Cloning and Characterization of a Saccharomyces cerevisiae Alkaline Ceramidase with Specificity for Dihydroceramide*

In a previous study, we reported that theSaccharomyces cerevisiae gene YPC1 encodes an alkaline ceramidase with a dual activity, catalyzing both hydrolysis and synthesis of yeast ceramide (Mao, C., Xu, R., Bielawska, A., and Obeid, L. M. (2000) J. Biol. Chem. 275, 6876–6884). In this study, we have identified a YPC1homologue in S. cerevisiae that also encodes an alkaline ceramidase. We show that these two ceramidases have different substrate specificity, such that YPC1p preferentially hydrolyzes phytoceramide, whereas the new ceramidase YDC1p hydrolyzes dihydroceramide preferentially and phytoceramide only slightly. Neither enzyme hydrolyzes unsaturated mammalian-type ceramide. In contrast to YPC1p, YDC1p had only minor in vitro reverse activity of catalyzing dihydroceramide formation from a free fatty acid and dihydrosphingosine and no activity with phytosphingosine. Overexpression of YDC1p had no reverse activity in non-stressed yeast cells, but like YPC1p suppressed the inhibition of growth by fumonisin B1 albeit more modestly. Deletion of YDC1 andYPC1 or both did not apparently affect growth, suggesting neither gene is essential. However, the Δydc1 deletion mutant but not the Δypc1 deletion mutant was sensitive to heat stress, indicating a role for dihydroceramide but not phytoceramide in heat stress responses, and suggesting that the two enzymes have distinct physiological functions.

Ceramide is a central molecule in the pathway of sphingolipid metabolism in mammalian cells (1,2). It is converted through the action of a desaturase (3) from dihydroceramide that is synthesized de novo from dihydrosphingosine and a fatty acyl-CoA 1 by (a CoA-dependent) ceramide synthase. Cer-amide can also be generated from sphingomyelin by sphingomyelinase or be glycosylated to yield more complex glycosphingolipids. Ceramide is also broken down by ceramidases to generate sphingosine and a fatty acid.
Metabolism of yeast ceramides appears to be similar to that of mammals (4). Phytoceramide, an equivalent of the mammalian ceramide, serves as a building block of yeast complex sphingolipids. It is synthesized de novo from a fatty acyl-CoA and phytosphingosine by (a CoA-dependent) ceramide synthase. Phytosphingosine is generated from dihydrosphingosine through hydroxylation by the action of dihydrosphingosine hydroxylase (5,6). Dihydrosphingosine is also acylated by a fatty acyl-CoA to generate dihydroceramide through the action of ceramide synthase. Phytoceramide (or dihydroceramide) accepts an inositol-phosphoryl group to yield inositol phosphorylceramide (IPC).
Ceramide, as the building block of complex sphingolipids in eukaryotic cells, not only is structurally essential for cell growth but also is important in modulating different cellular events including apoptosis, growth arrest, and stress responses (see reviews Refs. 2 and 7-10). Importantly, its breakdown product sphingosine and the subsequent metabolite sphingosine-1-P are also signaling molecules (11). As a signaling molecule or a donor of signaling molecules, ceramide must be tightly regulated in order to carry out multiple functions in cells. Indeed, levels of ceramide in cells change in response to different physiological environments or to different stimuli including growth factors, cytokines (12), heat (4,13), and proapoptosis agents (14). Changes in ceramide levels in cells involve different enzymes. Among these enzymes ceramidases are critical in controlling levels of ceramide in mammalian cells.
Several mammalian ceramidases have been purified, characterized biochemically, and subsequently cloned (15)(16)(17). They are classified as acid, neutral, and alkaline ceramidases according to their pH optimum. The acid ceramidase is localized to the lysosomes, and it is believed to be responsible mainly for housekeeping catabolism of membrane ceramide. Other membrane-bound ceramidases, which are considered as neutral-or alkaline-type enzymes, are believed to be in involved in signaling processes. For example, it was shown that mammalian alkaline ceramidase is activated by platelet-derived growth factor (12) and that both alkaline and neutral ceramidases are activated by interleukin 1␤ at a low concentration through tyrosine phosphorylation (18).
We recently cloned an alkaline ceramidase YPC1p from Saccharomyces cerevisiae (19). YPC1p preferentially deacylated phytoceramide to yield a free fatty acid and phytosphingosine. It also slightly deacylated dihydroceramide to generate a free fatty acid and dihydrosphingosine, but it did not act on unsaturated ceramide. Importantly, this alkaline ceramidase had a reverse activity of catalyzing formation of phytoceramide from a free fatty acid and phytosphingosine in vitro and in cells. Formation of phytoceramide by the reverse activity of YPC1p appears to be an important alternative pathway for the synthesis of phytoceramide when the CoA-dependent ceramide synthase is inhibited in S. cerevisiae. Identification of YPC1p suggests that breakdown of ceramides is conserved between yeast and mammals.
Another alkaline ceramidase was recently cloned from Pseudomonas and mycobacterium (20). This enzyme also had reverse activity but had no sequence homology to the yeast enzyme YPC1p, indicating they belong to different classes of ceramidases. In fact there was no homologous yeast sequence to the Pseudomonas ceramidase. Interestingly the Pseudomonas alkaline ceramidase shared a similarity in protein sequence and many biochemical properties with the mouse neutral ceramidase, which was recently purified and cloned from mouse liver (16,21) and a human mitochondrial ceramidase cloned from human kidney (22).
On the other hand, a search of protein data bases for homology to YPC1p identified that another yeast putative protein encoded by the open reading frame YPL087w has 53% identity to YPC1p. In this report, we present evidence that this protein (YDC1p) is another membrane-bound alkaline ceramidase, with specific activity toward dihydroceramide. We demonstrate that this protein also has reverse activity albeit significantly less than that of YPC1p. We also show that the deletion mutant of YDC1, but not YPC1, endows sensitivity to heat stress. These data suggest that the breakdown of phytoceramide and dihydroceramide in S. cerevisiae is catalyzed by different enzymes that have different physiologic functions.

EXPERIMENTAL PROCEDURES
Yeast and Bacterial Strains-Yeast strains used in this study are listed in Table I and cultured and maintained as described (19). Epicurean coli XL-1 and SURE strains were used for most gene manipulations. E. coli Top10 strain was used for expression of yeast proteins under control of P bad promoter in the pBAD/His vector as described (19).
Construction of Mutants Deficient in YDC1 or Both YDC1 and YPC1-A copy of the wild type YDC1 gene in the diploid strain JK9-3d a/␣ was replaced by a disruption module consisting of the Kluyveromyces lactis URA3 gene (Research Genetics, Inc.) flanked by the 5Ј end (nucleotide 16 -46) and 3Ј end (nucleotide 892-926) portions of the YDC1 coding region. Briefly, the disruption module amplified by PCR was transformed into diploid cells by a lithium acetate method. The diploid cells containing the ydc1 disruption module (deletion allele) were sporulated, and the resulting tetrads were dissected as described (19). All 10 tetrads dissected gave 4 viable spores, indicating that the YDC1 gene is not essential for viability of the yeast. The haploid strain (⌬ydc1) harboring the ydc1 deletion allele was selected by prototrophs for uracil, and deletion of YDC1 was verified by PCR using a forward primer (5Ј CAAGAATTTAAGCAAAAGGATATATCATAGAACCTA-GTG 3Ј) upstream of the YDC1 coding region and the reverse primer (the K. lactis internal 3Ј primer (Research Genetics, Inc.)) located in the middle of the URA3 gene. By using these two primers, the PCR would yield a product if the disruption module is correctly integrated into the chromosome and replaces the YDC1 gene. Diploid strain harboring both ypc1 and ydc1 deletion alleles was created by crossing the two haploid strains ⌬ydc1 and ⌬ypc1. Sporulation and tetrad dissection of the diploid strain yielded 4 normal spores, indicating that double deletions of both YPC1 and YDC1 are viable. The haploid strain harboring deletion of both ydc1 and ypc1 was selected by geneticin resistance followed by prototrophs for uracil. Since the URA3 gene interferes with heat stress study, the wild type URA3 gene which replaced the wild type YDC1 allele in both ⌬ydc1 and ⌬ypc1⌬ydc1 strains was replaced by a truncated URA3 mutant that lacks the BbrP1 (restriction enzyme) fragment. Replacement of the wild type URA3 gene was counter-selected by 5-fluoroorotic acid and was verified by PCR.

Plasmid Construction for Protein Expression in Yeast-
The open reading frame of the gene YDC1 was amplified by PCR (1 cycle of 94°C for 2 min; 30 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 1 min; 1 cycle of 72°C for 10 min) using the precision plus Taq polymerase (Stratagene, Inc.) and the primers 5Ј CGGGGTACCATGCTGTTCAG-CTGGCCTTATCCAG 3Ј (forward) and 5Ј CGGGAATTCTTAGTTATT-CTTTTTTGTTTCATCATCTACC 3Ј (reverse). These primers contain the restriction sites (underlined) KpnI and EcoRI, respectively, for the cloning purposes. The amplified product was digested by restriction enzymes KpnI and EcoRI and cloned into the KpnI and EcoRI sites of the vector pYES2 to yield the construct pYES2-YPC1, thus expressing YDC1p under control of the promoter Gal1. To facilitate detecting and purifying YDC1p, the coding region of the gene YDC1 was tagged with the sequence (underlined) encoding the FLAG epitope peptide by PCR using the forward primer 5Ј CGGGGTACCATGGACTACAAGGACGA-CGATGATAAGCTGTTCAGCTGGCCTTATCCAG 3Ј and the reverse primer 5Ј CGGGAATTCTTAGTTATTCTTTTTTGTTTCATCATCTACC 3Ј as described (19). The PCR product was digested with KpnI and EcoRI and cloned into the vector pYES2 as described above. The resulting construct pYES2/YDC1-FLAG that expresses the FLAG-tagged protein YDC1p under the control of the promoter Gal1, the construct pYES2-YPC1, and the empty vector pYES2 were introduced into the strain ⌬yor1 by the lithium acetate method after correctness of these constructs was ensured by sequencing. The strain containing pYES2, pYES2-YPC1, or pYES2/YDC1-FLAG was grown and maintained in SC-ura medium with 2% glucose. Expression of YDC1p or the tagged YDC1p (YDC1p-FLAG) was induced in SC-ura medium with 2% galactose, and YDC1p-FLAG was detected by Western analysis using a monoclonal antibody against FLAG peptide as described. Microsomes were prepared from the cells expressing YDC1p-FLAG as described (19). Proteins were solubilized from the microsomes with 0.25% Triton X-100 in the lysis buffer (25 mM Tris-HCl, pH 7.4, containing 5 mM CaCl 2 , 150 mM NaCl, and 20 g/ml CLAP), and YDC1p-FLAG was purified by an anti-FLAG affinity column (Sigma) as recommended by the manufacturer.
Cellular Localization Using GFP Tagging-The YPC1 and YDC1 coding sequences without the stop codon were amplified from the yeast genomic DNA and cloned into the GFPuv-containing vector pYES2-GFPuv in frame with the GFPuv coding sequence. The constructs pYES2-YPC1-GFPuv and pYES2-YDC1-GFPuv that express the YPC1-GFPuv and YDC1-GFPuv fusion proteins, respectively, were introduced into the yeast strain JK9-3d. Expression of the YPC1-GFPuv and YDC1-GFPuv fusion proteins was induced in the SC-ura medium containing 2% galactose for 6 -12 h and was detected by Western blot analysis using the anti-GFP antibody. The fluorescence of the yeast cells was examined under a Zeiss fluorescent microscope and recorded by a CCD camera.
Expression of YDC1p in E. coli-The coding sequence of the gene YDC1 was excised by restriction enzymes KpnI and EcoRI from the plasmid pYES2-YDC1 and cloned into KpnI and EcoRI sites of the vector pBAD/His B (Invitrogen) that was digested by the same enzymes to create pBAD/His-YDC1, thus expressing the polyhistidine (His) and Xpress tagged YDC1p under control of the araBAD promoter (P BAD ). pBAD/His and pBAD/His-YDC1 were introduced into the E. coli strain TOP10 by electroporation. The tagged YDC1p was expressed and purified as described (23).
Preparation of Radiolabeled Ceramides-[ 3 H]Ceramide and phytoceramide were synthesized as follows: N-[9,10-3 H]D-erythro-C 16 -ceramide and N-[9,10-3 H]D-ribo-C 16 -phytoceramide were prepared by acylation of the respective sphingoid bases with [9,10-3 H]palmitoyl chloride generated in situ from [9,10-3 H]palmitic acid as described (24).  (19). To measure ceramidase activity, 20 l (approximately 150 g of proteins) of microsomes or a purified protein was added to substrates, and reactions were incubated at 30°C for 40 -90 min. Reactions were stopped by adding 300 l of methanol: chloroform (2:1) and dried under a SpeedVac. Lipids were resolved by TLC; the reaction product palmitic acid or dihydrosphingosine was identified, scraped, and measured by a scintillation counter. Both the protein concentration and the time of incubation were within the linear range for the assay. One unit of ceramidase was defined as the amount of the enzyme needed to release 1 pmol of palmitic acid or dihydrosphingosine per min.
The reverse activity of ceramidase was measured using [ 3 H]palmitic acid (0.3 nmol) and phytosphingosine (or dihydrosphingosine) (5 nmol) as substrates as described (19). Microsomes were added to the above substrates and incubated at 30°C for 2 h. Both protein concentration and time of incubation were within the linear range for the assay. Phospholipid determination on microsomes showed that different microsomal preparations with equal amount of proteins contained equal amount of lipids. One unit of the reverse activity was defined as the amount of the enzyme needed to form 1 pmol of phytoceramide (or dihydroceramide) per min.
Protein Analysis-Proteins were separated by SDS-PAGE and were detected by Coomassie staining or Western blotting analysis by following standard procedures.
Heat Tolerance Study-Exponentially growing cells of yeast strains were incubated at 50°C for 40 min. The heat-treated cells and those set at room temperature were plated onto YPD plates and incubated at 30°C for 2-3 days. The plates were photographed and colony-forming units (CFU) were determined. The post-heat stress viability was defined as a percentage of post-heat stress CFU to those prior to heat stress.

Identification of Another Yeast
Ceramidase-To find potential homologues for the YPC1 gene, we searched the Saccharomyces Genome Database. We identified a putative protein, encoded by the open reading frame YPL087w, that had 52% identity to the yeast alkaline ceramidase YPC1p over the entire protein sequence. This protein had 317 amino acids, with a predicted pI of 6.77. Several highly conserved regions were found between the two proteins (Fig. 1A). Similar to YPC1p, the homologue was a very hydrophobic protein and shared a similar hydropathy profile with YPC1p (Fig. 1B). Both proteins were predicted to have several transmembrane domains (Fig.  1C), suggesting that this homologue is also an integral membrane protein. Both proteins had an ER retention sequence (KKXX, X represents any amino acid residue) at their carboxyl termini, suggesting that they may be localized to the ER. Based on similarities of protein sequence and hydropathy profiles, this protein was predicted to be another ceramidase.
To investigate whether the homologous gene encodes for a ceramidase activity, we cloned its coding sequence into the vector pYES2 under the control of the Gal1 inducible promoter. Sequencing confirmed that the coding region was identical to that reported in the yeast genomic data base. The vector pYES2 and the new construct were transformed into the yeast strain ⌬yor1, and gene expression was induced by galactose. Microsomes were prepared and assayed for ceramidase activity using different ceramides as substrates. Interestingly overexpression of the homologous protein caused a substantial increase in ceramidase activity toward C 6 -dihydroceramide ( Fig. 2A), but only a slight increase in the activity toward phytoceramide The conserved regions between the two proteins are shaded and boxed. Phosphorylation sites of cAMP-dependent kinase ( or ), protein kinase C (q), and tyrosine kinase (Ⅺ) were predicted by the PROSITE program. ER retention sequences (*) at carboxyl termini are predicted by the PSORT II program. B, protein hydropathy profile is plotted according to the Kyte/Doolittle method. C, transmembrane domains are predicted by the Agros method. (Fig. 2B), and no activity toward unsaturated ceramide (Fig.  2C). In contrast to the homologue, microsomes from cells overexpressing YPC1p hydrolyzed phytoceramide preferentially over dihydroceramide (Fig. 2, A and B). These data suggest that this homologue also encodes for a ceramidase activity; however, it had different substrate specificity. We thus renamed this gene YDC1 as yeast dihydroceramidase.
Tagged and Purified YDC1 Encodes a Ceramidase Activity-We demonstrated above that YDC1p encodes a ceramidase activity mainly hydrolyzing dihydroceramide. However, we have not ruled out the possibility that YDC1p is a regulator of ceramidase activity. To verify that YDC1p is itself a ceramidase, we went on to express the YDC1p in E. coli. To facilitate detection and purification, we tagged the YDC1p with a polyhistidine (His) tag. The His-tagged YDC1p was expressed in E. coli as analyzed by Western blot (data not shown). The Histagged YDC1p did not have ceramidase activity, probably due to a post-translational modification required for the activity that E. coli lacks. Therefore, we elected to express the YDC1p in yeast cells. To facilitate purifying the YDC1p, we tagged it with an epitope tag FLAG. Expression of the tagged YDC1p was induced by galactose after the expressing construct pYES2-YDC1-FLAG was introduced into yeast cells. Microsomes were prepared from the cells expressing YDC1-FLAG or containing the empty vector pYES2-FLAG. Proteins were extracted from microsomes with 0.25% Triton X-100 and applied to an anti-FLAG affinity column. After washing, the FLAGtagged protein was eluted by a buffer containing FLAG peptide (10 g/ml). The eluates were resolved by SDS-PAGE, and proteins were detected by Coomassie staining (Fig. 3A) and Western blot analysis (Fig. 3B) using the anti-FLAG antibody. A protein band with an apparent molecular mass of 37 kDa and a cluster of protein bands with apparent molecular mass ranging from 75 to 200 kDa were revealed by Coomassie staining as well as by Western blotting in the YDC1-FLAG eluate but not in the vector control eluate. The eluates were assayed for ceramidase activity. Fig. 3C shows that the purified FLAG-tagged YDC1p, but not the vector control eluate, had ceramidase activity toward C 6 -dihydroceramide. The high molecular weight protein cluster could be aggregated YDC1-FLAG since it is a very hydrophobic protein. Alternatively, it could be YDC1-FLAG associated complexes. However, since YDC1p is highly homologous to the alkaline ceramidase YPC1p, and purified YDC1p had ceramidase activity toward C 6 -dihydroceramide, it is most likely that the YDC1 gene encodes a ceramidase and less likely that it is a regulator of enzymatic activity.
YDC1p Is Also an Alkaline Ceramidase-In our previous study, we demonstrated that YPC1p is an alkaline ceramidase, with an optimal pH of 9.5 (19). To study whether YDC1p has the same or a different pH optimum, we measured the ceramidase activity of microsomal preparations from cells overexpressing YDC1p using C 6 -dihydroceramide as a substrate at different pH. Fig. 4 shows that YDC1p has very low activity at acidic pH, moderate activity at neutral pH, and the highest activity at alkaline pH. These data suggest that similar to YPC1p, YDCp also belongs to the alkaline ceramidase family.
YDC1p Functions as a Ceramidase in Cells-In vitro YDC1p showed ceramidase activity. We next wanted to know whether in cells YDC1p has the same ceramidase activity. Ceramidase activity in cells was evaluated using [ 3 H]C 6 -dihydroceramide labeled at the C-4 and C-5 positions of the dihydrosphingosine moiety. Similar to YPC1p, overexpression of YDC1p enhanced breakdown of [ 3 H]C 6 -dihydroceramide, thus leading to accumulation of dihydrosphingosine (DHS), phytosphingosine (PHS), DHS-1-P, PHS-1-P, and the glycerolipids phosphatidylethanolamine, phosphatidylcholine, and phosphatidylinositol (Fig. 5), suggesting that YDC1p has endogenous ceramidase activity in cells.
YDC1p Has the Reverse Activity of Synthesizing Dihydroceramide from a Fatty Acid and Dihydrosphingosine-To investigate if the YDC1p has this reverse activity similar to its homologue YPC1p, microsomes prepared from cells containing the empty vector (pYES2) or overexpressing YDC1p were assayed for the reverse activity. Microsomes were incubated with [ 3 H]palmitic acid and phytosphingosine or palmitic acid and dihydrosphingosine at 30°C for 2 h. The product, phytocera-FIG. 3. The purified YDC1p expressed in yeast has ceramidase activity. Proteins were extracted from microsomes prepared from the cells with an empty vector or containing the FLAG tagged YDC1p and were purified using an anti-FLAG affinity column as described under "Experimental Procedures." Eluates were resolved by SDS-PAGE, and proteins were detected by Coomassie staining (A) or by Western blotting using anti-FLAG antibody (B). The eluates were also assayed for ceramidase activity as described under "Experimental Procedures"(C). C6dh-cer, C 6 -dihydroceramide. mide or dihydroceramide, was analyzed by TLC and quantitated by a scintillation counter. Fig. 6A shows that microsomes from YDC1p-overexpressing cells had no increase in the reverse activity compared with the vector control cells when phytosphingosine and palmitic acid were used as substrates. In contrast, the YPC1-overexpressing cells had 30-fold higher activity than vector control cells. When dihydrosphingosine and palmitic acid were used as substrates microsomes from YDC1 cells had 3-fold higher reverse activity than those from vector control cells (Fig. 6B), whereas YPC1 cells had 30 times higher activity than control cells. These results suggest that in yeast cells, YPC1p has a major reverse activity of ceramidase and can use both DHS and PHS as substrates, whereas YDC1p has only a minor reverse activity and can only use DHS as substrate.
Overexpression of YDC1p Rescues from Fumonisin B1 Inhibition-Overexpression of YPC1p endowed resistance to fumonisin B1 because of its reverse activity to synthesize ceramide. We demonstrated above that YDC1p had the reverse activity in vitro when dihydrosphingosine and fatty acid were used as substrates. We wondered whether in the presence of fumonisin B1, overexpression of YDC1p could have the reverse activity in cells. Cells containing pYES2-YDC1, the vector, or pYES2-YPC1 (as a positive control) were grown on SC-ura plates containing 450 M fumonisin B1. Fig. 7A shows that YDC1p also endowed resistance to fumonisin B1 albeit somewhat less than YPC1p. This result shows that YDC1p could have the reverse activity of ceramidase in cells under stress conditions but was not as effective as YPC1p.
Fumonisin B1 induced cytotoxicity in yeast could be due to blockage of synthesis of ceramides and complex sphingolipids, or due to accumulation of long chain bases and their phosphates, or both. We have previously shown that a large proportion of exogenous long chain bases were phosphorylated by long chain base kinases to yield long chain base phosphates that are toxic to yeast cells (26). YPC1p or YDC1p could endow resistance to fumonisin B1 in part by using accumulated long chain bases to synthesize the yeast ceramides and attenuate the cytotoxicity of the long chain base phosphates. To test this, we evaluated if overexpression of either ceramidase could endow resistance to the long chain bases phytosphingosine and dihydrosphingosine by diverting them to phytoceramide and dihydroceramide using the reverse activity. The strains containing the vector pYES2, pYES2-YPC1, and pYES2-YDC1 were spotted onto SC-ura plates containing 25 M phytosphingosine and 2% galactose, and growth inhibition of the different strains was examined by the serial cell dilution method described above. Fig. 7B shows that the YPC1 strain is more resistant to phytosphingosine than the vector strain, whereas the YDC1 strain is only slightly more resistant to phytosphingosine, suggesting that the YPC1 strain, and to a much lesser extent, the YDC1 strain using its reversal action of ceramidase, converts phytosphingosine to phytoceramide in cells when phytosphingosine is in excess. This result is consistent with the in vitro study, such that YPC1p, but not YDC1p, can use phytosphingosine as substrate to synthesize yeast ceramide. On the other hand, neither the vector strain nor the YPC1 or YDC1 strains were sensitive to DHS (up to 70 M). This is compatible with our other data 2 indicating that PHS but not DHS mediates yeast growth arrest.
Deletion of Both YDC1 and YPC1 Eliminates All Basal Ceramidase Activity toward Phytoceramide and Dihydroceramide-To study the physiologic functions of YPC1p and YDC1p, we made deletion mutants of YPC1, YDC1, or both. First, we examined how deletion of these two genes affects metabolism of sphingolipids. We labeled the mutants and their parental strain with [ 3 H]palmitic acid and analyzed total sphingolipids by TLC. Fig. 8 shows that deletion of either YPC1 or YDC1 caused an increase in IPC, MIPC, and M(IP) 2 C and a decrease in DHS-1-P and PHS-1-P compatible with their function as ceramidases. Deletion of both YPC1 and YDC1 had an additive effect on metabolism of sphingolipids. We speculated that complex sphingolipids increase in the deletion mutants because ceramidase activity in these mutants is significantly decreased or totally abolished. Therefore, we measured ceramidase activity of these mutant strains. Microsomes prepared from these strains and assayed for ceramidase activity (Table II) showed that deletion of YPC1 eliminated most of the basal ceramidase activity for phytoceramide but slightly reduced the activity for dihydroceramide. In contrast, deletion of YDC1 eliminated most of the activity toward dihydroceramide but only slightly reduced the activity toward phytoceramide. However, deletion of both YPC1 and YDC1 completely removed the activity toward both phytoceramide and dihydroceramide. These data suggest that YPC1p and YDC1p indeed have their respective substrate specificity in cells and are the only enzymes responsible for catabolism of the yeast ceramides (dihydroceramide and phytoceramide).
Deletion of Both YDC1 and YPC1 Eliminates Most of the Reverse Activity of Ceramidase-We also measured reverse activity of ceramidase in the deletion mutants. Table III shows that deletion of YPC1 eliminated most of the reverse activity when either phytosphingosine or dihydrosphingosine was used as substrates. Deletion of YDC1 showed no change in the activity when phytosphingosine and palmitic acid were used as substrates but showed slight reduction in the activity when dihydrosphingosine and palmitic acid were used as substrates. Deletion of both YPC1 and YDC1 knocked out most of the activity when either phytosphingosine or dihydrosphingosine were used as substrates along with palmitic acid. These results suggest that the reversal action of ceramidase in yeast cells is mainly carried out by YPC1p and to a much lesser extent by YDC1p.
Both YPC1p and YDC1p Are Localized to ER-Both YPC1p and YDC1p have an ER retention sequence, suggesting they may be localized to the ER. We examined their localization by tagging YPC1p and YDC1p with a green fluorescent protein (GFPuv). Western blotting analysis using an anti-GFP antibody was performed on extracts from cells expressing GFPuv, GFPuv-YPC1p, or GFPuv-YDC1p fusion proteins as described under "Experimental Procedures." Fig. 9A shows that free GF-Puv was detected only in the 100,000 ϫ g supernatant, whereas both GFPuv-YPC1p and GFPuv-YDC1p were detected only in the 100,000-g pellet, suggesting that both YPC1p and YDC1p are membrane proteins as predicted, and their localization in cells was not affected by GFPuv tagging. The fluorescent pattern of exponentially growing cells expressing the GFPuv-YPC1p or YDC1p fusion was recorded by a CCD camera. Fig.  9B shows that GFPuv-tagged YPC1p and YDC1p have a typical ER pattern shared by another lipid enzyme DHS-1-P phosphatase YSR2, suggesting they are indeed localized to the ER where most enzymes of sphingolipid metabolism are localized.
Deletion of YDC1, but Not YPC1, Causes a Decrease in Tolerance to Heat Stress-Yeast cells deficient in YPC1p, YDC1p, or both are viable and have normal growth rates in both rich medium and defined medium under permissive temperatures, suggesting that neither YPC1p nor YDC1p is essential. The yeast ceramides and other sphingolipids have been implicated in the response to heat stress, whereby their levels are elevated in heat-stressed cells (13,27). Increased ceramides were suggested to come from the de novo pathway (27). We previously showed that elevation of dihydroceramide or phytoceramide imparted on yeast cells a sensitivity to heat stress, whereas elevation of dihydrosphingosine-1-P has been implicated in heat resistance (26). Deletion of either YPC1 or YDC1 caused an increase in synthesis of complex sphingolipids due to decreased breakdown of ceramides. To study whether the pathway of breakdown of ceramide is involved in the process of heat stress, we investigated heat stress responses in the mutants deficient in YPC1p, YDC1p, or both. Cells were incubated at 50°C for 40 min, and post-heat viability was determined by CFU on YPD plates. Fig. 10 shows that the post-heat stress viability of both the ⌬ydc1 and ⌬ypc1⌬ydc1 mutant strains was lower than that of either the ⌬ypc1 mutant or the wild type strains. The ⌬ypc1 strain that lacks the ability to break down phytoceramide had the same response as the wild type JK9-3d to heat. On the other hand, the ⌬ydc1 that lacks the ability to break down dihydroceramide was highly sensitive to heat stress. The ⌬ypc1⌬ydc1 strain, like ⌬ydc1, was also more sensitive to heat stress. These results suggest that accumulation of dihydroceramide but not phytoceramide may be responsible for mediating sensitivity to heat stress. Alternatively, loss of breakdown product in the ⌬ydc1 strain (dihydrosphingosine), but not in the ⌬ypc1 strain (phytosphingosine), could be responsible for mediating this heat sensitivity. DISCUSSION In this study, we identified and cloned a homologue to the S. cerevisiae phytoceramidase (YPC1p). We demonstrate that this homologue is also an alkaline ceramidase that hydrolyzes dihydroceramide specifically, and we named it yeast dihydroceramidase (YDC1p). YDC1p shows reverse activity of ceramidase in vitro; however, unlike YPC1p, YDC1p has a minor reverse activity and only with dihydrosphingosine as substrate. In cells, YDC1p acts as a ceramidase under normal culture conditions, but the ceramidase action can be reversed in the presence of fumonisin B1. Both YPC1p and YDC1p have an ER retention sequence at their carboxyl termini, and we verify their localization to the ER by GFP tagging. We also demonstrate that deletion of YDC1 but not YPC1 renders cells sensitive to heat stress, suggesting that dihydroceramide (or dihydrosphingosine), but not phytoceramide (phytosphingosine), in yeast may have a distinct role in response to heat stress.
Several ceramidases have been cloned from mammalian cells. These include an acid lysosomal ceramidase (15), a mouse liver neutral ceramidase (16), and its homologue a human kidney mitochondrial ceramidase (17). These latter two enzymes are homologous to the Pseudomonas alkaline ceramidase (20). We have also recently identified and cloned a human homologue of the yeast alkaline ceramidases. Comparison of protein sequence revealed that our yeast ceramidases and their human homologue are completely distinct from the acid lysosomal ceramidase as well as the mouse neutral ceramidase and the human mitochondrial ceramidase, which in turn are also distinct from the acid ceramidase. Therefore, based on protein sequence, it is not yet possible to predict a substrate-binding site.
As far as substrate specificity, the yeast ceramidases prefer yeast-saturated ceramides as their substrates, whereas all the other neutral and alkaline ceramidases described to date prefer unsaturated ceramide (21,22). The bacterial ceramidase also uses mammalian type unsaturated ceramide as substrate; therefore, the enzyme may have a role in bacterial invasiveness. In fact such a role was raised in the study of Ohnishi et al. (28) where the Pseudomonas ceramidase was implicated in hydrolyzing skin ceramides in atopic dermatitis. Therefore, whether these different ceramidases have distinct physiologic roles needs to be further studied.
Protein motif prediction using the PROSITE program revealed that both YPC1p and YDC1p have putative cAMP protein kinase tk;1phosphorylation sites and protein kinase C phosphorylation sites, and YDC1p has a tyrosine kinase phosphorylation site (Fig.  1A). It has been shown that activity of mammalian neutral and alkaline ceramidases is regulated by protein phosphorylation. Whether the yeast ceramidases are phosphorylated and the phosphorylation modifies their enzymatic activity awaits further study.
It is believed that ceramides are synthesized in the ER. Localizations of YPC1p and YDC1p to the ER suggest that YPC1p and YDC1p have immediate access to the yeast ceramide as soon as it is formed. The benefit of the rapid access of these enzymes to the ceramides could be to regulate turnover of ceramides most efficiently because ceramide levels are crucial for the well being of yeast cells. YDC1p and YPC1p are the only ceramidases in yeast cells, so the ER should be an important pool for generating dihydrosphingosine and phytosphingosine, the products of these enzymes. We previously showed that dihydrosphingosine-1-P phosphatases (YSR2 and YSR3) are also localized to the ER. These data suggest that the ER is the center of both synthesis and breakdown of yeast sphingolipids.
In mammalian cells, it is believed that sphingosine is derived from deacylation of ceramide by ceramidase and not from de novo biosynthesis. Its subsequent metabolite sphingosine-1-P is also one of the breakdown products of ceramide. Sphingosine and sphingosine-1-P have been implicated in opposite cellular actions, such that sphingosine suppresses cell growth and sphingosine-1-P promotes cell proliferation. In yeast, sphingo- FIG. 9. Both YPC1p and YDC1p are localized to ER. YDC1p and YPC1p were tagged with a fluorescent GFPuv as described under "Experimental Procedures." Expression of the tagged proteins was induced in SC-ura medium containing 2% galactose. Cells were disrupted by glass bead collision, and whole cell lysate was fractionated into cytosolic and membrane fractions after removal of nuclei. Proteins from the two fractions were separated by SDS-PAGE, GFPuv, and the GFPuv fused YPC1p and YDC1p were detected by Western blot analysis using anti-GFP antibody (A). Fluorescence of cells was examined under a fluorescent microscopy and recorded by a digital camera (B).

TABLE II
Ceramidase activities in the wild type, ⌬ypc1, ⌬ydc1, and ⌬ypc1 ⌬ydc1 strains Microsomes were prepared from each strain and were assayed for ceramidase activity towards phytoceramide and dihydroceramide as described under "Experimental Procedures." Data are the mean of one experiment performed in duplicate and a representative of at least three independent experiments.  Table II were assayed for reverse activity of ceramidase using palmitic acid and dihyrosphingosine or phytosphingosine as substrates as described under "Experimental Procedures." Data are the mean of one experiment performed in duplicate and a representative of at least three independent experiments. sine has not been detected, and both dihydrosphingosine and phytosphingosine are synthesized de novo. In our previous (19) and current studies, we clearly demonstrate that both dihydrosphingosine and phytosphingosine are also the breakdown products of dihydroceramide and phytoceramide, respectively, and these breakdown products can be phosphorylated by kinases to form long chain base phosphates. The long chain base phosphates are decreased in cells lacking either YPC1p or YDC1p, indicating that these long chain base phosphates are in part from breakdown of sphingolipids. Deletion of YPC1, YDC1, or both does not affect growth under normal conditions, suggesting yeast ceramidases may not be mediators of cell growth and proliferation. However, deletion of YDC1, but not YPC1, sensitizes cells to heat stress. YDC1p and YPC1p use dihydroceramide and phytoceramide, respectively, as substrate in cells. These results suggest that dihydroceramide but not phytoceramide is responsible for the modulation of the heat stress response. In a previous study we demonstrated that overexpression of dihydrosphingosine-1-P phosphatase (YSR2) also sensitizes cells to heat stress (26). Deletion of YDC1 is similar to overexpression of YSR2 in that both have the same effect on metabolism of sphingolipids as follows: causing an increase in dihydroceramide and a decrease in dihydrosphingosine-1-P. Thus, we conclude that changes in dihydroceramide, dihydrosphingosine-1-P, or both are responsible for the heat stress sensitivity. FIG. 10. Deletion of YDC1 confers a sensitivity to heat. Exponentially growing cells from different strains were diluted to the same density (4 ϫ 10 5 ) using YPD medium, and one portion of cells was heat-treated at 50°C for 40 min, and the other portion was set at room temperature. The heat-treated cells were plated onto YPD plates after being cooled to room temperature and were incubated at 30°C for 2 days. The non-stressed cells were diluted 100 times and plated onto YPD plates. CFUs were counted. The post-heat shock viability of the strains was defined as the percentage of post-heat shock CFUs over prior heat shock CFUs.