Disturbance of sphingolipid biosynthesis abrogates the signaling of Mss4, phosphatidylinositol-4-phosphate 5-kinase, in yeast.

The functional relationships between phosphoinositides and sphingolipids have not been well characterized to date. ISP-1/myriocin is a potent inhibitor of sphingolipid biosynthesis and induces severe growth defects in eukaryotic cells because of the sphingolipid deprivation. We characterized a novel multicopy suppressor gene of ISP-1-mediated cell death in yeast, MSS4. MSS4 encodes a phosphatidylinositol-4-phosphate 5-kinase that synthesizes phosphatidylinositol (4,5)-bisphosphate (PI4,5P(2)). We demonstrate here that ISP-1 treatment of yeast causes defects both in the activity and subcellular localization of Mss4. The effect of the Mss4 defect on the downstream signaling was examined, because interaction between the Mss4 product, PI4,5P(2), and the pleckstrin-homology domain of Rom2 mediates recruitment of Rom2 to the membrane, which is the crucial step for subsequent Rho1/2 activation. Indeed, failure of Rom2 recruitment was observed in ISP-1-treated cells as well as in csg2-deleted cells, which have reduced mannosylated inositolphosphorylceramide. These data suggested that proper sphingolipids are required for the signaling pathway involving Mss4.

Sphingolipids consist of a hydrophilic head group attaching to a long-chain base (LCB) 1 -containing hydrophobic portion. Even though sphingolipids are known to be essential constituents of membranes, recent study showed the role of sphingolipids, including LCB in various biological events, e.g. in cell growth, endocytosis, stress response, and intracellular signaling (1)(2)(3). Budding yeast, Saccharomyces cerevisiae, has a sim-ilar species of LCB to that in mammalian cells, and studies of LCB metabolic pathways in yeast have greatly contributed to our current understanding of the biosynthetic pathway of LCB in mammalian cells (4).
The de novo synthesis of LCB in yeast is initiated by serinepalmitoyltransferase, and this biosynthetic pathway is followed as shown in Fig. 1A. ISP-1/myriocin inhibits the serinepalmitoyltransferase activity, and this inhibition plays a major role in the function of ISP-1 as an immunosuppressant (5). As a result, ISP-1 induces apoptosis of cytotoxic T cells (6) and yeast cell death (7). This compound is especially useful when one would like to achieve different severity of the LCB biosynthetic defect dependent on the concentration of ISP-1 for treatment. LCB biosynthetic defect could be rescued by the further addition of phytosphingosine (PHS) to the medium. We have recently succeeded in isolating multicopy suppressor genes against ISP-1 by titrating its concentration for treatment in yeast. One such gene, SLI1, encodes ISP-1-inactivating enzyme, accounting for its strong resistance activity (8). Other type of ISP-1-resistant gene, SLI2/YPK1, encodes a protein kinase. Overexpression of Ypk1 or Pkh1, an upstream kinase of Ypk1 overcame ISP-1-mediated yeast death (7), and Pkh1 was also shown to be activated by PHS in vitro (9). Therefore, the connection of sphingolipids is implicated to the Pkh1-Ypk1 signaling pathway (10 -12).
In this study, we isolated a novel ISP-1-resistant gene, SLI6/ MSS4. MSS4 was originally isolated as a suppressor of an stt4 mutation that has a defect in type III phosphatidylinositol 4-kinase (13). MSS4 encodes type I phosphatidylinositol-4phosphate 5-kinase (PI4P5K) (Fig. 1B) (14,15). Mutation of the MSS4 gene causes aberrant filamentous actin localization and subsequent cell death (15). Recent reports suggested the involvement of Mss4 in the Rho1-mediated signaling pathway (14,16). In this pathway, Rom2 regulates Rho1/2 as a GDP/ GTP exchanging factor. The colocalization of Rom2 with Rho1/2 to the plasma membrane is expected to be important for its function. Such localization of Rom2 is regulated through its PH domain, which shows preferential binding to PI4,5P 2 . Failure of proper PI4,5P 2 synthesis due to the mutation of STT4 or MSS4 causes impaired localization of Rom2 from the plasma membrane to the cytosolic compartment. Thus, the binding of Rom2 to PI4,5P 2 is regarded as a key event for the plasma membrane localization and resultant activation of Rho1/2, which is followed by the up-regulation of downstream molecules (17). Mss4 was also reported to primarily localize to the plasma membrane, although phosphorylation-dependent shuttling between the plasma membrane and the nucleus was suggested as its possible regulatory mechanism (18). Following Rho1/2 activation through PI4,5P 2 synthesis, Pkc1 is activated by Rho1 to activate the Mpk1 MAP kinase cascade (19) in-* This work was supported in part by Grants-in-aid for Scientific Research 13470488 (to Y. K.) and 13877373 (to Y. K.) from the Japan Society for the Promotion of Science, Grants-in-aid for Scientific Research on Priority Areas 12140202 (to Y. K.) from the Ministry of Education, Culture, Sports and Technology, CREST, and RIKEN. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  volved in cell wall integrity pathway (20). It is also reported that overexpression of the activated form of PKC1 suppressed aberrant actin localization caused by rho1 mutations (21). Thus, it is thought that the Rho1-mediated signaling pathway is needed for proper actin organization.
An indirect functional relationship between sphingolipids and phosphatidylinositol phosphates has been suggested based on analyses of the csg2⌬ strain. The calcium-sensitive phenotype of an MIPC biosynthesis mutant, csg2⌬ was suppressed by PI4P5K, mss4 mutation ( Fig. 1A) (22), or the mutation of genes involved in LCB biosynthesis, e.g. FAS2, LCB1/2, SUR2, and TSC10 (Fig. 1A). Here we show that disturbance of sphingolipid biosynthesis affected the function of Mss4 both in terms of enzymatic activity and plasma membrane localization, indicating a novel type of relationship connecting sphingolipids and phosphatidylinositol phosphates.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, and Media-The S. cerevisiae strains used in this study are listed in Table I. KMY1006 is a laboratory strain. TKY strains were derived from KMY1006 or AAY522 by transformation of PCR fragments containing the kanMX4 gene. BY4741 and plc1⌬-BY4741 were obtained from the Saccharomyces Genome Deletion Project. Plasmids used in this study are listed in Table II. Rich medium (YPD) and synthetic defined minimal medium (SD) supplemented with appropriate nutrients for plasmid maintenance were used.
Molecular Cloning of ISP-1-resistant Gene-Multicopy suppressor screening for ISP-1-mediated yeast death was carried out as reported previously, except NcoI digestion of the library to remove the clones carrying the SLI1 (8) and SLI2 (7) genes.
Determination of ISP-1 Resistance-Cells were grown at 30°C in SD-selective medium for 16 -18 h. Cultured cells were diluted to A 600 ϭ 0.2 and then grown at 30°C in pre-heated SD-selective medium for 1 h. Cells (2 l each) were then spotted with 2-fold serial dilution on SD plates containing ISP-1 (500 ng/ml for KMY1006 and TKY11, 150 ng/ml for TKY16, or 750 ng/ml for BY4741). The plates were incubated at 30°C for 3 days.
Visualizing EGFP-tagged Proteins-The EGFP-Mss4 fusion protein was visualized by confocal microscopy (Fluoview, Olympus, Japan) without fixation. When visualizing EGFP-Mss4 fusion protein in ISP-1-treated cells, cells were grown at 30°C in SD-selective medium for 16 -18 h. Cultured cells were diluted to A 600 ϭ 0.2 and then grown at 30°C in pre-heated SD-selective medium for 2 h. The samples were then treated for 3 h with methanol as a control, 500 ng/ml ISP-1, or 5 M PHS. This condition for ISP-1 treatment was chosen to allow quantification of viable yeast cells, which was assessed by both the colony forming assay and dye-exclusion assay. Yeast cells were collected by centrifugation and observed without fixation. To obtain higher sensitivity, GFP-Rom2 fusion protein was visualized by confocal microscopy (LSM510 Carl Zeiss).
Isolation of Detergent-resistant Membrane Fraction-Isolation of the detergent-resistant membrane fraction was carried out as described previously (25,26). Cells (40 A 600 units) were grown with or without 500 ng/ml ISP-1 for 3 h before harvesting. Cells were collected, washed with cold water twice, then stored at Ϫ80°C. The cell pellet was lysed in 500 l of TNE buffer (50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 5 mM EDTA) with a protease inhibitor mixture (Nacalai Tesque, Japan) by vortexing for 10 min at 4°C. The cell lysate was separated from cell debris by centrifugation at 500 ϫ g for 5 min. After separation, the lysate (500 l) was incubated with Triton X-100 (final 1%) for 30 min on ice and adjusted to 40% Optiprep by addition of 1 ml of Optiprep solution (Axis-shield, Oslo, Norway). The solution was overlaid with 2.4 ml of 30% Optiprep in TNE with 0.1% Triton X-100, and 1.1 ml of TNE with 0.1% Triton X-100. The samples were centrifuged at 268,000 ϫ g for 2 h using an MLS-50 rotor (Beckman Coulter), and five fractions were collected from the top. The volumes of these five fractions were 1 ml, 500 l, 1.5 ml, 500 l, and 1.5 ml collected successively from the top of the gradient. The second fraction and the fourth fraction contained the borders of layers. Twenty microliters of these fractions was analyzed by Western blotting.
Alternative method for isolation of detergent-resistant membrane fraction using CHAPS for detergent was modified from Umebayashi et al. (27). Cells were grown with or without 500 ng/ml ISP-1 for 3 h before harvesting. Yeast cells (40 A 600 units) were collected by centrifugation and resuspended in 275 l of TNE with protease inhibitors. This suspension was vortexed with glass beads for 10 min at 4°C for lysis. This cell lysate was separated from cell debris by centrifugation at 500 ϫ g for 5 min. The cleared lysate (175 l) was then mixed with 175 l of TNE buffer containing 40 mM CHAPS (Sigma-Aldrich) and incubated at 4°C for 30 min. This solution was centrifuged at 1,500 ϫ g for 5 min, and 330 l of supernatant was mixed with 770 l of 50% Optiprep solution (Axis-shield, Oslo)/TNE/20 mM CHAPS, then adjusted to the concentration of 35% Optiprep. The solution was overlaid with 2.4 ml of 30% Optiprep in TNE with 20 mM CHAPS, and 1.5 ml of TNE with 20 mM CHAPS. The samples were centrifuged at 268,000 ϫ g for 7.5 h using an MLS-50 rotor (Beckman Coulter, Fullerton), and five fractions were collected from the top. The volumes of these five fractions were 1.2 ml, 600 l, 1 ml, 1.2 ml, and 1 ml collected successively. The second fraction and the fourth fraction contained the borders of layers. Twenty microliters of these fractions was analyzed by Western blotting.
Actin Staining-The method for staining of actin filaments was described before (28). Yeast pre-cultures were grown at 30°C in SDselective medium for 16 -18 h. Cultured cells were diluted to A 600 ϭ 0.2 and then grown at 30°C in pre-heated SD-selective medium for 2 h. Then, the cells were treated with ISP-1 (500 ng/ml) for 3 h. The cells were fixed at room temperature for 30 min with formaldehyde (final 5%) in the culture medium, washed, and resuspended in phosphate-buffered saline. Staining was performed with rhodamine-phalloidin (Molecular Probes) for 2 h. The cells were washed three times before microscopic observation. Actin filaments were observed by confocal microscopy (Fluoview, Olympus, Japan).

MSS4, a PI4P5 Kinase Gene Was Isolated as Novel Multicopy
Suppressor for ISP-1-mediated Yeast Death-To gain a better understanding of the physiological roles of sphingolipids, we screened a yeast genomic library to obtain novel multicopy suppressor genes against cell death due to sphingolipid deprivation induced by ISP-1, which is a potent inhibitor of sphingolipid biosynthesis (5). A novel resistance gene, SLI6, was identified as the MSS4 gene ( Fig. 2A), which encodes the PI4P5K activity that produces PI4,5P 2 (14,15). The ISP-1 resistance of Mss4 in KMY1006 strain was also observed in the other strain BY4741 (data not shown). We first assessed kinase-dead mutants to determine whether or not the PI4P5K activity of Mss4 is required for ISP-1 resistance. Overexpression of kinase-dead mutants, Mss4 D598A and Mss4 D636A , did not suppress ISP-1-induced cell death ( Fig. 2A) even though these proteins were almost equally expressed compared with the wild-type Mss4 (Fig. 2B). The loss of activity in these two Mss4 mutants, Mss4 D598A and Mss4 D636A , was confirmed in the in vitro PI4P5K assay (Fig. 2C). These results indicate that proper PI4P5K activity is required for the ISP-1 resistance of Mss4. Moreover, overexpression of the mouse gene for PIP5K I␣ also caused similar resistance (data not shown). Thus we were interested in the molecule involved in ISP-1 resistance, because it should shed light on the relationship between these two pathways. It was reported that Plc1 hydrolyzes PI4,5P 2 to inositol 1,4,5-triphosphate and diacylglycerol (29), each of which are also well known as second messengers of signal transductions. To investigate whether the subsequent synthesis of inositol 1,4,5-triphosphate and diacylglycerol confers the ISP-1 resistance of Mss4, the ISP-1 resistance of plc1⌬ strain was tested, because both an increased PI4,5P 2 level and decreased levels of inositol 1,4,5-triphosphate and diacylglycerol are expected in this mutant. We used the plc1⌬ strain under a BY4741 background (plc1⌬-BY4741) obtained from the Saccharomyces Deletion Project, because this plc1⌬-BY4741 strain did not show a slow growth phenotype in SD medium. When compared with the parental strain, the plc1⌬-BY4741 strain showed marked ISP-1 resistance (Fig. 2D). Interestingly, overexpression of STT4, which encodes an enzyme that produces Mss4 substrate, PI4P (30) (Fig. 1B), had no effect on ISP-1 resistance (data not shown). These data indicated that the level of PI4,5P 2 is important for ISP-1 resistance, rather than PI4P, inositol 1,4,5-triphosphate and diacylglycerol.
Overexpression of Mss4 Did Not Restore the Decrease of Sphingolipid Biosynthesis by ISP-1-ISP-1 inhibits serinepalmitoyltransferase, and thus ISP-1 treatment of yeast cells results in a decrease in biosynthesis of sphingolipids, which are required for cellular viability. Therefore ISP-1 resistance due to Mss4 overexpression may involve the recovery of sphingolipid biosynthesis. To test this possibility, de novo biosynthesized sphingolipids were labeled with [ 3 H]serine, isolated, and analyzed on a TLC plate in Mss4-overexpressing cells. As shown in Fig. 3, de novo biosynthesis of sphingolipids was almost identical to that in control cells in either the absence or presence of ISP-1 in the medium. These results suggest that the ISP-1 resistance of Mss4 is not caused by the recovery of the sphingolipid biosynthesis.
Compromised in Vitro PI4P5K Activity of Mss4 Extracted from ISP-1-treated Cells-It was reported that the activities of Mss4 and mammalian PI4P5Ks are up-regulated by lipid molecules such as phosphatidic acid (14,31,32). These reports prompted us to investigate the effect of ISP-1-mediated changes in sphingolipids on the PI4P5K activity of Mss4. Interestingly, Mss4 isolated from the ISP-1-treated cells showed reduced PI4P5K activity (Fig. 4B). Moreover, the reduction found in the ISP-1-treated cells was almost restored to the control level by the further addition of PHS to the culture medium in a concentration range sufficient to compensate the loss of sphingolipids in ISP-1-treated cells. These data suggest that the decrease in one or more intracellular sphingolipids caused by ISP-1 treatment is responsible for the compromised ISP-1 Treatment Caused Abnormal Localization of Mss4 -Even though the loss of activity caused by ISP-1 treatment was demonstrated above, the mechanism of its action remained obscure. Thus, we also examined the effect of ISP-1 on the function of Mss4 in a cell biological assay. Proper subcellular localization of Mss4 to the plasma membrane was shown to be important, because loss of Mss4 activity in vivo was demonstrated in a nuclear-targeting, temperature-sensitive mutant, mss4-1 (14,18), although the mechanism of plasma membrane localization is not well understood. To investigate the effect of sphingolipid withdrawal on the localization of Mss4, the EGFPtagged Mss4 was expressed in yeast cells. The EGFP-Mss4 fusion protein used here had almost the same ISP-1 resistance as Mss4 when overexpressed (data not shown). EGFP-Mss4 expressed using a single copy vector was clearly localized to the plasma membranes, as reported previously (14). However, in the presence of ISP-1, the major part of the EGFP signal disappeared from the plasma membrane (Fig. 5A). Interestingly, this abnormal localization was normalized by the further addition of PHS even in the presence of ISP-1. It is important to note that we observed an EGFP-Mss4 signal after a 3-h treatment with ISP-1 when the growth phenotype of yeast was not different from that of untreated cells (7). The copy number of the vector seemed to be important, because the plasma membrane localization was still observed when EGFP-Mss4 was overexpressed using a multicopy vector regardless of the ISP-1 treatment. This sphingolipid-mediated change in localization prompted us to study the possible distribution of Mss4 into a sphingolipid-enriched fraction known as lipid microdomains or Rafts (33). To study this, we prepared a 1% Triton X-100 detergent-resistant membrane fraction and analyzed the Mss4 distribution into the floating fractions after density-gradient ultracentrifugation. In this assay, the majority of Mss4 was not fractionated into the detergent-resistant membrane fraction but into the detergent-soluble membrane fraction, which is the so-called "non-raft" membrane fraction (Fig. 5B). This result was consistent with another preparation using 20 mM CHAPS to prepare detergent-resistant membrane (Fig.  5C). Furthermore, biochemical distribution of Mss4 in such assay was not altered upon ISP-1 treatment of yeast cell membrane. Collectively, these data imply that even though Mss4 is not likely distributed in the sphingolipid-enriched membrane, sphingolipid withdrawal induced by ISP-1 treatment results in  blotting (B). The immunoprecipitated samples were subjected to the PI4P5K assay. The product, PI4,5P 2 was analyzed on a TLC plate in triplicate experiments as indicated by the numbers (C). D, disruption of the PLC1 gene caused ISP-1 resistance. 2-fold serial dilutions of exponentially growing parental strain (BY4741) and plc1⌬-BY4741 were plated on SD plates containing 750 ng/ml ISP-1 and incubated for 3 days at 30°C.

FIG. 3. Expression of Mss4 did not affect sphingolipid biosynthesis in yeast.
Sphingolipids were metabolically labeled with [ 3 H]serine and extracted, then separated on TLC plates as described (8). Sphingolipid biosynthesis of Mss4-overexpressing cells (lanes 3 and 4), Rho2-overexpressiong cells (lanes 5 and 6) and wild-type cells (KMY1006) (lanes 1 and 2) in the presence (lanes 2, 4, and 6) or absence (lanes 1, 3, and 5) of 500 ng/ml ISP-1. Effects of Sphingolipid Biosynthesis Mutants on the Plasma Membrane Localization of Mss4 -The above findings suggested that the proper level of sphingolipid(s) is important for Mss4 function. However, a number of sphingolipid species are expected to be deficient in ISP-1-treated yeast. Moreover, the abnormal localization of EGFP-Mss4 may be a side effect of toxic diminution of sphingolipids by ISP-1, and the recovery from the abnormal localization by Mss4 overexpression may be due to suppression of the toxic effect. To clarify these points, we also used deletion strains that affect the sphingolipid biosynthesis but not cell growth, such as sur2⌬, csg2⌬, and ipt1⌬. Sur2 catalyzes the conversion from dihydrosphingosine to PHS (34), and the sur2⌬ strain does not have PHS-containing sphingolipids (Fig. 1A). Csg2 function is implicated in MIPC synthesis, and thus csg2⌬ mutation causes an extreme reduction in MIPC and M(IP) 2 C (35,36). Ipt1 is involved in M(IP) 2 C synthesis (37), and ipt1⌬ does not have M(IP) 2 C. In sur2⌬ and ipt1⌬ strains, EGFP-Mss4 was mainly localized to the plasma membranes, as shown in the wild-type cells, when EGFP-Mss4 was expressed from a single copy vector (Fig. 6A). However, the greater part of EGFP-Mss4 was spread in the cytoplasm in the csg2⌬ strain. This abnormal localization was restored by the overexpression of EGFP-Mss4 using the multicopy vector, as is the case of the ISP-1 treatment (data not shown). When the ISP-1 resistance of these strains was compared, csg2⌬ was the most sensitive strain among the three (Fig. 6B). Taken together, these data suggest the involvement of complex sphingolipid such as MIPC in the normal function of Mss4. Because Mss4 is essential for the normal growth of yeast cells, Mss4 inactivation by ISP-1 seemed to be responsible for ISP-mediated yeast cytotoxicity at least in part, along with the reported events affected by the change in LCB biosynthesis (26, 38 -42).
Rho1 and Rho2 Are Involved in the ISP-1 Resistance Downstream of Mss4 -It has been reported that Mss4 is involved in the signaling cascade to activate MAP kinase through the activation of Rho1/2 of small G proteins. Such a signal transduction cascade was revealed utilizing a multicopy suppression system of temperature-sensitive mss4 mutation phenotypes by downstream Rho2 (15). Rho2 and its homologue Rho1 are known to be regulated by PI4,5P 2 via Rom2, which is a major GDP/GTP exchanging factor of Rho1 and Rho2 (17,43). In this context, PI4,5P 2 binding to the PH domain of Rom2 causes translocation of Rom2 to the inner membrane of bud tips where Rho proteins are localized. Thus, we assessed the effect of multicopy expression of components involved in this pathway. In such experiments, overexpression of "con- A, exponentially growing KMY1006 cells harboring pRS415-EGFP-MSS4 (single copy) or YEp351-EGFP-MSS4 (multicopy) were treated with 500 ng/ml ISP-1 and 5 M PHS for 3 h. The EGFP-Mss4 protein was visualized by confocal laser microscopy. B and C, exponentially growing KMY1006 cells harboring pRS415-HA 2 -MSS4 were treated with or without 500 ng/ml ISP-1 for 3 h and harvested. The cell lysates were incubated with 1% Triton X-100 (B) or 20 mM CHAPS (C) and fractionated as described under "Experimental Procedures." Aliquots of fractions were subjected to SDS-PAGE. Mss4, Pma1, and Pgk1 were visualized by Western blotting. Fractions 1-2 contain the detergent-resistant membrane (DRM) fractions. Pma1 was used as a marker protein localized in the detergent-resistant membrane. Pgk1 was used as a marker for the cytosolic proteins.
stitutively active" form of Pkc1, Rho2, and Rom2 caused ISP-1 resistance, whereas Mpk1 and Rho1 did not (Fig. 7A). This is consistent with a former report showing that only Rho2 but not Rho1 activated the downstream signaling pathway upon overexpression (44). To determine the contribution of Rho1 to ISP-1 resistance, the effect of overexpression of Rom2 on the ISP-1 resistance was therefore tested under the rho2⌬ background. In this condition, Rom2 is expected to use Rho1 as its sole downstream molecule. Overexpression of Rom2 in the rho2⌬ cells still caused ISP-1 resistance (Fig.  7B). Furthermore, overexpression of Mss4 in the rho2⌬ cells caused ISP-1 resistance. These data suggest that both Rho1 and Rho2 along with Rom2 are involved in the ISP-1 resistance downstream of Mss4. Rho1 and Rho2 activate Pkc1 and the resultant MAP kinase cascade activation involving Mpk1/ Slt2 (44,45) to control cell wall integrity (46). We also assessed the role of Mpk1 in ISP-1 resistance. Even though deletion of MPK1 caused hypersensitivity to ISP-1, probably due to the defect in cell wall integrity, overexpression of Mss4 or Rho2 still elevated ISP-1 resistance even under the mpk1⌬ background (Fig. 7C). These data suggest that the Mpk1 pathway downstream of Pkc1 is not the sole downstream factor in the Mss4-Pkc1-mediated ISP-1 resistance.
Change in Mss4 Localization Caused a Defect in Rom2 Localization-In a recent report, it was postulated that the bud tip recruitment of Rom2 is a key event to activate Rho1/2 downstream of phosphoinositide kinases Stt4 and Mss4 (17). We therefore examined the Rom2 subcellular localization upon ISP-1 treatment, as an indicator of the in situ activation of Rho1/2 downstream of Mss4. As expected from the mislocalized Mss4 in ISP-1-treated yeast, bud tip/neck recruitment of Rom2-GFP was disrupted in ISP-1-treated cells (Fig. 8A). This finding was further confirmed by the cytosolic localization of Rom2-GFP in the csg2⌬ strain (Fig. 8A). These data are consistent with the role of PI4,5P 2 in Rho1/2 activation. Therefore, we concluded that Mss4 mislocalization due to ISP-1 treatment caused a defect in the Rho1/2 activation in situ. We also examined whether or not ISP-1 treatment caused gross disorganization of filamentous-actin in yeast, because it is known that Mss4 regulates actin organization through the Tor2-Pkc1 signaling pathway (16,21). Rhodamine-phalloidin staining of ISP-1-treated cells showed the loss of actin cables without an obvious effect on the actin patches. Overexpression of Mss4 or Rho2 partly restored the actin cable formation (Fig. 8B). These data indicate that treatment of ISP-1 affects events downstream of PI4,5P 2 , and thus Rho1/2, as a result of Mss4 mislocalization. DISCUSSION In this study, we have isolated MSS4, a gene for PI4P5K in yeast, as a novel resistance gene for ISP-1, a potent inhibitor of sphingolipid biosynthesis. This was of particular interest, because other than pioneering work reported by Beeler and Dunn (22), no evidence has been reported about the interrelationship between sphingolipid and phosphoinositide signaling in detail. In the present study, we showed that the ISP-1 resistance of Mss4 is closely related to the intercellular PI4,5P 2 biosynthetic pathway. A proper level of sphingolipid(s) was required for proper function of Mss4, both on activity and cellular localization. Moreover, signal transduction event downstream of Mss4 was affected by the change in the sphingolipid(s). A, 2-fold serial dilutions of the exponentially growing yeast cells overexpressing Mss4, Rom2, Rho1, Rho2, Mpk1, and "constitutively active" Pkc1 or having a control vector were spotted onto SD plates containing 500 ng/ml ISP-1 and incubated for 3 days at 30°C. B, 2-fold serial dilutions of the exponentially growing rho2⌬ (TKY11) cells overexpressing Mss4, Rom2, and Rho2 or having a control vector together with wild-type strain with a control vector were spotted onto the SD plates containing 500 ng/ml ISP-1 and incubated for 3 days at 30°C. C, 2-fold serial dilutions of the exponentially growing mpk1⌬ (TKY16) cells overexpressing Mss4, Rom2, and Rho2 or having a control vector together with wild-type strain with a control vector were spotted onto the SD plates containing 150 ng/ml ISP-1 and incubated for 3 days at 30°C.
Consist with the ISP-1 resistance of Mss4, overexpression of reported downstream molecules of Mss4, such as Rom2, Rho2, and Pkc1, also resulted in resistance to ISP-1. This indicates that activation of the signaling pathway downstream of PI4,5P 2 indeed confers resistance to ISP-1. This result was in agreement with previous reports that activation-loop kinase, Pkh1 (7) and its downstream AGC kinases Ypk1 (7) and Pkc1 (47) are multicopy suppressors of ISP-1. Because lethal phenotype of YPK1/2 double mutation could be restored by overexpression of constitutively activated Pkc1 (11,48), one or more target molecules of Pkc1 partly overlap with that of Ypk1. We thus assessed the interrelationship between Mss4 and Ypk1. Indeed, slow growth phenotype of ypk1⌬ mutant was rescued by Mss4 overexpression possibly due to the Pkc1 activation (data not shown). We also observed no increase in the Mss4 activity in Ypk1-overexpressed yeast, suggesting that the relationship between Ypk1 and Mss4 in the signaling pathways could be in parallel. Taken together, Mss4 presumably activates Pkc1 through Rho activation to confer ISP-1 resistance due to the redundancy of target molecule(s) of Ypk1 and Pkc1. A relationship between Mss4 and Mpk1-mediated cell wall integrity pathway has been reported (21). We also examined the effect of Mss4 overexpression in the mpk1⌬ background. Although the mpk1⌬ strain showed marked enhancement of the ISP-1 sensitivity, overexpression of any of Mss4, Rom2, and Rho2 caused reasonable acquisition of ISP-1 resistance when compared with vector control. This finding was further confirmed by the fact that the addition of sorbitol to the medium to maintain osmolarity failed to confer ISP-1 resistance (data not shown). Therefore, target molecules of Ypk1 and Pkc1 in the ISP-1-resistant pathway seemed somewhat distinct from the Mpk1-mediated cell wall integrity pathway, at least in part. Recently, the important roles of the complex of signaling proteins designated TORC2 (TOR complex 2, which involves Tor2, Lst8, Kog1, Avo1, Avo2, and Tsc11) was reported (49). TORC2 signals through a pathway consisting of Rom2, Rho1, and the Pkc1-mediated Mpk1, which also probably includes the functions of Bck1, Mkk1, and Mss4, to the actin cytoskeleton. Thus, it is reasonable to assume that the "Mpk1-dependent" ISP-1 resistance of Mss4 may utilize such a signaling complex, if there is one. If that is the case, a change in sphingolipids might have some effect on the TORC2 complex as well.
Significance of ISP-1 treatment could be to disrupt the signaling pathway downstream of PI4,5P 2 . Therefore, we examined the effect of ISP-1 treatment and/or csg2 deletion on the reported cellular events downstream of PI4,5P 2 . In fact, the bud-neck localization of Rom2, reported to be a GDP/GTP exchanging factor for Rho1/2 (43), was disturbed upon ISP-1 treatment or csg2 deletion. This result could be interpreted as a consequence of Mss4 mislocalization, because the PH domain of Rom2 binds to PI4,5P 2 and is required for proper localization (17). The phenotype of filamentous actin organization was also consistent with reported PI4,5P 2 mutations, because ISP-1 treatment disrupted actin cable formation, as reported for the mutations of Mss4 (14,15), Rho1 (50), or an actin-binding protein, profilin (51). Mss4 overexpression corrected the disrupted actin cable.
Among the mutants examined, csg2⌬ showed abnormal localization of Mss4-EGFP out of the plasma membrane (Fig.  1A). The biosynthesis of MIPC from IPC and GDP-mannose is defective in csg2⌬ strain; thus it accumulates IPC and lacks MIPC due to the metabolic defect (35,36). Similar mislocalization of the Mss4 was observed in ISP-1-treated cells, which is expected to cause loss of both IPCs and MIPC. Therefore the factor related to the mislocalization of Mss4 could be the loss of MIPC, even though we could not rule out the effect of molecules related to or metabolized from MIPC. Regardless, MIPC or/and molecule(s) related to it are required for the efficient anchoring of Mss4 in the plasma membrane, where PI4P is available for PI4P5K activity (18). At present, it is not known how such lipid molecule(s) could work as an anchor for Mss4, because MIPC is likely to be located in the outer leaflet of the plasma membrane (52). Unlike the well established phosphoinositide-PH domain interaction (53), a rather complex environment (possibly some "microdomain" structure) may be required for MIPC or related molecules to function as such anchor. In any case, the regulatory mechanism of Mss4 is expected to be rather novel, and identification of the possible domain/region required for such lipid-mediated anchoring of Mss4 will be a project of great interest in the future. We are currently studying the putative "domain" requirement of Mss4 for its proper plasma membrane localization. Identification of such domain would help to understand the question whether sphingolipids serve a structural role or a signaling role in bringing Mss4 to the membrane.
Present data imply that ISP-1 treatment of yeast seems to have a significant negative impact on Mss4 both in terms of activity and intracellular localization. Because MSS4 is known to be a lethal gene when disrupted (13), disturbance of Mss4 could be one of the causes of ISP-1-mediated cell death, even though it is not likely the sole cause. In ISP-1-treated yeast, the amount of Mss4 becomes limited, and thus overexpression of Mss4 rescues yeast from ISP-1 treatment. This is likely to be the reason why we were able to isolate MSS4 as a novel multicopy suppressor of ISP-1.
At present, no functional relationship between sphingolipidand PIP-mediated signaling has been reported in mammalian cells. The present findings in yeast may be evolutionally con- Rho2 (c and f), or having a control vector (a and d) were treated with 500 ng/ml ISP-1 (d-f) or the methanol vehicle (a-c) for 3 h and fixed. The filamentous actin was stained with rhodamine-phalloidin and visualized by confocal laser microscopy. The small arrowheads indicate actin cables, whereas big arrows indicate cortical actin patches. All images were created with the same sensitivity setting to ensure overall fluorescence signals were within the similar range. served in mammalian cells, because at least similar molecules are conserved in both systems (54,55).