Phosphorylation of yeast plasma membrane H+-ATPase by casein kinase I.

The plasma membrane H+-ATPase of Saccharomyces cerevisiae is subject to phosphorylation by a casein kinase I activity in vitro. We show this casein kinase I activity to result from the combined function of YCK1 and YCK2, two highly similar and plasma membrane-associated casein kinase I homologues. First, H+-ATPase phosphorylation is severely impaired in the plasma membrane of YCK-deficient yeast strains. Furthermore, the wild-type level of the phosphoprotein is restored by the addition of purified mammalian casein kinase I to the mutant membranes. We used the H+-ATPase as well as a synthetic peptide substrate that contains a phosphorylation site for casein kinase I to compare kinase activity in membranes prepared from yeast cells grown in the presence or absence of glucose. The addition of glucose results in increased H+-ATPase activity which is associated with a decline in the phosphorylation level of the enzyme. Mutations in both YCK1 and YCK2 affect this regulation, suggesting that H+-ATPase activity is modulated by glucose via a combination of a “down-regulating” casein kinase I activity and another, yet uncharacterized, “up-regulating” kinase activity. Biochemical mapping of phosphorylated H+-ATPase identifies a major phosphopeptide that contains a consensus phosphorylation site (Ser-507) for casein kinase I. Site-directed mutagenesis of this consensus sequence indicates that Glu-504 is important for glucose-induced decrease in the apparent Km for ATP.

When glucose is added to starved yeast cells, there is a transient rise in cAMP which induces a protein phosphorylation cascade (1). Among the many metabolic changes observed within minutes after glucose addition is an increased plasma membrane H ϩ -ATPase activity (2), which is encoded by the essential PMA1 gene (3). This observation initially led to the hypothesis that the cAMP-dependent protein kinase might be involved in regulation of the PMA1 protein during growth in glucose, an hypothesis consistent with the finding that H ϩ -ATPase activity increases after cAMP is added to starved Schizosaccharomyces pombe cells (4) or to Saccharomyces cer-evisiae mutants defective in cAMP synthesis (5). It is now clear, however, that the regulation of H ϩ -ATPase activity by glucose is still observed in strains that are deficient in cAMP-dependent kinase activity (6), arguing against a direct role of the cAMP-dependent protein kinase in the phosphorylation of PMA1. Glucose-dependent H ϩ -ATPase activity was also found to be altered by the addition of inhibitors of protein kinase C or Ca 2ϩ -calmodulin-dependent kinase to yeast cells (7). Although suggesting a role for these kinases in the phosphorylation of PMA1, these in vivo approaches do not make it possible to distinguish secondary from direct effects.
The C-terminal region of PMA1 contains putative phosphorylation sites for Ca 2ϩ -calmodulin-dependent kinase and/or protein kinase C. Mutations of these sites render the enzyme unresponsive to glucose (8), consistent with the hypothesis that the C terminus constitutes an inhibitory domain whose interaction with the active site is regulated by glucose via protein kinase-mediated phosphorylation. That the H ϩ -ATPase activity is indeed regulated by phosphorylation has recently been confirmed by the finding that in vivo phosphorylation of PMA1 is associated with increased H ϩ -ATPase activity during growth in glucose (9). However the identity of the kinase(s) involved in this regulation and the target sites in PMA1 remain to be identified.
On the other hand, PMA1 is known to be phosphorylated by a plasma membrane-bound kinase both in vitro and in vivo (10,11). Accordingly, treatment of PMA1 with acid phosphatase leads to a decrease in H ϩ -ATPase activity (12). Although PMA1 is a major in vitro substrate for a protein kinase of the plasma membrane-associated casein kinase type I (12), there is no demonstration of a functional relationship between casein kinase I (CK1) 1 activity and glucose regulation of H ϩ -ATPase activity.
In this study we show that the plasma membrane-bound CK1 activity described by Kolarov et al. (12) results from the combined function of YCK1 and YCK2, two yeast homologues of mammalian CK1 which together are essential to mitotic growth (13). We also provide evidence for the regulation of CK1 activity by glucose and show that loss of YCK function results in impaired regulation of H ϩ -ATPase activity by glucose. Finally we show that PMA1 is phosphorylated in vitro at a consensus site for CK1, which is located in the putative MgATP-binding domain of the enzyme.
Construction of pma1E504A and pma1S507A Mutant Strains-Sitedirected mutagenesis of the 2.5-kilobase HindIII fragment of the PMA1 gene was made in pALTER-1 using the Promega (Madison, Wi) mutagenesis kit and either the oligonucleotide 5Ј-GAAGCCAATGCAG-CAACCTTG-3Ј for the E504A mutation or the oligonucleotide 5Ј-GAAC-CCTCTAGCAGCCAATTC-3Ј for the S507A mutation (nucleotide changes introduced in the mutagenic oligonucleotides are underlined). The pma1 fragments containing the mutations were liberated from pALTER-1 and utilized to replace the corresponding fragment of PMA1 cloned as a 4.6-kilobase HindIII/XbaI fragment in the single copy plasmid pRS315. That the resulting pma1E504A and pma1S507A mutant genes only contained the introduced mutation was confirmed by DNA sequencing of the complete coding region.
The pma1E504A and pma1S507A mutant genes were introduced into the YAK2 strain by transformation using the lithium acetate method. Yeast strains expressing the mutated H ϩ -ATPase were selected on glucose medium containing 5-fluoroorotic acid as described (49). The pma1E504A and pma1S507A alleles were isolated from the corresponding mutant strains and sequenced to check the presence of the expected mutation and to exclude any secondary mutation that might have occurred during yeast transformation.
Growth Sensitivity to Hygromycin B-Cells of the LRB341, LRB343, and LRB346 strains were grown at 28°C for 3 days on rich YPD medium containing 2% (w/v) yeast extract (Difco), 2% (w/v) agar (Sigma), 2% (w/v) glucose, and 25, 50, 75, or 100 g/ml hygromycin B (Sigma). To assess the additional effect of KCl, the cells were plated on solid medium containing 100 g/ml hygromycin B and KCl at a concentration ranging from 25 to 100 mM.
Purification of Plasma Membranes-Cells grown aerobically at 28°C in 2% (w/v) yeast extract (KAT, Ohli, Hamburg), 5.8% (w/v) glucose were harvested in the late exponential phase of growth (250 ϫ 10 6 cells/ml) and washed three times in cold water. Half of the cells were then resuspended in either 250 mM sorbitol (glucose-starved cells) or 250 mM glucose (glucose-activated cells) and incubated at 28°C for 15 min. The cells were then centrifuged and ground at 4°C with glass beads in 10 mM Tris acetate, pH 10.5, 1 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, and either 250 mM glucose (activated cells) or sorbitol (starved cells). The next steps of plasma membrane purification are described in Goffeau and Dufour (15) except that the two Triton X-100 washes were omitted. For simplicity's sake, we shall use the term "sorbitol membranes" to designate plasma membranes from glucosestarved cells and the term "glucose membranes" when referring to plasma membranes from glucose-activated cells. Protein concentrations were measured by the method of Peterson (16), using bovine serum albumin as the standard.
ATPase Activity Measurements-ATPase activity was assayed by incubating a volume of plasma membrane suspension corresponding to 5 g of protein for 5 min at 30°C in a reaction mixture (100 l final volume) containing 6 mM ATP (sodium salt, Sigma), 9 mM MgCl 2 , 10 mM NaN 3 , 50 mM MES-NaOH at pH 6.0. The reaction was stopped by addition of 0.1% SDS. Released P i was measured by colorimetry of the molybdate-P i complex as described (15). Each assay was performed in the presence or absence of 100 M vanadate, and the difference between the two measurements was used to estimate ATPase activity. Measurements of apparent K m for ATP and V max were determined using an ATP-regenerating system (49). Typical Michaelis-Menten kinetics were not obtained in those assays since ATP hydrolysis was found to deviate from a first-order kinetics. Similar complexity has been illustrated by the recent kinetic analysis of some pma1 mutant enzymes (50).
Phosphorylation of Plasma Membranes-Purified glucose membranes or sorbitol membranes were resuspended in 50 l of reaction mixture containing 0.25 mg/ml protein, 6 mM MgCl 2 , 10 mM NaN 3 , 50 mM buffer (pH 6.0 and 6.6 MES-NaOH or pH 7.2 and 8.5 MOPS-NaOH), and 50 M [␥-32 P]ATP (2000 cpm/pmol). The phosphorylation reaction was carried out for 15 min at 30°C. The reaction was stopped by adding 13 l of 5 ϫ concentrated sample buffer (17) and boiling the samples for 3 min. Aliquots of 12.5 g of proteins from each sample were electrophoresed on Tricine/SDS-polyacrylamide gels in the buffer system described in Ref. 18. The 8% polyacrylamide (w/v) separating gels were prepared from a solution containing 30% (w/v) acrylamide and 0.8% (w/v) bisacrylamide in water. Gels run overnight at constant voltage (50 V) were stained with Serva Blue, dried, and autoradiographed on X-OMAT film (Kodak).
Sorbitol membranes or glucose membranes from strain ⌺1278b were resuspended at 0.5 mg/ml in a cold phosphorylation medium containing 100 mM buffer (MES-NaOH pH 6.0 or MOPS-NaOH pH 7.2), 12 mM MgCl 2 , 2 mM EGTA, 2 mM DTT, 16 M PK-I peptide inhibitor, 20 M calmidazolium and incubated with 100 M ATP for 15 min at 30°C. Then 20 l of each cold-phosphorylation reaction mixture was added to 20 l of a reaction mixture containing 2 mg/ml of the specified peptide and 200 M [␥-32 P]ATP. The incubation was continued for an additional 15 min at 30°C. A 25-l aliquot was then spotted onto P81 phosphocellulose paper (Whatman). Each filter was washed five times for 10 min with 0.5% (v/v) phosphoric acid, air-dried, and its radioactivity measured by scintillation counting.
Phosphorylation of the CK1 Peptide by Plasma Membranes from Wild-type and CK1-defective Strains-Sorbitol or glucose membranes from strain ⌺1278b, W303, LRB343, or LRB346 were resuspended to 0.5 mg/ml in the reaction mixture described above and cold-phosphorylated for 10 min at 30°C at pH 7.2. Then 20 l of each reaction mixture was added to 20 l of a reaction mixture containing 100 M [␥-32 P]ATP and the CK1 peptide at 2 mg/ml. The incubation was continued for an additional 10 min at 30°C. A 25-l aliquot was then spotted on P81 phosphocellulose paper and analyzed as described above.
Phosphorylation of Plasma Membrane H ϩ -ATPase by Mammalian CK1-Sorbitol membranes from strain LRB346 were resuspended to 0.5 mg/ml in a medium containing 50 mM MOPS-NaOH, 12 mM MgCl 2 , 1 mM DTT, and 8 milliunits of CK1. CK I was purified from porcine spleen (specific activity, 0.4 units/mg; one unit of CK1 incorporates 1 mol of [ 32 P]phosphate per min into casein (2 mg/ml) at 37°C (21)). The reaction was started by adding [␥-32 P]ATP to a final concentration of 100 M and was carried out at 30°C. At the times indicated, aliquots containing 6.25 g of protein were boiled for 3 min in Laemmli sample buffer, electrophoresed, and analyzed by autoradiography as described above.
Kinetics of CK1 Peptide Phosphorylation-Sorbitol and glucose membranes from strain ⌺1278b were cold-phosphorylated as described above and then depleted of ATP by two successive washes and centrifugations at 100,000 ϫ g for 1 h in a Beckman table-top ultracentrifuge. Membranes were resuspended in washing buffer (5 mM MOPS-NaOH, pH 7.2), and 6.25 g of protein was added to the reaction mixture containing 50 mM MOPS-NaOH, pH 7.2, 6 mM MgCl 2 , 1 mM EGTA, 1 mM DTT, 8 M PK-I peptide inhibitor, 10 M calmidazolium, and 1 mg/ml CK-1 peptide. The reaction was started by addition of [␥-32 P]ATP (5-400 M). The reaction was carried out for 3 min at 30°C and stopped by spotting 20-l aliquots onto P81 phosphocellulose filters as described above. Blanks were incubated under the same conditions without the peptide. All reactions were done in duplicate.
Purification of Peptides from Phosphorylated Plasma Membrane Proteins-Protein samples (2 mg of plasma membrane proteins from glucose-starved LRB341 cells) were resuspended to a final concentration of 0.6 mg/ml in a reaction buffer (6 mM MgCl 2 , 10 mM NaN 3 , 50 mM MES-NaOH, pH 6.0) and labeled with 100 M [␥-32 P]ATP (20000 cpm/ pmol) at 30°C for 15 min. The reaction was stopped by adding 5 ϫ Laemmli sample buffer and by boiling for 3 min. The solubilized membranes were deposited in a well 7 cm long and 3 mm wide and electrophoresed as described above. Membrane proteins were blotted overnight at 50 V onto polyvinylidene difluoride membranes (Millipore) in 25 mM Tris, 192 mM glycine, 20% (v/v) methanol. After autoradiography, the heavily phosphorylated bands corresponding to the H ϩ -ATPase and to an unidentified 50-kDa protein were excised and cut into 1-mm 2 pieces. Free active sites on the polyvinylidene difluoride membranes were blocked with a 0.2% (w/v) aqueous solution of polyvinylpyrrolidone (PVP-K40, Merck), and excess polyvinylpyrrolidone was washed away with 2% (v/v) methanol. The pieces were then resuspended in 100 l of a solution containing 25 mM Tris-HCl, 1 mM EDTA at pH 8.2, and blotted proteins were digested with 3 g of endo-Lys-C (sequencing grade, Boehringer Mannheim) for 24 h at 37°C. The supernatant and three additional washes of the pieces with digestion buffer were pooled, concentrated in a C-18 SepPac cartridge (Amicon), and eluted with 70% (v/v) isopropyl alcohol, 0.1% trifluoroacetic acid. The organic solvent was evaporated, and the purified peptides were injected into a C-18 reverse-phase HPLC column (Vydac, 218TP52) mounted on an ABI dual pump. A linear gradient was formed by mixing solvent B (50% acetonitrile, (J. T. Baker Inc.), 1 mM NaH 2 PO 4 ) with solvent A (0.05% trifluoroacetic acid (ABI), 1 mM NaH 2 PO 4 ) and gradually increasing the proportion of solvent B (from 4 to 98% v/v) as the mixture was run through the column, at a flow rate of 425 l/min, over a 150-min period. The absorbance of the eluent was recorded at 214 nm, and the fractions corresponding to absorbance peaks were collected and dried in a Speed Vac (Savant). Radioactive peptides were solubilized in 70% (v/v) isopro-pyl alcohol, 0.1% trifluoroacetic acid, blotted onto Polybrene membranes, and subjected to automated Edman degradation in an ABI 477 sequencer.

Time and pH Dependence of Yeast Plasma Membrane
Phosphorylation-Plasma membranes were prepared from glucosestarved cells of the wild-type strain ⌺1278b. Fig. 1 shows that the addition of [␥-32 P]ATP to the purified membranes resulted in the intense labeling of a 100-kDa protein, which was identified as PMA1, the catalytic subunit of the plasma membrane H ϩ -ATPase (3). Two additional proteins, a 50-kDa protein (p50) and a small polypeptide (SPP), were also phosphorylated in vitro. The phosphorylation level of PMA1 increased linearly with time (over a 15-min incubation) and was more pronounced at pH 6.0 than at pH 7.2, whereas p50 and SPP were better phosphorylated at pH 7.2. The phosphorylation level of all three polypeptides increased approximately linearly with protein concentrations in the range of 0.125 to 1.5 mg/ml (data not shown).
Effect of Growth Conditions on the Level of Plasma Membrane Phosphorylation-We examined the phosphorylation pattern of plasma membranes isolated from LRB341 cells that had been incubated in glucose instead of sorbitol before membrane homogenization. The wild-type strain LRB341 which has a genetic background different from that of ⌺1278b is the parental strain of the yck mutants (21,22) used in this study (see below). Fig. 2A shows that the major proteins phosphorylated in sorbitol membranes of LRB341 cells were the same three as observed in ⌺1278b cells. Phosphorylation of PMA1 in both strains was also higher at pH 6.0 than at a more alkaline pH as shown in Fig. 1. The three phosphoproteins showed different pH optima for phosphorylation. The phosphorylated residue(s) of PMA1 appeared to be more accessible at pH 6.0 (the optimum pH for ATPase activity) than at more alkaline pH values. This result suggests that the sequence including the phosphorylation site(s) is prone to undergo pH-dependent conformational changes, such as those reported by Blanpain  Glucose membranes or sorbitol membranes were purified from wild-type strain LRB341 and mutant strains LRB343 and LRB346. The membranes were incubated for 15 min at 30°C as in the experiment depicted in Fig. 1, although a broader pH range (pH 6.0 and 6.6 MES-NaOH or pH 7.2 and 8.5 MOPS-NaOH) was examined. A, YCK1 YCK2 strain LRB341. B, YCK1 ⌬yck2 strain LRB343. C, ⌬yck1 yck2 ts strain LRB346. The YCK1 and YCK2 genes encode two yeast isoforms of mammalian CK1 (13). The yck2 ts mutation affects the catalytic kinase domain of YCK2, conferring growth thermosensitivity to the mutant cells. Interestingly, the phosphorylation levels of all three phosphoproteins were found to be higher in sorbitol membranes than in glucose membranes ( Fig. 2A). Thus, preincubating wild-type yeast cells in glucose had inhibitory effects on plasma membrane phosphorylation.
Use of Specific Synthetic Substrates for Characterizing the Plasma Membrane-bound Kinase Activity-Peptides containing phosphorylation sites for the mammalian p90 rsk , mitogenactivated protein kinase, protein kinase A, protein kinase C, CK1, and CK2 (Fig. 3A) were used as phosphorylation substrates to further characterize the kinase activity associated to yeast plasma membranes.
In our initial experiments, in which the peptides and [␥-32 P]ATP were added directly to membranes of the wild-type strain ⌺1278b, only small differences in phosphorylation were observed because of the high background of endogenous phosphorylation of the membrane proteins (data not shown). To lower this endogenous background, the membranes were preincubated with 100 M ATP at 30°C for 10 min, before the addition of the various peptide substrates and [␥-32 P]ATP. Under these conditions only the peptide DDEESITRR, a specific synthetic substrate of CK1 (20), was phosphorylated (Fig. 3B).
Using the CK1 peptide as a specific phosphorylation substrate, we showed that CK1 activity decreased in glucose membranes (61% remaining relative to the amount seen in the sorbitol membranes). In both membranes, CK1 peptide phos-phorylation was higher at pH 7.2 than at pH 6.0 (Fig. 3B, see also Fig. 4).
Phosphorylation of the CK1 Peptide by Plasma Membranes Defective in CK1 Activity-In the yeast S. cerevisiae, three isoforms of CK1 are currently recognized. Two of these, YCK1 and YCK2, are tightly associated with the plasma membrane and together are required for cell viability (13,25). In contrast, the third casein kinase I isoform is found predominantly in the nucleus (25). We therefore examined the phosphorylation of the CK1 peptide by plasma membranes isolated from yck mutant strains. The YCK1 ⌬yck2 strain LRB343 contains a null mutation of the YCK2 gene (⌬yck2), whereas the ⌬yck1 yck2-1 ts strain LRB346 bears a gene deletion of YCK1 (⌬yck1) and another mutation in YCK2 (yck2-1 ts ) that confers growth thermosensitivity to the mutant cells (14).
In comparison to the wild-type W303 and ⌺1278b strains (YCK1 YCK2), a 2-fold decrease in the phosphorylation level of the CK1 peptide was found in sorbitol membranes of the YCK1 ⌬yck2 strain LRB343 lacking YCK2 (Fig. 4). This result suggests that YCK1 and YCK2 contribute equally well to the CK1 activity seen in wild-type membranes. Strikingly, the amount of 32 P incorporated in the CK1 peptide by glucose membranes of the LRB343 strain was slightly higher than the amount measured in sorbitol membranes of the same mutant strain. In contrast, no phosphorylation of the CK1 peptide could be detected in either membranes of the ⌬yck1 yck2-1 ts strain LRB346.
Taken together, these results indicate that the plasma membrane kinase activity which specifically phosphorylates the CK1 peptide is mediated by YCK1 and YCK2. Consistently, these two similar and functionally interchangeable isoforms of CK1 are tightly associated to the plasma membrane through posttranslational prenylation of their C terminus (25).
Phosphorylation of PMA1 in Plasma Membranes of CK1deficient Yeast Strains-To test the possibility that YCK1 and YCK2 recognize the plasma membrane H ϩ -ATPase as an in vitro substrate, we examined the phosphorylation level of PMA1 in plasma membranes isolated from the yck mutant strains LRB343 and LRB346.  4. Phosphorylation of the CK1 peptide by plasma membranes from CK1-defective cells. Sorbitol membranes or glucose membranes (0.5 mg/ml protein) were purified from the wild-type strains W303 and ⌺1278b (YCK1 YCK2) and from the yck mutant strains LRB343 (YCK1 ⌬yck2) and LRB346 (⌬yck1 yck2 ts ). The membranes were incubated at pH 7.2 in the presence of ATP for 10 min at 30°C before being diluted 1:2 and incubated with the CK1 peptide and [␥-32 P]ATP for 10 min at 30°C as described under "Materials and Methods." In sorbitol membranes of the YCK1 ⌬yck2 strain LRB343 (Fig. 2B), loss of YCK2 function had little effect on the level of PMA1 phosphorylation in comparison to membranes of the wild-type YCK1 YCK2 strain LRB341 ( Fig. 2A). Strikingly, the high phosphorylation level of PMA1 was maintained in glucose membranes of the mutant strain although it decreased in the parental strain. Loss of YCK2 function has no effect on pH dependence of PMA1 phosphorylation in either membrane. Concerning the remaining other two proteins that are phosphorylated by wild-type membranes (see Fig. 1), we found that the phosphorylation of p50 was quite low in either membrane of the mutant strain, whereas the phosphorylation level of the small polypeptide (SPP) was higher in glucose membranes than in sorbitol membranes (Fig. 2B). Fig. 2C shows the phosphorylation of plasma membranes from the ⌬yck1 yck2-1 ts strain LRB346, which is defective in CK1 activity (see Fig. 4). Only SPP exhibited some appreciable phosphorylation and only in sorbitol membranes. A very weak phosphorylation of PMA1 and p50 could be detected after a longer exposure of the gels (data not shown). Similar results were obtained if the mutant cells were incubated at the more permissive temperature of 22°C (data not shown).
These results indicate that the in vitro phosphorylation of PMA1 is due solely to CK1 activity. Both YCK1 and YCK2 contribute to this kinase activity which increases upon glucose starvation, as already shown by using the CK1 peptide as a phosphorylation substrate.
In vitro Phosphorylation of PMA1 by Purified Mammalian Casein Kinase I-The addition of purified mammalian CK1 (20) to sorbitol membranes of the CK1-deficient strain LRB346 (⌬yck1 yck2-1 ts ) restored wild-type level of PMA1 phosphorylation in the mutant membranes. As shown in Fig. 5, PMA1 was readily phosphorylated by the exogenous kinase. The level of PMA1 phosphorylation reached a plateau 10 min after the addition of [␥-32 P]ATP and then slightly decreased after 30 min of incubation. This is another evidence confirming that PMA1 is subject to phosphorylation by yeast CK1 homologues in vitro. This result also indicates that PMA1 contains a species-conserved consensus motif for phosphorylation by CK1.
Regulation of Yeast CK1 Activity by Glucose-We have shown that the phosphorylation level of the CK1 peptide increases upon glucose starvation (Fig. 3B). As a first step for characterizing this regulation of CK1 activity, ATP dependence of the CK1 peptide phosphorylation were compared under initial rate conditions in sorbitol membranes and glucose membranes of the wild-type strain ⌺1278b. The membranes were first cold-phosphorylated and then depleted of ATP before fixed concentrations of the CK1 peptide were added and incubated in the presence of increasing concentrations (5-400 M) of [␥-32 P]ATP (blanks were without the CK1 peptide).
The double-reciprocal plot of the data shown in Fig. 6 revealed that the apparent K m for ATP was similar for sorbitol and glucose membranes. This apparent K m value obtained is close to that described by Vancura et al. (26) for soluble YCK2. The predicted V max , however, was about two times higher for sorbitol membranes than glucose membranes (Fig. 6). One explanation for decreased phosphorylation of the synthetic CK1 peptide at the onset of glucose metabolism is that glucose somehow induces down-regulation of CK1 activity.
H ϩ -ATPase Activity in Plasma Membranes of yck Mutant Strains-We have shown that the level of PMA1 phosphorylation in glucose membranes is low relative to sorbitol membranes ( Fig. 2A). On the other hand, glucose is known to stimulate H ϩ -ATPase activity (2). To determine whether this regulation is a function of the extent of phosphorylation, we analyzed the effects of yck mutations on H ϩ -ATPase activity.
The addition of glucose to starved cells of the wild-type strain LRB341 results in a 3-fold increase of H ϩ -ATPase activity (Fig.  7), as previously shown (2). In contrast, there is no effect of glucose on H ϩ -ATPase activity in the YCK1 ⌬yck2 strain LRB343 (Fig. 7), which shows abnormally high levels of PMA1 phosphorylation both in the presence or absence of glucose (see Fig. 2). This result suggests that CK1-mediated phosphorylation may prevent PMA1 from being activated by glucose. When CK1 activity is strongly decreased by the double ⌬yck1 yck2 ts mutation, glucose-starved cells (strain LRB346) show increased H ϩ -ATPase activity (Fig. 7), comparable with the value obtained with glucose-incubated cells of the wild-type strain LRB341. However, the addition of glucose to the ⌬yck1 yck2 ts mutant cells results in further activation of the H ϩ -ATPase activity, suggesting another regulation distinct from CK1-mediated phosphorylation.
Effect of yck Mutations on Sensitivity of Yeast Cells to Hygromycin B-Hygromycin B is an antibiotic that affects protein translation. The transport of this aminoglucoside into yeast cells is thought to require the plasma membrane electric potential generated through the PMA1 function. This belief primarily stems from the correlation between increased growth resistance to hygromycin B and a depolarization of cellular membrane potential due to specific mutations in PMA1 (27,28). Fig. 8 shows that the ⌬yck1 yck2 ts strain LRB346 failed to grow in the presence of 100 g/ml hygromycin B. However, growth was restored by the addition of 100 mM KCl to the growth medium. This sensitivity to low concentrations of hygromycin B can be explained if increased H ϩ -ATPase activity in the ⌬yck1 yck2 ts cells (see Fig. 7) leads to enhanced uptake of the drug. Although there is no evidence for membrane potential defects in the mutant cells, the alleviating effect of KCl is consistent with a hyperpolarization of plasma membrane potential.
Identification of a CK1 Phosphorylation Site in PMA1-In vitro phosphorylation of PMA1 occurs exclusively on serine residues (11,12). We attempted to identify those residues in order to gain structural information about the mechanisms involved in CK1-mediated regulation of H ϩ -ATPase activity. Fig. 9 shows the presence of three radioactive peaks in the reverse phase-HPLC pattern obtained after endo-Lys-C cleavage of phosphorylated PMA1. Automated Edman degradation of peak 1 failed to yield an identifiable sequence, for lack of a sufficient amount of peptide (data not shown). Peak 3 was a complex mixture of at least three peptides present in various amounts. The more abundant peptide (120 pmol) contained no casein kinase I consensus phosphorylation motif and masked the sequences of the other peptides present in lower amounts in the same peak. Sequencing the major absorbance peak 2 revealed two peptides contiguous in PMA1. The main peptide (100 pmol) sequence was 483 TVEEDHPIPEDVHENYENK 501 . The sequence of the less abundant peptide (20 pmol) was 502 VAELASRGFRALGVARK 518 . Numbers indicate the positions of the first and last residues of the peptide relative to the PMA1 sequence (3). The second peptide contains a consensus phosphorylation motif for CK1 (20,29,30), with only one serine residue at position 507 (Ser-507) and, three positions away toward the N terminus, a glutamate residue (Glu-504). In an independent control experiment, we purified and sequenced a phosphopeptide from the 50-kDa protein (p50) which is phosphorylated in vitro (Fig. 1). The peptide also contains a glutamate residue at position Ϫ3 with respect to the only serine residue present in the sequence (data not shown).
Biochemical mapping of phosphorylated PMA1 identified Ser-507 as an in vitro target for CK1, consistent with the finding that CK1-phosphorylation of PMA1 occurs on Ser residues in vitro (11,12). Interestingly, Glu-504 and Ser-507 are located in the ATP-binding domain of the enzyme. The effect of phosphorylating Ser-507 on H ϩ -ATPase activity was determined by replacing the wild-type PMA1 gene with pma1-E504A or pma1-S507A, two mutant alleles with, respectively, Glu-504 or Ser-507 replaced by Ala. Neither PMA1 phosphorylation nor regulation of H ϩ -ATPase activity by glucose was found to be dependent on Ser-507 (data not shown). Surprisingly, the pma1-E504A mutant enzyme showed a smaller reduction in the apparent K m for ATP, which was induced by glucose (a 3-fold decrease in comparison to a 12-fold decrease for the wild-type enzyme). However, the V max values of the pma1-E504A H ϩ -ATPase were not significantly different from  9. Reverse phase-HPLC chromatogram of peptides from in vitro phosphorylated PMA1. Sorbitol membranes from the YCK1 YCK2 strain LRB341 were phosphorylated in vitro, electrophoresed, and blotted onto polyvinylidene difluoride membranes. The band corresponding to PMA1 was excised and digested with Endo-Lys-C. Digestion products were injected into a C18 column, and peptides were eluted by a linear gradient of solvents A and B. Solvent A was 0.05% trifluoroacetic acid, 1 mM NaH 2 PO4 and solvent B was 50% acetonitrile, 0.045% trifluoroacetic acid, 1 mM NaH 2 PO 4 . The absorbance was monitored at 214 nm. Numbers indicate radioactive peaks as discussed in the text. those of the wild-type enzyme, in both glucose and sorbitol membranes (Table I). DISCUSSION A CK1 activity has been shown in vitro to partly copurify with and to phosphorylate the plasma membrane H ϩ -ATPase of S. cerevisiae (12). The recent isolation of yeast homologues of CK1 (13) provided new genetic tools for further investigating the role of CK1-mediated phosphorylation on H ϩ -ATPase activity.
We first show that the phosphorylation of PMA1 is severely affected by mutations of both YCK1 and YCK2, two similar and plasma membrane-associated isoforms of CK1 (13,25). We also show that the wild-type phosphorylation level is restored by the addition of mammalian CK1 to the mutant membranes. These results confirm that PMA1 is subject to phosphorylation by a plasma membrane-associated CK1 which in S. cerevisiae is encoded by YCK1 and YCK2.
PMA1 is the first substrate of CK1 to be described in yeast. In mammals, CK1 is a ubiquitous enzyme that phosphorylates a large number of cell proteins, including signaling molecules such as the plasma membrane-bound receptors for insulin (31) and tumor necrosis factor (32), the regulatory subunit (inhibitor-2) of protein phosphatase-1 (33), the cAMP response element CREM (34), the SV40 large T antigen (35), and the nuclear p53 protein (36). These reports suggest a role for CK1 in signal transduction events. Unlike the cAMP-dependent protein kinase A or the multimeric CK2 (37,38), it appears unlikely that CK1 activity is regulated by a second messenger or through protein-protein interactions with other regulatory subunits. Brockman and Anderson (39) have reported a regulatory effect of phosphatidylinositol bisphosphate on the activity of the 37-kDa casein kinase I of red blood cells, although this result has been questioned by others (40). Studies on CK1␦ from rat testis (41) and CKI1 of S. pombe (42) have raised the possibility that CK1 activity might be regulated through the phosphorylation of its C terminus which acts as an autoinhibitory domain.
In S. cerevisiae, SNF4 is a positive effector of the SNF1 protein kinase (43), which is required for carbon catabolite derepression (44). Overexpression of YCK1 or YCK2 allows growth of snf1 or snf4 mutants on raffinose without restoring invertase activity (13). This observation suggests that increased dosages of the YCK genes may result in more efficient hexose metabolism without affecting the glucose repression mechanism directly (13). Interestingly, we have found that the YCK kinases are activated upon glucose starvation by using a synthetic peptide (CK1 peptide) as a phosphorylation substrate.
The CK1 peptide contains a phosphorylation site for CK1 and is phosphorylated specifically by YCK1 and YCK2 as phos-phorylation could no longer be detected in a yeast strain deficient in both YCK kinases. In particular, loss of YCK2 function results in a 2-fold reduction of CK1 peptide phosphorylation by sorbitol membranes, consistent with YCK1 and YCK2 contributing equally well to plasma membrane CK1 activity.
The addition of glucose to starved yeast cells results in decreased CK1 activity as assayed with the specific CK1 peptide substrate. Although the molecular mechanisms of this downregulation are unknown, we can imagine that, by analogy to the situation in mammals (41), YCK1 and YCK2 are regulated by phosphorylation of their C terminus, which may serve to create an autoinhibitory domain. Alternatively, the amount of these kinases may be regulated at the plasma membrane, through selective prenylation. The possibility that one YCK isoform may be involved in the inactivation of the second isoform is suggested by the finding that glucose has no inhibitory effect on CK1 activity in a mutant strain that only expresses YCK1. Actually, this CK1 activity is comparable with the activity measured in glucose membranes of wild-type cells.
As shown for the CK1 peptide, the phosphorylation level of PMA1 decreases during growth in glucose. It is known that ATPase activity is regulated at the plasma membrane in response to glucose (2). Accordingly, we have found that a decline in CK1-mediated phosphorylation of PMA1 is associated with increased H ϩ -ATPase activity during glucose metabolism.
When compared with the wild-type strain, H ϩ -ATPase activity increases 3-fold in starved cells of a CK1-deficient strain. Furthermore, glucose has no effect on H ϩ -ATPase activity in a mutant strain that maintains abnormally high levels of PMA1 phosphorylation, due to the loss of YCK2 function. All together these results suggest several mechanisms for the regulation of the H ϩ -ATPase. One possibility is that H ϩ -ATPase activity is inhibited through phosphorylation by the YCK kinases whose functions are stimulated upon glucose starvation. It is also possible that YCK-mediated phosphorylation of PMA1 may prevent the enzyme from being activated by glucose, through another kinase-dependent phosphorylation. This second explanation is supported by our results that the H ϩ -ATPase activity of a YCK-deficient strain is still stimulated by glucose and by the finding that increased H ϩ -ATPase activity correlates with the phosphorylation of specific sites in vivo (9).
It appears likely that PMA1 contains at least two antagonistic regulatory sites (Fig. 10). YCK-mediated phosphorylation of one site inhibits H ϩ -ATPase activity; phosphorylation of the second site by a different protein kinase, however, influences the H ϩ -ATPase activity. When one of the two YCK isoforms is missing, glucose has no effect on H ϩ -ATPase activity, showing the dominance of the inhibitory phosphorylation over the activating one. This model is probably too simple since more than one target site may be phosphorylated by each kinase. Moreover, it does not consider other ways by which PMA1 might be regulated, such as dephosphorylation of CK1-phosphorylated sites by a protein phosphatase (12,14).
We have identified the Ser-507 residue of PMA1 as a major site for Yckp-mediated phosphorylation. The sequence surrounding Ser-507 conforms with a bona fide consensus motif for CK1 (20,29,30) and is fully conserved in fungal H ϩ -ATPases but not in plant enzymes (45). However, mutagenesis analysis revealed that the overall level of PMA1 phosphorylation is not affected by the pma1S507A mutation, which replaces Ser-507 with Ala. It is therefore likely that PMA1 contains more than one CK1 phosphorylation site, consistent with the identification of three phosphopeptides by biochemical mapping.
Interestingly enough, we have found that the Glu-504 residue of PMA1 is important for regulation of H ϩ -ATPase activity by glucose. The pma1E504A mutation has a strong inhibitory  (48). Interestingly, the P536L mutation was previously isolated as a suppressor of the double mutation S911A/T912A which affects glucose-induced activation of the H ϩ -ATPase (8). These and other results strongly indicate that the ATP-binding region of the H ϩ -ATPase is prone to structural rearrangements and interacts with the C terminus whose phosphorylation may control H ϩ -ATPase activity. It is therefore possible that Glu-504 mediates the regulation of H ϩ -ATPase activity by glucose either directly, through the control of ATP binding, or indirectly, via the interaction between the catalytic domain and the C-terminal regulatory domain of the enzyme. The plasma membrane H ϩ -ATPase is rate-limiting for yeast growth, which is optimal at neutral intracellular pH and decreases as the cell interior becomes more acidic (47). One of the major regulatory functions of the H ϩ -ATPase is the control of intracellular pH by ejection of protons from the cytosol. Active nutrient transport is another mechanism of growth control since the proton gradient generated by the enzyme is the driv-ing force for nutrient uptake. As glucose metabolism produces intracellular acidification, the increase in ATPase activity can be rationalized as the basis of the need for intracellular pH homeostasis (47). Expression of yeast PMA1 in mouse fibroblasts increases intracellular pH and induces cell proliferation, suggesting that similar mechanisms may be involved in growth control by proton transport in all eukaryotic cells (47). A common early response to stimuli that activate cell proliferation is an intracellular alkalinization which, in animal cells, is caused by the activation of a Na ϩ ,H ϩ exchange system. An increase in intracellular pH is known to activate glycolysis. Consistently, one of the most prominent properties characteristic of rapidly growing cancer cells is their capacity to sustain high rates of aerobic glycolysis. Type II hexokinase, which is expressed at high levels in such cells and bound to the outer mitochondrial membrane, plays a key role in the aberrant glucose metabolism (51). Ouabain, an inhibitor of the Na ϩ ,K ϩ -ATPase, inhibits glycolysis in tumor cells. Moreover, a protein kinase of Ehrlich ascites tumor cells phosphorylates the Na ϩ ,K ϩ -ATPase, which causes decreased Na ϩ pump efficiency (52). The pH-sensing mechanism related to growth control of eukaryotic cells remains to be determined (47). However, the nature of these pH sensor(s) could be investigated by a genetic approach using yeast as a model system.
In conclusion, we have shown that PMA1 is phosphorylated by two plasma membrane-associated isoforms of CK1 in vitro. This is the first evidence of a biochemical function for CK1 in yeast. CK1-phosphorylation of PMA1 also provides a coherent starting basis for further in vivo studies of the regulation of H ϩ -ATPase activity and may prove to be a useful system for understanding the role of CK1-mediated proton transport for cell growth. PMA1 is regulated through phosphorylation by a plasma membranebound CK1 and another unknown protein kinase. Steady-state kinase activity is represented by a double arrow, with the plus or minus sign indicating activation of inhibition of phosphorylation, respectively. Kinase modulating signals are represented above and below the arrows, and their overall effect on H ϩ -ATPase activity is shown in the last column (a wavy line indicates no effect). The YCK1 and YCK2 wild-type strain is represented in the first row; the YCK1 ⌬yck2 strain which only expresses the YCK1 isoform is shown in the second row, and the ⌬yck1 yck2-1 ts strain which only expresses a mutant version of YCK2 is shown in the third row.