Phosphorylation by Pho85 Cyclin-dependent Kinase Acts as a Signal for the Down-regulation of the Yeast Sphingoid Long-chain Base Kinase Lcb4 during the Stationary Phase*

Sphingoid long-chain base 1-phosphates (LCBPs) act as bioactive lipid molecules in eukaryotic cells. In yeast, LCBPs are synthesized mainly by the long-chain base kinase Lcb4p. Until now, the regulatory mechanism for Lcb4p has been unclear. In the present study, we found that Lcb4p is post-translationally modified by phosphorylation. Using a protein kinase mutant yeast collection, we further demonstrated that the cyclin-dependent kinase Pho85p is involved in this phosphorylation. Pho85p functions in a number of cellular processes, especially in response to environmental changes. Two of 10 Pho85p cyclins, Pcl1p and Pcl2p had overlapping functions in the phosphorylation of Lcb4p. Site-directed mutagenesis identified the phosphorylation sites in Lcb4p as Ser451 and Ser455. Additionally, pulse-chase experiments revealed that Lcb4p is degraded via the ubiquitin-dependent pathway. The protein was stabilized in Δpho85 cells, suggesting that phosphorylation acts as a signal for the degradation. Lcb4p is down-regulated in the stationary phase of cell growth, and both phosphorylation and ubiquitination appear to be important for this process. Moreover, we demonstrated that Lcb4p is delivered to the vacuole for degradation via the multivesicular body. Since forced accumulation of LCBPs results in prolonged growth during the stationary phase, down-regulation of Lcb4p may be physiologically important for proper cellular responses to nutrient deprivation.

Sphingosine kinase catalyzes the phosphorylation of sphingosine to form sphingosine 1-phosphate (S1P). 1 In mammalian cells, S1P regulates diverse biological processes including proliferation, differentiation, migration, and apoptosis, both as an extracellular mediator and an intracellular second messenger (1)(2)(3)(4). The extracellular effects of S1P are mediated via the S1P/Edg family of plasma membrane G-protein-coupled receptors, whereas its intracellular targets remain undetermined.
To date, two sphingosine kinase isoforms, SPHK1 and SPHK2, have been identified.
The yeast Saccharomyces cerevisiae does not contain S1P, but, instead, contains two other long-chain base 1-phosphates (LCBPs), dihydrosphingosine 1-phosphate and phytosphingosine 1-phosphate. These LCBPs are synthesized by two longchain base (LCB) kinases, Lcb4p and Lcb5p, which are homologous to the mammalian sphingosine kinases. Most of the LCB kinase activity is attributable to Lcb4p, since deletion of the LCB4 gene results in a drastic reduction in activity (to 3% of wild-type), yet deleting LCB5 has little effect (5).
Like their mammalian counterparts, LCBPs in yeast also function as bioactive lipid molecules active in diverse cellular processes including calcium mobilization, heat stress, and cell cycle regulation (6 -10). Unregulated, increased signaling from LCBPs, resulting from genetically driven accumulation, has been shown to induce cell death (11,12), although the mechanisms involved are unknown.
LCBPs appear to have a particular role in yeast cell cycle progression from G 1 to S phase. Yeast cells grow in glucosecontaining medium by utilizing the glucose through fermentation; however, if glucose becomes limited, the cells can utilize ethanol through respiration, a transition called the diauxic shift. After the postdiauxic phase, wild-type cells undergo G 1 cell cycle arrest and enter into the stationary phase. However, ⌬dpl1 cells, which lack the lyase that metabolizes LCBP, resulting in its accumulation, exhibit a delay of the G 1 arrest (13). Conversely, mutants with diminished sphingosine kinase activity enter the stationary phase without apparent postdiauxic growth (14). Involvement of LCBPs in G 1 -to-S phase progression can also be inferred by the observation that ⌬lcb4 ⌬lcb5 cells have a defect in their ability to recover from G 1 arrest caused by heat shock (11).
Mammalian sphingosine kinase is activated by several external stimuli including platelet-derived growth factor (15), tumor necrosis factor ␣ (16), phorbol esters (17,18), and the cross-linking of Fc␥R1 and Fc⑀R1 (19,20). Recent studies have shown that activation of SPHK1 is mediated by phosphorylation, most likely by ERK1/2 (21,22). On the other hand, information regarding the regulation of the yeast LCB kinases has been limited. Reportedly, LCB kinase activity is low during early log phase, increases just before the diauxic shift, and then gradually decreases in the stationary phase (14), although the mechanism involved in these changes remains unclear. Here, we demonstrate that the down-regulation of LCB kinase activity observed during the stationary phase is due to the degradation of Lcb4p. We further reveal the precise elements of this degradation pathway, including the phosphorylation of Lcb4p by Pho85 cyclindependent kinase (CDK) and its subsequent ubiquitination, multivesicular body (MVB) sorting, and vacuolar degradation.
The plasmid pUG34, a yeast expression vector constructed to produce a fusion protein with an N-terminal enhanced green fluorescent protein (EGFP) under the control of the MET25 promoter, was a gift from J. H. Hegemann. To construct EGFP-tagged LCB4, the LCB4 region was amplified using the plasmid pWK78 and primers 5Ј-AAG-GATCCGTGGTGCAGAAAAAACTTAGGGC-3Ј and 5Ј-AAGCTTGACG-CAACTTCCAAGTGAAT-3Ј. The amplified fragment was digested by BamHI and SalI and cloned into the BamHI-SalI site of pUG34 to generate pWK63.
Dephosphorylation by Alkaline Phosphatase-Cell lysates were prepared as described previously (25). Cells equivalent to 2.0 A 600 were recovered by centrifugation, suspended in 100 l of 0.2 N NaOH containing 0.5% 2-mercaptoethanol, and incubated for 15 min on ice. Sam- ples were then treated with 1 ml of cold acetone and incubated for 30 min at Ϫ20°C. After centrifugation (15,000 ϫ g, 5 min, 4°C), the pellets were suspended in 150 l of buffer A (50 mM Tris-HCl (pH 8.0), 1 mM EDTA, and 1% SDS). Lysates (70 g of protein) were then diluted with 1.9 ml of buffer B (50 mM Tris-HCl (pH 8.0), 1 mM MgCl 2 , and 0.5% Triton X-100) and centrifuged at 15,000 ϫ g for 5 min at 4°C to remove insoluble proteins. The supernatant was divided into two fractions, and one fraction was treated with 2.5 units of E. coli alkaline phosphatase (TOYOBO, Osaka, Japan). Both fractions were then incubated for 1 h at 37°C. Proteins (10 g) were separated by SDS-PAGE, and the Lcb4p proteins were detected by immunoblotting using affinity-purified anti-Lcb4p antibodies.
Pulse-Chase and Immunoprecipitation-Yeast cells were grown to early log phase at 30°C in SC medium lacking methionine and cysteine. Cells were pulse-labeled with [ 35 S]methionine/[ 35 S]cysteine (EXPRESS TM protein labeling mix; 1,000 Ci/mmol; PerkinElmer Life Sciences) at 25 Ci/10 7 cells for 15 min and chased with cold methionine (final 0.5 mg/ml) and cysteine (final 0.1 mg/ml). Aliquots were taken at the indicated times and mixed with NaN 3 . After the cell samples were washed with SC medium containing 10 mM NaN 3 , cell lysates were prepared as described above. Incorporated radioactivity was counted by a liquid scintillation system, LSC-3600 (Aloka, Tokyo, Japan). Cell lysates with equal amounts of radioactivity were diluted with 1 ml of buffer D (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.1 mM EDTA, and 0.1% Lubrol PX). Lcb4p was then immunoprecipitated using anti-Lcb4p antiserum and protein A-Sepharose (Amersham Biosciences). After washing the samples with 1 ml of buffer D and with 1 ml of 10 mM Tris-HCl (pH 8.0), bound proteins were eluted with buffer E (125 mM Tris-HCl (pH 6.8), 4% SDS, 20% glycerol, 10 mM 2-mercaptoethanol, and a trace amount of bromphenol blue), boiled for 3 min, and separated by SDS-PAGE. Gels were fixed, treated with Amplify Fluorographic Reagent TM (Amersham Biosciences), and dried. Radioactivities associated with Lcb4p were quantified using a Bio-Imaging Analyzer BAS-2500.
Fluorescence Microscopy-Cells were first grown in SC medium lacking histidine, and the expression of EGFP and EGFP-Lcb4p was induced by changing the medium to SC medium lacking both histidine and methionine. Cells were washed with phosphate-buffered saline, mounted on slides, and analyzed by fluorescence microscopy AxioSkop 2 plus (Zeiss).

Lcb4p Is Modified by Phosphorylation-Since
LCBPs are signaling molecules, their synthesis is expected to be tightly regulated. However, until now, the regulatory mechanism behind the function of Lcb4p, a major synthesizing enzyme for LCBPs, has been unclear. To examine the post-translational modification of Lcb4p, we prepared antibodies against the fulllength recombinant protein. This antibody detected Lcb4p as a 73-kDa protein in an immunoblot of lysates prepared from wild-type cells; this protein was absent in lysates of the ⌬lcb4 cells (Fig. 1A). Moreover, when Lcb4p was overexpressed from the pWK79 plasmid, a large increase in this band was observed, confirming the specificity of this antibody. When a longer electrophoresis was performed, a lower, 72-kDa band became separated from the major 73-kDa band (Fig. 1B). Upon treatment with alkaline phosphatase, the 73-kDa band was shifted to the position of the 72-kDa band, indicating that Lcb4p is modified by phosphorylation (Fig. 1C).
Pho85 Kinase Is Involved in the Phosphorylation of Lcb4p-Genome sequencing analysis predicted the existence of 117 protein kinases in S. cerevisiae. Of these, 102 genes are nonessential for viability. To identify the protein kinase responsible for phosphorylating Lcb4p, we prepared total lysates from mutants for each of these nonessential protein kinases and investigated the phosphorylation status of Lcb4p by immunoblotting. Although the major 73-kDa Lcb4p band was detected in 101 mutants, the 72-kDa band was detected only in mutants for the PHO85 gene (⌬pho85), which encodes a member of the CDKs ( Fig. 2A). Pho85p plays important roles in a number of cellular responses including carbon source utilization, glycogen metabolism, morphogenesis, cell cycle progression, and phosphate metabolism (27,28). Upon treatment with alkaline phosphatase, no characteristic shift of the Lcb4p band was observed in the ⌬pho85 cells, confirming that phosphorylation of Lcb4p was abolished in these mutants (Fig. 2B). In contrast, despite their involvement in the phosphorylation and activation of the human sphingosine kinase SPHK1 (22), the MAP kinases appear unlikely to be involved in the phosphorylation of Lcb4p, since all six MAP kinase mutants (⌬fus3, ⌬kss1, ⌬hog1, ⌬slt2, ⌬mlp1, and ⌬smk1) exhibited normal phosphorylation of Lcb4p (Fig. 2C).
Pcl1p and Pcl2p Are Redundant Cyclins Responsible for the Phosphorylation of Lcb4p by Pho85p-As for other CDKs, association with cyclins is essential for the enzyme activity of Pho85p, and binding of particular cyclins to Pho85p confers substrate specificity on the enzyme. Ten Pho85p cyclins (Pcls) have been identified (29). To identify the cyclins involved in the phosphorylation of Lcb4p, we performed immunoblotting of Lcb4p on total lysates prepared from mutants for each Pcl. The phosphorylation of Lcb4p was normal in nine Pcl mutants (⌬pcl1, ⌬pho80, ⌬clg1, ⌬pcl5, ⌬pcl6, ⌬pcl7, ⌬pcl8, ⌬pcl9, and ⌬pcl10); however, in the ⌬pcl2 mutants, the phosphorylation was partially reduced (Fig. 3A). Pcl1p and Pcl2p have overlapping functions in G 1 cell cycle progression (30). Therefore, we constructed ⌬pcl1 ⌬pcl2 double mutants and investigated the phosphorylation of Lcb4p in these mutants. As shown in Fig.  3B, the ⌬pcl1 ⌬pcl2 double mutants completely lacked the ability to phosphorylate Lcb4p. Thus, Pcl1p and Pcl2p are redundant cyclins associated with Pho85p that are necessary for the phosphorylation of Lcb4p.
Lcb4p Is Phosphorylated at Several Sites, Including Ser 451 and Ser 455 -Like other CDKs, Pho85p is a proline-directed kinase; its target site for phosphorylation is a Ser or Thr followed by a Pro residue (31). Lcb4p contains six possible Pho85p phosphorylation sites (Ser 150 , Ser 158 , Ser 451 , Ser 455 , Ser 473 , and Thr 550 ) (Fig. 4A). To identify the specific phosphorylation sites, we substituted each Ser or Thr residue with Ala using site-directed mutagenesis. These mutant forms of Lcb4p were expressed in the ⌬lcb4 cells and detected by immunoblotting. Four of these mutants (Lcb4p-S150A, -S158A, -S473A, and -T550A) did not show any significant differences in gel mobility compared with wild-type Lcb4p. In contrast, the Lcb4p-S451A migrated slightly faster than wild-type Lcb4p, and Lcb4p-S455A exhibited even faster mobility (Fig. 4B). Further separation of the proteins revealed that both Lcb4p-S451A and Lcb4p-S455A bands were doublets of phosphorylated and nonphosphorylated forms (Fig. 4C). Thus, some phosphorylation sites are still active in Lcb4p-S451A and Lcb4p-S455A. These results suggest that Lcb4p is phosphorylated at several sites, including Ser 451 and Ser 455 .
Phosphorylation Is Not Required for the Enzyme Activity of Lcb4p-To examine whether phosphorylation of Lcb4p affects its activity, we performed an in vitro sphingosine kinase assay. Consistent with a previous report (5), almost all of the sphingosine kinase activity was attributable to Lcb4p, since the activity in the ⌬lcb4 cells (reflecting Lcb5p activity) was only 2% of the total activity in wild-type cells (Fig. 5A). There was no distinguishable change in activity between wild-type and ⌬pho85 cells (Fig. 5A). We also compared the activities of wildtype Lcb4p and Lcb4p-S455A expressed from plasmids in the ⌬lcb4 mutants. The activity of the Lcb4p expressed from the plasmid was slightly higher than that expressed from chromo-somes due to differences in the protein levels (Fig. 5B). There was no loss of activity observed with the Lcb4p-S455A as compared with the wild-type (Fig. 5A). Thus, the kinase activity of Lcb4p is not regulated by its phosphorylation.
Phosphorylation Affects the Stability of Lcb4p-Phosphorylation by Pho85p is known to act as a signal for degradation in some substrates, such as Gcn4p and Sic1p (28). Therefore, we next examined whether phosphorylation also affects the stability of Lcb4p, using a pulse-chase experiment. Wild-type and ⌬pho85 cells were labeled with [ 35 S]methionine/cysteine for 15 min and then incubated with excess cold methionine and cysteine for 1, 2, 4, 6, and 8 h. We found that in wild-type cells Lcb4p was degraded, with a half-life of about 3 h (Fig. 6, A and  B). This degradation appeared to be mediated via the ubiquitin pathway, since Lcb4p was stabilized in ⌬doa4 cells, which have reduced amounts of ubiquitin (32). Although the expression level of Lcb4p was decreased due to some unknown reason,
Lcb4p was also markedly stabilized in the ⌬pho85 cells, indicating that phosphorylation of Lcb4p functions in determining its stability.
We also examined the amounts of Lcb4p in cells in various growth phases. Abundant amounts of Lcb4p were detected in wild-type cells in log phase. However, these levels began to decrease in cells reaching stationary phase (cell density of A 600 ϳ9.0), and most of the phosphorylated Lcb4p disappeared over the next 2 h (Fig. 6C). It is important to note that the nonphosphorylated Lcb4p appeared to be unchanged. The disappearance of the phosphorylated Lcb4p was due to degradation, since significant amounts of Lcb4p were detected in the ⌬doa4 cells, even 16 and 40 h after reaching stationary phase (Fig. 6D). Likewise, degradation of Lcb4p was retarded in ⌬pho85 cells in stationary phase. These results indicate that phosphorylation plays an important role in the degradation of Lcb4p, especially during the stationary phase.
Ubiquitinated Lcb4p Is Degraded in the Vacuole via the Multivesicular Body-The stabilization of Lcb4p in the ⌬doa4 cells suggested that the protein is ubiquitinated prior to degradation. Indeed, when overproduced, Lcb4p was readily detectable by immunoblotting with an anti-ubiquitin antibody (Fig. 7A). The ubiquitinated Lcb4p was mainly observed at 92 kDa, about 20 kDa higher than the intact Lcb4p, although faint bands at 100 and 110 kDa were also detected. Considering the molecular mass of ubiquitin (8.6 kDa), Lcb4p may be mainly modified by a chain of two or three ubiquitin subunits or monoubiquitinated at two or three residues. The amounts of ubiquitinated Lcb4p as well as those of nonubiquitinated Lcb4p were not increased but, rather, were slightly decreased by treatment with MG132, a proteasome inhibitor (Fig. 7A). Thus, the proteasome appears not to be involved in the degradation of Lcb4p.
In contrast, the effects of phosphorylation on ubiquitination were more evident. In the ⌬pho85 cells, the amount of ubiquitinated Lcb4p was decreased to ϳ35% of that in wild-type cells (Fig. 7B). These results suggest that phosphorylation functions to increase the efficiency of ubiquitination.
Ubiquitination, especially monoubiquitination, is a known signal for sorting proteins into the MVB (or late endosome), where the biosynthetic pathway for the vacuolar proteins and FIG. 6. Lcb4p is degraded in a ubiquitin-and phosphorylationdependent manner. A, SEY6210 (wild-type), SIY120 (⌬pho85), and SIY157 (⌬doa4) cells were labeled with [ 35 S]methionine/cysteine for 15 min at 30°C and then incubated with cold methionine and cysteine for 1, 2, 4, 6, and 8 h. Total lysates were prepared, subjected to immunoprecipitation using anti-Lcb4p antibodies, and separated by SDS-PAGE. Gels were fixed, treated with the fluorescent reagent Amplify TM , dried, and exposed to x-ray film. B, radioactivities associated with Lcb4p in A were quantified and are expressed as a percentage of those at the 1-h chase point. C, SEY6210 cells grown overnight were diluted into fresh YPD medium at a density of A 600 ϭ 0.15 and incubated at 30°C. At each indicated time point, the cell density was measured, an aliquot of cells (2 ϫ 10 7 cells) was collected. Total cell lysates were prepared from the aliquots, and fixed amounts of protein (10 g) were separated by SDS-PAGE and subjected to immunoblotting with anti-Lcb4p antibodies. D, SEY6210, SIY120, and SIY157 cells were har- the endocytic pathway converge. A subgroup of vacuolar protein sorting (Vps) proteins are known to be required for maintaining functional MVB as well as sorting ubiquitinated proteins into the internal vesicles of the MVB. These proteins include Vps27p and three endosomal sorting complexes required for transport (ESCRT) complexes: the ESCRT-I (Vps23p/Vps28p/Vps37p), the ESCRT-II (Vps22p/Vps25p/ Vps36p), and ESCRT-III (Vps2p/Vps20p/Vps24p/Snf7p) (33)(34)(35). To investigate whether the degradation of Lcb4p is mediated via the MVB, cells carrying mutations in these VPS genes, in either log or stationary phase, were subjected to immunoblotting using anti-Lcb4p antibodies. No down-regulation of Lcb4p was observed in the stationary phase in those vps mutants exhibiting defective sorting into the MVB (Fig. 8). In contrast, other vps mutants with unaffected MVB sorting, ⌬vps41, ⌬vps38, and ⌬vps35, had reduced amounts of Lcb4p in the stationary phase similar to wild type controls (Fig. 8). These results provide evidence that the degradation of Lcb4p is mediated via the MVB.
After being sorted into the MVB, proteins are normally delivered to the vacuole for degradation. To monitor the localization of Lcb4p, we constructed an EGFP-Lcb4p fusion protein under the control of a MET25 promoter and examined its expression in cells by immunofluorescence microscopy. EGFP itself was detected at the cytosol of wild-type cells in either log (3 h after the induction) or stationary phase (24 h after the induction). However, EGFP-Lcb4p was observed at the cell perimeter of wild-type cells in log phase, but in cells in stationary phase, staining was detected in the vacuole (Fig. 9A). In contrast, EGFP-Lcb4p was apparent at the cell perimeter, even in the stationary phase, in both the ⌬doa4 and ⌬pho85 cells (Fig. 9A). These results suggested that Lcb4p is transported to the vacuole for degradation and that ubiquitination and phosphorylation are involved in this process.
The vacuole contains six proteases, including proteinase A, which is encoded by PEP4, proteinase B encoded by PRB1, and carboxypeptidase Y encoded by PRC1. In addition to functioning as protein-degrading enzymes, these three proteinases, especially proteinase A, play important roles in the maturation (processing) of other proteases in the vacuole (36). Therefore, the ⌬pep4 mutation and, more strongly, the ⌬pep4 ⌬prb1 or ⌬pep4 ⌬prc1 double mutation, inhibit the total protease activity in the vacuole. We investigated the degradation of Lcb4p during the stationary phase in these mutants. The amounts of Lcb4p were slightly increased in the ⌬pep4 cells compared with wild-type cells (Fig. 9B). Moreover, the degradation was almost completely abolished in both the ⌬pep4 ⌬prb1 and the ⌬pep4 ⌬prc1 cells (Fig. 9B), indicating that Lcb4p is indeed degraded in the vacuole. Note that only the phosphorylated Lcb4p levels were increased by mutations in these vacuolar protease genes, whereas nonphosphorylated Lcb4p levels remained unchanged. These results confirm that phosphorylation is important for the down-regulation of Lcb4p in the stationary phase. DISCUSSION In the present study, we have demonstrated that Lcb4p is modified by phosphorylation. Although phosphorylation is not required either for the activity or the membrane localization of Lcb4p (data not shown), it does affect the stability. A previous study reported that in yeast, the sphingosine kinase activity changes throughout the growth phases and that the activity is low during early log phase, increases just before the diauxic shift, and then gradually decreases during the stationary phase (14). Although we could not reproduce the change in activity in log phase, at least in the background strain used, we did find that the decrease in the activity during stationary phase was caused by the down-regulation of Lcb4p. Growth arrest in the stationary phase seems to be influenced by intracellular amounts of LCBPs. The ⌬dpl1 cells, which accumulate LCBPs, exhibit prolonged proliferation while approaching stationary phase (13). Conversely, yeast cells lacking sphingosine kinase activity stop proliferating soon after exiting from log phase (13). Therefore, the down-regulation of Lcb4p during the stationary phase and, thus, the phosphorylation of Lcb4p may play physiologically important roles in the cellular response to nutrient deprivation.
CDKs, including Pho85p, require an association with a cyclin partner to be activated. To date, 10 cyclins have been identified as Pho85p cyclins (29). Of these, we found that Pcl1p and Pcl2p have overlapping functions in the phosphorylation of Lcb4p. Since both Pcl1p and Pcl2p are expressed in the late G 1 phase of the cell cycle (37,38), it is reasonable to consider that Lcb4p is phosphorylated in late G 1 . It must be noted that LCBPs possess the regulatory function of stimulating G 1 -to-S cell cycle progression. Upon nutrient depletion, G 1 cell cycle arrest in the stationary phase is delayed in the ⌬dpl1 cells (13). Similarly, in mammalian cells, the overexpression of SPHK1, a homologue of Lcb4p, results in a decreased ratio of G 0 /G 1 cells and an increase in S phase cells, regardless of normal or limited serum conditions (39). Moreover, ⌬lcb4 ⌬lcb5 cells have a defect in the recovery from G 1 to S phase following heat shock (11). Since phosphorylation acts as a degradation signal for Lcb4p, it is conceivable that the phosphorylation in the late G 1 phase contributes to the down-regulation of LCBPs after exiting the G 1 phase.
Cyclins are known to function in substrate recognition. The binding site of Lcb4p to Pcl1p/Pcl2p has so far not been identified. However, we found that Lcb4p tagged at its C terminus was not phosphorylated (data not shown). Moreover, the truncation of the 10 C-terminal residues also abolished the phosphorylation (data not shown). Thus, it is possible that the C terminus of Lcb4p contains a binding site for the cyclins.
Results are divided as to the localization of Lcb4p. In our fluorescence microscopic analysis, Lcb4p was observed on the cell perimeter (Fig. 9). However, a previous report demonstrated that Lcb4p C-terminally tagged with hemagglutinin (Lcb4p-HA) was apparent as punctate spots and as diffuse patches, which were co-localized with late Golgi and endosome markers (40). The majority of Lcb4p was also co-sedimented with late Golgi and late endosome markers in a sucrose density gradient (40). Plasma membrane localization of Lcb4p-HA has also been observed, especially when overproduced (40). Yet another researcher, also using Lcb4p-HA, observed punctate staining as well as ringlike staining, which coincided with an endoplasmic reticulum (ER) marker (41). In that report, the majority of Lcb4p was co-fractionated with the ER marker in a sucrose density gradient (41). Although no exact cause for such differences is clear, it is possible that the C-terminal tagging affects the localization of Lcb4p. In fact, in our immunofluorescence microscopic analysis, punctate spots and diffuse patches were observed for Lcb4p-HA but not for the Lcb4p lacking the HA tag (data not shown). Sucrose gradient fractionation detecting endogenous Lcb4p demonstrated that Lcb4p is localized mainly at the ER, but also at the plasma membrane. 2 On the other hand, Lcb4p (without any tag) overproduced from a 2 plasmid under its own promoter was distributed between the ER and the plasma membrane almost equally. 2 Interestingly, the ER localization of Lcb4p was restricted to the cortical ER. 2 In yeast, the ER is typically stained as two ring structures, one at the nucleus (nuclear ER) and the other close to the plasma membrane (cortical ER) (42). Thus, the staining of EGFP-Lcb4p at the cell perimeter, observed in Fig. 9, may correspond to the plasma membrane and the cortical ER.
Considering the difference in the molecular masses (ϳ1 kDa) between alkaline phosphatase-treated and nontreated Lcb4p, we speculate that Lcb4p is multiply phosphorylated (probably at five or six sites). We created a Ser/Thr to Ala mutant for each of six putative Pho85p phosphorylation sites, yet we could detect no difference in gel mobility for four of these mutant forms of Lcb4p (Lcb4p-S150A, -S158A, -S473A, and -T550A) compared with that of wild-type Lcb4p. However, since it is difficult to distinguish a reduction of one phosphorylation by gel mobility shift, the possibility that these sites are phosphorylated cannot be excluded by only this experiment. On the other hand, Lcb4p-Ser 451 and Lcb4p-Ser 455 mutants exhibited apparently faster gel mobility than the wild-type Lcb4p, suggesting that Ser 451 and Ser 455 are normally phosphorylated.
In the present study, we also demonstrated that Lcb4p is degraded in the vacuole via the MVB and that ubiquitination serves as a signal for sorting Lcb4p into the MVB. At the limiting membrane of the MVB, ubiquitinated proteins are invaginated into vesicles (33,43,44). Fusion of the limiting membrane of the MVB with the vacuolar membrane results in delivery of the luminal MVB vesicles to the hydrolytic interior of the vacuole, where they are degraded. Lcb4p is a peripheral membrane protein attached to the cytosolic surface of the membrane. Thus, the internalization step at the MVB is required for Lcb4p to access the lumen of the vacuole.
Recently, it was reported that the mammalian sphingosine kinase SPHK1 is also phosphorylated and that ERK1/2 seems to be involved, since inhibitors of ERK1/2 activation inhibited the phosphorylation of SPHK1, and ERK2 phosphorylated SPHK1 in vitro (22). In the case of Lcb4p, any involvement of the MAP kinases is unlikely, since none of the mutants for the four MAP kinase cascades in yeast exhibited phosphorylation defects (Fig. 2C). However, interestingly, both the phosphorylation site of Lcb4p and that of SPHK1 possess common characters. First, the phosphorylated Ser residues of both SPHK1 and Lcb4p are followed by Pro. Moreover, the phosphorylation site of SPHK1 (Ser 225 ) is aligned very closely to that of Lcb4p (Ser 455 ); indeed, these residues are matched in some alignments. Although phosphorylation induces activation and membrane translocation of SPHK1, such is not the case for Lcb4p. However, it is possible that there are common regulatory features. For example, there remains a possibility that phospho-rylation also leads to the down-regulation of SPHK1. Alternatively, phosphorylation of these residues could cause interactions with common factors. It is known that WW domain-containing proteins specifically recognize phosphorylated Ser/Thr-Pro sequence (45). WW domains contain 38 -40 amino acid residues and play important roles in various cellular functions (46). For instance, Nedd4 in mammalian cells and Rsp5p in yeast are E3 ubiquitin ligases (47,48). Rsp5p, then, is a candidate for the enzyme that is responsible for the ubiquitination of phosphorylated Lcb4p. Interestingly, Lcb4p contains a Lys cluster (10 Lys residues exist from Lys 411 to Lys 441 ) near the phosphorylation sites (Ser 455 and Ser 451 ). This Lys cluster is a likely candidate for ubiquitination. It is possible that phosphorylation induces a conformational change in Lcb4p, facilitating the ubiquitination of these Lys residues. Recent studies have also demonstrated that members of the peptidylprolyl-cis/ trans-isomerase families, Pin1 in mammals and Ess1p in yeast (49,50), contain a WW motif and isomerize only Pro residues following phosphorylated Ser/Thr residues. In many cases, phosphorylation by Pro-directed Ser/Thr kinases such as MAP kinases and CDKs initiates its function only after the following Pro residue is isomerized by a member of the Pin1 family. Thus, further studies are required to elucidate the precise molecular mechanism that links the phosphorylation of Lcb4p to the degradation pathway.