TEDS Site Phosphorylation of the Yeast Myosins I Is Required for Ligand-induced but Not for Constitutive Endocytosis of the G Protein-coupled Receptor Ste2p*

The yeast myosins I Myo3p and Myo5p have well established functions in the polarization of the actin cytoskeleton and in the endocytic uptake of the G protein-coupled receptor Ste2p. A number of results suggest that phosphorylation of the conserved TEDS serine of the myosin I motor head by the Cdc42p activated p21-activated kinases Ste20p and Cla4p is required for the organization of the actin cytoskeleton. However, the role of this signaling cascade in the endocytic uptake has not been investigated. Interestingly, we find that Myo5p TEDS site phosphorylation is not required for slow, constitutive endocytosis of Ste2p, but it is essential for rapid, ligand-induced internalization of the receptor. Our results strongly suggest that a kinase activates the myosins I to sustain fast endocytic uptake. Surprisingly, however, despite the fact that only p21-activated kinases are known to phosphorylate the conserved TEDS site, we find that these kinases are not essential for ligand-induced internalization of Ste2p. Our observations indicate that a different signaling cascade, involving the yeast homologues of the mammalian PDK1 (3-phosphoinositide-dependent-protein kinase-1), Phk1p and Pkh2p, and serum and glucocorticoid-induced kinase, Ypk1p and Ypk2p, activate Myo3p and Myo5p for their endocytic function.

Myosins I constitute a well characterized and ubiquitous group of unconventional myosins, which participate in a variety of cellular processes, including endocytosis, phagocytosis, cell motility, secretion, and cell polarity (1,2). As other myosins, myosins I bear an N-terminal actin-activated ATPase, which can translocate actin filaments in vitro (3). A short positively charged tail that binds acidic phospholipids and targets the myosin to the appropriate cellular membranes follows the conserved motor domain (1,2,4,5). The fungal and the protozoal myosins I bear an additional C-terminal extension that participates in the activation of Arp2/3-dependent actin polymerization (6 -10). In the yeast Saccharomyces cerevisiae, two highly homologous genes encode myosins I: MYO3 and MYO5. Deletion of either gene does not result in any obvious phenotype, whereas, depending on the strain background, a double knock-out is lethal or very sick, suggesting functional redundancy (11,12). Genetic analysis indicates that the yeast myosins I participate in the polarization of the actin cytoskeleton and in endocytosis. Cortical actin patches, which concentrate in the daughter cells in wild type budding yeast, partially redistribute to mother cells in a double myo3⌬ myo5⌬ null mutant generated in a permissive strain background (11,12). Besides, Myo3p and Myo5p have a well established function in the ligand-induced internalization of the ␣-factor pheromone receptor Ste2p. Ste2p is a G protein-coupled receptor that triggers the mating response in the presence of ␣-factor (13). Binding of the pheromone to its receptor accelerates its internalization and degradation rate as part of a pheromone desensitization program (14). Temperature-sensitive myosin I mutants are blocked in their capacity to internalize radioactively labeled ␣-factor immediately upon a shift to restrictive temperature (11,12). The endocytic ␣-factor uptake requires the most C-terminal domain of the myosins I, which participates in the induction of Arp2/3-dependent actin polymerization (8), and it might also require its motor activity (12).
A number of results suggest that the actin-activated ATPase of the protozoal and the fungal myosins I is induced by phosphorylation of a conserved serine or threonine positioned on a surface loop that contacts the actin filament (15). Phosphorylation of this residue induces a conformational change that accelerates phosphate release during ATP hydrolysis in vitro (16). The unconventional mammalian myosins VI also bear a serine or a threonine at this position, and therefore, they might also be regulated by phosphorylation (2). Other myosins have a glutamate or an aspartate at this site, suggesting that a negative charge is required there for full myosin ATPase activity (17). Because only threonine, glutamate, aspartate, or serine is present at this position in the myosin motor domain, this site is named the TEDS site (17).
A number of results illustrate the physiological relevance of the myosin I TEDS site phosphorylation. In vivo phosphorylation of this residue has been demonstrated for the protozoal myosins I (18,19), and mutation of the TEDS site to alanine of the Dictyostelium myoB results in a protein unable to complement a double myoA Ϫ /myoB Ϫ null mutant (20,21). Consistently, the yeast MYO3 TEDS site serine to alanine mutant (myo3-S357A) fails to complement the synthetic lethal phenotype of a double myo3⌬ myo5⌬ null strain (20,21). Interestingly, however, not all fungal myosin I functions require phosphorylation of this residue and/or full ATPase activity, since a TEDS site to alanine mutation in the Aspergillus nidulans MYOA only causes slight defects when compared with the null mutation (22,23).
Purification of the protozoal myosin I TEDS site kinase identified a member of the p21-activated kinase (PAK) 6 family (24,25). PAKs are ubiquitous kinases activated by acidic phospholipids and by the small Rho-like GTPases Cdc42 and Rac (24 -30). Among many cellular tasks, PAKs participate in the organization of the actin cytoskeleton and in the development of cell polarity (31). Three genes encode PAKs expressed during vegetative growth in yeast: STE20, CLA4, and SKM1. Deletion of STE20 and CLA4 within the same cell is synthetically lethal, suggesting that they share an essential function (28,(32)(33)(34)(35). The cellular role of the third PAK Skm1p is still unknown (36). Numerous results demonstrate that the yeast myosins I are targets of the Cdc42p-activated PAKs controlling actin assembly and polarization (see Ref. 37 and references therein). Cla4p and Ste20p phosphorylate the Myo3p TEDS site in vitro (21), and, most strikingly, a Myo3p bearing a TEDS serine to aspartate substitution (Myo3-S357Dp) suppresses the actin depolarization defects of a cdc42-1 mutant (9).
In the present work, we have investigated the role of the yeast myosin I TEDS site phosphorylation in the endocytic uptake. Interestingly, we find that myosin I phosphorylation is required for fast, ligand-induced internalization of Ste2p but not for slow, constitutive endocytosis of the receptor. Further, we show that a Myo5p mutant that mimics the TEDS site phosphorylated state significantly accelerates constitutive Ste2p internalization. Our results suggest that activation of the myosin I ATPase can control the endocytic uptake rate in yeast and that a kinase or kinases activate the myosins I to sustain fast internalization. Surprisingly, although only PAKs are known to phosphorylate the conserved myosin TEDS site, we found that sustained activity of Cdc42p and the yeast PAKs Ste20p, Cla4p, and Skm1p is dispensable for ␣-factor-induced Ste2p internalization. Our results suggest that a different signal-ing cascade that involves the yeast PDK1 (3-phosphoinositide-dependent protein kinase-1) and serum and glucocorticoid-induced kinase (SGK) homologues (Pkh1/2p and Ypk1/2p, respectively) activates Myo3p and Myo5p for their endocytic function.

MATERIALS AND METHODS
Strains, Genetic Techniques, and Two-hybrid Analysis-Yeast strains used in this study are listed in Table 1. Unless otherwise mentioned, strains without plasmids were grown in complete YPD medium and strains with plasmids were selected on SDC minimal medium (14). Sporulation, tetrad dissection, and scoring of genetic markers were performed as described (38). Transformation of yeast was accomplished by the lithium acetate method (39). Detailed information about yeast strain generation is available by request. Dot spots were prepared from fresh saturated cultures. Cells were diluted to 10 7 cells/ml, and 5 l of 4 ϫ 1-10 serial dilutions were spotted on plates with the adequate solid medium. Cells were grown for 2 days at the indicated temperature. The interaction trap two-hybrid system was used (40). Plasmids pEG202, pJG4-5, pRFHM-1, and pSH18-34 and the strain EGY48 were obtained from R. Brent (Harvard Medical School, Boston, MA). To measure ␤-galactosidase activity, EGY48 bearing the lexAop-lacZ reporter plasmid pSH18-34 was co-transformed with the appropriate pEG202-and pJG4-5-derived plasmids and streaked out on X-gal-containing SGCϪHisϪTrpϪUra plates. Pictures were taken after 2-3 days at 30°C.
DNA Techniques and Plasmid Construction-Plasmids used in this study and their relevant features are listed in Table 2. DNA manipulations were performed as described (41). Details for plasmid construction are available under request.
Ste2p Internalization Assays and Carboxypeptidase Y (CPY) Pulse and Chase- 35 S-␣-factor uptake assays to analyze ligand-induced internalization of Ste2p were performed as described (14). For experiments with temperature-sensitive mutants, cells were preincubated in 37°C prewarmed YPD for 30 min before the addition of 35 S-␣-factor. For the experiments with the analogue-sensitive mutant, cells were preincu-  (42)) or mock-treated. Uptake assays were performed at least three times, and the mean and S.D. values were calculated per time point. The S.D. values were less than 10% of the value. For the analysis of constitutive endocytosis, surface-exposed Ste2p was measured with 35 S-radiolabeled ␣-factor as described (14). CPY pulse and chase was performed as described (43) using 35 S-labeling mix (Tran 35 S-label; ICN Biomedicals) and a polyclonal antibody against CPY (12) for immunoprecipitation. In Vivo 32 PO 4 3Ϫ Labeling, in Vitro Kinase Assays, Protein Extracts, Immunoprecipitations, and Immunoblots-To determine protein expression, cells grown to log phase were harvested and glass bead-lysed in IP buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA) containing protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml pepstatin A, 1 g/ml leupeptin, 1 g/ml antipain). Protein concentration was determined with a Bio-Rad protein assay. Immunoblots were performed as described (44). For in vivo 32 PO 4 3Ϫ labeling of yeast, 10 8 log phase cells were harvested, resuspended in low phosphate SD medium (50 M KH 2 PO 4 ) and incubated at 23°C for 5 h. Cells were harvested and incubated for 30 min at room temperature in 1 ml of low phosphate SD medium containing 1 mCi/ml radioactive 32 PO 4 3Ϫ (ICN Biomedicals, Irvine, CA). Cells were harvested, and glass bead-lysed in 100 l of IP buffer containing protease inhibitors and phosphatase inhibitors (10 mM sodium pyrophosphate, 10 mM NaN 3 , 10 mM NaF, 0.4 mM EDTA, 0.4 mM NaVO 3 , 0.4 mM Na 3 VO 4 , 2 M cyclosporin A, 0.5 M okadaic acid). After the addition of Triton X-100 to 1%, unbroken cells and debris were removed, and Myo5p was immunoprecipitated using an anti-Myo5p polyclonal serum (44) prebound to Protein A-Sepharose (Amersham Biosciences) or IgG-Sepharose (Amersham Biosciences). Beads were recovered, and associated proteins were separated using a 10% SDS-polyacrylamide gel. Proteins were transferred to a nitrocellulose filter and analyzed by autoradiography and immunoblot using a rat monoclonal anti-HA antibody (3F10; Chemicon Hofheim).
For in vitro kinase assays, a Myo5p fragment containing the wild type or the S357A mutated TEDS site fused to GST was purified from Escherichia coli BL21 bearing pGST-myo5-TEDS and pGST-myo5-TEDS-SA as described (6), except that 50 mM Tris/HCl, pH 7.5, 120 mM NaCl, 2 mM EDTA, 1% Triton X-100 was used for solubilization. 10 g of total protein was precipitated with 40 l of 50% glutathione-Sepharose beads (v/v) (Amersham Biosciences). Beads were washed with 50 mM Tris/HCl, pH 7.5, 150 mM NaCl buffer and resuspended in 30 l of phosphorylation buffer (150 mM Tris-HCl, pH 8, 150 mM NaCl, 20 mM glutathione, 1 mM dithiothreitol, and 10 mM MgCl 2 ). Purified GST-Ypk1p or wild type or mutant GST-Ypk2p (100 ng (for experiments with [␥-32 P]ATP) or 300 ng (for experiments with [␥-33 P]ATP)) were added to 10 l of GST-myo5-TEDS-or GST-myo5-TEDS-SA-coated beads in a final 20-l volume in the presence of 1 mM ATP and 4 Ci of [␥-32 P]ATP or [␥-33 P]ATP. After a 30-min incubation at 30°C, the reaction was stopped by the addition of SDS sample buffer to a final concentration of 0.8% and ATP to a final concentration of 10 mM. The resulting samples were resolved on 10% SDS-polyacrylamide gels. Coomassie Blue-stained gels were analyzed by autoradiography. GST-Ypk1p, GST-

TABLE 2 Plasmids
All plasmids listed in this table carry an ori and an Amp R and bear a yeast centromeric origin of replication except for the pJG4-5 and pEG202 series that bear a yeast 2 origin of replication and the pGST series that do not bear a yeast origin of replication. aa, amino acids.

Plasmid
Yeast marker Insert Source/Reference This study Ypk2p, GST-Ypk2-H459Yp, and GST-Ypk2-K373Ap were purified from yeast YC123 as described (45,46). Fluorescence Microscopy Techniques-Phalloidin staining was performed on 4% formaldehyde-fixed cells in the presence 300 nM TRITCphalloidin (Sigma) as described (47). Samples were visualized using a Zeiss Axiovert 35 fluorescence microscope. For the experiments with the analogue-sensitive mutants, cells were incubated for 1 h in the presence of 100 M 1NM-PP1 or mock-treated before fixation. The percentage of unpolarized cells was calculated by counting the number of small budded cells exhibiting more than two actin patches in the mother cell. More than 100 cells were examined per experiment. For staining of cells with FM4-64, SCMIG275 transformed with plasmids encoding the green fluorescent protein-tagged wild type and mutant Myo5p were grown to 10 7 cells/ml, harvested, resuspended in 25 l of YPD, and incubated on ice for 15 min. FM4-64 (Molecular Probes, Inc., Eugene, OR) was added to 8 M, and cells were further incubated on ice for 15 min. Cells were harvested at 4°C and resuspended in 50 l of ice-cold 2% alginate in 50 mM glycine, pH 6.2, containing 8 M FM4 -64. 2 l of sample were placed on an ice-cold slide and carefully covered with a coverslip. 2 l of 50 mM CaCl 2 was added to the sample from every side of the coverslip to solidify the alginate. Cells were immediately visualized using a Leica TCS SP confocal microscope equipped with Ne-He and argon lasers.

ProtA-Myo5p Purification and MALDI-TOF Analysis of TEDS Site Phosphorylation-N-terminal
Protein A-tagged Myo5p was purified from SCMIG275 transformed with p33ProtAMYO5. 10 11 yeast log phase cells were harvested and resuspended in 100 l of IP buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA) containing protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1 g/ml bestatin, 1 g/ml leupeptin, 1 g/ml antipain) and glass bead-lysed. 1 ml of IP, 1% Triton X-100 (IPT) containing protease inhibitors was added/g of wet pellet, mixed, and incubated for 10 min on ice. Unbroken cells and debris were eliminated by centrifugation at 3,600 rpm at 4°C for 10 min and twice at 10,000 rpm at 4°C for 10 min. The supernatant was recovered and incubated for 2 h at 4°C in the presence of IgG-Sepharose (Amersham Biosciences). Beads were washed with IPT and IP, and Myo5p was released with 10 units/ml TEV protease (Invitrogen) in 50 mM Tris-HCl, pH 8, 0.5 mM EDTA for 2 h at 16°C. Myo5p was eluted from the beads in the presence of 0.5 M NaCl. Coomassie Blue staining of an SDS-polyacrylamide gel of IgG-Sepharose precipitates from yeast expressing ProtA-Myo5p demonstrated the presence of two polypeptides of ϳ130 and 150 kDa (Fig. 1C) that could be decorated with rabbit PAP (anti-peroxidase antibodies raised in rabbits with horseradish peroxidase; DAKO) (data not shown). TEV protease digestion of the beads released 132-and 116-kDa polypeptides (Fig. 1C) that most likely correspond to the full-length Myo5p and a truncated form lacking a 16-kDa C-terminal polypeptide (8). Consistently, the upper but not the lower band was recognized by an anti-Myo5p polyclonal antibody raised against the C-terminal portion of the protein (data not shown) (44). For MALDI-TOF analysis, purified protein was separated in a 10% SDSpolyacrylamide gel, and ϳ5 g of ProtA-Myo5p was sliced from the gel. Protein bands were digested in gel with trypsin (Promega), and phosphopeptides were enriched using Ga 3ϩ IMAC minispin columns (Pierce). Peptides from in-gel digests were adjusted to 10% acetic acid and incubated with the beads of a minicolumn for 10 min at room temperature. The columns were washed with 0.1% acetic acid, pH 3, and 0.1% acetic acid, 10% acetonitrile, pH 3, to remove nonspecifically bound peptides. Bound peptides were eluted with 150 mM ammonium bicarbonate, pH 9. For MALDI-TOF analysis (48), the IMAC-enriched samples were eluted from C18 ZipTips (Millipore) directly onto the sample plate with 1 l of 5 mg/ml ␣-cyano-4-hydroxycinnamic acid in 50% acetonitrile, 3% formic acid and were then overlaid with 1 l of 25 mM ammonium citrate solution to enhance the ionization efficiency of phosphopeptides. Positive ion mass spectra were acquired on a Reflex III (Bruker Daltonik) for peptide mass determination or on an Ultraflex TOF/TOF (Bruker Daltonik) for peptide fragmentation analysis by PSD/LIFT.

RESULTS
The Myo5p TEDS Site Is Phosphorylated in Vivo-Despite all data supporting the physiological relevance of the myosins I TEDS site phosphorylation, phosphorylation at this position in vivo has only been demonstrated for the protozoal myosins I (18,19). Since some of the fungal myosins I do not require phosphorylation to fulfill their cellular tasks (22,23), we decided to first investigate if the S. cerevisiae Myo5p TEDS site was phosphorylated in vivo. In order to address this matter, the wild type Myo5p or a mutant bearing a TEDS serine to alanine substitution (Myo5-S357Ap) was immunoprecipitated from 32 PO 4 3Ϫ -radiolabeled yeast cells. An equivalent phosphorylation signal could be detected for samples expressing the wild type and mutant Myo5p. No signal could be detected when the anti-Myo5p antibody was omitted in the immunoprecipitation or when the proteins were immunoprecipitated from cells lacking MYO5 (Fig. 1A). Interestingly, C-terminal truncation of Myo5p resulted in loss of the radioactive signal (Fig. 1B), suggesting that if the Myo5p TEDS site was phosphorylated, phosphorylation requires an intact Myo5p tail.
To further investigate if the Myo5p TEDS site was phosphorylated in vivo, we purified it from yeast expressing a Protein A N-terminal Phosphorylation of the Yeast Myosin I TEDS Site Is Required for the Organization of the Actin Cytoskeleton-Experimental evidence from Wu and co-workers (21) demonstrated that mutation of the Myo3p TEDS serine to alanine results in a protein (Myo3-S357Ap) that cannot complement the lethality of a double myo3⌬ myo5⌬ knock-out, indicating that phosphorylation of the TEDS site might be required for at least the essential myosin I function in yeast. However, instability or mislocalization of the protein might have caused the loss of function in these experiments. On the other hand, the contribution of the TEDS site phosphorylation to any particular myosin I function could not be examined under these circumstances. Interestingly, we found that the analogous mutation in MYO5 (MYO5-S357A) could indeed complement the lethality of the myo3⌬ myo5⌬ double knock ( Fig. 2A), although it caused a temperature-sensitive growth defect (Fig. 2B). This result is consistent with our previous observations suggesting that Myo5p plays a predominant role in the cell when compared with Myo3p (12). No growth defect could be detected when the serine 357 was mutated to a negative charged amino acid (myo5-S357E) (Fig. 2, A and B). This result demonstrated that the unphosphorylated myosin I is sufficient to sustain growth, at least under optimal conditions. To investigate if TEDS site phosphorylation was required for the polarization of the actin cytoskeleton, myo3⌬ myo5-S357A cells were grown at permissive temperature (23°C), fixed, and stained with TRITC-phalloidin to visualize filamentous actin. Almost 80% of the myo3⌬ myo5-S357A small budded yeast cells exhibited a clearly depolarized actin cytoskeleton with more than two actin patches present at the mother cells (Fig. 2, C and D). Under the same conditions, only 1% of small budded yeast expressing the wild type Myo5p presented a depolarized cytoskeleton (Fig. 2, C and D). These results suggested that phosphorylation of the myosin I TEDS site is required to sustain a properly polarized actin cytoskeleton in yeast. Interestingly, nearly 30% of the small budded myo3⌬ myo5-S357E cells also exhibited a depolarized cytoskeleton (Fig. 2, C and D), suggesting that dephosphorylation of the TEDS site is also required for an accurate cytoskeletal organization.
Phosphorylation of the Myosin I TEDS Site Is Required for Ligandinduced but Not for Constitutive Internalization of Ste2p-To investigate if the myosin I endocytic function also requires TEDS site phos-phorylation, we examined the kinetics of constitutive and ligandinduced endocytosis of Ste2p (14) in cells expressing the Myo5p TEDS site mutants as the only source of myosin I (myo3⌬ myo5-S357E and myo3⌬ myo5-S357A). As shown in Fig. 3A, the myo3⌬ myo5-S357A strain exhibited a strong defect in its capacity to internalize Ste2p in the presence of ␣-factor when compared with cells expressing the wild type Myo5p. Strikingly, however, constitutive internalization of the receptor appeared unaffected in these cells (Fig. 3B). The prominent ␣-factor uptake defect installed in the myo3⌬ myo5-S357A strain was not the result of cell sickness, since biosynthetic transport and maturation of the vacuolar protease carboxypeptidase Y (43) was unaffected (Fig. 3C). Further, immunoblot analysis of the wild type and the mutant Myo5p (Fig. 4A) and confocal fluorescence microscopy of strains expressing green fluorescent protein-tagged versions of these proteins demon- strated that the observed phenotypes were not due to instability or mislocalization (Fig. 4B). These data indicated that TEDS site phosphorylation of the myosin I is required in yeast for fast, ligand-induced internalization of Ste2p but not for its slow, constitutive uptake. Interestingly, we also found that cells expressing the Myo5p mutant mimicking the phosphorylated state (myo3⌬ myo5-S357E) displayed a slight but significant acceleration of the constitutive Ste2p uptake kinetics when compared with cells expressing the wild type Myo5p (Fig. 3B). This result suggested that phosphorylation of the myosin I TEDS site can control the endocytic uptake rate, possibly through activation of the myosin I ATPase. No acceleration could be detected, however, when the internalization assay was done in the presence of the pheromone (Fig.  3A). Other endocytic factors might limit the fast ligand-induced Ste2p internalization rate. Actually, although the initial uptake kinetics of the myo3⌬ myo5-S357E strain was nearly identical to that of the wild type, mutant cells only managed to internalize 70% of the total cell-bound ␣-factor as opposed to 100% for cells expressing wild type Myo5p (Fig.  3A). Depletion of the available endocytic machinery in cells with constitutively accelerated endocytosis or excessive recycling of Ste2p to the plasma membrane could account for the observed phenotype in the myo3⌬ myo5-S357E cells (49).

Sustained Cdc42p Signaling through the PAKs Is Not Essential for Ligand-induced Ste2p
Internalization-Our data demonstrated that TEDS site phosphorylation was necessary not only for the polarization of the actin cytoskeleton but also for ligand-induced Ste2p internalization. Since only PAKs are known to phosphorylate the conserved myo-  sin TEDS site, we decided to investigate if the Cdc42p/PAK/myosin I signaling pathway, which is required in yeast to polarize the actin cytoskeleton, also functioned to activate Myo3p and Myo5p for their endocytic function. For that purpose, we first created a temperaturesensitive kinase-dead PAK mutant by deleting the chromosomal genes encoding SKM1 and STE20 and by introducing a H685Y mutation in a highly conserved residue of the CLA4 core kinase domain (50). As expected, the cla4-H685Y ste20⌬ skm1⌬ strain (pak-ts) was temperature-sensitive for growth (Fig. 5B). At permissive temperature (24°C), the growth rate of pak-ts cells was comparable with the wild type (Fig.  5B), although some of the mutant cells presented elongated buds similar to those described in the cla4-75 ste20⌬ mutant (Fig. 5C) (33). Upon a 30-min shift to 37°C, pak-ts cells stopped dividing and adopted a round morphology analogous to that described for other cla4 ste20 mutants at this temperature (Fig. 5C) (35,42). Also, consistent with previous publications (35,42), our pak-ts mutant was unable to repolarize its actin cytoskeleton at elevated temperature (Fig. 5, D and E). In wild type yeast, heat stress causes rapid depolarization of the actin cytoskeleton, which only recovers 2-3 h after prolonged incubation at elevated temperature (51). Our results demonstrated that a temperature shift to 37°C tightly inactivated the Cla4-H685Yp in vivo.
To investigate if PAKs were required to sustain ligand-induced Ste2p internalization, we performed an ␣-factor uptake assay using the cla4-H685Y ste20⌬ skm1⌬ mutant upon a shift to restrictive temperature. Surprisingly, upon 30-min preincubation at 37°C, the ␣-factor internalization kinetics of the mutant cells was identical to that of the wild type (Fig. 5F). In addition, we failed to detect any ␣-factor uptake defect in yeast strains bearing four different cdc42 mutations (cdc42-1, cdc42-118, cdc42-123, and cdc42-129) (Fig. 6) that are known to specifically interfere with Cdc42p signaling to the PAKs (52). Most strikingly, the cdc42-1 mutation that specifically hinders the signaling pathway to the myosins I (9) did not alter the ␣-factor uptake kinetics at restrictive temperature. Additional preincubation at restrictive temperature of the pak-ts and cdc42 mutants resulted in very low ␣-factor binding to the cells, maybe reflecting a defect in Ste2p traffic to the plasma membrane (53,54).
To further inspect the endocytic uptake phenotype of a conditional pak mutant under the very same experimental circumstances that clearly installed an actin polarity defect, we decided to use a chemically sensitive CLA4 allele, cla4-as3 (42). The Cla4-as3p bears two mutations that enlarge the ATP binding pocket of the kinase without significantly affecting its activity in vivo (42). These mutations allow binding of the membrane-permeable pyrimidine analogue 1NM-PP1, which can then efficiently and specifically inactivate the Cla4-as3p kinase (42). It was previously shown that the in vivo activity of Cla4-as3p is inhibited upon 1-h incubation in the presence of 25 M 1NM-PP1 (42). Under the same conditions, no uptake defect was observed in the ste20⌬ skm1⌬ cla4-as3 (pak-as3) mutant when compared with a wild type or an isogenic ste20⌬ skm1⌬ CLA4 strain (data not shown). Further, increasing the concentration of 1NM-PP1 to 100 M did not alter the internalization kinetics of the conditional pak-as3 mutant (Fig. 7A). In contrast to the uptake kinetics, the actin distribution in the 100 M 1NM-PP1-treated pak-as3 cells was similar to that observed in the myo3⌬ myo5S357A strain  (pak-ts) (SCMIG574) strain or the isogenic wild type (SCMIG50). C, Nomarski optics micrographs of the strains described in B grown at 23°C to logarithmic phase and incubated for 30 min at the indicated temperatures. Bar, 1 m. D, fluorescence micrographs showing the strains described in B, fixed and stained with TRITC-phalloidin to visualize filamentous actin. Cells were grown at restrictive temperature for 3 h prior to fixation. Bar, 1 m. E, graph showing the percentage of budded cells exhibiting more than two actin patches per mother cell for the experiment described in D. F, ligand-induced Ste2p internalization kinetics of the strain described in B at the restrictive temperature (37°C). Cells were preincubated for 30 min at 37°C prior to the addition of 35 S-labeled ␣-factor. The graphs plot percentage of internalized versus cell-associated counts.
(Figs. 2C and 7B). Upon 1-h incubation in the presence of the drug, pak-as3 cells appeared big and round, and more than 80% of the small budded yeast exhibited a depolarized actin cytoskeleton (Fig. 7, B and C). This terminal phenotype was strikingly different from that previously described for the ste20⌬ cla4-as3 double mutant (42). Deletion of SKM1 and/or treatment of the cells with a higher 1NM-PP1 concentration may account for the observed differences. The actin and morphology defects installed in the pak-as3 cells were dependent on the presence of the cla4-as3 mutation and the drug, since they did not appear in the mock-treated pak-as3 strain or in the 1NM-PP1-treated ste20⌬ skm1⌬ CLA4 cells (Fig. 7, B and C).
The data strongly indicated that ligand-induced endocytosis of Ste2p is not abolished under conditions that tightly block the Cdc42p-activated PAK signaling pathway, and the data suggested that kinases other than PAKs can also activate the myosins I for their endocytic function.
A Signaling Cascade Involving the Yeast Homologues of the Mammalian PDK1 and SGK Might Activate the Myosins I for Their Endocytic Function-Searching for proteins required for ligand-induced Ste2p internalization that might physically and genetically interact with the myosins I, we found the essential and functionally redundant Pkh1p and Pkh2p, the yeast homologues of the mammalian PDK1 (55). Sphingoid base-activated Pkh1p and Pkh2p are thought to initiate a signaling cascade that modulates the endocytic uptake in yeast (55). Interestingly, we found that deletion of the chromosomal copy of MYO5 caused a synthetic growth defect in the temperature-sensitive pkh1-ts pkh2⌬ strain (55) (Fig. 8A). The synthetic growth defect could not be observed when MYO5 was deleted in other mutant backgrounds with similar growth and uptake defects (8), suggesting an intimate functional interaction between the myosins I and the yeast PDK1 homologues. In agreement with this view, we also observed a physical interaction between Pkh1p and Pkh2p and the Myo5p head and tail domains in a two-hybrid assay (Fig. 8B). These interactions seemed to be specific, since we could not observe significant transcriptional activation of the ␤-galactosidase reporter when we tested interactions between the Myo5p fragments and other kinases involved in endocytosis (Prk1p and Yck1p (56, 57)) (Fig. 8B).
The yeast Pkh1p and Pkh2p and the mammalian PDK1 are known to activate a number of downstream kinases that bear a consensus motif (Thr-Phe-Cys-Gly-Thr-X-Glu-Tyr, where Thr is the phosphorylable residue and X is any amino acid). Such kinases include members of the protein kinase C, protein kinase A, and SGK families (58 -61). Pkh1p and Pkh2p are also able to phosphorylate in vitro the yeast Pkc1p (protein kinase C) and the functionally redundant highly similar SGK homologues Ypk1p and Ypk2p (Fig. 8C) (55,(62)(63)(64). The Myo3p and Myo5p TEDS site does not match the PDK1 consensus motif. Therefore, we suspected that these molecular motors might not be direct targets of Pkh1p and Pkh2p. These kinases might rather work to recruit and to activate the actual TEDS site kinase. To identify kinases that might mediate between Pkh1/2p and the myosins I, we searched for synthetic interactions between the myo5⌬ null mutation and mutations in genes encoding the known Pkh1/2p targets Pkc1p, Ypk1p, and Ypk2p (pkc1-ts (55,63,64), ypk1⌬, and ypk2⌬). Interestingly, although we failed to observe synthetic growth defects in the double mutants, we detected a strong and a mild synthetic ␣-factor uptake defect when MYO5 was deleted in the ypk2⌬ and ypk1⌬ backgrounds, respectively. The synthetic uptake defect observed in the ypk2 myo5 mutants could still be observed when the myo5⌬ null mutation was substituted with the myo5-S357A point mutation, thus suggesting that the lack of Ypk2p interfered with the Myo5p ATPase function. Only additive uptake  defects were observed when the myo5⌬ mutation was combined with a pkc1-ts mutation or with a partial loss of function mutation in another gene required for early events in the endocytic uptake (sla2-n) (Fig. 9). These results suggested that the yeast homologues of the mammalian SGK and Myo5p functionally cooperate in the endocytic uptake of Ste2p. Consistent with this view, we found that Ypk2p and, to a lesser extent, Ypk1p interacted with Myo5p in the two-hybrid assay (Fig. 10A). No significant activation of the reporter was triggered when Pkc1p was used in the assay (Fig. 10A).
To test if Ypk2p or Ypk1p directly and specifically phosphorylate the Myo5p TEDS site, an in vitro kinase assay was performed using purified components. For that purpose, a recombinant purified Myo5p fragment fused to GST bearing the TEDS site (amino acids 322-391) (GST-TEDS) was incubated in the presence of [␥-32 P]ATP and purified Ypk2p or Ypk1p fused to GST. In the presence of Ypk2p, the TEDS construct was covalently labeled with radioactive phosphate (Fig. 10B). Phosphorylation occurred in the TEDS serine, since mutation of the Myo5p Ser 357 to alanine or glutamic acid completely abrogated transfer of the phosphate to the TEDS construct (Fig.  10B). Further, MALDI-TOF signals that correspond to masses of phosphorylated GSVYHVPLNIVQADAVR (1917,64) and RGSVYH-VPLNIVQADAVR (2073.72), could be detected in samples treated with GST-Ypk2p but not in untreated samples (Fig. 10C). An extra band probably corresponding to the autophosphorylated Ypk2p could also be observed on the autoradiography of all samples containing this kinase. Neither autophosphorylation nor TEDS phos-phorylation could be detected in the samples containing purified Ypk1p (Fig. 10B), suggesting that this kinase might not be active under the conditions assayed. TEDS site phosphorylation in the presence of purified GST-Ypk2p required an active enzyme, since mutation of conserved residues in the kinase domain (His 459 and Lys 373 ) (50,65) completely abolished the phosphorylation of the Myo5p TEDS site (Fig. 10B).
All of these data strongly indicated that kinases other than PAKs can also phosphorylate the conserved myosin I TEDS site. Further, the results suggested that the yeast myosins I are targets of the Pkh1/2p and Ypk1/2p signaling cascade that modulates endocytosis in yeast.

TEDS Site Phosphorylation of the Myosins I and the Control of the Endocytic Uptake Rate in Yeast-
Our results demonstrated that a mutant Myo5p mimicking the constitutively TEDS site-dephosphorylated isoform (Myo5-S357Ap) was unable to rescue the actin polarization defects of a myo3⌬ myo5⌬ double knock-out. In addition, the mutant failed to sustain fast, ligand-induced internalization of the ␣-factor receptor Ste2p. Since the mutant was stable and it was properly localized, the data indicated that TEDS site phosphorylation is required for at least two myosin I functions in budding yeast. These observations are in striking agreement with the results from Novak and Titus (20) showing that substitution of the Dictyostelium discoideum myoB TEDS FIGURE 8. Myo5p physically and genetically interacts with the yeast homologues of the mammalian PDK1, Pkh1p, and Pkh2p. A, deletion of MYO5 (myo5⌬) causes a synthetic growth defect in a temperature-sensitive pkh1-ts pkh2⌬ background. Dot spots grown at the indicated temperatures of a myo5⌬ pkh1-ts pkh2⌬ strain (SCMIG605) bearing either an empty plasmid (Ycplac33) (myo5⌬) or the same plasmid encoding the wild type MYO5 (p33MYO5) (MYO5). B, two hybrid assays on X-gal-containing plates to test for physical interaction between the Myo5p head (H) (pEG202myo5-H), the Myo5p tail (T) (pEG202myo5-T), or an unrelated control protein (C) (pEG202BICOID) as baits and the indicated yeast kinases cloned in pJG4-5 as preys. The empty pJG4-5 was used as control. C, known targets of the yeast homologues of the mammalian PDK1, Pkh1p and Pkh1p, include the yeast Pkc1p and the yeast homologues of the mammalian serum and SGK Ypk1p and Ypk2p. FIGURE 9. Deletion of MYO5 (myo5⌬) causes a strong synthetic ␣-factor uptake defect in a null ypk2⌬ mutant. Shown is 35 S-labeled ␣-factor-induced Ste2p internalization kinetics of myo5⌬ (SCMIG275), pkc1-ts myo5⌬ (SCMIG617), ypk1⌬ myo5⌬ (SCMIG699), ypk2⌬ myo5⌬ (SCMIG700), and sla2-n myo5⌬ (SCMIG546) strains bearing either an empty plasmid (Ycplac33) or the same strains bearing the plasmid encoding the wild type MYO5 (p33MYO5) (wt, pkc1-ts, ypk1⌬, ypk2⌬, and sla2⌬, respectively) at 30°C. The graphs plot percentage of internalized versus cell-associated counts.
serine by alanine generates a protein that is properly localized but cannot complement the endocytic and actin organization defects of the myoA Ϫ /myoB Ϫ double mutant. However, in agreement with what was demonstrated for the A. nidulans MYOA (22,23), we also found that phosphorylation of the myosins I was not essential to sustain slow constitutive endocytosis. Also in agreement with the results in A. nidulans (66), we found that the basal ATPase activity of the unphosphorylated myosin I might be rate-limiting in this process, since mutants that mimicked the phosphorylated state significantly accelerate constitutive endocytosis. Since Myo5p TEDS site phosphorylation is essential for rapid ␣-factor-induced internalization of Ste2p, our results suggest that fungi can use TEDS site phosphorylation and dephosphorylation to control the endocytic uptake rate in response to intracellular or extracellular signals.
Two simple models might explain the observation that myosin I TEDS site phosphorylation is required for ligand-induced but not for constitutive internalization of Ste2p. In the first model, a slow and a fast endocytic pathway would co-exist, only the last one being sustained by the phosphorylated myosins I. Ligand binding might then simply target Ste2p to the fast endocytic pathway. A second model to explain the observed results would propose that certain intracellular or extracellular signals might be able to modulate the endocytic uptake rate by inducing signaling cascades that phosphorylate and activate Myo3p and Myo5p. Interestingly, modulation of TEDS site phosphorylation by extracellular signals have been described in D. discoideum. In this organism, treatment with the chemoattractant cAMP causes a transient 3-fold increase in the amount of myoB phosphorylated at the TEDS site (19). Analysis of this possibility will now require careful inspection of the Myo5p TEDS site phosphorylation state under conditions that induce endocytosis. Unfortunately, this is not a trivial experiment, given that, in contrast to the D. discoideum myoB (19), Myo5p is phosphorylated at other sites besides serine 357 (Fig. 1), and the expected increase in the amount of the phosphorylated isoform might be subtle and transient.
Different Signaling Pathways Lead to Phosphorylation and Activation of the Yeast Myosins I for Distinct Cellular Functions-Mutations in genes encoding myosins I cause multiple defects in eukaryotic cells. Although the primary cause leading to such phenotypes is not always fully understood, it is reasonable to think that myosins I participate in different cellular tasks, which might be regulated by distinct intracellular and/or extracellular signals. To date, however, only PAKs have been identified as TEDS site kinases, and only the Cdc42p/PAK/myosins I signaling pathway required for the polarization of the actin cytoskeleton in yeast was described (see Introduction). Reinforcing the physiological relevance of this signaling cascade, we showed that Myo5p was phosphorylated in vivo and that a MYO5 TEDS site serine to alanine mutant (myo5-S357A) failed to complement the actin polarization defect of a myosin I null strain. On the other hand, our results clearly showed that even though myosin I TEDS site phosphorylation was required for ligand-induced Ste2p internalization, tight in vivo inactivation of the Cdc42p/PAKs signaling pathway did not result in any significant uptake defect, not even under experimental conditions that clearly installed severe actin defects. These data demonstrated that sustained Cdc42p/ PAK activity is not essential to support fast endocytic uptake in yeast, and they suggested that a different signaling cascade can activate Myo3p and Myo5p for their endocytic function.
Consistent with this view, we found that Myo5p physically and functionally interacts with different components of a signaling pathway that is known to control endocytosis in yeast (55,63): the yeast homologues of the mammalian PDK1, Pkh1p, and Pkh2p and their downstream targets Ypk2p and Ypk1p. Further, we demonstrated that purified Ypk2p specifically phosphorylates the Myo5p TEDS site in vitro.
The mammalian PDK1 stands at a pivotal point in cell signaling, and it mediates a multitude of cellular responses following extracellular stimulation by peptide growth factors and hormones. Besides cell death and survival, deregulation of PDK1 influences a wide variety of processes, including cell growth, motility, and differentiation (67). In yeast, Pkh1/2p have been shown to influence cell wall biogenesis (64) and endocytosis (55,63). Our data provide now evidence for the first molecular mechanism whereby Pkh1/2p and Ypk1/2p can modulate the endocytic uptake rate by controlling the myosin I TEDS site phosphorylation state and, thereby, the myosin I ATPase activity. Nevertheless, it should be mentioned that myosins I are probably not the only Ypk1/2p targets required for the endocytic uptake, since expression of the Myo5-S357Ep mutant, which mimics the constitutively phosphorylated state, failed to suppress the uptake defects of ypk1/2 mutants (data not shown).
Taken together, we provide for the first time robust evidence suggesting that kinases other than PAKs can also phosphorylate the conserved TEDS site of the unconventional myosins and different signaling cascades can activate the myosins I to fulfill distinct cellular tasks.