Sphingolipids Signal Heat Stress-induced Ubiquitin-dependent Proteolysis*

Sphingolipids are essential eukaryotic membrane lipids that are structurally and metabolically conserved through evolution. Sphingolipids have also been proposed to regulate eukaryotic stress responses as novel second messengers. Here we show that, inSaccharomyces cerevisiae, phytosphingosine, a putative sphingolipid second messenger, mediates heat stress signaling and activates ubiquitin-dependent proteolysis via the endocytosis vacuolar degradation and 26 S proteasome pathways. Inactivation of serine palmitoyltransferase, a key enzyme in generating endogenous phytosphingosine, prevents proteolysis during heat stress. Addition of phytosphingosine bypasses the requirement for serine palmitoyltransferase and restores proteolysis. Phytosphingosine-induced proteolysis requires multiubiquitin chain formation through the stress-responsive lysine 63 residue of ubiquitin. We propose that heat stress increases phytosphingosine and activates ubiquitin-dependent proteolysis.

Despite the growing number of cellular functions regulated by sphingolipids and their derivatives, the molecular mechanisms by which these regulatory functions are executed in mammalian cells are poorly understood (5)(6)(7). Being ubiquitous and essential components of eukaryotic plasma membranes, sphingolipids are evolutionarily conserved from yeast to humans (8,9). The yeast Saccharomyces cerevisiae has several advantages as a model system to study sphingolipid-mediated cellular regulation. First, the basic structure and metabolism of sphingolipids are conserved between yeast and mammals, and yet yeast has only three major species of sphingolipids, whereas mammalian cells may contain more than 300 distinct molecular species (10). Second, yeast is a genetically tractable organism whose genome has been completely sequenced (11). Lastly, many yeast genes encoding sphingolipid biosynthetic and metabolic enzymes have been recently identified (12). The controlled expression of these genes can be exploited to modulate intracellular levels of certain sphingolipids and their derivatives.
The LCB1 and LCB2 genes encode serine palmitoyltransferase, which catalyzes the first committed step in yeast sphingolipid biosynthesis, the condensation of L-serine and palmitoyl-CoA to produce 3-ketodihydrosphingosine (KDS) 1 (13,14). Recently, heat stress was shown to increase cellular levels of sphingoid bases (dihydrosphingosine (DHS) and phytosphingosine (PHS)) and ceramides with little effect on the levels of complex sphingolipids (15,16). This increase in sphingoid bases is blocked in an lcb1 mutant, suggesting the de novo synthesis of sphingoid bases by serine palmitoyltransferase.
This suggests potential involvement of sphingolipids in stress response, but the biological consequences of such an increase in the levels of cellular sphingoid bases and ceramides have been obscure. Moreover, these studies could not differentiate whether sphingolipids maintain plasma membrane integrity or are involved in relaying stress signals to downstream effectors. In this study, we set out to understand the role of sphingolipids in yeast stress response and the mechanism by which these lipids exert biological effects. This study led us to identify an essential function for serine palmitoyl transferase (SPT) and the de novo synthesis of PHS in mediating stress effects on nutrient permeases via a mechanism involving ubiquitin-dependent proteolysis.
Media and Genetic Manipulations-Recipes for media and yeast genetic methods followed standard protocols (25). For solid media, after media were autoclaved and cooled down, sphingolipid derivatives were added to the media together with Tergitol (type NP-40; Sigma) to the final concentration of 0.05% for even distribution of the lipids (26). Nitrogen starvation medium is 0.67% yeast nitrogen base without amino acids and ammonium sulfate (Difco) plus 2% glucose, and carbon starvation medium is synthetic complete (SC) minus glucose.
Tetrad dissection was performed after incubation of yeast cells on a sporulation medium for three or more days at 30°C and resuspending and incubating the yeast cells in 100 l of Zymolyase 20T (ICN Biomedicals, Inc.) for 10 min at room temperature.
Yeast transformation followed the protocol developed by Gietz et al. (27). Gene disruptions were carried out as described previously (28). In short, open reading frames were replaced by the PCR products consisting of the G418 R gene cassette flanked by 43-base pair sequence homology to targeted genes on both sides of an open reading frame. Each gene disruption was confirmed by PCR, which was designed to amplify the specific chimeric junction of the target gene and the G418 R cassette.
Preparation of Sphingolipid Derivatives-PHS and stearylamine (STA) were purchased from Sigma, DHS was from Biomol, and KDS and C 2 -phytoceramide (PHC), which was N-acetylated from PHS, were gifts from Dr. Alicja Bielawska (Medical University of South Carolina, Charleston, SC). The quality of these sphingolipid derivatives was controlled by thin-layer chromatography for apparent homogeneity. Stock solutions were made in ethanol at 20 mM, stored at Ϫ20°C, and warmed up before use to redissolve precipitates. Tritiated L-histidine, L-leucine, L-tryptophan, L-serine, and uracil (1 mCi/ml each) were purchased from American Radiolabeled Chemicals Inc.
Yeast Growth Assays-Measurement of yeast growth was carried out as described previously (29). Briefly, in liquid culture, an overnight culture of cells at exponential growth phase was diluted into fresh medium containing indicated lipids or ethanol as a control. While incubating in a shaking incubator at 30°C, growth was monitored at a given time by measuring absorbance at 600 nm (A 600 ), and the numbers were converted into cell density (cells/ml), using a pre-configured conversion table, when it was necessary. On solid medium, a small amount of cells from a single colony were streaked by three successive uses of toothpicks, or exponential-phase cells in liquid culture were plated onto medium. Plates were incubated at 30°C for 2 days and photographed for our record.
Sphingolipid Analysis-Isogenic wild-type and lcb1-100 mutant cells were grown at 24°C until exponential phase, harvested, and resuspended in media at either 24 or 39°C. After 15 min, an equal number of cells were harvested, processed for high performance liquid chromatographic analysis, and normalized by a phosphate assay as described previously (15).
Northern Analysis-Log-phase cells were treated with the indicated sphingolipid derivatives or ethanol for 2 h, and total RNA was prepared using the RNeasy Mini Kit (QIAGEN). 20 g of total RNA per lane were loaded onto 1% agarose-formaldehyde gel, and transfer and hybridization were carried out as described previously (30). Probes were prepared with a Random Primed DNA Labeling Kit (Roche Molecular Biochemicals). Radioactive bands were visualized by autoradiography.
Western Analysis-Cells were treated as in the Northern analysis procedure. Protein was extracted as described previously (31), and the concentration was determined by the Bradford method. 10 (for Fur4 and Cpr1) or 30 (for Deg1-␤Gal, ubiquitin, and ubiquitin-Myc) g of total proteins per lane were resolved by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and routinely stained with Ponceau S for checking even loading and transfer. Western blotting was performed according to a previously published procedure (32). The membranes were incubated with a rabbit polyclonal anti-Fur4 antiserum (1:20,000 dilution; gift of Dr. R. Haguenauer-Tsapis; see Refs. 22 and 33), a rabbit polyclonal anti-Cpr1 antiserum (1:2,000 dilution; see Ref. 34), a mouse monoclonal anti-␤-galactosidase (1:2,500 dilution; Promega), a rabbit polyclonal anti-ubiquitin antibody (1:1,000 dilution; Sigma), or a mouse monoclonal anti-c-Myc antibody (1:1,000 dilution; Babco) for 1 h at room temperature. Membranes were washed three times with phosphate-buffered saline (PBS) and incubated with the following proper secondary antibodies: goat anti-rabbit IgG (1:5,000 dilution; Jackson Immunoresearch) or goat anti-mouse IgG (1:5,000 dilution; Jackson Immunoresearch) for 1 h at room temperature. Membranes were again washed three times with PBS, and protein bands were visualized using ECL reagents (Amersham Pharmacia Biotech).

RESULTS AND DISCUSSION
In this study we investigated whether sphingolipids mediate stress signals in S. cerevisiae. We have found that PHS specifically inhibits import of various nutrients, such as uracil, into yeast cells. 2 Here we determined the mechanisms by which PHS inhibits uracil transport. Uracil import is mediated by the uracil permease (Fur4), a member of the major facilitator superfamily (36). Fur4 activity is known to be down-regulated upon heat stress and starvation (22,37), suggesting Fur4 is a target of stress response pathways. We tested whether the inhibitory effect of PHS on uracil permease activity is due to a decrease in levels of the Fur4 protein. When yeast cells were exposed to PHS (20 M), Fur4 protein levels started to decrease by 15 min and were almost undetectable by 2 h (Fig. 1A). In contrast, PHS did not alter the level of a control protein (Cpr1) (Fig. 1A). In contrast to PHS, other closely related lipids including DHS, KDS, C 2 -PHC, and STA had little or no effect on Fur4 (Fig. 1B).
Next, we determined whether the effect of PHS on Fur4 activity was physiologically relevant. Previous studies have demonstrated that PHS levels increase upon heat stress (15,16) and that heat stress inhibits Fur4 activity (37). We tested whether PHS is necessary for Fur4 degradation in response to heat stress. To address this, we utilized an S. cerevisiae strain containing a conditional mutation in the LCB1 gene (lcb1-100), which encodes SPT, an essential enzyme that catalyzes the first and rate-limiting step in de novo synthesis of PHS (13). The lcb1-100 strain is a temperature-sensitive (ts) mutant strain, whose growth is arrested at 37°C and slow even at the permissive temperature (24°C) (18). This suggests that SPT activity in the lcb1-100 strain is impaired to some degree even at 24°C. The ts phenotype is reversible if the incubation temperature is returned from 37 to 24°C or if the growth medium is supplemented with 5 mM PHS when the mutant is grown at 37°C, indicating that the growth defects of the lcb1-100 strain are due to a defect in sphingolipid synthesis. Because the lcb1-100 allele was originally isolated as an endocytosis mutant (end8 -1) and later shown to be a temperature-sensitive allele of the LCB1 gene, we first tested whether this mutation affects SPT activity and PHS generation. In wild-type cells 15 min of heat stress increased cellular levels of PHS severalfold. In contrast, in the lcb1-100 mutant cells, heat stress had little effect on PHS levels ( Fig. 2A). We next evaluated Fur4 degradation in this model. With a short time lag in wild-type cells, Fur4 degradation followed the rise in PHS levels such that upon heat stress, Fur4 levels dropped soon after the increase in PHS and completely disappeared by 30 min of heat stress (Fig.  2B). In contrast, in the lcb-100 strain upon heat stress, which 2 N. Chung, Y. A. Hannun, and L. M. Obeid, unpublished data.

FIG. 1. PHS decreases the levels of the Fur4 uracil permease.
A, exponential growth phase cells of wild-type yeast strain JK9 -3d containing a plasmid expressing the uracil permease, Fur4 (22), were treated with 20 M PHS (Sigma) for the indicated times. Ethanol was used as a vehicle for PHS and did not affect the levels of Fur4 over the same time period (data not shown). Cpr1 was used as a loading control, which also shows the absence of nonspecific protein degradation by PHS. B, specificity of Fur4 decrease by PHS. Immunoblotting was performed after the treatment with the indicated lipid (20 M) for 2 h. PHC 2 , C 2 -phyloceramide.
blocked the generation of PHS by inactivating SPT, the Fur4 protein was stable over a prolonged incubation. These findings suggest that production of PHS by SPT is required for Fur4 degradation. To establish that the function of Lcb1 is to supply PHS, we tested whether the addition of exogenous PHS to lcb1-100 mutant cells would restore Fur4 degradation. SPT was first inactivated at 37°C for 30 min to deplete endogenous PHS, and exogenous PHS was then added under conditions non-permissive for SPT activity (37°C). As shown in Fig. 2C, exogenous PHS restored Fur4 degradation in lcb1-100 mutant cells, albeit not to the level of that in LCB1 wild-type cells.
Next, we investigated the mechanism by which PHS reduces Fur4 protein. By Northern blot analysis, PHS does not repress transcription of the FUR4 gene (Fig. 3A). Nutrient permeases, including the uracil permease, are known to undergo proteolytic degradation (38,39). The uracil permease can be ubiquitinated in its cytoplasmic tail, which signals endocytosis to the vacuole and degradation (37). The ubiquitin ligase, Npi1, and the deubiquitinase, Doa4, are required for Fur4 endocytosis and degradation (22,40). In both npi1 and doa4 mutant yeast strains, the Fur4 protein was no longer degraded upon addition of PHS (Fig. 3B). Moreover, the npi1 and doa4 mutant strains were also PHS-resistant. Thus, Npi1 and Doa4 function downstream of PHS in the pathway regulating Fur4 stability.
We next addressed whether PHS activates ubiquitination of Fur4 or regulates a step in Fur4 endocytosis independent of the ubiquitination. To resolve this issue, we analyzed the effects of PHS on the Deg1-␤Gal protein. Deg1-␤Gal is a fusion protein between the amino terminus of the ␣2 transcriptional repressor (which contains a degradation signal for ubiquitin-mediated proteolysis) and the Escherichia coli enzyme ␤-galactosidase (23). This fusion protein is cytoplasmic and is subject to degradation by the 26 S proteasome pathway, which is independent of endocytosis (38,39). As shown in Fig. 4, PHS (20 M) induced degradation of Deg1-␤Gal. To confirm that PHS-induced degradation of Deg1-␤Gal was via the proteasome pathway, we analyzed the fate of the Deg1-␤Gal protein in a doa1 mutant strain. DOA1 encodes a regulatory component of the proteasome pathway and is required for degradation of the ␣2 repressor (23). As shown in Fig. 4, the Deg1-␤Gal protein was not degraded when doa1 mutant cells were exposed to PHS. Thus, PHS also stimulates protein degradation through the proteasome pathway, supporting the hypothesis that PHS regulates ubiquitination.
We addressed how PHS regulates ubiquitination and degradation of Fur4 and Deg1-␤Gal. Diverse types of stress are known to increase the intracellular ubiquitin pool via transcriptional activation of the polyubiquitin gene UBI4 (35). PHS induced UBI4 transcription; however, by Western blot analysis, the levels of ubiquitin were largely unchanged by PHS (data not shown), and Fur4 and Deg1-␤Gal were still degraded upon addition of PHS in ubi4⌬ mutant cells at a rate similar to UBI4 wild-type cells (data not shown). We then examined other aspects of the ubiquitin system that may have a role in PHSinduced proteolysis. Ubiquitination involves linkage of the carboxyl-terminal glycine residue of ubiquitin to the ⑀-amino group of a lysine residue in a target protein. Monoubiquitinated proteins can be further ubiquitinated through any of three lysine residues on the first ubiquitin; Lys-29, Lys-48, or Lys-63 (24). Polyubiquitination serves to increase the efficiency and rate of proteolysis (40). We tested whether PHS increases polyubiquitination by measuring the formation of ubiquitinubiquitin (ub-ub) dimers from monomeric ubiquitin in the presence of PHS. As shown in Fig. 5A, PHS increased ub-ub dimer formation in a dose-dependent manner. We next measured PHS-induced ub-ub dimer formation through specific lysine residues by utilizing ubiquitin mutants in which two of three lysines were mutated to arginines. Interestingly, PHS increased polyubiquitination at all three lysine residues (Fig.  5B). Among these lysine residues, Lys-63 is suggested to be involved in stress responses (24) and polyubiquitination of the Fur4 permease (40), making it a likely candidate target for PHS. To determine whether polyubiquitination through Lys-63 is required for PHS-induced proteolysis, we used mutant yeast strains that express either wild-type ubiquitin or a Lys-63-Arg mutant (K63R) as the sole source of ubiquitin. PHS-induced FIG. 2. De novo PHS synthesis by SPT is necessary and sufficient for the decrease in Fur4 levels. A, PHS levels in the lcb1-100 serine palmitoyltransferase mutant strain. lcb1-100 mutant and isogenic wild-type cells were grown to exponential phase at 24°C, and one-half of the culture was maintained at the permissive temperature (24°C; open bars) and the other half was transferred to non-permissive temperature (37°C; shaded bars). An equal number of cells were harvested, sphingolipids were extracted and analyzed by reverse phase high performance liquid chromatography, and values were normalized with a phosphate assay as described previously (15). C 20 -and C 18 -PHS indicate two major PHS species found in yeast. B, Fur4 decrease requires wild-type SPT upon heat stress. Cells were treated as in A, and immunoblotting was performed. C, exogenous PHS can restore the defect in Fur4 decrease in the lcb1-100 mutant strain. lcb1-100 mutant cells were first grown at 24°C to exponential phase, transferred to 37°C to inactivate SPT, and immediately harvested or further incubated for 60 min with the indicated concentrations of PHS while maintaining the temperature at 37°C to block generation of endogenous PHS.

FIG. 3. PHS induces Fur4 degradation via ubiquitin-dependent proteolysis.
A, Northern analysis. JK9 -3d cells were treated as in Fig. 1A. The ACT1 mRNA is shown as a loading control. B and C, Fur4 immunoblotting was performed with an npi1 ubiquitin ligase mutant strain and isogenic control strain (19) and doa4⌬ deubiquitinase mutant strain and isogenic control strain, JK9 -3d. Sensitivity to PHS was measured by streaking cells on yeast extract peptone dextrose (YPD)/ Tergitol medium (1% yeast extract, 2% peptone, 2% glucose, 2% agar, and 0.05% Tergitol, type NP-40), containing PHS or ethanol, with growth at 30°C for 2 days. proteolysis of both Fur4 and Deg1-␤Gal was severely impaired in K63R ubiquitin mutant cells compared with wild-type cells (Fig. 5C). Because K63R ubiquitin, like wild-type ubiquitin, can be used in a monoubiquitination reaction, this demonstrates that polyubiquitination through Lys-63 is critical for the PHSinduced proteolysis. We conclude that PHS-induced degradation of Fur4 and Deg1-␤Gal is executed by a PHS-stimulated polyubiquitination activity that extends through the stressresponsive Lys-63 residue of ubiquitin.
Sphingolipids have been implicated in stress responses, but it has been unclear whether these lipids play an essential structural role or function as signal transducers. In this report we demonstrate an essential role for sphingolipids in signaling during heat stress response by showing that PHS targets both the Fur4 protein and the ␣2 repressor for degradation through two different ubiquitin-mediated proteolysis pathways. We propose a model in which heat stress activates PHS synthesis by SPT, which in turn signals to activate the ubiquitin system. Considering the ancient origin and evolutionary conservation of both sphingolipids and the ubiquitin system, this stresssphingolipid-ubiquitin mechanism is likely not limited to yeast cells. FIG. 4. PHS stimulates the 26 S proteasome pathway. Cells of yeast strain JK9 -3d containing a plasmid expressing the Deg1-␤Gal fusion protein were treated with 20 M PHS for the indicated times, and immunoblotting was carried out. Growth phenotype was measured as in Fig. 3.

FIG. 5. PHS-induced proteolysis requires polyubiquitination through residue Lys-63 of ubiquitin.
A, dose-dependent increase of polyubiquitination activity by PHS. The wild-type strain, JK9 -3d, was transformed with a plasmid expressing a ubiquitin tagged at its carboxyl terminus with the c-myc epitope. Cells were treated with PHS for 2 h. B, stimulation of polyubiquitination activity through specific lysine residues by PHS. JK9 -3d cells expressing mutant ubiquitins were treated with 20 M PHS for 2 h. Where indicated, two of three lysine residues in the ubiquitin are mutated to arginine (24). KKK indicates three wild-type lysines at the positions 29, 48, and 63, KRR indicates Lys-29 is wild-type, RKR indicates Lys-48 is wild-type, and so forth. Therefore, use of these plasmids allows cellular activity mediating polyubiquitination through a particular lysine residue to be measured. Also, because the carboxyl termini of these ubiquitins are capped with c-Myc epitope, they cannot be linked to any endogenous proteins. C, requirement of Lys-63 of ubiquitin for PHS-induced proteolysis. Strain SUB280 is deleted for all ubiquitin genes (ubi1⌬, ubi2⌬, ubi3⌬, and ubi4⌬) and contains a plasmid (pUB39) expressing a single copy of wild-type (WT) ubiquitin (20), and SUB413 is identical to strain SUB280 except it contains a plasmid expressing K63R mutant ubiquitin instead of wild-type ubiquitin (21). They were transformed with plasmids expressing Fur4 or Deg1-␤Gal. Cells were treated with 20 M PHS for the indicated times.