The Sphingoid Long Chain Base Phytosphingosine Activates AGC-type Protein Kinases in Saccharomyces cerevisiae Including Ypk1, Ypk2, and Sch9*

The Pkh1 protein kinase of Saccharomyces cerevisiae, a homolog of the mammalian 3-phosphoinositide-dependent kinase (PDK1), regulates downstream AGC-type protein kinases including Ypk1/2 and Pkc1, which control cell wall integrity, growth, and other processes. Phytosphingosine (PHS), a sphingoid long chain base, is hypothesized to be a lipid activator of Pkh1 and thereby controls the activity of Ypk1/2. Here we present biochemical evidence supporting this hypothesis, and in addition we demonstrate that PHS also stimulates autophosphorylation and activation of Ypk1/2. Greatest stimulation of Ypk1/2 phosphorylation and activity are achieved by inclusion of both PHS and Pkh1 in an in vitro kinase reaction. We also demonstrate for the first time that Pkh1 phosphorylates the Sch9 protein kinase in vitro and that such phosphorylation is stimulated by PHS. This is the first biochemical demonstration of Sch9 activators, and the results further support roles for long chain bases in heat stress resistance in addition to implying roles in chronological aging and cell size determination, since Sch9 functions in these processes. Thus, our data support a model in which PHS, rather than simply being an upstream activator of Pkh1, also activates kinases that are downstream targets of Pkh1 including Ypk1/2 and Sch9.

Sphingolipids play many vital roles in eucaryotic cells where they are structural components of membranes, act as regulators of signal transduction pathways, and associate with sterols to form lipid rafts (1,2). Sphingosine 1-phosphate and ceramide are the best characterized sphingolipid signaling molecules, with sphingosine 1-phosphate promoting growth and preventing apoptosis and ceramide performing an opposing role by promoting apoptosis (3,4). In contrast to the large body of knowledge for ceramide and sphingosine 1-phosphate, the long chain base (LCB) 1 sphingosine, derived from breakdown of ceramide and then used to make sphingosine 1-phosphate, is a poorly characterized signaling molecule in mammals (5). The opposite situation exists in the budding yeast Saccharomyces cerevisiae where very little is known about the signaling functions of ceramide and long chain base phosphates, but signaling functions for the long chain bases phytosphingosine (PHS) and dihydrosphingosine (DHS), the homologs of sphingosine, are being revealed. Here we present evidence that PHS is a signaling molecule that activates the protein kinase Pkh1, a homolog of the mammalian 3-phosphoinositide-dependent protein kinase (PDK1), which then activates downstream protein kinases including Ypk1, Ypk2, and Sch9.
Many different growth and survival factors activate PDK1 in mammals so that it phosphorylates and activates downstream protein kinases, several of which are members of the protein kinase A, protein kinase G, and protein kinase C (AGC) kinase family that controls multiple cellular processes (6). Homologs of PDK1 are widespread in nature, including two in S. cerevisiae, Pkh1 and Pkh2 (7). Deletion of either PKH1 or PKH2 has little or no effect on cell growth, but deletion of both is lethal, indicating that they perform one or more essential functions. Pkh1 and Pkh2 share about 50% amino acid identity and about 70% amino acid similarity with human PDK1. In addition, a human PDK1 cDNA complements the growth defect of a yeast pkh1⌬ pkh2⌬ double deletion mutant, demonstrating functional similarity of the human and yeast PDKs (7). The best characterized downstream substrate of Pkh1 and Pkh2 is the protein kinase Pkc1 (8), which regulates a mitogen-activated protein kinase module that controls cell wall integrity (9,10). Pkc1 controls many other cellular processes, but the mechanisms are less well characterized (11).
In addition to Pkc1, S. cerevisiae contains three other AGC kinase family members, Ypk1, Ypk2/Ykr2, and Sch9, that have been suggested to be substrates of Pkh1/2 based on the similarity of the amino acid sequence of their activation loop, including the threonine (termed the PDK1 site) phosphorylated in Pkc1 by Pkh1/2 (7). Ypk1 was first shown both in vitro and in vivo to be a substrate for Pkh1 (7). Some data support the idea that Pkh1 preferentially phosphorylates Ypk1, whereas Pkh2 phosphorylates Ypk2 (12). Ypk1 and its paralog, Ypk2, perform one or more functions necessary for growth, possibly a function in cell wall integrity (12,13). Human serum-and glucocorticoid-inducible kinase (SGK) is a functional homolog of Ypk1 (7).
In their seminal studies Casamayor et al. (7) noted that 3-phosphoinositides, which are the upstream signal for activating mammalian PDK1, did not stimulate the activity of Pkh1, probably because it lacks a pleckstrin homology (PH) domain, such as is found in PDK1 and Akt/protein kinase B. Thus, the activating signal for Pkh1/2 has remained unknown. The first clue about the nature of the activating signal came from an analysis of multicopy suppressor genes that enabled yeast cells to grow in the presence of an inhibitory concentration of myriocin (14), an antibiotic that inhibits serine palmitoyltransferase, the first enzyme in the sphingolipid biosynthetic pathway (15). YPK1 was identified as such a suppressor gene, and indirect evidence suggested that its phosphorylation was stimulated by a sphingolipid, possibly by PHS. Because Pkh1 was known to phosphorylate Ypk1 and since 3-phosphoinositides did not stimulate phosphorylation (7), Sun et al. (14) suggested that yeast use sphingolipids instead of phosphoinositides as activating lipids.
The first clue that LCBs were the activating sphingolipid came from studies of a temperature-sensitive strain (lcb1-100) having a block in endocytosis at 37°C (16). LCB1 encodes a subunit of serine palmitoyltransferase and at the restrictive temperature an lcb1-100 strain stops making sphingolipids (17), and instead of producing a transient increase in LCBs, as occurs in wild type cells during heat shock (18,19), the concentration of LCBs drops rapidly (20). The endocytosis defect was found to be suppressed by multiple copies of PKC1, YCK2, PKH1, or PKH2 (16,21,22). How multiple copies of these genes restore endocytosis is unclear, but the data suggested that the activity of these kinases was stimulated by some sphingolipid, and it was found using immunoprecipitated HA-tagged proteins that LCBs gave a 2-3-fold stimulation in Pkc1 phosphorylation by Pkh1 or Pkh2 (16).
To further our understanding of the signaling pathways and cellular processes regulated by LCBs we set out using purified proteins to directly demonstrate that LCBs stimulate phosphorylation and activation of Ypk1, Ypk2, and Sch9 by Pkh1. In addition to showing that LCBs stimulate Pkh1 activity, we also found that LCBs stimulate autophosphorylation and activation of Ypk1, Ypk2, and Sch9. These results show that rather than acting solely as an upstream activator of signaling pathways regulated by Pkh1, LCBs also act downstream to activate kinase substrates of Pkh1. Our data are the first biochemical demonstration of upstream regulators of Sch9, an important protein kinase, related to mammalian Akt/protein kinase B, that plays poorly defined roles in heat stress resistance (23,24), chronological aging (25), Ty1 transposition (26), cell size (27), entry into (28) and exit from stationary phase (29), homologous recombination in ribosomal gene hot spots (30), and adaptation to changes in nutrients (31).
Purification of His 6 -tagged Proteins-Yeast cells transformed with a multicopy vector expressing the desired gene under control of the GAL1 promoter were grown overnight at 30°C to an A 600 nm of 0.4 in 1 liter of S-Ura medium containing 2% sucrose and 0.1% glucose. Protein overproduction was induced by adding galactose to 2% and incubating the culture for 8 -10 h. A cell-free protein extract was prepared as described previously (8) and loaded onto a 0.8 ϫ 4-cm Poly-Prep column (Bio-Rad) containing 1 ml of nickel nitrilotriacetic acid-agarose (Qiagen, Valencia, CA). The flow-thru was collected and passed through the column again followed by 5 ml of wash buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 20 mM imidazole, 20% glycerol. His 6 -tagged proteins were eluted with 3 ml of buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 200 mM imidazole, 20% glycerol) except His 6 -Pkh2, which was eluted with 3 ml of a slightly different buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 150 mM imidazole). Six 0.5-ml fractions were collected and analyzed by SDS-PAGE and immunoblotting (32). Three fractions with the highest level of tagged protein were pooled, separated into aliquots, and stored at Ϫ80°C. His 6 -and 3HA-tagged proteins were also purified by immunoprecipitation from cell-free yeast extracts as described previously (16).
In Vitro Protein Kinase Assay-An appropriate amount of kinase and substrate, determined in preliminary experiments, were combined in 18 l of phosphorylation buffer and warmed to 30°C before starting the reaction by adding 2 l of ATP solution (1 mM ATP, 4 Ci of [␥-32 P]ATP). The final component concentrations were 50 mM MOPS, pH 7.5, 1 mM dithiothreitol, 10 mM magnesium acetate and 100 M ATP. The reaction was stopped by the addition of Laemmli buffer followed by SDS-PAGE and phosphorimage analysis to quantify phosphorylation of substrate.
Two-step Ypk1/Ypk2 Kinase Activity Assay-Affinity-purified His 6 -Ypk1 or His 6 -Ypk2 (1 g) were activated by incubation with His 6 -Pkh1 (50 ng) and 50 M PHS (stock prepared in 95% EtOH and diluted to a final concentration of 1%) in 200 l of kinase assay buffer (50 mM MOPS, pH 7.5, 1 mM dithiothreitol, and 10 mM magnesium acetate) plus 100 M ATP at 30°C. After 30 min the reactions were chilled on ice, and 1 ml of immunoprecipitation buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40) plus 2 g of anti-His antibody (Amersham, #27-4710-0) and 10 l of 50% protein A beads were added to each tube. Tubes were rotated at 4°C overnight, and the immunoprecipitates were washed 3 times with immunoprecipitation buffer and 2 times with kinase assay buffer before suspension in 50 l of kinase assay buffer. The kinase activity of Ypk1 or Ypk2 was assayed by adding 30 l of a kinase buffer with 2.5 M PKI-tide (Alexis Biochemicals, CA), 1 M microcystin-LR (Alexis Biochemicals), 10 mM magnesium acetate, 100 M [␥-32 P]ATP (4Ci), and 100 M Cross-tide (GRPRTSSFAEG, BIOSOURCE Int.) as substrate (7). After incubation for 15 min at 30°C, the reaction mixture was centrifuged for 20 s at top speed in a microcentrifuge, and 20 l of supernatant fluid was spotted onto phosphocellulose paper (2.5 cm circles, Whatman P81) which was processed as described previously (33). The pellets were treated with Laemmli buffer followed by SDS-PAGE to analyze the amount of Ypk1 or Ypk2 in each reaction. No effort was made to separate Ypk1 and Pkh1 because control reactions showed that Pkh1 did not phosphorylate Cross-tide. One unit of kinase activity is defined as the amount of protein required to catalyze phosphorylation of 1 nmol of Cross-tide/min. Two-step Sch9 Kinase Activity Assay-3HA-Sch9 was isolated by immunoprecipitation using an anti-HA monoclonal antibody (Anti-Rat, Roche Applied Science) and IgG-Sepharose beads (Sigma). In the first step of the reaction, immunoprecipitated 3HA-Sch9 (50 ng) was incubated at 30°C with 2.5 ng of His 6 -Pkh1 or His 6 -Pkh2 in a 20-l reaction containing or lacking 50 M PHS (stock prepared in 95% EtOH and diluted to a final concentration of 1%) and kinase assay buffer (final concentrations were 40 mM MOPS, pH 7.5, 1 mM dithiothreitol, and 10 mM magnesium acetate). After 30 min the reactions were centrifuged at 3000 rpm for 3 min in a microcentrifuge at 4°C. The supernatant was discarded, and the beads were resuspended in 500 l of kinase buffer followed by centrifugation. This washing procedure was repeated a total of three times.
In the second step of the reaction, the resuspended beads were incubated at 30°C for 30 min in a 20-l reaction containing 1.5 g of Lsp1 (the molar ratio of Sch9:Lsp1 was 1:100), 50 M PHS, or 1% ethanol (final concentration), and ATP (4 Ci of [␥-32 P]ATP and 100 M ATP). The reaction was stopped by the addition of Laemmli buffer followed by SDS-PAGE and phosphorimage analysis to quantify the phosphorylation of Lsp1.
Miscellaneous Assays-To quantify purified His 6 -tagged yeast proteins, serial dilutions were resolved on SDS-PAGE along with dilutions of a BSA standard (Bio-Rad). A scanned image of the Coomassie Brilliant Blue-stained gel was analyzed, and the protein quantities of specific bands were determined by using ImageQuant software.

RESULTS
LCBs Stimulate Phosphorylation of Ypk1 and Ypk2-Available published data imply that phosphorylation of Ypk1 and Ypk2 by Pkh1 and Pkh2 is stimulated by LCBs. However, there has been no direct biochemical demonstration of this hypothesis. To determine whether LCBs stimulate Pkh1 and phosphorylate the downstream kinases Ypk1 and Ypk2, we performed in vitro kinase assays using His 6 -tagged, partially purified protein kinases. We focus primarily on Pkh1, since it has functional overlap with Pkh2 and appears to be more important (12). We found that phosphorylation of Ypk1 increased linearly for about 40 min when both PHS and Pkh1 were included in the kinase reaction (Fig. 1A). Less phosphorylation was observed when either PHS or Pkh1 were omitted from the reaction, and the lowest level of phosphorylation was observed when only Ypk1 was present in the reaction.
To prove that phosphorylation of Ypk1 was due to Pkh1 and not to a contaminating protein kinase, a Pkh1 kinase dead variant (KD, K154R) was examined and found to give about the same level of phosphorylation as in a kinase reaction lacking Pkh1 but containing Ypk1 and PHS (Fig. 1, compare panels A and B). Weak phosphorylation in the Ypk1-only reaction is either due to autophosphorylation or to a contaminating kinase (Fig. 1A), since it was also seen when Ypk1 was incubated with a Pkh1 kinase dead variant (Fig. 1B). We conclude from these results that PHS produces a 5-8-fold stimulation of Ypk1 phosphorylation by Pkh1. In addition, PHS produces an approximate 2-fold stimulation of Ypk1 autophosphorylation either in the absence of wild type Pkh1 (Fig. 1A) or in the presence of the kinase dead mutant (Fig. 1B).
Qualitatively similar results were observed when Ypk2 was used as a substrate for Pkh1 (Fig. 1C). Phosphorylation of Ypk2 was barely stimulated by Pkh1 but was stimulated about 2-fold when PHS was added to the reaction. Adding both PHS and Pkh1 to the reaction gave a further 2-fold stimulation in phosphorylation for a combined 4-fold stimulation. To prove that Pkh1 and not a contaminating protein kinase was phosphorylated Ypk2, a Pkh1 kinase dead variant (KD, K208R) was compared with the wild type enzyme. The reaction containing the kinase dead variant did show some phosphorylation of Ypk2 that was stimulated 2-fold by PHS (Fig. 1D). Weak phosphorylation in the Ypk2 reaction is either due to autophosphorylation or to a contaminating kinase, which seems to be slightly stimulated by PHS.
Panels E and F in Fig. 1 show representative phosphorimages of the 32 P-labeled Ypk1 and Ypk2 proteins produced in kinase reactions. Reactions containing PHS or Pkh1 show two Ypk1 bands of about equal concentration. In reactions containing both PHS and Pkh1 there is an increase in the amount of the upper Ypk1 band, indicating that this species of Ypk1 protein is more phosphorylated than is the lower, faster migrating species. Ypk2 appears as a broader more diffuse band that becomes more phosphorylated in the reaction containing both PHS and Ypk2. The data shown in panels A-D (Fig. 1) represent values for the lower plus the upper band of Ypk1 and the broad band of Ypk2.
Pkh1-mediated phosphorylation of Ypk1 was examined in more detail by varying the concentration and type of LCB added to the kinase reaction. PHS at 50 M stimulated phosphorylation of Ypk1 slightly better than 10 M, although 100 M barely stimulated ( Fig. 2A). DHS, lacking the 4-OH group found on PHS, was tested as the biological erythro isomer and the non-biological threo isomer. Surprisingly, the threo isomer stimulated Ypk1 phosphorylation better than the erythro isomer (Fig. 2, compare panels B and C), suggesting that DHS may not be as significant as PHS in activating Ypk1 phosphorylation in vivo. The phosphorylated version of DHS, DHS-1-P, showed no stimulation of phosphorylation and neither did the acylated version of sphingosine, C 6 -ceramide (Fig. 2, D and E), indicating that the free amino and C1-OH groups on PHS are important for kinase stimulation. Stearylamine, a long chain amine, stimulated phosphorylation, but unlike the LCBs, there was no inhibition of phosphorylation at the highest concentration used. This difference suggests that stearylamine and LCBs work in different ways to stimulate phosphorylation. Finally, sphingosine, the LCB found in mammalian but not fungal sphingolipids, was the best stimulator of Ypk1 phosphorylation (Fig. 2G), giving almost twice the stimulation as 50 M PHS.
LCBs and Pkh1 Activate Ypk1 and Ypk2 in Vitro-We next examined what effects LCBs and Pkh1 have on Ypk1 kinase activity by using the artificial substrate, Cross-tide, used previously to assay activation of Ypk1 by Pkh1 and Pkh2 (7,12,32,34). The activation assay was done is two steps. In step one, Ypk1 or Ypk2 was incubated with or without Pkh1 or PHS in the presence of ATP so that Ypk1 or Ypk2 could be activated by phosphorylation. In the second step, Ypk1 or Ypk2 was immunoprecipitated, and the precipitate was used in a kinase assay to quantify the transfer of 32 P from [␥-32 P]ATP to Cross-tide. Ypk1 was activated 4-fold by either PHS or Pkh1, and inclusion of both in the reaction produced an 8-fold, additive increase in Ypk1 activity (Fig. 3). Activation by Pkh1 required its kinase activity, since the K154R kinase dead variant did not activate Ypk1. It should be noted that Pkh1 was also immunoprecipitated, but control reactions showed that it did not phosphorylate Cross-tide.
Ypk2 was activated 2-3-fold by PHS but was not activated by Pkh1 alone. However, when both PHS and Pkh1 were added to the reaction, Ypk1 was activated about 4-fold (Fig. 3). Like Ypk1, activation of Ypk2 required Pkh1 kinase activity, since the K154R kinase dead variant did not activate Ypk2.
LCBs, Pkh1, and Pkh2 Stimulate Phosphorylation of Sch9 -Based upon the sequence similarity of the PDK1 and PDK2 sites in Ypk1 and Ypk2 to the corresponding regions in Sch9, it was suggested that Sch9 is activated by Pkh1 and Pkh2 (7). However, this hypothesis has never been tested directly. Because phosphorylation of Ypk1, Ypk2, and Pkc1 by Pkh1 and Pkh2 has been demonstrated to be stimulated by PHS, we also wanted to see if phosphorylation of Sch9 is directly stimulated by PHS or indirectly via stimulation of Pkh1 or Pkh2.
To determine whether Sch9 is a substrate of Pkh1 and Pkh2 and if phosphorylation is stimulated by LCBs, we performed in vitro kinase assays using His 6 -tagged partially purified protein kinases. First, we found that Sch9 phosphorylation was stimulated about 2-fold by PHS and about 2.5-fold by Pkh1 (Fig. 4). Inclusion of both PHS and Pkh1 in the kinase reaction stimulated phosphorylation of Sch9 4-fold. Substitution of a Pkh1 kinase dead variant (KD, K154R) in place of the wild type enzyme resulted in a level of phosphorylation that was not significantly different from a reaction that contained only Sch9 (Fig. 4, compare lanes 1 and 5), indicating that the kinase In the first step of a two-step reaction, affinity-purified His 6 -Ypk1 or His 6 -Ypk2 were activated by incubation with ATP in the presence or absence of His 6 -Pkh1 and plus or minus 50 M PHS. In the second step, His 6 -tagged proteins were immunoprecipitated, and the kinase activity of Ypk1 was determined in the presence of [␥-32 P]ATP and Cross-tide, a peptide substrate. Phosphorylation of Cross-tide was quantified, and one unit (U) of activity equals the amount of Ypk1 or Ypk2 catalyzing phosphorylation of 1 nmol of Cross-tide/min. Values represent the mean Ϯ S.D. for n ϭ 3, and those that are statistically different (p Ͻ 0.05, Student's t test) are indicated with an asterisk (compare the samples with one versus to those with two asterisks, and compare the samples with two versus to those with three asterisks). activity of Pkh1 is responsible for stimulation of Sch9 phosphorylation rather than the activity of a contaminating kinase. The addition of PHS to the reaction containing the Pkh1 (KD) variant seemed to produce more stimulation of Sch9 phosphorylation than when the KD variant was absent (Fig. 4, compare  lanes 2 and 5). This could be due to a physical interaction between Sch9 and the KD variant, which makes Sch9 slightly more responsive to PHS. We conclude from these results that Pkh1 and PHS independently stimulate phosphorylation of Sch9.
Pkh2 was also examined to determine whether it could phosphorylate Sch9 and whether PHS stimulated phosphorylation. PHS gave about a 2-fold stimulation of phosphorylation (Fig. 4,  compare lanes 1 and 2), whereas Pkh2 gave about a 4-fold stimulation (Fig. 4, compare lanes 1 and 3). Phosphorylation was stimulated between 8-and 9-fold in the presence of both PHS and Pkh2, indicating an additive effect on phosphorylation (Fig. 4, compare lanes 1 and 4). The kinase dead version of Pkh2 (K208R) did not stimulate phosphorylation of Sch9 by itself (Fig. 4 compare lanes 1 and 5), indicating that the kinase activity of Pkh2 was responsible for phosphorylation of Sch9.
The other control reaction containing both the kinase dead variant of Pkh2 and PHS produced about the same level of phosphorylation as did the reaction containing wild type Sch9 plus PHS (Fig. 4, compare lanes 2 and 6). These results demonstrate that Pkh2 and PHS independently stimulate phosphorylation of Sch9 and that the largest stimulation occurs when both are present in the kinase reaction.
PHS and Pkh1 Stimulate the Kinase Activity of Sch9 -We wanted to determine whether the increase in phosphorylation of Sch9 produced by LCBs and Pkh1 or Pkh2 also increased Sch9 protein kinase activity. However, no substrate of Sch9 is known at this time. Previously we showed that Pil1 and Lsp1 are substrates of Pkh1 and Pkh2 (32), and it seemed possible that Sch9 might also phosphorylate Pil1 and Lsp1 since Sch9 is related to Pkh1 and Pkh2. Indeed, we found that Sch9 phosphorylates both Pil1 and Lsp1 in vitro and that Lsp1 is a slight better substrate (data not shown). We do not know if Lsp1 is a true, physiological substrate of Sch9, but it is a substrate in vitro, and this enabled us to examine the effects of PHS and Pkh1 on Sch9.
Using Lsp1 as a substrate, a two-step kinase assay was used to determine the effect of PHS and Pkh1 or Pkh2 on Sch9 activity. The assay was set up using a concentration of Sch9 that gave a very low level of phosphorylated Lsp1. In the first step of the reaction immunoprecipitated 3HA-Sch9 bound to-Sepharose beads was incubated in a kinase reaction containing or lacking PHS and containing or lacking His 6 -Pkh1 or His 6 -Pkh2. The Sepharose beads were then removed from the kinase reaction, washed, and added to a second kinase reaction containing His 6 -Lsp1 and [ 32 P]ATP. Preliminary experiments showed that for Sch9 to become activated and to be able to phosphorylate Lsp1, PHS had to be included in the second step of the kinase reaction (data not shown). Therefore, for the results shown in Fig. 5, PHS was either present or absent in the first step of the reaction, as indicated in the figure, and was present in all reactions during the second step of the assay. The addition of PHS to the first step of the kinase reaction did not enhance Sch9 kinase activity above the low background seen in the reaction containing only Sch9 (Fig. 5). The addition of Pkh1 to the first step of the kinase reaction stimulated Sch9 kinase activity 2-fold, whereas inclusion of both PHS and Pkh1 produced a 4-fold stimulation of activity. Stimulation of Sch9 activity was not due to a contaminating kinase because a Pkh1 kinase dead variant did not stimulate Sch9 activity (Fig. 5). In contrast to the results with Pkh1, we found that Pkh2 did not stimulate the protein kinase activity of Sch9, at least under the reaction conditions used and with Lsp1 as a substrate (Fig. 5).
Role of the PDK1 and PDK2 Phosphorylation Sites of Ypk1 in Vivo-We attempted to determine which residue(s) in Ypk1 and Ypk2 was phosphorylated in vitro by Pkh1 in response to PHS. Based upon previous studies (7) we expected the PDK1 (Thr-504) site to be phosphorylated by Pkh1 in response to PHS and perhaps also the PDK2 (Thr-662) site. However, when either or both residues in Ypk1 was changed to alanine or aspartic acid there was no consistent change in the level of phosphorylation obtained in vitro using Pkh1 plus or minus PHS (data not shown). Further experiments revealed that the single PDK1 or PDK2 site mutants and the double site variants were phosphorylated at multiple sites in our in vitro kinase assay. Ongoing studies are looking for the phosphorylation sites.
As an alternative to examining the role of Ypk1 and PHS in phosphorylation of the PDK1 and PDK2 sites in vitro, we examined the ability of single PDK1 and PDK2 site mutants and double mutants to complement three phenotypes of ypk1⌬ cells. For these experiments YPK1 and mutant alleles were cloned into the multi-copy vector pYES2/NTA under control of the GAL1 galactose-inducible yeast promoter. This promoter is repressed by glucose; however, basal expression in cells plated on YPD plates is sufficient to examine the activity of Ypk1, since ypk1⌬ cells carrying plasmid-borne YPK1 behave like wild type cells (Fig. 6). Results similar to those shown in Fig. 6 were also obtained using synthetic medium with galactose as the carbon source in place of glucose (data not shown).
The slow growth characteristic of ypk1⌬ cells was the first phenotype examined (12,35,36). This phenotype was not complemented by a plasmid carrying the PDK1 site mutants T504A or T504D (Fig. 6, YPD). The T662A PDK2 site mutant also did not complement, but the T662D mutant did restore growth to the wild type level. The double T504D,T662D mutant did not complement the growth defect, indicating that Thr-504 (PDK1 site) is essential for the Ypk1 function or functions that govern growth rate.
Second, we examined PDK1 and PDK2 site mutants for complementation of the hygromycin B sensitivity of ypk1⌬ cells (12). As in the case of the slow growth phenotype, only the PDK2 T662D mutant complemented the drug sensitivity phenotype (Fig. 6, YPD ϩ Hygromycin B).
Finally, we examined the ability of the PDK1 and PDK2 site mutants to promote growth of ypk1⌬ cells in the presence of a low concentration of myriocin, an inhibitor of the first step in sphingolipid synthesis mediated by serine palmitoyltransferase (37). Because Ypk1 and presumably Ypk2 operate downstream of an LCB-controlled signaling pathway, ypk1⌬ cells should be more sensitive to a low concentration of myriocin than wild type cells and, indeed, they are (Fig. 6, YPD ϩ  Myriocin). The only mutant that restored myriocin resistance to ypk1⌬ cells was the PDK2 T662D mutant, which gave nearly the wild type level of resistance.
We conclude from these experiments that the PDK1 phosphorylation site (Thr-504) is essential for Ypk1 function in vivo and that an acidic residue does not substitute for the essential phosphothreonine. On the other hand changing the PDK2 phosphorylation site (Thr-662) to an acidic residue but not to an alanine does not interfere with Ypk1 function, indicating that an acidic residue or phosphothreonine at this site is probably essential, at least for the phenotypes tested.

DISCUSSION
The data presented here demonstrate that PHS stimulates Pkh1 to phosphorylate and activate the downstream kinases Ypk1, Ypk2, and Sch9. These data, thus, verify previous predictions that PHS acts upstream of Pkh1 to regulate its activity toward downstream substrate kinases (7, 12-14, 16, 22). Interestingly, we also find that PHS stimulates the activity of Ypk1, Ypk2 (Fig. 3), and Sch9 (for the data shown in Fig. 5 PHS was included in the second step of the assay because no Sch9 activity was observed when PHS was omitted). Thus, our data support a model (Fig. 7) in which PHS acts to stimulate the activity of both Pkh1 and downstream kinases Ypk1, Ypk2, and Sch9.
Our data showing that Pkh1 and PHS activate Sch9 are the first biochemical demonstration of Sch9 activators. Previous genetic studies suggested that the G-protein coupled receptor, Gpr1, and its associated G␣ subunit, Gpa2, were upstream regulators of Sch9 activity (38), but another report concludes that Gpr1/Gpa2 do not regulate Sch9 function (39). Regulation of Sch9 by PHS is consistent with the fact that PHS and other LCBs transiently increase immediately after a heat shock (18,19), and they could mediate at least part of the known role of Sch9 in heat stress tolerance (23,24,40). Our data suggest that PHS or another LCB plays a role in chronological aging (25), entry into (28) and exit from stationary phase (29), and cell size (27) since Sch9 functions in these processes.
The effect of different LCBs on Pkh1-mediated phosphorylation of Ypk1 was examined to determine which LCBs are physiologically important and to establish lipid specificity. At a concentration of 50 M, PHS stimulated phosphorylation of Ypk1 four times better than did 50 M erythro-DHS, suggesting that PHS is more likely to be regulating Ypk1 activity in vivo than is DHS. We have previously estimated that the intracellular concentration of LCBs is in the range of 50 M (32), so the stimulation of Ypk1 phosphorylation and activity that we observe with 50 M PHS is well within the physiological range. Heat shock produces a transient increase in LCBs (18,19), which we would expect to activate Pkh1, Ypk1, Ypk2, and Sch9 based upon the results presented here. The heat-induced increase in LCBs would also activate Pkh1 and Pkh2, which can activate Pkc1 based upon previously published data (8,16). The limited stimulation of Ypk1 by erythro-DHS suggests that LCBs do not act in a general way like a detergent to activate Pkh1, Ypk1, Ypk2, or Sch9 but rather that PHS binds to a specific domain in these proteins. Specific binding versus a hydrophobic or detergent-like effect is also suggested by the difference observed between LCBs and stearylamine, a long chain amine that lacks the hydroxyl groups found on LCBs. At the highest concentrations tested (50 -100 M), all LCBs inhibited Ypk1 phosphorylation compared with lower concentrations, whereas stearylamine produced no inhibition at the highest concentration tested (Fig. 2, 100 M). DHS-1-phosphate and ceramide did not stimulate phosphorylation. These compounds lack both a free amino and an OH on C1, suggesting that both groups are important for stimulating the activity of the protein kinases examined here.
Many members of the AGC protein kinase family have one or more lipid binding domain. For example, PDK1, the mammalian homolog of Pkh1 and Pkh2, has a pleckstrin homology (PH) domain that binds 3-phosphoinositides as do mammalian Akt/ protein kinase B (41). Mammalian PKCs have unique domains that bind a variety of lipids (42). Of the yeast AGC kinases, only Pkc1 and Sch9 are predicted by the SMART algorithm (43) to contain lipid binding domains. It is unclear whether lipids actually bind to the predicted Pkc1 domains (11), whereas the predicted C2 domain in Sch9 has not been examined for lipid binding. Because some AGC kinases possess lipid binding domains and since Pkh1, Pkh2, Ypk1, and Ypk2 are stimulated by PHS, it would seem very likely that these kinases have a PHS binding domain.
Attempts to determine whether PHS stimulated phosphorylation of the PDK1 (Thr-504) or the PDK2 (Thr-662) site or both sites of Ypk1 in vitro were inconclusive because mutation of either or both sites to alanine did not reduce the phosphoryla-tion level produced by Pkh1 in the presence of PHS. Further analyses revealed that Ypk1 was phosphorylated at other sites yet to be identified in addition to the PDK1 and PDK2 sites. Similar unidentified phosphorylation sites have recently been reported (34). These data suggesting multiple phosphorylation sites were obtained with Ypk1 and Pkh1 produced in yeast, and they differ from the results reported for glutathione S-transferase fusion proteins made in human 293 embryonic kidney cells (7). Using purified glutathione S-transferase proteins it was demonstrated in vitro that glutathione S-transferase -Pkh1 phosphorylated the PDK1 site in the activation loop of Ypk1 but did not phosphorylate the PDK2 site in the hydrophobic domain. These differences in results suggest that the kinases made in yeast contain modifications that change their ability to be phosphorylated in vitro and that additional phosphorylation sites remain to be identified.
As an alternative to the in vitro analysis of the PDK1 and PDK2 sites, we examined their roles in vivo by determining the ability of mutant alleles to complement the slow growth phenotype and the hygromycin B and myriocin-sensitive phenotypes of ypk1⌬ cells. Mutating the PDK1 site (Thr-504) in the activation loop to alanine or aspartic acid to try and mimic the negative charge of a phosphorylated threonine generated alleles that did not complement any of the three phenotypes (Fig.  6). Mutating the PDK2 site (Thr-662) to aspartic acid but not to alanine produced an allele that complemented all three phenotypes. Combining the T504A or T504D mutation with the T662D mutation resulted in an inactive allele. These results are consistent with the in vitro kinase data showing that mutation of Thr-504 in Ypk1 blocked phosphorylation by glutathione S-transferase-Pkh1 (7). We conclude from these data that Thr-504 in the activation loop is essential for Ypk1 function and that phosphorylation cannot be bypassed by substitution with an acidic residue nor by the T662D mutation. A similar conclusion has recently been reached by others (34).
One explanation for the ability of the T662D mutant allele to complement the three phenotypes of ypk1⌬ cells is that the negative charge of the Asp-662 residue facilitates binding to the PDKI-interacting fragment-binding pocket of Pkh1 (44) and thereby allows Ypk1 to dock to Pkh1. The existence of a PDKI-interacting fragment-binding pocket in Pkh1 has been demonstrated experimentally (45). In mammals this docking mechanism promotes autophosphorylation and activation of PDKI (46), which then phosphorylates the bound AGC kinase at the PDK1 site in the activation loop. The same activation mechanism has been described for serum-and glucocorticoidinduced protein kinase (SDK) (47), the mammalian homolog of Ypk1, and for other mammalian AGC kinases including RSK (46), S6K (47), PRK2 (48,49), and probably for some PKC isoforms (49,50), Thus, the Asp-662-containing form of Ypk1 FIG. 6. The T662D PDK2 site variant of Ypk1 complements mutant phenotypes of ypk1⌬ cells. Wild type (BY4741) and ypk1⌬ cells (BY4741 ypk1: KAN) untransformed or transformed with the pYES2/NTA vector or the vector carrying the indicated YPK1 alleles were grown to saturation in YPD medium. 5-Fold serial dilutions were spotted (4.5 l) onto YPD plates, which were photographed after incubation for 2 days at 30°C. Plates contained hygromycin B (25 g/ml) or myriocin (1.5 g/ml) as indicated.

FIG. 7. Regulation of AGC protein kinases by PHS in S. cerevisiae.
The results presented in this paper support the idea that PHS stimulates phosphorylation of Pkh1, which then activates substrate kinases including Ypk1, Ypk2, and Sch9. PHS also stimulates autophosphorylation and activation of Ypk1, Ypk2, and Sch9; PHS stimulation of Pkc1 autophosphorylation has not been examined. Published data establish that PHS stimulates Pkh1 and Pkh2 (not shown) to phosphorylate Pkc1 (16). Phosphorylation is thought to occur in the activation loop at the PDK1 site of Ypk1/2, Sch9, and Pkc1. would likely bind to Pkh1, thereby stimulating it to autophosphorylate and become active. Pkh1 would then activate the bound Ypk1 by phosphorylating Thr-504. The ability of an acidic residue, aspartic or glutamic acid, to mimic phosphorylation at the PDK2 site has been noted in some mammalian AGC kinases including PRK1/PRN, PRK2, and atypical PKC isoforms (51,52).
Restoration of myriocin resistance in ypk1⌬ cells by the T662D allele is a particularly significant result because it supports the hypothesis that PHS plays both a direct role in stimulating Ypk1 activity and an indirect role via stimulation of Pkh1 as shown in Fig. 7. Myriocin inhibits serine palmitoyltransferase, the first enzyme in sphingolipid synthesis, and at high concentrations blocks sphingolipid synthesis. The low concentration of myriocin used in our experiments does not limit growth of wild type cells but does limit growth of ypk1⌬ cells, presumable because of a reduced level of LCBs and a consequent reduction in Pkh1/2 activity and activity of downstream kinases including Ypk1, Ypk2, and Pkc1. Restoration of myriocin resistance to ypk1⌬ cells by the T662D but not the T662A allele has recently been reported (34).
How might PHS stimulate Ypk1 and Pkh1 activity? Based upon what is known about mammalian members of the AGC kinase family, it seems that the PDK1 site in the activation loop of Ypk1 would need to be phosphorylated in order for the enzyme to have activity or to have increased activity. Thus, a possible role for PHS in vitro in the absence of Pkh1 would be to stimulate autophosphorylation of Ypk1/2 at the PDK1 site to yield a partially active enzyme (Fig. 3). Stimulation by Pkh1 in the absence of PHS would most likely also result from it phosphorylating the PDK1 site in Ypk1/2, since this is what Casamayor et al. (7) showed previously. The combination of PHS and Pkh1 gives the highest stimulation of Ypk1/2 activity (Fig. 3), and this could be due to phosphorylation of the PDK2 site or some unidentified site, perhaps the "turn motif" lying between the protein kinase and the hydrophobic motifs (53). It is not clear whether Pkh1 or Ypk1 would phosphorylate the PDK2 or the unidentified site. A similar scenario can be envisioned for Sch9. This model is speculative and is meant only as a starting point for further experiments to better understand how PHS stimulates Ypk1 activity.
In vivo most PHS is thought to be present in membranes with the polar NH 3 group on C2 and the OH on C1 exposed to solvent, allowing them to be bound by a protein. In this regard PHS would act in a manner similar to the 3-phosphoinisitides, which serve to tether proteins to a membrane and, in the case of PDK1 and Akt/protein kinase B, to bring the proteins close together. This could be the case for Pkh1/2 and Ypk1/2, although there is no indication that Pkh1/2 reside on membranes (12). One study suggested that Ypk1 bound to the plasma membrane (14), but two others reported a cytosolic location (34,54). A limitation of the localization studies is that the proteins contained an epitope tag and were usually overexpressed, since all of these AGC kinases are present at a low concentration in yeast (55). If PHS does serve to tether Pkh1/2 and/or Ypk1/2 to membranes, it may be a transient localization. Alternatively, it is possible that some fraction of PHS is soluble and not membrane-bound, and it is this fraction that binds to Pkh1, Ypk1/2, and Sch9.
There are reports that mammalian LCBs, particularly the major one sphingosine, regulate the activity of protein kinases. Bokoch et al. (56) showed that sphingosine stimulates autophosphorylation and activation of p21-activated kinase (PAK1) in vitro as well as does the small G-protein Cdc42, a known activator of PAK1. It has also been shown that sphingosine activates PDK1 both in vitro and in vivo (57). Other reports from Hakomori and co-workers (58) demonstrate that a cleavage product of PKC␦, containing the protein kinase domain, is stimulated by sphingosine to phosphorylate 14-3-3 proteins (58). Thus, these reports when combined with our results show that several members of the yeast and mammalian AGC protein kinase family are stimulated by LCBs, and they suggest that LCBs are more widespread regulators of protein kinases than is currently appreciated. The fact that mammalian p21activated kinase, which is a member of the STE kinase family, is stimulated by LCBs suggests that S. cerevisiae may contain non-AGC kinases that are regulated by LCBs.