdIns(4,5)P 2 (data not shown). Of the activity seen against PtdIns(3,5)P 2 , this was principally against phosphate in the D-3-position, as, under initial rate conditions, Ͼ95% of the phosphate released was from this position (data not shown). Notably, the specific activity of Sac1p phosphatase was very low in comparison with the phosphatase activity we have characterized for the Inp52p 5-phosphatase domain (data not shown). We also tested a Sac domain construct from Sac1p that produced phosphatase activity similar to that seen in the full-length construct (data not shown). It has been demonstrated that the Sac domain from Inp53p could catalyze the conversion of PtdIns(3)P, PtdIns(4)P, and PtdIns(3,5)P 2 , but not that of not PtdIns(4,5)P 2 , to PtdIns. This pattern of specificity is similar to the profile shown here for Sac1p.
Mutation of Sac1p at Leu-246 Confers Drug Sensitivity-We have previously isolated the mds1-sac1 allele consisting of two mutations within the SAC1 gene, F97L and L246P (3). Both of the mds1 mutations occur within the highly conserved Sac domain motifs (Fig. 5A), although these mutations do not occur within the proposed catalytic RXNCXDCLDRTN motif (11). It appeared likely that one of the two mds1 mutations may be sufficient to produce the mds1 phenotypes by destroying Sac1p phosphatase activity. To investigate this possibility, sac1 constructs with only one of the mds1 mutations, F97L or L246P, were produced. To test for suppression of sac1 phenotypes, wild-type and mutant sac1 constructs were transformed into sac1-deleted strains, and we examined the ability of these strains to grow on medium containing inhibitory concentrations of various drugs with or without inositol. Some sac1 alleles confer inositol auxotrophy upon mutant strains (3,4); indeed, complete deletion of sac1 confers inositol auxotrophy on strains, whereas the mds1-sac1 allele has been show to cause no defect in inositol metabolism (3). The data in Fig. 5B confirm these findings within the strain FY833.sac1⌬URA3 and show FIG. 4. Sac1p has phosphatidylinositide phosphatase activity. GST-tagged Sac1p was expressed and purified from yeast and tested for phosphatase activity against 32 P-labeled PtdIns(3)P, PtdIns(4)P, PtdIns(3,5)P 2 , and PtdIns(4,5)P 2 . The protein showed activity principally against monophosphorylated phosphatidylinositides, and no activity could be detected for PtdIns(4,5)P 2 (data not shown).  SYN 1, synaptojanin-1. B, FY833.sac1⌬URA3 yeast mutants were grown on SD plates with or without a restrictive concentration of novobiocin (nov; 1.8 mg/ml) or inositol (ino). Growth (ϩϩϩ) occurred on novobiocin only when cells were transformed with vectors containing wild-type SAC1 or mutant sac1-F97L. Cells transformed with vector containing the sac1-L246P or mds1-sac1 construct did not grow on novobiocin, indicating that the L246P mutation is responsible for the phenotype. Inositol auxotrophy, not a phenotype of mds1-sac1 alleles, was suppressed buy all versions of SAC1.
that wild-type SAC1 and mutant sac1-F97L genes conferred wild-type drug resistance, whereas mutant sac1-L246P and mds1-sac1 genes did not restore drug resistance. Additionally, all the mutant alleles tested restored inositol prototrophy. Thus, the L246P mutation is likely to be solely responsible for the multiple-drug sensitivity seen in mds1 mutants.
The mds1-sac1 Mutations Do Not Destroy Phosphatase Activity-To establish whether the mutation responsible for drug sensitivity was also responsible for the changes in lipid levels, SAC1 containing either one or both of the mds1 mutations, F97L or L246P, was cloned into a GST-tagged vector for expression in yeast. These vectors were transformed into yeast, and proteins were extracted, purified, and tested for phosphatase activity as described above. As shown in Fig. 6, the in vitro activity of mutants Sac1p-F97L, Sac1p-L246P, and Sac1p-mds1 did not differ from that of wild-type Sac1p (see Fig. 4 for comparison). This was rather surprising since the inability to rescue the drug sensitivity suggested that mutation at Leu-246 would destroy, reduce, or at least alter specificity of the activity of the Sac1p phosphatase to produce the observed multipledrug sensitivity.
The mds1-sac1 Mutations Confer Sac1p Phosphatase Regulatory Defects-To further assess the consequences of the mutations within Sac1p, [ 3 H]inositol-labeled yeast cells containing a completely deleted SAC1 gene were transformed with wildtype and mutant sac1 clones. Cells were grown in YPD medium, and as described above, total cell phospholipids were then extracted and deacylated to allow separation by HPLC. The complemented mutant showed restored levels of lipid phos-phoinositides ( Fig. 7A) with both wild-type and mutant sac1 genes. However, we did observe that mutant cells containing the sac1 genes with the L246P mutation showed slightly less restoration of the level of lipid phosphoinositides (Fig. 7). We also [ 3 H]inositol-labeled yeast mutants grown in SD medium and again extracted total phospholipids for examination by HPLC. Under these conditions, sac1-L246P and mds1-sac1 were unable to restore the levels of all of the lipid phosphoinositides to those seen with wild-type and sac1-F97L clones (Fig. 7, B and C). Indeed, although in minimal medium the level of PtdIns(4)P in the sac1 mutant increased only ϳ2-fold (Fig. 7B), both Sac1p-L246P and double-mutant Sac1p were completely unable to reverse this increase. It appears that the phosphatase activity of the SAC1P-L246P mutants was only slightly decreased in mutants grown in YPD medium, whereas in minimal medium, the activity was more dramatically affected. This observation was confirmed when we failed to observe multiple-drug sensitivity in sac1 deletion strains transformed with Sac1p-L246P clones when grown on YPD medium (data not shown); drug sensitivity was only seen on minimal medium (Fig. 5B). DISCUSSION The study here demonstrates that in Sac1p-depleted cells, the regulation of lipid phosphoinositides is defective and that this affects all phosphoinositides, including PtdIns(4,5)P 2 . The work also indicates that Sac domains from Inp52p and Inp53p exhibit phosphatase activity that can suppress drug sensitivity and phosphoinositide regulation defects in sac1 mutants. In  7. Mutant mds1-sac1 and sac1-L246P continue to show aberrant phosphatidylinositide levels. Shown are data from HPLC analysis of [ 3 H]inositol-labeled lipids from sac1 deletion mutant FY833.sac1⌬URA3 transformed with pRS313.SAC1, pRS313. SAC1F97L, pRS313.SAC1L246P, or pRS313.MDS1 grown in YPD/ glucose medium (A) or SD/glucose medium (B and C) (% total lipid phosphoinositides). Constructs containing the L246P mutation appeared to have slightly reduced phosphatase activity when grown in YPD medium; however, when grown in SD medium, sac1 deletion strains transformed with these mutants showed an inability to restore the levels of the phosphoinositides. addition, we have demonstrated that Sac1p exhibits lipid phosphoinositide activity and that the enzyme will dephosphorylate monophosphorylated phosphatidylinositol and PtdIns(3,5)P 2 , but not PtdIns(4,5)P 2 . Most significantly, we have shown that mutations conferring sac1 phenotypes, which had been predicted to be a result of loss of phosphatase function, show an unchanged in vitro phosphatase activity and specificity. This appeared to be the case in vivo; however, we went on to identify conditions under which the mutant protein conferred phosphoinositide defects and drug sensitivity. We therefore conclude that the conserved motifs within the Sac domain mutated in Mds1p-Sac1p confer critical in vivo regulation of Sac1p activity.
As Sac1p appears to exhibit a similar phosphatase activity to Sac domains from Inp52p and Inp53p, it is perhaps not surprising that these proteins are able to suppress sac1 mutant phenotypes. The fact that Inp51p is unable to suppress sac1 defects is not unexpected. Although clearly containing a well defined Sac domain, Inp51p contains a variety of amino acid substitutions (in the proposed catalytic RXNCXDCLDRTN motif) that might be expected to destroy activity. The evidence from this work and that of Ref. 11 confirms this prediction, although this remains to be proven formally. Fig4p, on the other hand, shows no significant mutations within the proposed catalytic motif or other motifs within the Sac domain (data not shown) and might therefore be expected to exhibit phosphatase activity. However, deletion of the gene does not cause significant defects in lipid phosphoinositide levels (11), perhaps indicating that the protein does not represent a major activity under these conditions. The protein is induced upon pheromone stimulation (26), and perhaps activity is regulated under these conditions much as demonstrated here for Sac1p. In contrast, the results obtained with synaptojanin were unexpected. The Sac domain of synaptojanin-1 probably shows phosphatase activity comparable to that of Inp52p and Inp53p (11), and we would have therefore expected synaptojanin to suppress sac1 defects. We have confirmed that the domain has phosphatase activity, which is significantly higher than that of Sac1p. 2 The confirmation of expression of the protein and characterization of the phosphatase activity and the lipid phosphoinositide specificity of synaptojanin should resolve these differences.
Characterization of the phosphatase activity and the lipid phosphoinositide specificity of Sac1p has established that Sac1p displays phosphatidylinositol monophosphate and some PtdIns(3,5)P 2 phosphatase activity. The protein clearly has an important role in the regulation of cellular lipid phosphoinositide levels as is seen by the 10-fold increase in the levels of PtdIns(4)P (in YPD medium). The substrate specificity of the protein would indicate that Sac1p is also important in regulating levels of PtdIns(3)P; however, as the levels of PtdIns(3)P increase only slightly on deletion of the gene, this perhaps indicates that other enzymes are also involved in the regulation of this lipid. Yeast cells also contain two homologues of the mammalian pTEN phosphatase (27,28), which has been demonstrated to dephosphorylate PtdIns(3)P as well as PtdIns(3,4,5)P 3 (29). Although the specificities of these pTEN homologues remain to be established, it is clear that they encode proteins with phosphatase activity (28). Thus, Sac1p may not be exclusively involved in the regulation of D-3-phosphorylated phosphoinositides Mutants also show significantly decreased levels of PtdIns(4,5)P 2 , which was surprising, especially considering the elevated levels of PtdIns(4)P and as Sac1p would not seem to directly regulate PtdIns(4,5)P 2 , having no detectable phosphatase activity for it. PtdIns(4)P 5-kinase activity, producing PtdIns(4,5)P 2 , must therefore be either tightly regulated (30) or compartmentally isolated from the PtdIns(4)P increase in sac1 mutants. Sac1p is located in the endoplasmic reticulum (ER) and Golgi apparatus in yeast (4), whereas Mss4p (31), the PtdIns(4)P 5-kinase (32,33), is proposed to be located on the plasma membrane (33). The reduction in the lipid seen could be due to the action of phosphatases (such as Inp52p) overexpressed to counteract the accumulation of PtdIns(4)P. This is perhaps confirmed by the observation that PtdIns(4,5)P 2 levels remain quite low in sac1 mutants complemented with Inp52p. The precise mechanisms by which these phosphatases regulate cellular lipid phosphoinositides remain to be established. However, these observations provide a model by which Sac1p is able to mediate its many effects.
Regulation of the actin cytoskeleton and thus cell wall chitin has been shown to occur via the effects of PtdIns(4,5)P 2 on some of the many actin-binding proteins such as profilin (34) and cofilin (35). The effects of SAC1 deletion on PtdIns(4,5)P 2 could well account for the actin defects seen in sac1 mutants (1-3). Sac1p mutants that confer reduced levels of PtdIns(4,5)P 2 show defects in actin cable polymerization and cortical actin patch localization, phenotypes also seen in mss4-depleted cells, which are severely compromised in their ability to synthesize PtdIns(4,5)P 2 (31). Guo et al. (11), who did not demonstrate significant changes in PtdIns(4,5)P 2 levels, proposed a novel role for PtdIns(3)P, PtdIns(4)P, or PtdIns(3,5)P 2 in regulating the actin cytoskeleton. This mechanism is unlikely to be via D-3-phosphorylated lipids, as deletion of VPS34, the only Pt-dIns 3-kinase in S. cerevisiae, affects vacuolar function (36) rather than actin function.
The gene encoding Sac1p was originally cloned after it was identified as a mutant suppressor of defects in Sec14p, the yeast Golgi phosphatidylinositol-phosphatidylcholine transport protein. The mechanism by which Sac1p inactivation bypasses the requirement for Sec14p function has been discussed previously (6,12,37), and it is proposed to involve diacylglycerol produced as a result of accumulation of PtdIns(4)P.
The effects of mutations in SAC1 on the lipid phosphoinositides may also mediate the effects seen on ER protein translocation and folding (5,6). ATP transport into the ER is essential for the multiple reactions occurring within the ER lumen to enable proteins to cross the ER membrane and to initiate transport (38). Cells containing deleted sac1 genes have a greatly reduced ability to transport ATP into the ER lumen, whereas cells overexpressing Sac1p have enhanced ATP transport (5). These observations are not directly due to Sac1p, which does not resemble transporters of any kind and has been demonstrated to have no intrinsic ATP transport activity (6). ATP transporters in the plasma membrane have been shown to be regulated by PtdIns(4)P (39,40), and it is possible that similar proteins in the ER could have reduced activity in the presence of elevated PtdIns(4)P levels. Additionally, PtdIns(4,5)P 2 has been shown to activate ER ATPases (41); the reduction in PtdIns(4,5)P 2 levels seen in sac1 mutants could account for the phenotypes seen.
The mechanisms of inositol auxotrophy and drug sensitivity seen in some sac1 mutants are perhaps more difficult to explain. The deletion mutants tested in this work and those in other studies (11,12) are inositol-auxotrophic and show changes in lipid phosphoinositides and drug sensitivity. The mds1-sac1 mutant can display drug sensitivity, but is not an inositol auxotroph. The sac1-22 mutant is an inositol auxotroph and drug-resistant (3) and only suppresses sec14-1 mutants when grown on inositol (36). Whether the phenotypes are in 2 R. Woscholski, manuscript in preparation. anyway linked or how Sac1p phosphatase defects conferring changes in phosphoinositides cause these phenotypes is not clear. It is evident, however, that mds1-sac1 mutations within highly conserved motifs of the Sac domain, presumed to be intrinsic for phosphatase activity, actually alter regulation of the enzyme activity. Thus, the protein must interact with other regulatory growth condition-dependent factors to mediate phosphatase activity. It is possible that these interactions may have implications for the regulation of inositol metabolism and for cellular drug resistance.
S. cerevisiae probably contains only four significant phosphorylated lipid phosphoinositides: PtdIns(3)P, PtdIns(3,5)P 2 , PtdIns(4)P, and PtdIns(4,5)P 2 . No PtdIns(3,4,5)P 3 has been detected (17,31), and although PtdIns(3,4)P 2 has been detected in S. cerevisiae (16), this is likely to be insignificant. Yet there appear to be many lipid phosphoinositide phosphatases with a variety of specificities. Having defined Sac1p as a regulated lipid phosphoinositide phosphatase, the identification of the factors controlling its activity and in what cellular location is the aim of future work.