STT4 Is an Essential Phosphatidylinositol 4-Kinase That Is a Target of Wortmannin in Saccharomyces cerevisiae *

Wortmannin is a natural product that inhibits signal transduction. One target of wortmannin in mammalian cells is the 110-kDa catalytic subunit of phosphatidylinositol 3-kinase (PI 3-kinase). We show that wortmannin is toxic to the yeastSaccharomyces cerevisiae and present genetic and biochemical evidence that a phosphatidylinositol 4-kinase (PI 4-kinase), STT4, is a target of wortmannin in yeast. In a strain background in which stt4 mutants are rescued by osmotic support with sorbitol, the toxic effects of wortmannin are similarly prevented by sorbitol. In contrast, in a different strain background, STT4 is essential under all conditions and wortmannin toxicity is not mitigated by sorbitol. Overexpression of STT4 confers wortmannin resistance, but overexpression of PIK1, a related PI 4-kinase, does not. In vitro, the PI 4-kinase activity of STT4, but not of PIK1, was potently inhibited by wortmannin. Overexpression of the phosphatidylinositol 4-phosphate 5-kinase homolog MSS4 conferred wortmannin resistance, as did deletion of phospholipase C-1. These observations support a model for a phosphatidylinositol metabolic cascade involving STT4, MSS4, and phospholipase C-1 and provide evidence that an essential product of this pathway is the lipid phosphatidylinositol 4,5-bisphosphate.

Small cell-permeable compounds have proven valuable tools to study signal transduction. For example, studies on the mechanism of action of the immunosuppressive antifungal natural products cyclosporin A and FK506 led to the identification of the protein phosphatase calcineurin as a critical calcium sensor during T-cell activation and physiological responses in yeast cells. Related studies of another natural product, the immunosuppressant rapamycin, revealed a role for the TOR kinase homologs in regulating cell proliferation in both yeast and mammalian cells (reviewed in Refs. 1

and 2).
Wortmannin is a hydrophobic steroid-related natural product of the fungus Talaromyces wortmannii (reviewed in Ref. 3). Wortmannin is an immunosuppressive and anti-inflammatory agent that inhibits signal transduction events in a variety of cell types. For example, wortmannin inhibits neutrophil and platelet activation by a variety of ligands, such as leukotriene B 4 , platelet-activating factor, N-formyl-Met-Leu-Phe, and thromboxane, which act via G-protein-coupled receptors (4,5).
In addition, wortmannin also blocks insulin stimulation of glucose uptake in adipocytes (6). Thus, wortmannin can block signal transduction events emanating from either G-proteincoupled or tyrosine kinase receptors, which signal through distinct pathways. Wortmannin does not affect immediate signal transduction events in these pathways, such as tyrosine phosphorylation, production of inositol trisphosphate (IP 3 ), 1 or Ca 2ϩ mobilization. That stimulation of protein kinase C by phorbol ester overcomes the inhibitory effects of wortmannin (7,8) suggests that wortmannin acts upstream of protein kinase C, possibly at a point at which G-protein-coupled and tyrosine kinase signaling pathways converge.
Several observations suggested that a relevant in vivo target of wortmannin might be an enzyme, namely the lipid kinase that phosphorylates the 3-position of phosphatidylinositol: PI 3-kinase. First, wortmannin blocks antigen-dependent stimulation of PI 3-kinase activity in rat basophils (4). Second, wortmannin markedly inhibits phosphatidylinositol 3,4,5-trisphosphate production in neutrophils stimulated with N-formyl-Met-Leu-Phe, consistent with a block in phosphorylation of phosphatidylinositol 4,5-bisphosphate (PI4,5P 2 ) by PI 3-kinase (5). Third, the ability of insulin to stimulate PI 3-kinase activity in adipocytes is inhibited by wortmannin (6). In mammalian cells, PI 3-kinase is a heterodimer composed of a 110-kDa catalytic subunit and an 85-kDa regulatory subunit that interacts with other signal transduction elements via SH2 domains (9). The activity of purified PI 3-kinase is potently inhibited in vivo and in vitro by wortmannin (4,5,10). Finally, using anti-wortmannin antibodies and protease digestion, it has been shown that wortmannin forms a covalent complex with an active site residue of the 110-kDa PI 3-kinase catalytic subunit, lysine 802 (4,11).
Although wortmannin potently inhibits the PI 3-lipid kinase with an IC 50 of 5 nM, other potential targets exist. For example, wortmannin inhibits a protein kinase, myosin light chain kinase, with an IC 50 of ϳ20 nM (12). In addition, wortmannin also inhibits DNA-dependent protein kinase, a member of the PI 3/4-kinase superfamily, with an IC 50 of 200 nM (13). Demethoxyviridin, a structural analog of wortmannin, inhibits a membrane associated PI 4-kinase from the fission yeast Schizosaccharomyces pombe (IC 50 ϭ 100 nM) (14), whose identity has not been established. Recently a wortmannin-sensitive, membrane-associated PI 4-kinase, shown previously to maintain hormone-regulated PI4P pools in mammalian cells (15)(16)(17), was cloned from a human cDNA library (18). Thus, wortmannin inhibits several protein and lipid kinases.
Previous studies of antifungal natural products reveal that both the mechanisms of action as well as drug targets are often highly conserved from unicellular to multicellular eukaryotes.
We report here our studies on the mechanism of action and targets of wortmannin in the budding yeast Saccharomyces cerevisiae. Our genetic and biochemical evidence indicates that a target of wortmannin in yeast is the PI 4-kinase STT4 (19). Conditions that rescue cells lacking STT4 also overcome wortmannin toxicity, overexpression of STT4 renders cells wortmannin-resistant, and the PI 4-kinase activity of STT4 is inhibited by wortmannin in vitro. The second yeast PI 4-kinase, PIK1 (20 -22), does not appear to be a target of wortmannin in vivo, and the PI 4-kinase activity of PIK1 is not sensitive to wortmannin in vitro. STT4 is tightly associated with a membrane-pellet fraction. In addition, we report that STT4 is essential in some yeast strains. Because wortmannin inhibition causes a nonspecific cell cycle arrest, STT4 function may be required throughout the cell cycle. Mammalian homologs of both yeast STT4 and PIK1 have been recently cloned (15,18,23). Curiously, the mammalian PIK1 homolog, and not the mammalian STT4 homolog, is wortmannin-sensitive in vitro.
We also find that overexpression of the yeast PI4P 5-kinase homolog MSS4 or deletion of the yeast gene encoding phospholipase C (PLC1) confer wortmannin resistance, supporting a model in which the inhibitory effects of wortmannin may be due to depletion of an essential PI4,5P 2 pool. Our studies underscore the utility of yeast as a model system to identify drug targets, suggest caution in the use of wortmannin as a specific inhibitor of mammalian PI 3-kinase, reveal that diverse members of the lipid/protein kinase superfamily are wortmanninsensitive, and provide evidence for a phosphatidylinositol metabolic cascade in yeast involving sequential action of STT4, MSS4, and PLC1 in which PI4,5P 2 is an essential lipid product.
DNA Manipulations and Plasmids-DNA manipulations were carried out as outlined in Ref. 26. Yeast were transformed by the lithium acetate/heat shock method (27). The following plasmids were used. 1) pYeF1 and pYeF2 (28) are 7.1-kilobase pair 2-m plasmids carrying the URA3 gene and the GAL promoter and were used to epitope-tag STT4 and PIK1.
2) pGALHA-STT4 -8 (HA-STT4) was created by ligating the STT4 gene into the NotI site of pYeF1 in frame with the HA epitope tag at the 5Ј end of the gene, placing the fusion gene under the control of the GAL promoter. The STT4 gene was amplified by PCR from a plasmid copy of STT4 using primers 5Ј-TTAAGCGGCCGCATGAGATTTACCAGAGGA-GATTG and 5Ј-TATAGCGGCCGCGTCAGTACGGAATGCCATTT-GAGCC.
PI Kinase Assays-For PIK1 assays, cells transformed with plasmid pGALPIK1-HA were inoculated in synthetic media lacking uracil supplemented with 2% raffinose, grown to OD 600 ϭ 0.5-0.8, and expression of PIK1-HA induced with 2% galactose for 2 h. As a control for protein expression and antibody specificity, parallel cultures were grown as above and supplemented with 2% glucose instead of galactose. For STT4 assays, the isogenic wild-type STT4 strain YS3-6D and the ⌬stt4 -16C strain were grown in YPD media. Preparation of cell extracts, immunoprecipitations, lipid kinase reactions, and TLC analysis were performed as described previously (32) with the following exceptions. Immunoprecipitations were carried out in equal amounts (by protein) of cell lysate in a total volume of 1 ml with 3 l of anti-HA mouse monoclonal 12 CAS (Babco) antibody. Where indicated, the immunoprecipitates were preincubated with the appropriate concentration of wortmannin for 10 min at 4°C, and the reactions were started by addition of 10 M [␥-32 P]ATP and 10 M MgCl 2 . All reactions were incubated for 20 min at 30°C with occasional gentle agitation; under these conditions, the reactions were linear for up to 30 min.
Cellular Fractionation-Strain JK9 -3da transformed with pGAL-HASTT4 was grown in S-raffinose minus uracil media to OD 600 ϭ 0.5-0.8, after which 2% galactose was added and incubated for 1 h. Cell-free lysates and P100 and S100 fractions were prepared as described (32). Cell fractions were analyzed by Western blot with the anti-HA monoclonal antibody (see above), followed by ECL detection (Amersham).

RESULTS
Wortmannin Is Toxic to Yeast-Wortmannin is a fungal metabolite that potently inhibits mammalian PI 3-kinase (33). We sought to identify wortmannin targets in yeast. Wortmannin inhibited the growth of yeast on synthetic minimal medium with a minimum inhibitory concentration of ϳ10 g/ml (Fig.  1A). Wortmannin treatment did not cause cell lysis or a specific cell cycle arrest (data not shown). In addition, wortmannin was not toxic to yeast cells grown on rich YPD media, possibly due to nonspecific drug binding by yeast extract components as has been observed with other toxins. Consistent with this interpretation, addition of yeast extract to synthetic media prevented wortmannin toxicity (data not shown).
Yeast Lacking PI 3-Kinase Are Wortmannin-sensitive-Wortmannin can inhibit the yeast PI 3-kinase VPS34p in vitro (24) at concentrations higher (IC 50 ϭ 3 M) than those that inhibit mammalian PI 3-kinase (IC 50 ϳ 5 nM) (33). However, yeast cells lacking VPS34 are both viable and remain wortmanninsensitive, with a minimum inhibitory concentration of ϳ1 g/ml (Fig. 1A). Thus, VPS34 is not the only target of wortmannin in yeast. The ϳ10-fold increased sensitivity of ⌬vps34 mutant cells to wortmannin compared with the isogenic VPS34 parent strain could result from an increase in the free drug concentration due to the loss of one drug binding target. Alternatively, this effect may be nonspecific because ⌬vps34 mutant cells have perturbed vacuolar function, grow slowly, and are temperature-sensitive, and thus might be inherently more drug-sensitive. The latter explanation is supported by the finding that end1 and vps3 mutations, which also result in vacuolar defects and temperature-sensitive slow growth, are similarly wortmannin-hypersensitive (data not shown).
Yeast PI 4-Kinase STT4 Is a Wortmannin Target in Vivo-Because yeast cells lacking VPS34 have no detectable PI 3kinase activity or PI3P, yet retain sensitivity to wortmannin, this drug must inhibit an additional target other than PI 3kinase in yeast. Recent studies demonstrate that wortmannin or its analog demethoxyviridin also inhibit PI 4-kinases from mammals and fission yeast (14,16,18). Two PI 4-kinases, PIK1 and STT4, have been identified in S. cerevisiae (19 -22). We tested whether the PIK1 or STT4 PI 4-kinases are targets for wortmannin in yeast.
PIK1 is an essential enzyme (21,22). A yeast strain expressing a temperature-sensitive PIK1 mutant (pik1-ts) was inhibited by wortmannin at permissive and semi-permissive temperatures with the same minimum inhibitory concentration as the isogenic wild-type PIK1 strain (Fig. 1A). These findings suggest that PIK1 is not a target of wortmannin.
STT4 is also essential for growth on standard yeast growth media, but the inviability of ⌬stt4 mutant cells can be remediated on rich media, such as YPD, by 1 M sorbitol (19) and more poorly on a synthetic medium with 1 M sorbitol (Fig. 1B, lower  left quadrant). To test if wortmannin inhibits STT4 in vivo, we asked whether 1 M sorbitol would rescue cells from wortmannin toxicity. Remarkably, in synthetic complete media, 1 M sorbitol rescued cells from the toxic effects of 10 g/ml wortmannin (Fig. 1B). Thus, conditions that allow growth of cells lacking STT4 (⌬stt4) also allow growth in the presence of wortmannin.
In several other yeast strain backgrounds, we and others 2 have found that ⌬stt4 mutant cells are inviable and are not rescued by 1 M sorbitol ( Fig. 2A and data not shown). Thus, in these other strain backgrounds, STT4 is essential under all conditions. Correspondingly, in strain backgrounds where STT4 is essential, 1 M sorbitol failed to rescue cells from wortmannin toxicity (Fig. 1B). In summary, in synthetic media lacking sorbitol, STT4 is essential and wortmannin is toxic in all strains. In one unusual strain, both the lethality of an ⌬stt4 mutation and wortmannin toxicity can be overcome by osmotic support with 1 M sorbitol. These findings provide genetic evidence that the PI 4-kinase STT4 is a target of wortmannin in vivo. STT4 is probably not the only wortmannin-sensitive target in yeast, because sorbitol did not rescue cells lacking STT4 from wortmannin toxicity (Fig. 1B, lower left quadrants).
To explore genetic differences between strain backgrounds with respect to the essential function of STT4 a nonisogenic cross was performed between the ⌬stt4 viable mutant and our wild-type laboratory strain (in the presence of sorbitol). Sporulation and dissection of 24 tetrads revealed 1 tetrad with two wild-type STT4 and two slow growth, sorbitol-dependent ⌬stt4 segregants (parental ditype), 5 tetrads consisting of two wildtype STT4, one slow growth sorbitol-dependent ⌬stt4 mutant, and one inviable segregant (tetratype), and 18 tetrads consisting of two wild-type STT4 spores and two inviable spores (nonparental ditype). These results are consistent with a minimum of three unlinked extragenic suppressors required for sorbitol rescue of the ⌬stt4 mutation.
STT4 Is Essential and Overexpression Confers Wortmannin Resistance-As mentioned above, STT4 is essential in our yeast laboratory strain background, and thus an STT4/⌬stt4 heterozygous diploid sporulates to yield tetrads containing two viable (STT4) and two inviable (⌬stt4) segregants ( Fig. 2A). An HA epitope-tagged version of the STT4 gene was cloned under the transcriptional control of the GAL promoter (pGALHA-STT4). Expression of STT4 complemented the ⌬stt4 mutation and restored growth even when cells were grown in glucose, which reduces expression from the GAL promoter (Fig. 2B). Rescue of the stt4::G418 mutant cells required the STT4 expression plasmid, as these cells were inviable on 5-fluoroorotic 2 S. Emr, D. Levin, and D. Voelker, personal communications.
acid medium, which is toxic to cells containing the plasmidborne URA3 marker. Interestingly, growth in glucose provided a level of STT4 expression that also conferred resistance to 10 g/ml wortmannin (Fig. 3). Overexpression of STT4 from the GAL promoter markedly inhibited growth, and thus we could not assess wortmannin toxicity under these conditions (data not shown). That STT4 overexpression is toxic suggests the activity of this enzyme may normally be regulated in vivo. Expression from the GAL promoter of an epitope-tagged form of the PIK1 PI 4-kinase, PIK1, complemented a pik1-ts mutation but did not confer wortmannin resistance and was not toxic when cells were grown with either glucose or galactose ( Fig. 3 and data not shown). These findings implicate the STT4 PI 4-kinase as a specific target of wortmannin in yeast.
PI 4-Kinase Activity of STT4 Is Wortmannin-sensitive in Vitro-To further assess the wortmannin sensitivity of STT4 and PIK1 in vitro, these enzymes were immunoprecipitated with an anti-STT4 polyclonal antiserum or with an HA epitopetagged version of PIK1 and an anti-HA monoclonal antibody. The immunoprecipitates were incubated with PI and [␥-32 P]ATP, and PI4P production was detected by borate thin layer chromatography and autoradiography. Immunoprecipitation from cells expressing STT4 yielded detectable PI 4-kinase activity (Fig. 4B, lane 2), whereas cells lacking STT4 (⌬stt4) yielded no PI 4-kinase activity (Fig. 4B, lane 4), establishing the specificity of the antisera for STT4. The PI 4-kinase activity of STT4 was inhibited ϳ80% and ϳ95% by 1 and 10 nM wortmannin, respectively (Fig. 4, B-D). In comparison, galactose induction of PIK1-HA led to a marked overexpression of PIK1 as detected by Western blot (Fig. 4A) and to a corresponding increase in HA-precipitable PI 4-kinase activity (Fig. 4B, compare lanes 5 and 7); however, the PI 4-kinase activity of PIK1-HA was completely resistant to wortmannin at a concentration of 1, 10, or 20 M (Fig. 4A, lane 6, and data not shown). This is in accord with previous observations that the PI 4kinase activity of PIK1 purified to homogeneity is not sensitive to wortmannin at high concentrations. 3 These findings provide biochemical evidence that the PI 4-kinase activity of STT4 is inhibited by wortmannin, and confirm our genetic evidence that STT4 is a target of wortmannin in yeast.
STT4 Is Associated with a Membrane-pellet Fraction in Yeast-To further characterize STT4, cells expressing HA-STT4 were fractionated into soluble (S 100 ) and insoluble (P 100 ) fractions and the amount of STT4 in these fractions was determined by Western blot. STT4 was exclusively detected in the particulate (P 100 ) fraction (Fig. 5). Similar fractionation results were obtained with endogenous STT4 detected with an anti-STT4 polyclonal antisera (data not shown). The strength and nature of the STT4 association to the P 100 fraction was tested by treating the cell lysate previous to centrifugation with agents known to disrupt membrane-protein or protein-protein interactions. STT4 was solubilized by 1% SDS; however, treatment with 2% Triton X-100 or with agents that interfere with protein-protein interactions, such as 0.5 M NaCl and 1.6 M urea, 3 J. Thorner, personal communication. FIG. 3. Overexpression of STT4 PI 4-kinase confers wortmannin resistance. Yeast strain JK9 -3da containing a 2 plasmid lacking (Vector, pYeF1) or expressing from the GAL1 promoter either the epitope-tagged PIK1 (PIK1-HA) or STT4 (HA-STT4) proteins was grown on minimal medium (SD-ura) without or with 10 g/ml wortmannin. Growth was for 72 h at 30°C.

FIG. 4. Wortmannin inhibits STT4 PI 4-kinase activity in vitro.
A, strain JK93da/pGAL-PIK1-HA was grown in S-raffinose-uracil media to OD 600 ϭ 1.0. To induce or repress PIK1-HA expression, 2% galactose or 2% glucose were added, respectively. After 4 h, cells were harvested and protein extracts were prepared and analyzed by Western blot with anti-HA monoclonal antibody. B, lipid kinase activity of STT4 and PIK1 was assayed in immunoprecipitates from strains YS3-6D (STT4) (lanes 2 and 3) and strain JK9 -3da/pGAL-PIK1-HA (lanes 5 and 6), respectively, in the absence (Ϫ), or in the presence (ϩ) of 1 M wortmannin. As controls for antibody specificity, immunoprecipitates from cell extracts from the ⌬stt4 mutant strain ⌬stt4 -16C (⌬, lane 4), and strain JK9 -3da/pGAL-PIK1-HA grown in glucose to repress PIK1-HA expression (lane 7) were also assayed. The migration positions of PI3P and PI4P standards (std) are indicated on the right. C, dose-dependent inhibition of STT4 by wortmannin. STT4 immunoprecipitates from the STT4 wild-type strain YS3-6D were assayed for PI 4-kinase activity in the absence (lane 3) or presence (lanes 4 -7) of the indicated concentrations of wortmannin. A control assay with preimmune sera is shown in lane 2. The migration positions of PI3P and PI4P standards (std) are indicated on the right. D, the percent inhibition of STT4 PI 4-kinase activity in B were quantified by phosphoimaging and plotted versus the concentration of wortmannin (nM).

FIG. 5. Subcellular distribution of STT4.
Cell-free lysates of JK9 -3da cells expressing HA-STT4 were centrifuged at 100,000 ϫ g to obtain insoluble (P) and soluble (S) fractions. Prior to centrifugation, lysates were either not treated (none) or were treated for 10 min at 4°C with 1% SDS, 2% Triton X-100, 0.5 M NaCl, or 1.6 M urea. The figure shows the Western blot of the resulting fractions with HA monoclonal antibody. Similar results were obtained with endogenous STT4 detected with a polyclonal antiserum (data not shown). The migration position of STT4 is indicated by the arrow to the right, and numbers to the left indicate mass in kDa. failed to extract STT4 from the particulate fraction (Fig. 5). These results indicate that STT4 is tightly associated with the pellet fraction, possibly via associations with both proteins and membranes.
MSS4 Overexpression or PLC1 Mutation Confer Wortmannin Resistance-The product of STT4, PI4P, can be further metabolized to form other important signaling molecules, including PI4,5P 2 , diacylglycerol, and IP 3 . Inhibition of STT4 could result in a depletion of all of these molecules. To address whether depletion of these metabolites may be responsible for the growth inhibitory effects of wortmannin, the effect of lipidmetabolizing enzymes on wortmannin sensitivity was investigated. Two genes in S. cerevisiae, MSS4 and FAB1, show marked identity with the human C isoform of phosphatidylinositol 4-phosphate 5-kinase (34). Because MSS4 is essential (30), we could not test the effects of deleting the gene. Expression of MSS4 from a multicopy 2 plasmid conferred wortmannin resistance, allowing JK9 -3da cells to grow on 7.5 g/ml wortmannin (Fig. 6A). In contrast, we found no evidence that the second yeast PI4P 5-kinase homolog FAB1 plays any role in wortmannin action. Thus, ⌬fab1 mutant cells are viable, grow poorly on minimal media, and were neither wortmannin-hypersensitive nor resistant (data not shown). Moreover, expression of FAB1 from a multicopy 2 plasmid did not alter wortmannin sensitivity (Fig. 6A, data not shown). These findings indicate a specific role for MSS4, but not for FAB1, in a wortmanninsensitive phosphatidylinositol cascade.
We also tested whether yeast PLC1 participates in the wortmannin-sensitive phosphatidylinositol cascade involving STT4 and MSS4. In fact, strains in which the PLC1 gene was deleted were viable and resistant to growth inhibition by wortmannin at 10 g/ml (Fig. 6B). Wortmannin resistance co-segregated with the ⌬plc1 mutation in a genetic cross (data not shown), and reintroduction ot the wild-type PLC1 gene complemented the plc1 deletion mutation to restore wortmannin sensitivity (Fig. 6B). These findings indicate that PLC1 plays a role in a wortmannin-sensitive phosphatidylinositol metabolic cascade in yeast, possibly by cleaving an essential product of STT4 and MSS4 such as PI4,5P 2 .

DISCUSSION
Our studies identify the PI 4-kinase STT4 as a target of the natural product wortmannin in the yeast S. cerevisiae. First, in a strain background in which the lethal phenotype of an stt4 deletion is remediated by sorbitol, wortmannin toxicity is similarly mitigated by sorbitol. In contrast, in other strain backgrounds STT4 is essential, and wortmannin is toxic, under all conditions. Second, we show that modest overexpression of STT4 confers wortmannin resistance, as does overexpression of MSS4 and deletion of PLC1. Finally, we demonstrate that the PI 4-kinase activity of STT4 is potently inhibited (IC 50 ϭ 1 nM) in vitro by wortmannin. Our findings suggest a model in which inhibition of STT4 PI 4-kinase activity by wortmannin leads to a lethal depletion of a PI4,5P 2 pool in yeast.
A target of wortmannin in mammalian cells is the 110-kDa catalytic subunit of PI 3-kinase. Wortmannin inhibits PI 3kinase with an IC 50 ϭ 1-5 nM and forms a covalent complex with the active site of the enzyme. Yeast cells express a single PI 3-kinase, VPS34, and an obvious question is whether the VPS34 PI 3-kinase is also a target of wortmannin in yeast. In previous studies, the PI 3-kinase activity of VPS34 was found to be relatively resistant to wortmannin in vitro, with an IC 50 ϭ 3 M (14, 24). Our finding that cells lacking VPS34 are still wortmannin-sensitive further supports the view that another protein is a target for wortmannin action in yeast. Why is the VPS34 PI 3-kinase less sensitive to wortmannin compared with the human PI 3-kinase? VPS34 is structurally and functionally distinct from the wortmannin-sensitive heterodimeric mammalian PI 3-kinase. A mammalian PI 3-kinase, distinct from the mammalian p85/p110 PI 3-kinase activated by growth factors, has recently been identified that is insensitive to wortmannin and is likely the mammalian homolog of the yeast VPS34 PI 3-kinase (35). Thus, it appears that yeast lack a homolog of the mammalian p85/p110 PI 3-kinase.
The STT4 gene was previously identified among a collection of mutants resistant to the protein kinase C inhibitor staurosporine (19). Disruption mutants lacking STT4 were inviable but, like pkc1 mutant cells, could be rescued by osmotic remediation with 1 M sorbitol. This finding led to the model that STT4 lies upstream of PKC in the yeast PKC pathway that regulates cell wall biosynthesis (19). In contrast, we and others 2 find that, in our and other strain backgrounds, STT4 is essential and cells lacking STT4 are not rescued by sorbitol (Fig. 2). We have confirmed that, in our strain background, pkc1 mutants are viable in the presence of 1 M sorbitol and inviable in its absence (data not shown). Thus, unlike PKC1, STT4 is essential under all conditions in our strain background. If STT4 does lie upstream of PKC1, it clearly must have additional functions. Alternatively, STT4 and PKC1 could regulate parallel pathways with shared and distinct functions.
The STT4 gene encodes an ϳ200-kDa protein with a predicted C-terminal region with homology to PI 3-kinases and PI FIG. 6. Lipid metabolizing enzymes effect wortmannin sensitivity. A, yeast strain JK9 -3da containing a 2 plasmid lacking (vector) or expressing the PI4P 5-kinase homologs MSS4 (2 MSS4) or FAB1 (2 FAB1) was grown on synthetic media lacking (0) or containing wortmannin (7.5 g/ml) for 5 days at 30°C. B, isogenic yeast strains lacking (⌬plc1) or expressing phospholipase C (PLC1) and containing a 2 plasmid lacking (vector) or expressing PLC1 (2 PLC1) were grown on synthetic media lacking (0) or containing wortmannin (10 g/ml) for 10 days at 24°C. C, a proposed model for the effects of wortmannin in yeast. Wortmannin inhibits STT4, decreasing pools of PI4P and thereby decreasing the availability of substrate for MSS4, which results in depletion of a PI4,5P 2 pool necessary for cell proliferation. Overexpression of MSS4 increases PI4,5P 2 and confers wortmannin resistance. Deletion of PLC1 prevents the cleavage of PI4,5P 2 into diacylglycerol (DAG) and IP 3 , increasing resistance to wortmannin. 4-kinases. In previous studies, extracts from ⌬stt4 mutant strains had a marked reduction in PI 4-kinase activity, providing evidence that STT4 was indeed a PI 4-kinase (19). We have immunoprecipitated STT4 and have found that the protein indeed has PI 4-kinase activity that is readily inhibited by wortmannin. In addition, we find that wortmannin is toxic to yeast and causes a nonspecific cell cycle arrest. These findings suggest that the PI 4-kinase activity of STT4 is essential. The essential product of STT4 could be PI4P itself, or a further metabolite such as PI4,5P 2 . Two genes in yeast, MSS4 and FAB1, have been shown to share marked identity with the mammalian C isoform of PI4P 5-kinase (36), which phosphorylates PI4P to produce PI4,5P 2 . MSS4 was isolated as a multicopy suppressor of the cell lysis defect of an stt4-ts mutants (30). MSS4 does not fully suppress the cell lysis defect of an stt4 deletion and does not suppress the staurosporine sensitivity of any stt4 mutant. We find that overexpression of MSS4 confers wortmannin resistance, providing further evidence that MSS4 acts downstream of STT4 (Fig. 6A). The homology of MSS4 to mammalian PI4P 5-kinase and its genetic interactions with STT4 suggest a model in which the product of STT4 lipid kinase activity, PI4P, is phosphorylated by MSS4 to produce PI4,5P 2 (Fig. 6C). Deletion of the gene encoding yeast PLC1, which breaks down PI4,5P 2 to produce diacylglycerol and IP 3 (31,37,38) also confers wortmannin resistance. Both overexpression of MSS4 and deletion of PLC1 would be predicted to increase PI4,5P 2 levels. That these mutations also confer wortmannin resistance implies that wortmannin may act to decrease essential pools of PI4,5P 2 by inhibiting STT4 and decreasing production of its precursor, PI4P (Fig. 6C).
PI4,5P 2 has been shown to play important signaling roles in mammalian cells, involved in phospholipase D activation (39), cytoskeletal dynamics (40), and integrin-mediated cell adhesion (41). PI4,5P 2 has been found to bind to the pleckstrin homology domain of several signaling molecules (42)(43)(44). Intriguingly, STT4 has a pleckstrin homology domain, opening the possibility that the pathway proposed in Fig. 6C may be autoregulated, either positively or negatively.
It has previously been reported that STT4 mutation results in a G 2 /M cell-cycle arrest at nonpermissive temperatures (19,30). Wortmannin treatment, however, results in a nonspecific cell cycle arrest. One possible explanation is that STT4 serves multiple functions, both at G 2 /M and at other points in the cell cycle. Consistent with this interpretation, depriving ⌬stt4 deletion cells of sorbitol resulted in a nonspecific cell cycle arrest (data not shown). Thus, one possible explanation is that while wortmannin inhibits all STT4 functions, the temperature-sensitive allele may be defective in only one function. Another possibility is that targets of wortmannin other than STT4 are required at different points during the cell cycle. Indeed, that an stt4 deleted strain is still wortmannin-sensitive and that STT4 overexpression confers relative but not absolute wortmannin resistance indicate that there are additional targets of wortmannin in yeast.
We find that the STT4 PI 4-kinase is associated with an insoluble membrane fraction in yeast. STT4 may serve to produce PI4P in a distinct membranous compartment of the cell. Yeast cells also express another essential PI 4-kinase, PIK1 (20 -22). Why do yeast cells express two different essential PI 4-kinases? One model is that these enzymes are localized to different intracellular compartments and serve to generate two different pools of PI metabolites, both of which are essential for cell function. Consistent with this view, we find that overexpression of PIK1 does not complement a ⌬stt4 mutation, and overexpression of STT4 does not complement conditional pik1 mutations. MSS4 overexpression also fails to rescue pik1 mu-tations (data not shown). Thus, STT4 and PIK1 play distinct roles and can not substitute for each other, even when overexpressed.
Mammalian homologs of both the yeast STT4 and PIK1 PI 4-kinases have recently been identified. The mammalian homolog of STT4 is PI4K␣ and the two share 50% identity in the kinase domain (15,23). The human homolog of PIK1 is PI4K␤, and the two share 42% identity in the kinase domain (18). Interestingly, PI4K␣, the STT4 homolog, is insensitive to micromolar concentrations of wortmannin, while PI4K␤, the PIK1 homolog, is inhibited by wortmannin (IC 50 ϭ ϳ50 nM) (18). This is in marked contrast to our findings that the yeast PIK1 PI 4-kinase is not inhibited by wortmannin, whereas yeast STT4 is potently inhibited by wortmannin. The autophosphorylation activity of the mammalian target of rapamycin (mTOR), a PI kinase homolog, has also recently been shown to be inhibited by submicromolar concentrations of wortmannin, suggesting another possible target in mammalian cells (45). Alignment of amino acid sequence of wortmannin-sensitive and wortmannin-insensitive enzymes fails to reveal any residues unique to either group. All have a high degree of identity in the kinase domain, and all contain an invariant lysine residue that is the site of cross-linking to wortmannin in the mammalian PI 3kinase. These observations suggest that subtle secondary structural differences dictate sensitivity to wortmannin inhibition.
The STT4 PI 4-kinase is a member of a larger family of proteins, the PI 3-kinase-related lipid/protein kinase superfamily. This family of proteins includes both PI 3-kinases and PI 4-kinases, the target of rapamycin proteins TOR1, TOR2, and their mammalian homolog RAFT1/FRAP/mTOR, the catalytic subunit of DNA-dependent protein kinase, and checkpoint control proteins, including TEL1 and MEC1 in yeast, and the ataxia telangiectasia protein (ATM) and its related protein ATR/FRP in mammals (see Ref. 46 for review). Several members of this family of unusual lipid and protein kinases are now known to be inhibited by wortmannin, including the mammalian PI 3-kinase, DNA-dependent protein kinase, mTOR, human PI4K␤, and the yeast PI 4-kinase STT4. Because of the growing number of members of this protein family, and their medical and pharmacological importance, wortmannin represents a valuable lead compound in the analysis of both their intracellular functions and in the rational design of other inhibitors that might more specifically target different family members.