Heat-induced elevation of ceramide in Saccharomyces cerevisiae via de novo synthesis.

Sphingolipid-related metabolites have been implicated as potential signaling molecules in many studies with mammalian cells as well as in some studies with yeast. Our previous work showed that sphingolipid-deficient strains of Saccharomyces cerevisiae are unable to resist a heat shock, indicating that sphingolipids are necessary for surviving heat stress. Recent evidence suggests that one role for the sphingolipid intermediate ceramide may be to act as a second messenger to signal accumulation of the thermoprotectant trehalose. We examine here the mechanism for generating the severalfold increase in ceramide observed during heat shock. As judged by compositional analysis and mass spectrometry, the major ceramides produced during heat shock are similar to those found in complex sphingolipids, a mixture of N-hydroxyhexacosanoyl C18 and C20 phytosphingosines. Since the most studied mechanism for ceramide generation in animal cells is via a phospholipase C-type sphingomyelin hydrolysis, we examined S. cerevisiae for an analogous enzyme. Using [3H]phytosphingosine and [3H]inositol-labeled yeast sphingolipids, a novel membrane-associated phospholipase C-type activity that generated ceramide from inositol-P-ceramide, mannosylinositol-P-ceramide, and mannose(inositol-P)2-ceramide was demonstrated. The sphingolipid head groups were concomitantly liberated with the expected stoichiometry. However, other data demonstrate that the ceramide generated during heat shock is not likely to be derived by breakdown of complex sphingolipids. For example, the water-soluble fraction of heat-shocked cells showed no increase in any of the sphingolipid head groups, which is inconsistent with complex sphingolipid hydrolysis. Rather, we find that de novo ceramide synthesis involving ceramide synthase appears to be responsible for heat-induced ceramide elevation. In support of this hypothesis, we find that the potent ceramide synthase inhibitor, australifungin, completely inhibits both the heat-induced increase in incorporation of [3H]sphinganine into ceramide as well as the heat-induced increase in ceramide as measured by mass. Thus, heat-induced ceramide most likely arises by temperature activation of the enzymes that generate ceramide precursors, activation of ceramide synthase itself, or both.

Widespread research, primarily in mammalian cells, has focused on sphingolipids as possible mediators of stress responses. Sphingolipid metabolites such as sphingosine, sphingosine-phosphate, and ceramide have been proposed as signaling molecules in a host of cellular processes (for recent reviews, see Refs. [1][2][3][4]; however, in most cases the precise molecular interactions in a sphingolipidmediated signaling cascade await definition. Ligand-activated sphingomyelinase activity acting on plasma membrane sphingomyelin is the most studied mechanism for generating the second messenger ceramide (5). Some reports suggest an alternative mechanism; altered ceramide synthase activity produces the ceramide increase seen in apoptosis (6) and arachidonic acid signaling in macrophages (7).
Saccharomyces cerevisiae is an exemplary organism in which to sort out the sphingolipid-related genes and proteins necessary for mounting a response to stress. Not only has the genome sequence been determined, but the sphingolipid composition is relatively simple as compared with mammals. The principal ceramide of S. cerevisiae, phytosphingosine (4-OH sphinganine) N-acylated with an ␣-OH C 26 fatty acid, is phosphodiester-linked to inositol (IPC), 1 mannosylinositol (MIPC), or inositolphosphorylmannosylinositol (M(IP) 2 C) (8). The synthesis of the Saccharomyces inositolphosphoceramides is schematized in Fig. 1, indicating inhibition of ceramide synthase by the potent antifungal agent australifungin (9).
Sphingolipids were first implicated in stress responses in S. cerevisiae when it was observed that strains unable to make sphingolipids failed to grow under conditions such as high temperature, high osmotic pressure, and low pH, whereas such strains could withstand these stresses when cultured so as to contain sphingolipids (10). Subsequent reports implicated sphingolipids in signaling roles in S. cerevisiae. N-Acetyl sphingosine (C2-ceramide) was shown by some (11,12) but not all (13) laboratories to inhibit growth (12) via the proposed activation of a protein phosphatase (11,12).
Shifting the temperature from 24 to 39°C is known to induce a variety of responses such as heat shock protein synthesis (14) and trehalose accumulation (see Ref. 15 and references cited therein). We recently demonstrated that such a temperature shift caused a severalfold elevation of ceramide (16) as well as a transient increases in sphinganine and phytosphingosine (17). Recent work also indicates that sphingolipids could be involved in the trehalose accumulation response, including the failure to accumulate trehalose in sphingolipid-deficient cells, exogenous sphinganine induction of trehalose accumulation in wild type cells, and exogenous sphinganine induction of the transcription of the TPS2 gene required for trehalose synthesis (17).
In this paper, we examine the biochemical basis for the heat-induced ceramide elevation. A recent study has raised questions concerning the validity of many reports of ceramide elevation based on using diacylglycerol kinase to measure ceramide (18). We have used direct chemical methods to establish the nature of the heatinduced ceramides in S. cerevisiae. We show that, although there exists a membrane-associated phosphodiesterase C-type enzyme activity that generates ceramide from yeast sphingolipids, it appears that the heat-induced ceramide increase results from de novo synthesis rather than from phosphodiesterase-mediated sphingolipid catabolism. This finding is different from the most studied explanation of ceramide elevation in mammalian cells, namely stimulated sphingomyelinase activity as part of a "sphingomyelin cycle" (5).

Ceramide Analysis in Unlabeled Cells
Free ceramides were extracted from cells, perbenzoylated, and quantitated after HPLC as described previously (23). Underivatized ceramide was sometimes directly analyzed. The initial chloroform/methanol (1:1) extract was filtered (Gelman 0.45 ACRO LC3S filters), dried, and dissolved in 200 l of chloroform. A 50-l aliquot was subjected to HPLC on a 0.45 ϫ 30-cm 5 Lichrosorb Si 60 column equilibrated with chloroform and eluted (1 ml/min) with 1 ml of chloroform followed by a 19-min linear gradient between chloroform and chloroform/methanol (3:1). Ceramide-III eluted at about 15 min and was detected with a Varex evaporative light scattering detector (drift tube temperature, 70°C; nitrogen flow rate, 1.1 liters/min).

Preparative Isolation and Analysis of Ceramides from
Heat-shocked Cells An overnight culture grown at 24°C (2 liters; A 650 ϭ 1.0) was warmed to 39°C and incubated with shaking for 40 min, followed by termination with trichloroacetic acid to 5%. The water-washed cell pellet was extracted with 60 ml of chloroform/methanol (1:1) at 50°C for 30 min and centrifuged while warm, and the supernatant was dried and suspended in 2 ml of chloroform. The extract was applied to a 6-ml silica gel column (Adsorbosil, 100 -200 mesh, Applied Science Labs) equilibrated with chloroform, and eluted with 9 ml of chloroform, 9 ml of chloroform/methanol (9:1), and 9 ml of chloroform/methanol (1:1). Ceramide-III assay by HPLC of the three fractions showed 0, 471, and 34 nmol, respectively. Fraction 2 was further purified by HPLC on a 0.45 ϫ 30-cm column of 5-m Lichrosorb Si60 (Merck) equilibrated with chloroform. The elution schedule (flow rate of 1 ml/min) was 1 min chloroform, 19 min linear gradient between chloroform and chloroform/methanol (75:25). The eluate was monitored with a Varex evaporative light scattering detector, and the fractions expected for ceramide-III were pooled, dried, and dissolved in chloroform. Final purification was on the same Lichrosorb column with isocratic elution with chloroform/methanol (9:1). The final ceramide fraction was analyzed for long chain base and fatty acid (24) as well as assayed by fast atom bombardment mass spectrometry. Part of the sample was benzoylated and separated by HPLC by a scaled up version of the protocol described above for quantitative ceramide analysis. Samples from each of the twin peaks were subjected to electron impact mass spectrometry.

Mass Spectrometry
Positive ion fast atom bombardment spectra of underivatized ceramides were measured with a Concept IH (Kratos) two-sector mass spectrometer equipped with an Ion Tech Ltd. saddle field gun operating with a xenon gas and set at a resolution of about 1500 at the acceleration voltage of 5.3 kV. Samples (ϳ2 nmol/2 l) in methylene chloride were added to 3-nitrobenzyl alcohol matrix (2-3 l) on a 7-mm diameter stainless steel probe tip. Spectra were acquired in a raw data mode, in a 100 -2100 atomic mass unit range, 3 s per decade, using external CsI mass calibration. Spectra representing the best response were averaged and digitally smoothed. Electron impact mass spectra of perbenzoylceramides were measured at 70 eV using a CONCEPT IH (Kratos) twosector instrument. Samples on the platinum wire were directly introduced to the ion source at 250°C. Spectra were acquired from 40 to 1400 atomic mass units, 3 s per decade, and the instrument was set to about 2000 atomic mass units resolving power.  [4, H]sphinganine (775 ϫ 10 6 dpm; prepared as described above) were dried in a sterile culture flask, sonic treated with 10 ml of culture medium, inoculated with a starter culture of strain YPH252 to a starting A 650 of 0.2, and shaken for 22 h at 30°C, the A 650 reaching 11. The cells were treated with trichloroacetic acid (5% final concentration), centrifuged, and washed twice with water. Lipids were extracted by treating each pellet with 2 ml of solvent B (diethylether, 95% ethanol, water, pyridine; 5:15:15:1 (v/v/v/v) containing 0.5 ml/liter concentrated ammonia) for 30 min at 60°C. Further purification proceeded by slightly different routes.

Preparation of [ 3 H]Phosphosphingolipids
The solvent B extract from the [4,5-3 H]sphinganine-labeled cells was added to a 1-ml column of BioRex 70 resin (H ϩ form, 200 -400 mesh, Bio-Rad) in a Pasteur pipette (packed in water and equilibrated with methanol) and washed with 3 ml of solvent B followed by 2 ml of methanol. The eluates, now free of long chain bases, were combined, dried, and dissolved in 1 ml of solvent B. Acyl ester lipids were deacylated by adding 1 ml of 0.2 N KOH in methanol and incubating for 30 min at room temperature. Further work up by adsorbtion to and elution from a Chelex resin C18 Celite mixture was as described previously (25). The sphingolipid fraction was dried and suspended in 1 ml of chloroform/methanol (1:1) and applied to a 3-ml column of silica gel (Adsorbosil, 100/200 mesh, Applied Sciences, Inc.) equilibrated with chloroform/methanol (1:1), and neutral catabolites of [4, H]sphinganine (ϳ11% of the radioactivity) were eluted with 6 ml of chloroform/ methanol (1:1). The [ 3 H]sphingolipids were eluted with 15 ml of solvent A containing three drops of concentrated ammonia/5 ml. Only radioactive sphingolipids were evident by thin layer chromatography (yield, 170 ϫ 10 6 dpm).
The solvent B extract from the [ 3 H]inositol-labeled cells was deacyl-FIG. 1. Sphingolipid synthesis in S. cerevisiae. The pathway reflects the available data; however, it is unclear whether long chain base 4-hydroxylation occurs at the level of the free long chain base(s) or at the level of ceramide. Also unclear is the nature of the substrate(s) for fatty acid ␣-hydroxylation. ated directly and purified over a C 18 -Celite column as above; this procedure was repeated once. Thin layer chromatography of the final base-stable Celite eluates showed all the radioactivity to be in various species of sphingolipids (yield, 99 ϫ 10 6 dpm).
Separation into molecular species was carried out by thin layer chromatography (Whatman HP-K 200 m; 10 ϫ 20 cm, solvent C). Each radioactive zone was eluted with solvent A, dried, dissolved in 1 ml of solvent A, and filtered (Acrodisc 4 CR 4 mm; 0.45-m filters, Gelman Sciences, Inc.). Purity and identity was verified by subjecting each molecular species to high performance thin layer chromatography as above. Unlabeled standards were incorporated in each lane and were detected by charring (25) after the radioactivity was located (BioScan apparatus).

Product Analysis of Putative Yeast Membrane Phospholipase C Action on [ 3 H]Phosphosphingolipids
The reaction mixtures (0.3 ml) consisted of 22 mM potassium phosphate, pH 7.0, 2 mM dithiothreitol, 5 mM MgCl 2 , 0.6% n-octyl-␤-glucopyranoside, 0.22 g of membrane (27) protein. Also added were 10 nmol of unlabeled sphingolipid (IPC-III, MIPC-III, or M(IP) 2 C-III) and equal amounts of radioactivity of the corresponding [ 3 H]inositol-and [ 3 H]sphinganine-labeled sphingolipids: IPC-III, 101,000 cpm; MIPC-III, 39,000 cpm; M(IP)2C-III, 200,000 cpm. The sphingolipids were dried and suspended by sonic treatment in the assay mixture before the addition of enzyme. After 60 min at 24°C, the reaction was terminated by 2-min heating (100°C) followed by adding 30 l of 0.5 M Na-EDTA, pH 7.1. Aliquots were chromatographed on silica gel paper (solvent C). One-cm zones were subjected to scintillation counting. Water-soluble products were at the origin, ceramides migrated near the solvent front, and sphingolipids migrated at these R F s: IPC-III, 0.65; MIPC-III, 0.54; M(IP) 2 C-III, 0.30. Separate 20-l aliquots of the reaction mixtures were chromatographed on silica gel paper with the solvent chloroform/methanol (19:1.5, v/v) along with ceramide-III standard. Radioactivity was determined for each lane as above. Water-soluble products as well as the phosphosphingolipids remain at the origin with this solvent, while ceramide-III migrates to an R F factor of 0.66.
To identify the water-soluble fragments of sphingolipids resulting from a putative phospholipase C reaction, 10 6 cpm each of [ 3 H]inositollabeled IPC-III and M(IP) 2 C-III were incubated as above except that no unlabeled sphingolipid was added, and the reaction was carried out for 4 h, adding an equal amount of membranes after 2 h. The reaction was terminated as above. To each sample was added 30 l of 0.5 M Na-EDTA, pH 7.1. Aliquots were chromatographed on silica gel paper (solvent C). Radioactivity in each lane was quantitated as described above. The polar product(s) at the origin derived from IPC-III and M(IP) 2 C-III were 44 and 16% of the total counts, respectively. Watersoluble products were separated by ion exchange chromatography by diluting the reaction mixtures with 5 ml of water followed by centrifugation and application of the supernatant to a 0.6 ϫ 81-cm column of AG1-X2 (200 -400 mesh bicarbonate form, Bio-Rad) equilibrated with 50 ml of 0.1 M ammonium bicarbonate. Elution was with 0.3 M ammonium bicarbonate, pH 7.9 (flow rate 1.22 ml/min), collecting 8.8-ml fractions. Radioactive peaks were located and quantitated, pooled, and dried in vacuo at 70°C to remove ammonium bicarbonate. Some of each pooled peak was dissolved in 0.1 ml of 0.2 M ammonium acetate, pH 8.6, treated with 30 l of Escherichia coli alkaline phosphatase (type III, Sigma, 0.33 units/l) for 3 h at room temperature, diluted with 2 ml of water, and chromatographed/quantitated on an AG1-X2 column exactly as described above. Another portion of the pooled peaks was dried and treated with 1 ml of 10 N NH 4 OH for 18 h at 150°C. The labeled ammonolysis products, mannosylinositol and inositol, were resolved by thin layer chromatography (Whatman HP-K) and developed with acetonitrile/water (3:1), followed by detection/quantitation with a BioScan apparatus.

Phospholipase C Assay with Yeast Membranes and Phosphosphingolipids
The reaction mixture was 25 mM potassium phosphate, pH 7.0, 2.5 mM dithiothreitol, 5 mM MgCl 2 , 0.6% n-octyl-␤-glucopyranoside, 33.3 M [ 3 H]sphingolipid (ϳ10 5 cpm), 100 -400 g of membrane protein in a 0.3-ml volume. The [ 3 H]sphingolipids were added to the reaction tube, dried, and suspended by sonic treatment with the assay mixture before the addition of enzyme. After incubation for 60 min at 24°C, the reaction was terminated and processed by two different methods.
Method A-Sodium dodecylsulfate (0.9 ml, 2.27%) was added to the reaction mixture. Ceramide was extracted from the mixture with two 2-ml portions of methyl tert-butylether. The pooled extracts were washed with 1 ml of water and dried, and radioactivity was measured. A small no enzyme blank reaction, equivalent to Ͻ1% apparent substrate breakdown, was subtracted from the membrane-containing samples to calculate the specific activity.
Method B-The reaction was stopped by the addition of 3 ml of chloroform/methanol (1:1) and centrifuged. The supernatant was added to a 1-ml column of AG4-X4 (acetate form, 100 -200 mesh, Bio-Rad; packed in water and equilibrated with methanol) and eluted with 3 ml of chloroform/methanol (1:1). The substrates bound to the resin. The eluted ceramide was dried and assayed for radioactivity. A no enzyme blank value was subtracted.

Turnover of Inositol-labeled Lipids
An overnight log phase culture was transferred to 15 ml of fresh medium containing 1.5 mCi of myo-[2-3 H]inositol (American Radiochemicals, Inc.) to give a starting A 650 ϭ 0.2 and cultured for 6 h at 24°C. The cells were rapidly resuspended in 15 ml of fresh nonradioactive medium and divided in three parts. One part (zero time) was centrifuged, and the pellet was quenched with 1 ml of 1 M HClO 4 . The other two parts were incubated at 24 and 39°C for 20 min followed by rapid centrifugation and quenching of the cell pellets with HClO 4 as above, yielding HClO 4 extracts and culture medium fractions, which were filtered (0.2-Teflon Acrodiscs CR, Gelman Sciences). After standing for 15 min at 0°C, the HClO 4 -treated cell pellets were frozen and thawed twice in dry ice/ethanol and centrifuged at 0°C. The supernatants were slowly neutralized (chlorophenol red pH indicator) at 0°C with 2.6 M KOH and centrifuged, and the final supernatants were reserved for ion exchange chromatographic analysis. The cell pellets were washed twice with water and extracted for sphingolipids with 1 ml of solvent B for 30 min at 60°C, followed by centrifugation while warm. The extract was dried and deacylated by treatment with 0.5 ml of monomethylamine reagent (28) for 30 min at 50°C. After evaporation of the reagent, the sample was dissolved in 1 ml of solvent A, and aliquots were chromatographed on silica gel paper (solvent C). One-cm zones from each lane were analyzed by scintillation counting to give three groups: the counts near the origin, representing deacylated phosphatidylinositol, the M(IP) 2 C zone, and the zones representing incompletely resolved MIPC plus IPC. Aliquots (0.5 ml) of the three neutralized HClO 4 extracts as well as the filtered culture media were diluted 10and 20-fold, respectively, with water and applied to 0.6 ϫ 81-cm columns of AG-1-X2 200 -400 mesh resin equilibrated with 0.1 M ammonium bicarbonate. Elution was carried out at 1.2 ml/min with 0.3 M ammonium bicarbonate, collecting 20 8.6-ml fractions, which were analyzed for radioactivity. Peak fraction numbers for labeled products were as follows: inositol, 2; glycerophosphorylinositol, 4; inositol-P and mannosylinositol-P, 6; mannose-(inositol-P) 2 , 11.

Incorporation of [ 3 H]Sphinganine into Ceramide and Phosphosphingolipids
About 7.5 A 650 units of mid-log phase cells were suspended in 5 ml of fresh medium containing 47 ϫ 10 6 cpm [ 3 H]sphinganine and incubated at 24 or 39°C. Aliquots (1.4 ml) were removed and terminated with 0.07 ml of 100% trichloroacetic acid and incubated at least 15 min on ice. The cells were centrifuged and washed twice with 1 ml of cold water, and the final pellet was extracted with 0.4 ml of solvent B for 30 min at 60°C. After centrifuging while still warm, the soluble extract was processed for further analysis.
For labeled ceramide analysis, [ 3 H]sphinganine was first removed by applying the solvent B extract to a 0.5-ml column of BioRex 70 (Hϩ) resin (200 -400 mesh) packed in water and equilibrated with methanol. Elution was carried out with 1.5 ml of solvent A and then with 1 ml of methanol. The combined eluates were dried and dissolved in 0.2 ml of solvent A and subjected to thin layer chromatography (Whatman K5 plates, solvent CHCl 3 /methanol, 19:1.5). To each lane, 3.5 nmol of ceramide-III was added. Following radioactivity determination (Bio-Scan apparatus), the plates were sprayed with 10% (w/v) CuSO 4 ⅐5H 2 O in 8% H 3 PO 4 and charred at 160°C (26) to locate the added ceramide standard. The total radioactivity in the ceramide zone was calculated from the BioRex 70 eluates and from the percentage distribution of radioactivity from the BioScan analysis of the thin layer plates.
For phosphosphingolipid analysis, the solvent B extract was dried and deacylated with 0.5 ml of monomethylamine reagent as above followed by thin layer chromatography (200-Whatman HP-K plates, solvent C). Each lane contained a mixture of sphingolipid standards, 1-2 g each of IPC-III, IPC-III, MIPC-III, and M(IP) 2 C-III. Following radioactivity determination (BioScan apparatus), the added sphingolipid standards were located by charring (26). The total radioactivity in the phosphosphingolipid zones was calculated from the deacylated solvent B extracts and from the percentage distribution of radioactivity (Bioscan apparatus) on the thin layer plates.

Effects of Australifungin and Cycloheximide on Conversion of [ 3 H]Sphinganine to Ceramide and Phosphosphingolipids
A log phase culture grown at 24°C was transferred to fresh medium without Tergitol (A 650 ϭ 1.7), and 4.5-ml samples were incubated for 10 min at 24°C after the addition of 3 l of australifungin (500 g/ml ethanol) or 3 l of ethanol to the controls. After the addition of 0.5 ml of 0.5% Tergitol containing 2.5 ϫ 10 8 dpm [ 3 H]sphinganine, incubation was continued at 24 and 39°C. At 20, 40, and 60 min, 1.4-ml aliquots were quenched with 0.07 ml of 100% trichloroacetic acid. Another experiment was carried out as above, except cycloheximide, 1 mM final concentration, was added at zero time. Labeled ceramides and phosphosphingolipids were extracted, processed, and analyzed by thin layer chromatography as described above.

Evidence for Heat Induction of Two Molecular Species of
Ceramide-S. cerevisiae cells were cultured at 24°C and switched to 39°C, and their extracted lipids were derivatized and separated by HPLC. It can be seen (Fig. 2) that a significant increase occurs at 39 but not at 24°C in a double peak that migrates similarly to ceramide prepared from autolysed commercial bakers' yeast. To positively identify these two peaks as ceramides, we isolated the putative ceramide fraction from heat-shocked cells and subjected the ceramide fraction and the two peaks formed after benzoylation and HPLC separation to compositional analysis and mass spectrometry. Hydrolysis of the putative ceramides yielded equimolar amounts of hydroxyhexacosanoic acid and C 18 and C 20 phytosphingosines. Fast atom bombardment mass spectrometry of the ceramide fraction showed two molecular ions with masses 712.7 and 740.7, consistent with N-hydroxyhexacosanoyl-C 18 phytosphingosine and N-hydroxyhexacosanoyl-C 20 phytosphingosine, respectively (Table I). After benzoylation and separation by HPLC, peaks I and II exhibited molecular ions with masses of 1155.8 and 1127.8, respectively, as expected for tetrabenzoyl-N-hydroxyhexacosanoyl-C 20 phytosphingosine and tetrabenzoyl-N-hydroxyhexacosanoyl-C 18 phytosphingosine, respectively (Table  I). These molecular species of ceramide are consistent with the ceramide found in complex yeast phosphosphingolipids (8).

Accumulation of Ceramide Is Rapid and Specific to Heat
Stress-We hypothesized that mutant strains lacking sphingolipids cannot grow at low pH, in high salt, or at high temperature (10) because they are unable to generate sphingolipid second messengers such as ceramide. To test this hypothesis, we examined wild type cells for changes in ceramide following stress. A rise in ceramide concentration can be detected after 10 min of heat treatment at 39°C. The rise peaks after 30 -40 min and is sustained for at least 2 h (Fig. 3). The 4 -5-fold heat-induced ceramide elevation appears to be specific to heat stress, since neither high osmotic pressure (Fig. 3A) nor low pH (Fig. 3B) at 24°C resulted in marked elevation of ceramide. Enzymatic Hydrolysis of Yeast Phosphosphingolipids Can Generate Ceramide-Activation of a sphingomyelinase activity seems to be responsible for the generation of ceramide in mammalian cells following various stimuli (1,2). Therefore, we sought evidence for the existence of a comparable phosphodiesterase, with phospholipase C specificity, that would hydrolyze yeast phosphoinositol sphingolipids to generate ceramide. Ceramide (N-hydroxyhexacosanoylphytosphingosine)-labeled sphingolipids were isolated from cells metabolically labeled with [ 3 H]sphinganine and used as substrates. In one enzyme assay (method A), the radiolabeled ceramide product was extracted with methyl tert-butylether. In an alternate assay (method B), the solvent-treated assay mixture was chromatographed on a small anion exchange column, which retained the acidic substrates but not the free ceramides.
[  2 C by incubation with crude membrane preparations in a reaction that absolutely required octyl glucoside and MgCl 2 (not shown). Ceramide release was reasonably linear with time, and protein concentration and optimum hydrolysis activity was obtained with 2-3 mol % of M(IP) 2 C in the mixed micelles, bulk concentration 33 M (Fig. 4). The pH optimum for the reaction was about 6 -6.5 (Fig. 4).
Product Analysis of Yeast Membrane Phosphodiesterase Action on [ 3 H]Phosphosphingolipids-Since phosphodiesterase activity toward yeast phosphosphingolipids has not been previously reported, it seemed essential to define the stoichiometry and nature of the products in order to establish whether hydrolysis is between the phosphorus and ceramide (phospholipase C) or between the phosphorus and the inositol (phospholipase D), the latter generating phosphoceramide that would have to undergo further hydrolysis by a phosphatase to yield ceramide. All of the evidence described below is consistent with membranes containing an enzyme(s) with phospholipase C-

FIG. 2. Assay of ceramide (Cer-III Std) in heat-shocked cells.
Aliquots (50 ml) of a culture grown at 24°C to an A 650 of 0.57 were removed at zero time and after 30 min of incubation at 24 or 39°C. Lipid extracts were benzoylated and subjected to HPLC as described under "Experimental Procedures."

TABLE I
Identification of ceramides isolated from heat-shocked cells A putative ceramide-III fraction was isolated from heat-shocked cells. Its fatty acid and long chain base composition was analyzed. Masses of the most abundant molecular ions were determined by fast atom bombardment (FAB) mass spectrometry. After perbenzoylation, peaks I and II (Fig. 2) were separated by HPLC, and these molecular species were subjected to electron impact (EI) mass spectrometry to determine their molecular ions. type phosphodiesterase activities catalyzing the following reactions. We first reacted sphingolipid substrates containing equal radioactivity in their ceramide and inositol portion and chromatographed the entire reaction mixture on silica gel-impregnated paper. In this system, the water soluble product(s) remain at the origin, while ceramide and other sphingolipids migrate at the R F s indicated under "Experimental Procedures." In the case of IPC-III, equal amounts of radioactivity were found in the ceramide and origin regions, consistent with phosphodiesterase activity (Table  II). In the case of M(IP) 2 C, in addition to ceramide, some MIPC was formed, requiring that some inositol-P be one of the polar products. The ratio of ceramide to total polar product radioactivity was as expected for the action of a phosphodiesterase(s) cleaving M(IP) 2 C to yield free ceramide as well as yielding equimolar amounts of inositol-P and MIPC (Table II). The observed stoichiometry is consistent with a phosphodiesterase acting on the phosphosphingolipids.
Evidence for a phospholipase C-type mechanism for sphingolipid hydrolysis was obtained by showing that the polar products had alkaline phosphatase-susceptible phosphomonoester groups. The polar products generated from [ 3 H]inositol-labeled sphingolipids treated with membranes were isolated by anion exchange chromatography and then treated with alkaline phosphatase. The water-soluble products generated from [ 3 H-inositol]IPC gave a major peak at the retention time expected for inositol monophosphate (52 ml) as well as a smaller peak where free inositol would emerge (Table III). Anion exchange chromatographic analysis after phosphatase treatment showed complete conversion of the putative inositol phosphate radioactivity to free inositol (Table III).
Analysis of the water-soluble products derived from [ 3 Hinositol]M(IP) 2 C disclosed a small amount of free inositol and two major products, the first with a net charge of about Ϫ2, consistent with inositol-P, mannosylinositol-P, or both, and the second with a net charge of about Ϫ3, consistent with inositol-P-mannosylinositol-P (Table III). The charge Ϫ2 product, when treated with phosphatase, completely migrated in the zero net charge region (Table III), consistent with the reaction(s) inositol-P 3 inositol ϩ P i and/or mannosylinositol-P 3 mannosylinositol ϩ P i . The second product in the net charge Ϫ3 region was completely converted by phosphatase to a product that

FIG. 4. Ceramide formation by reaction of [ 3 H]M(IP) 2 C and [ 3 H]IPC with putative membrane phospholipase C. [ 3 H]M(IP) 2 C was reacted with yeast membranes and processed as per "Experimental
Procedures." A and B, method B; C, method A; D, [ 3 H]IPC-III was the substrate and was processed as described in the legend to Table II.   (Table III), consistent with the reaction inositol-P-mannose-inositol-P 3 inositol-Pmannose-inositol ϩ P i . The identities of the two radioactive major M(IP) 2 C products (Table III) were further established by subjecting each to ammonolysis conditions that hydrolyze all phosphate bonds, leaving the mannose-inositol glycosidic bond intact (29). The radioactive ammonolysis products were resolved by TLC with the radioactivity being accounted for as inositol and mannosylinositol. Table IV shows that each major peak gave the same ratio of mannosylinositol to inositol counts as the original M(IP) 2 C. These data are consistent with the interpretation that the charge Ϫ3 peak (Table III) composition was inositol-P-mannose-inositol-P and that the charge Ϫ2 peak consisted of about equal amounts of inositol-P and mannoseinositol-P. The digestion of M(IP) 2 C (Table III) was carried out for much longer than the experiment described in Table IV (see "Experimental Procedures"), probably accounting for a higher proportion of the reaction M(IP) 2 C 3 MIPC ϩ inositol-P.

Does Sphingolipid Breakdown by a Phospholipase C Type Enzyme Account for Heat-induced Ceramide Elevation?-Since
we established that S. cerevisiae has phospholipase(s) C that can generate ceramide from sphingolipids, we looked for evidence that catabolism of phosphosphingolipids might be responsible for heat-induced increases in ceramide. Because ceramide increases represent only a few percent of the total potential ceramide in sphingolipids, it was not practical to look for a heat-induced decrease in sphingolipid levels. We therefore looked for the increases in the free, water-soluble inositolcontaining sphingolipid head groups that would be generated concomitant with ceramide formation. We cultured cells at 24°C in the presence of [ 3 H]inositol to label the sphingolipids, followed by incubation in fresh medium for 20 min at 24 and 39°C. Water-soluble substances were extracted from the cells with perchloric acid followed by lipid extraction. The distribution of radioactive lipids was obtained after paper chromatography. The aqueous cellular fractions as well as the culture medium were subjected to anion exchange chromatography to look for peaks of radioactivity that increase in the 39 versus 24°C samples at elution volumes expected for the sphingolipid head groups (Table III). Based on the radioactivity in the total sphingolipids (Table V) at zero time (4,077,000 cpm) and the increase in ceramide expected to result from heat treatment (Fig. 3) equivalent to breakdown of about 6% of the sphingolipid (see legend to Table V), we can calculate that the acidic sphingolipid catabolites should increase by about 258,000 cpm in the 39°C sample. One-tenth of that value could be readily detected, but as can be seen from the data (Table V), there is no increase in radioactivity at 39°C compared with 24°C, in either the perchloric acid cell extract or the culture medium. It thus appears unlikely that the ceramide increase following heat shock is due to breakdown of sphingolipids. We cannot rule out the possibility of an extremely rapid catabolism of sphingolipid head groups, which would prevent their detection.
Further support for this conclusion was obtained with a mutant (RCD113) defective in M(IP) 2 C synthesis (22), which   H]inositol for 6 h at 24°C, and then with fresh unlabeled medium for 20 min at either 24 or 39°C. The cellular HClO 4 -soluble fraction and the culture medium were analyzed by anion exchange chromatography to separate and quantify potential watersoluble sphingolipid head groups arising from sphingolipid breakdown. Sphingolipids, extracted from the perchloric acid-treated cells, were separated and measured after paper chromatography. Increase expected b in acidic catabolite(s) assuming ceramide elevation at 39°C is due to sphingolipid breakdown 258 a Sum of radioactivity in the eluate regions expected for inositol-P, mannose-inositol-P, and inositol-P-mannose-inositol-P.
b Ceramide increase at 39°C is 95 pmol/A 600 unit. Total sphingolipid inositol is 1500 pmol/A 600 unit. Therefore, if all of the additional ceramide at 39°C were generated by the catabolism of sphingolipid, one would expect to generate acidic catabolites amounting to (95/ 1500)(4.077 ϫ 10 6 ) ϭ 258,000 cpm.

TABLE III
Analysis of the water-soluble products formed by the action of yeast membranes on IPC and M(IP) 2 C [ 3 H]IPC-III and M(IP) 2 C-III were incubated with membranes, and the water-soluble products were resolved by anion exchange column chromatography as described under "Experimental Procedures." The resulting peaks were pooled, treated with alkaline phosphatase, and subjected to the same anion exchange chromatographic system. makes no detectable M(IP) 2 C due to the deletion of the IPT1 gene but accumulates increased levels of MIPC. Strain RCD113 gave a heat-induced ceramide response equivalent to its cognate wild type strain (Table VI). Thus, M(IP) 2 C breakdown cannot be the source of the elevated ceramide level observed during heat shock. Is de Novo Synthesis of Ceramide Responsible for the Heatinduced Increase?-In view of all of the results, we considered whether increased de novo synthesis accounts for heat-induced increases in ceramide. We showed previously (17) that a change from 24 to 39°C results in a temporary elevation of the concentration of sphinganine and phytosphingosine. We therefore determined if exogenous long chain base alone could increase ceramide levels. When phytosphingosine was added to cultures at 24°C, ceramide did not increase to the level achieved by raising the temperature to 39°C, although phytosphingosine, when added at 39°C, increased ceramide levels somewhat (Fig.  5). A similar experiment carried out with 50 M DL-sphinganine at 24°C induced little increase in ceramide above the untreated control (data not shown). We conclude that increased long chain base synthesis alone is insufficient to account for the ceramide increase observed at 39°C.
Further evidence that de novo ceramide synthesis is responsible for heat-induced ceramide accumulation was sought by studying the incorporation of [ 3 H]sphinganine into ceramide at 24 and 39°C and in the presence of a ceramide synthase inhibitor. The rate and extent of ceramide-III labeling was higher at 39 than 24°C (Fig. 6A). Australifungin, a potent antifungal agent and inhibitor of ceramide synthase (9), abolished labeling of ceramide at either temperature (Fig. 6A), indicating that the ceramide synthase responsible for ceramide labeling in this experiment was sensitive to this antibiotic. To further implicate ceramide synthase in the temperature-induced accumulation of ceramide, we determined if australifungin would prevent the temperature-induced increase in ceramide as measured by mass. Australifungin at 0.3 g/ml totally prevented the temperature-induced accumulation of ceramide (Table VII). In fact, the ceramide level fell below the control probably because it was further metabolized to complex sphingolipids. It thus seems likely that ceramide synthase plays a key role in the temperature-induced accumulation of ceramide.
If the 39°C induced increase in ceramide were due to increased synthesis of the enzyme ceramide synthase, then the enhanced conversion of [ 3 H]sphinganine to ceramide should be blocked by cycloheximide. Cycloheximide (1 mM) did not inhibit ceramide synthesis at either temperature ( Fig. 6B) but actually increased the extent of labeling; thus, it is unlikely that increased synthesis of ceramide synthase or any other protein mediates the temperature-induced synthesis of ceramide.
Another explanation for enhanced ceramide accumulation is a reduction in the rate of conversion of ceramide to complex sphingolipids following a temperature shift, for example, by inhibition of IPC synthase (Fig. 1). However, the temperature shift from 24 to 39°C does not inhibit but rather increases total radiolabeling of complex sphingolipids by [ 3 H]sphinganine, although not to the extent that it enhances ceramide labeling; as expected, sphingolipid labeling is abolished by australifungin (Fig. 6C).
Finally, although australifungin is a potent ceramide syn-   thase inhibitor, it has not been studied enough to know if other reactions are affected. Specifically, we sought to rule out the possibility that it might inhibit enzymatic hydrolysis of complex sphingolipids yielding ceramide. Australifungin, at concentrations up to 10-fold higher than those that abolished in vivo ceramide synthesis (Fig. 5, Table VI), was without effect on the membrane-catalyzed hydrolysis of IPC and M(IP) 2 C (not shown). We conclude from this experiment and all of the other data presented in this section that increased de novo ceramide synthesis accounts for the heat-induced ceramide elevation.

DISCUSSION
De Novo Synthesis Is Responsible for Heat-induced Elevation of Ceramide-Our experiments demonstrate that de novo synthesis, not breakdown of sphingolipids, is the primary mechanism for generating an increased level of ceramide following a shift from 24 to 39°C. To show that de novo synthesis of ceramide via ceramide synthase (Fig. 1) was essential for a increased ceramide, we employed australifungin, a potent antifungal drug shown to be an inhibitor of ceramide synthase (9). The heat-induced increase in ceramide as measured directly by mass as well as by incorporation of [ 3 H]sphinganine into ceramide, were completely inhibited by australifungin (Table VII, Fig. 6), thus identifying the essentiality of ceramide synthase for the heat-induced ceramide accumulation.
The question of what reaction(s) leading to increased ceramide are affected by temperature elevation is complex and largely unanswered. One possible explanation could be differential temperature effects on the rate constants of the enzymatic reactions leading to ceramide synthesis and to its further metabolism. An additional explanation is that elevated temperature could lead to alterations of the amount and/or structure of one or more enzymes of ceramide metabolism. Increased enzyme synthesis does not play a role as judged by the lack of effect of cycloheximide on ceramide synthesis (Fig. 6B). A heatinduced decreased rate of ceramide conversion to IPC and more complex sphingolipids, which could explain ceramide accumulation, is also not evident from the experiment measuring the incorporation of [ 3 H]sphinganine into sphingolipids (Fig. 6C). Thus, the enzymes that are likely candidates for temperature regulation are those involved in generating ceramide precursors as well as ceramide synthase itself. Earlier work showed (17) that heat shock causes a rapid and temporary rise in the concentration of sphinganine and phytosphingosine, precursors of ceramide. However, increased exogenous long chain base by itself appears to be inadequate to account for increased ceramide, since exogenous long chain bases added in excess at 24°C did not stimulate ceramide concentrations to the level achieved at 39°C (Fig. 5). The mechanism of heat-induced ceramide accumulation merits further analysis, especially since exogenous sphinganine activates trehalose accumulation at 24°C via gene activation (17), thus implicating yeast sphingolipids in a well known stress response.
Ligand activation of a sphingomyelinase, the "sphingomyelin cycle" (5), has been the most studied reaction to account for stress-induced ceramide generation in animal cells. However, two studies with animal cells implicated ceramide synthase in ceramide generation. Daunorubicin-induced apoptosis and ceramide elevation were prevented by the ceramide synthase inhibitor, fumonisin B1 (6); however, other workers claim sphingomyelin hydrolysis is associated with daunorubicin-induced apoptosis (30). Fumonisin has been reported to inhibit ceramide elevation associated with macrophage activation (7). Two studies in animals cells have observed heat shock-induced elevation in ceramide by as yet undefined mechanisms (31,32). Future work needs to be directed at the unknown mechanism(s) of temperature regulation of ceramide synthase activ-ity in both animal cells and yeast.
Phospholipase C Type Activity in Yeast Utilizing Phosphoinositol-containing Sphingolipids-Generation of ceramides by activation of a sphingomyelinase, Choline-P-ceramide ϩ H 2 O 3 Choline-P ϩ ceramide REACTION 5 is the predominant paradigm in mammalian cells for the formation of mediators in various signaling pathways with diverse outcomes (1)(2)(3)(4). We therefore looked for a comparable phosphodiesterase activity in yeast to explain the heat-induced increase in ceramide. Our data suggest the existence of one or more phosphodiesterases in S. cerevisiae membranes capable of catalyzing the hydrolysis of yeast sphingolipids to yield ceramide, R-Inositol-P-ceramide ϩ H 2 O 3 R-inositol-P ϩ ceramide REACTION 6 where R represents hydrogen, mannose, or inositol-P-(mannose). With sphingolipids containing [ 3 H]ceramide as well as [ 3 H]inositol in the head groups, the above stoichiometry was demonstrated (Tables II-IV). Furthermore, the ceramides from heat-shocked cells were isolated, and their chemical composition was confirmed by chemical analysis and by mass spectrometry (Table I). These analyses as well as ceramide analysis in the various heat shock experiments were all performed by chemical methods, a noteworthy observation in light of the recent challenge of the validity of ceramide measurements made by many investigators employing the enzyme diacylglycerol kinase for ceramide analysis (18).
Another phosphodiesterase with activity toward substrates containing ceramide has been described in S. cerevisiae. Ella et al. (13) characterized a sphingomyelinase activity, partially purified from S. cerevisiae membranes, capable of generating ceramide from sphingomyelin. However, this enzyme preparation had no activity toward yeast phosphoinositol sphingolipids. This enzyme was dependent on a divalent cation, as was the sphingolipase activity we describe, but was inhibited by octyl glucoside and other detergents unlike our enzyme activity. It should be noted that sphingomyelin has not yet been reported to occur in S. cerevisiae. The sphingomyelinase of Ella et al. (13) is not likely to be a phosphoinositol sphingolipid hydrolase.
Several groups (33)(34)(35)(36) have reported on a putative phospholipase C gene (PLC1) in S. cerevisiae, which upon deletion, results reportedly in differing phenotypes such as lethality or very slow, temperature-sensitive growth, etc. Plc1p was purified as a soluble enzyme after overexpression and was shown to catalyze the following reactions (36). We tested a sample of this enzyme (generously supplied by Dr. Jeremy Thorner) and found it to be without effect on yeast phosphoinositol sphingolipids when assayed (data not shown) as described (36). Furthermore, we assayed hydrolysis of [ 3 H]IPC and [ 3 H]M(IP) 2 C with membranes prepared from strain YJF132 (36) carrying a plc1 deletion as well as from the cognate wild type strain YJF131 (36). The plc1-deleted strain had about 50% of the sphingolipid hydrolase wild type activities. Whether the lowered activity is related to the very slow growth of the mutant strain and/or some indirect effect of the lack of plc1p on the regulation of sphingolipid phospholipase activity is unclear. Nonetheless, it is clear that substantial sphingolipid hydrolase activity remains in the plc1 deletion strain, and thus plc1 is unlikely to code for sphingolipid hydrolase activity. We previously examined the turnover of [ 3 H]inositol labeled lipids with uniformly labeled S. cerevisiae cells transferred to nonradioactive growth medium. Although a large decrease in the phosphatidylinositol pool could be readily observed, consonant with its conversion to sphingolipids and extracellular glycerophosphoinositol, no decrease in the total sphingolipid pool consistent with sphingolipid turnover was observed after many hours except for that interpreted as the conversion of IPC plus MIPC to M(IP) 2 C (37). If heat-induced ceramide elevation were due to sphingolipid breakdown, then only quite small sphingolipid changes could be expected. To obtain in vivo evidence of a small sphingolipid turnover (about 6% of the total sphingolipid) that might be associated with the heat-induced ceramide increase, the approach taken was to look for the appearance of [ 3 H]inositol-labeled, water-soluble sphingolipid head groups that could result from the action of a phospholipase C on the various yeast sphingolipids. No more than 3% of the expected products could be found (Table V), making it unlikely that sphingolipid hydrolysis was generating ceramide. One cannot exclude an extremely rapid catabolism of sphingolipid head groups masking any accumulation. In conclusion, the absence of sphingolipid head group accumulation and the observed australifungin-sensitive nature of ceramide elevation make de novo ceramide synthesis the most likely mechanism for heat-induced ceramide elevation.