Yeast 1,3- b -Glucan Synthase Activity Is Inhibited by Phytosphingosine Localized to the Endoplasmic Reticulum*

1,3- b - D -Glucan, a major filamentous component of the cell wall in the budding yeast Saccharomyces cerevisiae , is synthesized by 1,3- b -glucan synthase (GS). Although a yeast gene whose product is required for GS activity in vitro, GNS1 , was isolated and characterized, its role in GS function has remained unknown. In the current study we show that D gns1 cells accumulate a non-com-petitive and non-proteinous inhibitor(s) in the membrane fraction. Investigations of inhibitory activity on GS revealed that the inhibitor(s) is mainly present in the sphingolipid fraction. It is shown that D gns1 cells contain phytosphingosine (PHS), an intermediate in the sphingolipid

In plant and fungi, remodeling of the cell wall is one of the essential processes for cell shape determination. Among cell wall components in the budding yeast Saccharomyces cerevisiae, 1,3-␤-D-glucan (glucan) is the main structural component responsible for the rigidity of the cell wall (1). Glucan is synthesized by a specific biosynthetic enzyme, 1,3-␤-glucan synthase (GS) 1 (EC 2.4.1.34) localized to the plasma membrane. Yeast GS has been extensively studied both genetically and biochemically, revealing spatial and temporal regulation of cell wall synthesis (1,2). Recent studies of yeast GS revealed that it is composed of at least two subunits: a putative catalytic subunit encoded by two related genes, FKS1 and FKS2, and predicted to be an intrinsic membrane protein with 16-membrane spanning domains (3)(4)(5) and a regulatory subunit, a peripheral membrane protein encoded by RHO1 (6 -8). Since GTP-bound Rho1p is required not only for cell wall synthesis but also for intracellular actin organization (9), signal transduction leading to Rho1p plays a key role in cell morphogenesis. Another gene, GNS1, was originally isolated as a positive component required for GS activity in vitro (10). GS activity is severely reduced in the membrane fraction of a ⌬gns1 mutant (10). GNS1 interacts genetically with FKS1: a ⌬gns1 ⌬fks1 double mutant grows more slowly and exhibits more reduced GS activity in the membrane fraction than single mutants (10). Although these results suggest that GNS1 is somehow involved in GS activity, the physiological function of GNS1 remained unsolved since the ⌬gns1 mutant has a normal glucan content (10).
Other lines of evidence revealed that GNS1 is allelic to ELO2, which is involved in fatty acid elongation and sphingolipid synthesis (11,12). A ⌬elo2 (⌬gns1) mutant is defective primarily in elongation of very long chain fatty acids. Since yeast sphingolipids are structural components of very long chain fatty acids, inability of ⌬elo2 cells to synthesize very long chain fatty acids results in alteration in the amounts of intermediates in sphingolipid metabolism (11).
In this study, we further investigated the reduced GS activity in the membrane fraction of the ⌬gns1 mutant. Our results indicate that a PHS accumulation in the ⌬gns1 mutant causes non-competitive inhibition of GS activity.
Preparation of the Membrane Fraction-Cells were grown at 25°C in 1 liter of medium in a 2-liter flask rotating in air incubator (Innova 4330) at 150 rpm until the A 600 of the culture reached one. All the following procedures were carried out at 4°C, unless otherwise stated. The cells were harvested, washed with 1 mM EDTA, and disrupted by vortexing 4 times for 2 min each with 5 ml of glass beads in 20 ml the breaking solution containing 0.5 M NaCl, 1 mM EDTA with 1 mM phenylmethylsulfonyl fluoride. After centrifugation at 1,500 ϫ g for 5 min, the supernatant was collected and transferred to 33 PC tubes (Hitachi). The membrane fraction was collected by centrifugation at 100,000 ϫ g for 30 min in an RP70T rotor (Hitachi) with Himac CP 65␤ (Hitachi). The resultant pellet was suspended with a membrane buffer containing 50 mM Tris-HCl, pH 7.5, 10 mM EDTA, 1 mM ␤-mercaptoethanol, and 33% glycerol, homogenized with a Dounce homogenizer, and stored at Ϫ80°C.
Solubilization and Purification of GS-Purification of GS was carried out by product entrapment as described previously (6) with some modifications. GS was solubilized from the membrane fraction by adding 0.2 M NaCl, 20 M GTP␥S, 5 mM dithiothreitol, 0.5% CHAPS, and 0.1% cholesteryl hemisuccinate. This suspension was left on ice for 20 min and centrifuged at 100,000 ϫ g for 30 min. The supernatant was collected ("the detergent fraction"), to which 2.5 mM UDP-Glc and 20 mM potassium fluoride were added, and was subsequently incubated for 45 min at 30°C. The white polymer of 1,3-␤-glucan was collected by a 5-min centrifugation at 1,500 ϫ g at 4°C. The pellet was washed 3 times with extraction buffer (4 M GTP␥S, 1 mM dithiothreitol, 0.5% CHAPS, and 0.1% cholesteryl hemisuccinate in membrane buffer) containing 5 mM UDP-Glc and was centrifuged at 4,750 ϫ g for 5 min at 4°C. The resultant pellet was centrifuged again at 421,000 ϫ g for 10 min at 4°C. The tight pellet was homogenized in the extraction buffer and the purified GS fraction was recovered in the supernatant after re-centrifugation at 421,000 ϫ g for 10 min. GS activity was measured as formerly reported (4).
Trypsin and Heat Treatments-To the detergent fraction that was solubilized with CHAPS and cholesteryl hemisuccinate, 0.1 mg/ml trypsin was added. After a 1-h incubation on ice, trypsin digestion was stopped by addition of 0.2 mg/ml trypsin inhibitor. The samples were then heated to 100°C for 20 min.
Cell organelles were fractionated on equilibrium density gradients according to the published procedures (27) with several modifications.
Cultures were grown at 25°C in 4 liters medium until the A 600 of the cultures reached one. Cells lysates made with glass beads were cleared of debris by centrifugation at 1,000 ϫ g for 5 min. The membrane fraction was prepared by centrifugation at 100,000 ϫ g for 2 h. The pellet was resuspended with 5 ml of STE 10 (10% sucrose in breaking solution with protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 2 mg/ml chymostatin, 1 mg/ml leupeptin, 0.5 mg/ml pepstatin, and 0.5 mg/ml aprotinin) and was layered on top of 30 ml of a 20 -60% linear sucrose gradient in breaking solution. Samples were centrifuged at 100,000 ϫ g for 20 h at 4°C in a P28S rotor (Hitachi), and fractions of 3 ml were collected from the top of the gradient. Each fraction was diluted 5-fold by breaking solution and was centrifuged at 100,000 ϫ g for 1 h at 4°C in an RP70T rotor (Hitachi). The pellet was suspended with the membrane buffer as described above. Plasma membrane ATPase activity (28), Golgi GDPase activity (29), NADPH-cytochrome c reductase activity (30), ␣-D-mannosidase activity (31), kynurenine activity (32), and invertase activity (19) were assayed in gradient fractions as described previously. Sphingolipid Extraction-Sphingolipids were extracted from [ 3 H]serine-labeled cells as described (11). Radioactive bands were quantified and visualized with FLA-2000 (Fuji Photo Film) using a tritium screen. Sphingolipids were extracted from non-labeled cells as follows. Cultures of 1 liter each were grown at 25°C until the A 600 of each culture reached one. Cells were washed with 1 mM EDTA, suspended in 15 ml of breaking solution as described above and disrupted with 5 ml of glass beads by vortexing 4 times for 2 min. After addition of 30 ml of methanol and 15 ml of CHCl 3 , lipids were extracted according to the method of Bligh and Dyer (33). After removal of cell debris and glass beads by centrifugation at 1,500 ϫ g for 20 min, 12 ml of H 2 O, and 15 ml of CHCl 3 were added to the supernatant. Protein was removed by centrifuging at 1,500 ϫ g for 20 min. The lower layer was washed 3 times with 90 ml of CHCl 3 , methanol, 0.1 M EDTA, 0.1 M EGTA, pH 7.0 (2:2:5, v/v), and was dried with a vacuum evaporator. Half of the lipid extract was subjected to mild alkaline methanolysis by incubating with 3 ml of the monomethylamine reagent as prepared by Clarke and Dawson (34) for 1 h at 52°C. The resultant alkali-stable lipids were dried under a N 2 stream, resuspended in 0.1 ml of chloroform/methanol (2:1 by volume), and applied to Whatman Linear K6D silica gel TLC plates. The plates were developed with CHCl 3 , methanol, 30% NH 4 OH (9:7:2, v/v) or CHCl 3 , methanol, 4.2 N NH 4 OH (9:7:2, v/v). We preferred the 30% NH 4 OH solvent system since mass spectrometric analysis revealed that the PHS fraction obtained using the 4.2 N NH 4 OH solvent system contained degraded products of PHS or DHS. The mass spectrometric analysis was performed with Ion Trap LC/MS esquire 3000 (Nippon Bruker Daltonics). The fluorescent intensities of sphingolipids were quantified and visualized with LAS-1000plus (Fuji Photo Film) after spraying with primuline (Pfaltz & Bauer Inc.).
Fluorescence Microscopy Procedures-Procedures for immunofluorescence microscopy were as described previously (35). Anti-HA (16B12, BAbCO), and anti-Kar2p antibodies were used as primary antibodies. Cells were observed under the Olympus BX-FLA microscope (Olympus, Tokyo).

A GS Inhibitor(s) Accumulated in the ⌬gns1
Mutant-To characterize the reduced GS activity in the membrane fraction of ⌬gns1 cells, we determined the K m and V max values of GS activity. A kinetic analysis of GS activity revealed that the ⌬gns1 mutant has a decreased V max value, one-fifth that of wild-type cells, while the K m value was not significantly changed (Table I). These results indicate that the reduced GS activity in the ⌬gns1 mutant is due to the reduced maximum velocity. At least three possibilities are raised based on this observation: (i) the amount of GS diminishes, (ii) an activator of To test the first possibility, we measured the total amounts of Fks1p and Rho1p in ⌬gns1 cells. Immunoblotting analyses demonstrated that the Fks1p and Rho1p levels in the membrane fraction of ⌬gns1 cells were indistinguishable from those of wild-type cells (data not shown). In ⌬gns1 cells, Fks1p and Rho1p exhibited a normal localization pattern (6): both are placed at the growing tip of the bud (data not shown). These results suggested that the amount of GS localized to the plasma membrane was not altered in ⌬gns1 cells.
We also examined whether or not Gns1p is a component of the GS complex using a strain expressing Gns1p tagged at its C terminus with 3 repeats of the influenza hemagglutinin (3HA) epitope (Gns1:3HAp). A low-copy plasmid expressing Gns1:3HAp was introduced to the ⌬gns1 strain. The membrane fraction of ⌬gns1 cells expressing Gns1:3HAp exhibited a normal level of GS activity (data not shown), showing that Gns1: 3HAp is fully functional. GS was purified by extraction from the membrane fraction followed by product entrapment (see "Experimental Procedures"). In the purified GS fraction with more than 300-fold increase in specific activity (Fig. 1A), the regulatory subunit, Rho1p, was greatly enriched while Gns1: 3HAp was apparently lost (Fig. 1B). Examinations of the subcellular localization of Gns1:3HAp by immunofluorescent mi-croscopy revealed that Gns1:3HAp was mainly localized to the ER (Fig. 1C), exhibiting a localization pattern different from those of Fks1p and Rho1p (6). These results strongly suggested that Gns1p is not a tightly bound component of the GS complex.
Although we did not rule out the possibility that Gns1p activates GS, we obtained strong evidence suggesting that the ⌬gns1 mutant contains a GS inhibitor(s). We solubilized the membrane fractions of wild-type and ⌬gns1 mutant cells with detergents and treated with trypsin and heat to digest proteinous components. It was found that the resultant trypsin digest prepared from ⌬gns1 cells had an inhibitory activity to GS, while that from the wild-type cells had little inhibitory effect ( Fig. 2A). Dixson plots on GS activity showed linearity at 0.2 and 0.3 mM UDP-glucose and that the two lines intersected on the x axis (Fig. 2B). Taken together, the membrane fraction of ⌬gns1 cells likely contains a non-proteinous and non-competitive inhibitor(s) of GS.
Subcellular Distribution of Inhibitor(s) in the ⌬gns1 Cells-We analyzed the subcellular distribution of the inhibitor(s) in ⌬gns1 cells. Because inhibitory activity was mainly

FIG. 2. Non-competitive inhibition of the detergent fraction prepared from ⌬gns1 cells.
A, GS activity after mixing the detergent fraction prepared from ⌬gns1 and wild-type cells. The membrane fraction was solubilized with detergent ("the detergent fraction"), and the detergent fraction was treated with trypsin and heat ("the trypsindigested detergent fraction"). GS activity of the detergent fraction of wild-type cells was measured in the presence of the trypsin-digested detergent fraction. B, kinetic assay of GS inhibition by the detergent fraction of the ⌬gns1 mutant. In the presence of 10 l of the wild-type detergent fraction, various volumes (I) of the trypsin-digested detergent fraction of the ⌬gns1 mutant were added. GS activity was measured in the presence of 0.2 or 0.3 mM UDP-glucose. recovered from the membrane fraction (data not shown), the membrane fraction of ⌬gns1 cells was further fractionated on a 20 -60% linear sucrose gradient by centrifugation. We first determined the distributions of marker enzymes to check organelle distributions along the gradient. ER membranes were distributed in lower density fractions, while the plasma membrane was found in higher density fractions both in wild-type and ⌬gns1 fractions (Fig. 3A).
Each fraction was solubilized with detergents, followed by treatment with trypsin and heat, and was subjected to the inhibition assay. Fractions around the peak of the plasma membrane exhibited little difference in the inhibitory activity between the wild-type and ⌬gns1 strains. In contrast, lighter fractions of ⌬gns1 prominently inhibited GS activity (Fig. 3C). These results suggested that the inhibitor(s) of the ⌬gns1 mutant is concentrated not in the plasma membrane, but in lighter fractions containing the peak of ER.
PHS Inhibits GS Activity-The ⌬gns1 mutant is defective in elongation of very long chain fatty acids and synthesis of sphingolipids with very long chain fatty acids (11). Since the inhibitor(s) accumulated in ⌬gns1 cells is non-proteinous and mainly localized to the membrane fraction, we hypothesized that the inhibitor(s) is a lipid. To test this idea, we examined whether the total lipid extracted from ⌬gns1 cells has an inhibitory activity. It was found that the total lipid extracted from the ⌬gns1 mutant strikingly inhibited GS activity (Fig. 4), while that extracted from wild-type cells had less inhibitory activity. Furthermore, Fig. 4 shows that the sphingolipid prepared from the total lipid of the ⌬gns1 mutant by mild alkaline had the same inhibitory activity. These results suggested that the inhibitor(s) accumulated in the ⌬gns1 cells is mainly present in the sphingolipid fraction.
In order to identify which sphingolipid(s) inhibits GS activity, we resolved the sphingolipid fraction from [ 3 H]serine-labeled cells by TLC (11). As previously reported (11), the amounts of ceramide, inositol phosphoceramide (IPC), mannosylinositol phosphoceramide (MIPC), and mannosyl diinositolphosphorylceramide (M(IP) 2 C) decreased in the ⌬gns1 mutant (Fig. 5A), while both PHS and DHS contents increased 30-fold as compared with those in wild-type cells. Measurements of GS activity in the presence of each sphingolipid demonstrated that the PHS fraction from ⌬gns1 cells strikingly inhibited GS activity (Fig. 5B). The DHS fraction prepared from ⌬gns1 cells also inhibited GS activity to some extent.
Mutations defective in sphingolipid biosynthesis have been widely studied in S. cerevisiae ( Ref. 36; Fig. 6A). We constructed ⌬sur2 and ⌬ipt1 mutants, both of which affect sphingolipid biosynthesis (37)(38)(39), and measured their sphingolipid compositions and GS activities. ⌬sur2 cells contained an increased amount of DHS, while ⌬ipt1 cells contained normal levels of PHS and DHS (Fig. 6B). GS activity was specifically reduced in the membrane fraction isolated from ⌬sur2 cells (Fig. 6C). These results also suggested that DHS has GS inhibitory activity.
Since DPL1 encodes a possible long-chain base-phosphate lyase that catabolizes sphingolipids (Ref. 40 and Fig. 6A), it is likely that overexpression of DPL1 results in reduced intracellular levels of PHS and DHS (41). In order to test this possi- bility, we introduced multiple copies of DPL1 to the ⌬gns1 strain. We found that DPL1 overexpression in ⌬gns1 cells caused a 50% decrease in the amounts of PHS and DHS (Fig.  7A) as well as a reduction in GS inhibition (Fig. 7B). This result further supported the idea that an accumulation of PHS or DHS or both causes reduced GS activity in the membrane fraction.
To examine whether PHS and/or DHS inhibit GS activity, we isolated sphingolipids from wild-type cells, and directly measured their effects on the activity of purified GS. Ceramide, IPC, MIPC, and M(IP) 2 C did not affect GS activity in the concentration ranges examined. In contrast, PHS and DHS highly inhibited GS activity. The IC 50 of PHS was about 0.5 mg/ml (Fig.  8A), which is approximately in the same range as the physiological concentration in the ⌬gns1 mutant but notably higher than that in wild-type strain (data not shown). Fig. 8B shows that PHS inhibited GS activity in a non-competitive fashion. DHS also inhibited GS activity non-competitively ( Fig. 8A and data not shown), but the intracellular level of DHS was much lower than that of PHS in wild-type cells (Figs. 5A and 7A), suggesting that in vivo PHS is the primary GS inhibitor. Taken together, PHS, the intermediate of sphingolipids that accumulated in the ⌬gns1 mutant, was judged to be a potent intrinsic inhibitor to GS.
PHS Is Localized to the ER-In order to investigate the localization of PHS, we measured the PHS contents in purified organelle membranes. The quality of the organelle preparation was monitored by marker enzyme distributions (see "Experimental Procedures"). It was found that each organelle preparation contained only small amounts of other organelles (Table  II). 250 g of sphingolipids extracted from each organelle membrane was subjected to TLC (Fig. 9A). As previously reported (24,42), IPC was mainly localized to the Golgi membrane, Wild-type (YPH500), ⌬gns1, ⌬sur2, and ⌬ipt1 mutant cells were grown at 25°C in 1 liter of medium, and the sphingolipid fraction was prepared. 0.5 mg of sphingolipid fraction (dry weight) was applied and developed. C, decrease in the GS activity of the membrane fraction. Membrane fraction was prepared from wild-type (YPH500), ⌬gns1, ⌬sur2, and ⌬ipt1 mutant cells grown at 25°C and GS activities were measured. The data represent the means and standard deviations of at least three experiments. while MIPC and M(IP) 2 C were concentrated in the plasma membrane. Lipid particles contained neither MIPC nor M(IP) 2 C. Measurements of the fluorescent intensity of the PHS fraction demonstrated that this sphingolipid was largely localized to the microsomal membrane fraction (Fig. 9B). The localization pattern of PHS in wild-type cells is the same as that of Gns1p (Ref. 12 and Fig. 1C) and was consistent with the distribution of the inhibitor accumulated in ⌬gns1 mutant cells. DISCUSSION It was previously found that the membrane fraction of ⌬gns1 cells exhibits reduced GS activity (10). In this study, several lines of evidence indicated that an accumulation of PHS in the ⌬gns1 mutant causes non-competitive inhibition of GS activity. First, ⌬gns1 cells accumulated a non-competitive and nonproteinous inhibitor(s) in the membrane fraction. Second, the ⌬gns1 mutant accumulated PHS, which was discovered to inhibit GS non-competitively. Among the six sphingolipids examined, PHS was clearly the most potent GS inhibitor. Third, the localization of PHS to the ER was identical to that of the inhibitor accumulated in ⌬gns1 cells. Fourth, DPL1 overexpression partially lowered the level of PHS accumulated in the ⌬gns1 cells, resulting in reduced inhibition of GS. Thus, our results are consistent with the idea that PHS negatively regulates GS activity.
PHS inhibits GS activity non-competitively in several possi-ble ways. First, PHS may repress the interaction between Fks1p and Rho1p. For instance, PHS may form a microdomain around Fks1p to inhibit the interaction with the prenylated form of Rho1p required for GS activity (43). It has been reported that the particular lipid fraction isolated from mammalian cells contains glycosylphosphatidylinositol-anchored proteins (44 -46), but does not include prenylated proteins (47). Likewise, a particular lipid microdomain containing PHS possibly excludes prenylated Rho1p, preventing its interaction with Fks1p. Alternatively, PHS may alter the environment of lipid bilayer, causing inactivation of GS. The physicochemical state of lipid bilayer plays an important role in the activities of a number of enzymes (48). Since wild-type cells contain PHS on the ER membrane, it is likely that PHS inhibits GS activity on the ER. Many membrane-bound proteins localized to the plasma membrane are synthesized on the ER and transported through the secretory pathway (49). Fks1p, a putative catalytic subunit of GS, accumulated in intracellular organelles when vesicular transport was blocked by sec mutations, 2 suggesting that Fks1p is transported along this pathway to the plasma membrane after its synthesis on the ER. Furthermore, the GS activity was reduced in the membrane fraction of sec12-1, sec16-2, and sec21-1 cells, 2 all of which are defective in transport from the ER to the Golgi at the restrictive temperature (50,51). Furthermore, the reduction in GS activity in the sec mutants is diminished by overexpression of DPL1. 2 These results raise a possibility that a basal level of PHS residing in the ER functions to prevent nascent GS from being activated in the ER. If this is the case, it is the first example suggesting that a sphingolipid is involved in the inactivation of an enzyme at a specific organelle. In the case of many yeast enzymes that remain inactive until transported to their specific organelles, activation of enzymes is brought about by a modification or cleavage of their precursors. N-Glycosylation is required for activation of invertase (52,53) and exo-␤-1,3-glucanase (54), both of which are then transported to the plasma membrane. The N-linked oligosaccharide has been shown to affect the activities of these enzymes by stabilizing protein-protein interactions or by altering the affinity to their substrates. Peptide cleavage is required for activation of vacuolar enzymes such as proteinase A (55) and carboxypeptidase Y (56,57). Further study will be necessary to test whether sphingolipid in fact affects activities of nascent GS transported through the ER.  9. PHS distribution in yeast organelles. A, TLC of sphingolipids. Organelles were isolated from yeast cells, and sphingolipids were prepared from each organelle membrane. 250 g of sphingolipids was applied in each lane. B, the amount of PHS in the organelles. After spraying the TLC plate with primuline, fluorescent intensities of sphingolipids were quantified with LAS-1000plus (Fuji Photo Film). Nuc, MS, Mit, Vac, LP, SV, and PM indicate nucleus, microsomes, mitochondria, vacuoles, lipid particles, secretory vesicles, and plasma membrane, respectively. The data represent the means and standard deviations of three experiments.