Protein phosphatase 1 (Glc7p) and its binding protein Reg1p are essential for the regulation of glucose repression pathways in Saccharomyces cerevisiae. In order to identify physiological substrates for the Glc7p–Reg1p complex, we examined the effects of deletion of the REG1 gene on the yeast phosphoproteome. Analysis by two‐dimensional phosphoprotein mapping identified two distinct proteins that were greatly increased in phosphate content in reg1Δ mutants. Mixed peptide sequencing identified these proteins as hexokinase II (Hxk2p) and the E1α subunit of pyruvate dehydrogenase. Consistent with increased phosphorylation of Hxk2p in response to REG1 deletion, fractionation of yeast extracts by anion‐exchange chromatography identified Hxk2p phosphatase activity in wild‐type strains that was selectively lost in the reg1Δ mutant. The phosphorylation state of Hxk2p and Hxk2p phosphatase activity was restored to wild‐type levels in the reg1Δ mutant by expression of a LexA–Reg1p fusion protein. In contrast, expression of LexA–Reg1p containing mutations at phenylalanine in the putative PP‐1C‐binding site motif (K/R)(X)(I/V)XF was unable to rescue Hxk2p dephosphorylation in intact yeast or restore Hxk2p phosphatase activity. These results demonstrate that Reg1p targets PP‐1C to dephosphorylate Hxk2p in vivo and that the motif (K/R)(X) (I/V)XF is necessary for its PP‐1 targeting function.
The catalytic subunit of protein phosphatase 1 (PP‐1C) is amongst the most conserved proteins in nature (Cohen and Cohen, 1989). The enzyme is ubiquitously expressed at high levels in all cell types with an intracellular protein concentration of ∼10 μM. In vitro, the free catalytic subunit displays indiscriminate substrate specificity that overlaps with other ubiquitously expressed family members, including PP‐2A and PP‐2B (Cohen and Cohen, 1989; Shenolickar and Nairn, 1991). However, in vivo, PP‐1C is associated with a number of regulatory proteins that either inhibit its activity or specifically target the phosphatase to a selective set of substrates or intracellular location (Hubbard and Cohen, 1993; Campos et al., 1996; Egloff et al., 1997; Damer et al., 1998). To date, at least 20 PP‐1 regulatory subunits have been identified in mammals and yeast; however, only three of these proteins, M110 (MYTP1), PTG (PPP1R5) and PP‐1GM, have been demonstrated to act in a targeting capacity both in intact cells and in vitro. The M110 subunit binds PP‐1C in vitro and increases its smooth muscle myosin phosphatase activity by a factor of 20 whilst suppressing non‐specific activity towards other substrates (Alessi et al., 1994; Shirazi et al., 1994; Simuzu et al., 1994; Haystead et al., 1995). Addition of both native and recombinant M110 to β‐escin‐permeabilized smooth muscles enhances the ability of PP‐1C to bring about relaxation following Ca2+‐induced contraction (Alessi et al., 1994; Shirazi et al., 1994; Haystead et al., 1995; Gailly et al., 1996; Johnson et al., 1996). In vitro, PTG binds and targets PP‐1C to glycogen, glycogen synthase and phosphorylase a (Doherty et al., 1996; Printen et al., 1997). Overexpression of PTG in hepatocytes results in hyperglycogen accumulation with a concomitant 4‐fold activation of glycogen synthase and 40% inhibition of phosphorylase activity (Berman et al., 1998). In skeletal muscle, the GM subunit has been proposed to target PP‐1C to regulate glycogen synthase dephosphorylation in response to adrenaline and insulin (for a review see Hubbard and Cohen, 1993). Phosphorylation of GM by protein kinase A (PKA) causes dissociation of PP‐1C and thereby prevents activation of glycogen synthase. Conversely, phosphorylation of GM at a second distinct site by the insulin‐stimulated protein kinase p90rsk (MAPKAP1) enhances association of the subunit with PP‐1C to promote dephosphorylation and activation of glycogen synthase (Dent et al., 1988). The physiologically relevant targets of the other PP‐1 regulatory subunits have yet to be defined. Indeed, because PP‐1C has been implicated in so many cellular events, identifying the substrates for any of its putative targeting subunits is a seemingly intractable problem.
In Saccharomyces cerevisiae, PP‐1C is encoded by the essential gene GLC7 (DIS2S1) and genetic analysis has shown that the phosphatase participates in the regulation of many processes, including glucose repression pathways, glycogen accumulation, sporulation, cell cycle progression and protein translation (for a review see Stark, 1996). Like its mammalian counterpart, Glc7p is regulated by interaction with many distinct regulatory targeting subunits. In S.cerevisiae, genetic studies have implicated the Glc7p‐binding protein Reg1p in the regulation of glucose repression pathways (Tu and Carlson, 1995; Huang et al., 1996). Like other yeast, S.cerevisiae thrive on a variety of carbon sources, but glucose and fructose are preferred. When one of these sugars is present, the enzymes required for gluconeogenic pathways and for the utilization of alternative carbon sources are synthesized at low rates or not at all (for a review see Gancedo, 1998). The precise mechanism by which glucose signals to the nucleus to repress genes controlling expression of enzymes involved in gluconeogenesis and alternative carbon source utilization is not known (Gancedo, 1998). Glucose itself or glucose 6‐phosphate (G6P) may be important mediators. Genetic and biochemical evidence suggests that Reg1p physically associates and functions with Glc7p in intact cells to regulate the glucose repression pathway. In two‐hybrid analysis, Reg1p binds Glc7p and both proteins can be co‐immunoprecipitated from yeast extracts (Tu and Carlson, 1995). Furthermore, deletion of REG1 in a wild‐type background results in inhibition of glucose repression, accumulation of intracellular glycogen, slow growth and increased cell size (Huang et al., 1996). Reg1p also appears to have selective functions in the regulation of glucose repression by specifically antagonizing the function of the Snf1 kinase, which is a negative regulator of the glucose repression pathways (Ludin et al., 1998). Reg1p shows a glucose‐dependent association with the Snf1 kinase and may provide a means for dephosphorylation and inactivation of the enzyme by Glc7p.
Here, we hypothesized that comparison of the wild‐type yeast phosphoproteome with reg1Δ mutants and strains containing Reg1p with point mutations in its putative PP‐1C‐binding site would identify physiologically relevant substrates of the Reg1p–Glc7p complex. This is because substrate targets would become selectively hyperphosphorylated as a result of an imbalance of the normal kinase–phosphatase equilibrium that governs the steady‐state phosphorylation of the protein(s) under basal conditions. To test this hypothesis, two‐dimensional SDS–PAGE phosphoprotein maps were prepared from 32P‐labeled wild‐type and mutant strains grown in the presence and absence of glucose. A major increase in phosphate content was observed in the glycolytic enzyme hexokinase II (Hxk2p) and the E1α subunit of pyruvate dehydrogenase in wild‐type strains grown in the absence of glucose, reg1Δ strains, and strains expressing LexA–Reg1p with point mutations in the putative PP‐1C‐binding site. Subsequent fractionation and assay with 32P‐labeled Hxk2p identified hexokinase phosphatase activity in the wild‐type strains that was absent in both reg1Δ and LexA–Reg1p point mutant strains. The effects of the mutations on the ability of Reg1p to repress glucose‐induced expression of the SUC2 gene was also tested. Wild‐type LexA–Reg1p was shown to complement the reg1Δ mutation effectively, reducing SUC2 expression 130‐fold in glucose‐grown cells, whereas the mutant forms of LexA–Reg1p reduced SUC2 expression only 3‐fold. These results demonstrate that Reg1p targets PP‐1C towards Hxk2p to promote its dephosphorylation in vivo and are consistent with previous work by Randez‐Gil et al. (1998), suggesting an involvement of Reg1p–Glc7p in the dephosphorylation of Hxk2p. In addition, our findings show the utility of functional proteomic analysis for the identification of physiological substrates for protein phosphatases.
The effects of REG1 deletion on the yeast phosphoproteome
To examine the phenotypic effects of disruption of the REG1 gene on the yeast phosphoproteome, reg1Δ mutant (MCY3278) and wild‐type (FY250) strains from the same genetic background were labeled to steady state with either [35S]methionine or [32P]orthophosphate and grown in the presence or absence of glucose. Following two‐dimensional SDS–PAGE and autoradiography, the autoradiograms from each radiolabeled strain were made into digital images (Figure 1). To determine whether REG1 deletion affected the overall pattern of protein expression, autoradiograms from [35S]methionine‐labeled yeast were compared using the BIORAD Melanie II 2D‐analysis program. This analysis did not reveal any obvious differences in the expression of any one protein as a result of REG1 deletion (compare Figure 1A with B, and C with D). Analysis by Melanie of the autoradiograms prepared from 32P‐labeled yeast identified ∼80 distinct phosphoproteins of varying abundance in each gel (Figure 1E–H). To identify phosphoproteins that were increased in phosphate content as a result of REG1 deletion, Melanie compared Figure 1E (wild type) with F (reg1Δ). Four groups of spots were identified to be increased in phosphate content in this analysis, labeled Hxk2p (a) and Hxk2p (b), E1, heat shock protein 60 (hsp60) and spot C (the nomenclature was derived following identification of the proteins by mixed peptide sequencing, see below). None of the other 80 or so phosphoproteins detected by Melanie in the two autoradiograms showed any significant differences in phosphate content, suggesting that spots Hxk2p (a), Hxk2p (b), hsp60, E1 and C may be target substrates for the Reg1p–Glc7p complex. To test the effects of glucose starvation on overall protein phosphorylation in the wild‐type strain, Melanie compared Figure 1G with E. This analysis showed that three of the proteins that were increased in phosphate content following REG1 deletion (Figure 1F) were also hyperphosphorylated in the wild type as a result of glucose starvation (Figures 1G and 2A). Importantly, comparison of Figure 1H with F showed that the absence of glucose did not cause additional proteins to be phosphorylated in the reg1Δ strain (Figure 2A). Since REG1 deletion was found to mimic the effects of glucose starvation on the yeast phosphoproteome, this finding further supports the hypothesis that Reg1p plays a critical role in the regulation of glucose‐sensing pathways.
To determine if deletion of the REG1 gene or the presence of raffinose result in significant alterations of intracellular ATP concentration, the nucleotide content of the wild‐type and reg1Δ strain was measured. This measurement would indicate whether the observed alterations in phosphorylation state in the reg1Δ strain, or following glucose starvation in the wild‐type case, were likely to be due to differences in either the cellular ATP pool or the specific activity of the γ‐phosphate of the nucleotide. However, in three separate measurements by reverse‐phase HPLC, the overall amounts of GTP, ATP, ADP and AMP recovered from both strains did not vary significantly (data not shown). Moreover, when grown in dextrose or raffinose, the specific activity of the γ‐phosphate of ATP from wild‐type or reg1Δ strains varied by <11.5 ± 3.5% (SDM, n = 3) with respect to one another.
Identification of phosphoproteins regulated by Reg1p and glucose by mixed peptide sequencing
To identify 32P‐labeled proteins that were phosphorylated in response to REG1 deletion and glucose starvation, the two‐dimensional gels were transferred to polyvinyl membrane (PVM) stained with amido black and the membranes autoradiographed. Eight phosphoproteins were selected for mixed peptide sequencing. Five were increased in phosphate content in response to REG1 deletion, namely Hxk2p spots a and b, hsp60, E1α and spot C. Three others, hsp71, elongation factor 1 (EF1) and initiation factor 5A (IF5A), were unaffected by the absence of Reg1p and were used as internal markers in mixed peptide sequencing to ensure that each gel was matched for protein loading (Figure 2B). Of the eight proteins that were selected for mixed peptide sequencing, all but one was identified unambiguously in the yeast database by the FASTF algorithm. The identified proteins included hsp71, hsp60, EF1, IF5A, the E1α subunit of pyruvate dehydrogenase (E1) and Hxk2p a and b (Table I). Spot C was not identified because it was below our sequencing sensitivity (<100 fmol). Expectation scores for the identified proteins ranged from 5.4e‐75 for Hxk2p spot b to 7.6e‐8 for the E1α subunit. In contrast, expectation scores for the next highest scoring non‐related proteins ranged from 0.001 to 1.2. The identification of Hxk2p in this present study is consistent with previous evidence that glucose‐induced dephosphorylation of Hxk2p in vivo is blocked in Glc7p point mutants (Glc7pT152K) and reg1Δ mutants (Randez‐Gil, 1998).
Interestingly, two‐dimensional SDS–PAGE resolved Hxk2p into two closely migrating species (a and b) regardless of growing conditions or REG1 deletion. This separation does not appear to be related to isoformic variation or differences in phosphate content (i.e. mono and diphospho forms). Hxk2p shares ∼70% homology with Hxkp1; however, from the alignments shown in Table I, two peptide sequences were recovered from both spot a and b that are unique to Hxk2p. One of the peptides, residues 304–314, has the sequence MSSGYYLGEIL in Hxk2p; the same region in Hxk1p contains the sequence MTSGYYLGELL. A second peptide, residues 321–332, has the sequence MYKQGFIFKNQD in Hxk2p; the same region in Hxk1p contains the sequence LNEKGLMLKDQD. Mono and diphospho forms of Hxk2p also seem unlikely because the wild‐type two‐dimensional gel also contained two closely resolving species of the protein that contained no [32P]phosphate (Figure 1A). The reason for the separation of Hxk2p into two species is therefore likely to be due to some other post‐translational modification on one of the Hxk2p molecules.
To quantitate further the overall change in phosphorylation of the sequenced proteins, each mixed peptide sequencing run was analyzed for PTH amino acid content. This analysis also allowed the amount of dephosphoprotein relative to the phosphorylated form to be determined in each case. Figure 2B shows that the total amount of each protein sequenced in each gel (phospho and dephospho combined) did not vary significantly between wild‐type and mutant or if either strain was grown in the absence of glucose. These findings demonstrate that deletion of Reg1p and glucose starvation promote phosphorylation of Hxk2p, E1α and hsp60, but do not change their expression at the protein level. Similarly, no difference in the amount of EF1β, IF5A and hsp71 was detected between wild‐type or mutant strains, demonstrating that the overall protein loading for each gel was equivalent.
Measurement of the effects of deletion of REG1 on hexokinase phosphatase activity
To test the effects of REG1 deletion on Hxk2p phosphatase activity in yeast, recombinant Hxk2p was prepared as a 32P‐labeled phosphoprotein substrate using PKA (Figure 3). Unlike its mammalian counterparts, yeast Hxk2p is phosphorylated in vivo at a single residue, Ser15, and phosphorylation of this amino acid is increased in response to glucose (Vojtek et al., 1990; Kriegel et al., 1993; Randez‐Gil et al., 1998). The identity of the protein kinase that phosphorylates Hxk2p in intact yeast is not known; however, Ser15 is phosphorylated by cAMP‐dependent PKA in vitro (Kreigel et al., 1994). To confirm that Reg1p–Glc7p directly dephosphorylates Hxk2p, we assayed hexokinase phosphatase activity in wild‐type and reg1Δ mutant cytosolic cell extracts. Assay of yeast cell extracts prepared from the particulate and cytosolic fraction with 32P‐labeled hexokinase revealed that >95% of the total hexokinase phosphatase activity in S.cerevisiae is present in the soluble fraction. This activity required up to 100 nM okadaic acid to inhibit it completely, indicating that it is a type 1 serine/threonine phosphatase (data not shown). To characterize yeast hexokinase phosphatase activity further, cell extracts were prepared from MCY3278 and FY250 strains, fractionated by anion‐exchange chromatography and fractions assayed for phosphatase activity using 32P‐labeled Hxk2p and phosphorylase a as the substrates (Figure 4A and B). Figure 4A shows that fractions from wild‐type yeast contained a major and a minor peak of Hxk2p phosphatase activity. Significantly, when fractions were assayed from the reg1Δ strain, the major Hxk2p phosphatase peak was absent. This result demonstrates that Reg1p targets PP‐1C (Glc7p) to dephosphorylate Hxk2p. When column fractions from both strains were assayed with phosphorylase a as the substrate, no significant difference in phosphatase activity was observed (Figure 4B). When assayed in the presence of 1 nM okadaic acid, all phosphorylase phosphatase activity was abolished in the column fractions, whereas hexokinase phosphatase activity was not affected (data not shown). This suggests that the majority of the phosphorylase phosphatase activity detected in the column fractions was due to PP‐2A. The finding that both the wild‐type and mutant strains contained phosphorylase phosphatase activity that was not affected by REG1 deletion suggests that Reg1p is selective in its actions towards Hxk2p. This finding is consistent with the general hypothesis of PP‐1 regulation in which its regulatory subunits target PP‐1C towards a selective set of substrates (Hubbard and Cohen, 1993). To determine the effects of REG1 deletion on PP‐1C expression in the yeast cytosol, anion‐exchange fractions were Western blotted with anti‐PP‐1C antibody (Figure 4C). In the wild‐type column fractions, cross‐reactivity with PP‐1C was observed in two major areas, fraction 4 (data not shown) and between fractions 26 and 40. In the reg1Δ fractions, cross‐reactivity was only observed in fraction 4, with no evidence of PP‐1C between fractions 26 and 40. The presence of PP‐1C in fractions 26–40 in the wild‐type case is therefore due to binding of the phosphatase to Reg1p. These findings further support the hypothesis that Reg1p targets PP‐1C to dephosphorylate Hxk2p in vivo. The finding that column fractions from the reg1Δ strain contained very little PP‐1C by Western blot analysis is consistent with observations by others (Hubbard and Cohen, 1993) and our laboratory (Campos et al., 1996; Damer et al., 1998) that the majority of cellular PP‐1 is localized in the particulate fraction. Our data suggest that in yeast, the Reg1p–Glc7p complex represents the major pool of PP‐1 that is present in the cytosol.
The effects of disruption of the Reg1p PP‐1C‐binding site on the yeast phosphoproteome
Recently, Barford and colleagues identified the motif (K/R)(V/I)XF, or (K/R)X(I/V)XF, in Reg1p and some other yeast PP‐1 subunits, as a putative PP‐1C‐binding site that is present in all known mammalian and yeast PP‐1 regulatory subunits (Egloff et al., 1997). To probe the functional significance of the (K/R)X(I/V)XF motif in Reg1p, point mutations were introduced into a LexA–Reg1p fusion protein at the phenylalanine residue (F468 in the Reg1p sequence). Two substitutions were made, F468D and F468R, to introduce a positive or negative charge into the binding site in place of the hydrophobic phenylalanine residue (note F468 is absolutely conserved in the primary sequences of all 20 of the mammalian and yeast PP‐1 regulatory subunits reported to date). Plasmids containing LexA–Reg1p, LexA–Reg1pF468R and LexA–Reg1pF468D fusions were transformed into the reg1Δ strain. The transformed strains were incubated with 32P or [35S]methionine and extracts prepared for two‐dimensional SDS–PAGE and autoradiography (Figure 5). Western analysis of LexA fusion protein immunoprecipitates prepared from each strain with LexA antibody confirmed that the fusion protein was expressed in all three cases (Figure 6A). Figure 5A–C shows that introduction of the LexA–Reg1p fusion proteins into the yeast proteome does not result in dramatic changes in the overall pattern of [35S]methionine‐labeled proteins when the gels were compared with one another. Significantly, however, Figure 5D shows that introduction of LexA–Reg1p into the reg1Δ strain completely rescues the normal pattern of phosphoprotein labeling that was observed in the FY250 strain earlier. In particular, the phosphate content of both Hxkp2 spots was restored to basal levels observed in the wild‐type state. In contrast, introduction of aspartic acid and arginine residues at F468 prevented reversion of the overall two‐dimensional phosphoprotein pattern back to the wild‐type phenotype (compare Figure 5E and F with D). In the case of the F468D mutation, the phosphorylation state of Hxk2p (a), Hxk2p (b) and E1α were increased 5.4 ± 1.2‐, 6.5 ± 2.2‐ and 3.2 ± 0.52‐fold (SDM, n = 2), respectively, compared with strains transformed with wild‐type LexA–Reg1p. Similar effects were also observed with the F468R mutants. None of the other 80 or so phosphoproteins detected in Figure 5D–F were altered significantly in phosphorylation state as a result of the transformations.
To test whether the two point mutations disrupted Hxk2p phosphatase activity of the Reg1p–Glc7p complex, cell extracts were fractionated by anion‐exchange chromatography and assayed with 32P‐labeled Hxk2p (Figure 6B). Figure 6B shows that transformation of LexA–Reg1p into the MCY3278 strain restores Hxk2p phosphatase activity. This finding further demonstrates that the LexA–Reg1p fusion protein is expressed as a fully functional Glc7p targeting subunit. However, when the column fractions from the F468D and F468R mutants were assayed with 32P‐labeled Hxk2p, no evidence of phosphatase activity towards the substrate was detected. These observations are consistent with the increased phosphorylation of Hxk2p observed in these mutants, and suggest that mutation of F468 is sufficient to inhibit Reg1p function in vivo. To examine further whether introduction of aspartic acid or arginine at F468 disrupts Reg1p binding to Glc7p, the LexA–Reg1p fusion protein immunoprecipitates were blotted for PP‐1C. Figure 6C shows that immunoprecipitates from LexA–Reg1p extracts contain Glc7p whereas those from the F468R and F468D mutants contained greatly reduced cross‐reactivity at 37 kDa. Two‐hybrid analysis further confirmed that the two point mutations also decreased interaction of Reg1p with Glc7p (Table II). These data demonstrate that the substitution at F468 was sufficient to disrupt binding of Reg1p to Glc7p in intact yeast. The finding that mutation of F468 to arginine or aspartic acid results in inhibition of Hxkp2 dephosphorylation and disrupts binding of Reg1p to Glc7p in intact yeast demonstrates that the motif (K/R)X(I/V)XF is likely to be essential to the targeting function of all PP‐1 regulatory subunits in vivo.
The effects of disruption of the Reg1p PP‐1C‐binding site on the pathways of glucose repression
The reg1Δ mutation relieves glucose repression of many genes, including SUC2, encoding invertase. To test the effect of mutations in the PP1‐binding site on the function of Reg1p in vivo, we tested LexA–Reg1pF468R and LexA–Reg1pF468D for their ability to restore glucose repression of SUC2 in a reg1Δ mutant. The wild‐type LexA–Reg1p effectively complemented the reg1Δ mutation and reduced SUC2 expression 130‐fold in glucose‐grown cells, whereas the mutant forms of LexA–Reg1p reduced SUC2 expression only 3‐fold (Table III). The mutant forms did not affect derepression of invertase activity in raffinose. Immunoblot analysis of this experiment indicated that the mutant proteins were present at levels ∼5‐fold lower than the wild‐type LexA–Reg1p (Table III). These findings indicate that mutation of the PP1‐binding site severely impairs, but probably does not completely abolish, Reg1p function in vivo in the glucose‐sensing pathway.
In this study, we have demonstrated that deletion of the Glc7p (PP‐1C)‐binding protein Reg1p in S.cerevisiae causes hyperphosphorylation of at least two enzymes governing the cellular fate of glucose, Hxk2p and E1α. Importantly, the effect of REG1 deletion on these proteins is mimicked by glucose starvation and is consistent with previous genetic and biochemical evidence demonstrating its essential role in the regulation of the glucose repression pathway. The hyperphosphorylation of Hxk2p that occurs in reg1Δ strains is consistent with the absence of Hxk2p phosphatase activity observed following anion‐exchange chromatography of extracts prepared from these cells. Importantly, the wild‐type phenotype was completely restored by expression of a LexA–Reg1p fusion protein in the reg1Δ strain. Collectively, these results demonstrate that Reg1p binds Glc7p to target the phosphatase to dephosphorylate Hxk2p both in vitro and in intact yeast. We also show that the putative PP‐1C‐binding motif identified in vitro by Egloff et al. (1996), (K/R)X(I/V)XF in yeast and (K/R)(I/V)XF in mammals, is a physiological PP‐1C‐binding site in vivo. Introduction of single point mutations at F468 was sufficient to disrupt binding of Reg1p to Glc7p and inhibit hexokinase phosphatase activity both in vitro and in vivo. Disruption of the PP‐1C‐binding site also significantly reduced the ability of the yeast to glucose repress, suggesting that the Glc7p–Reg1p complex is required in the glucose signaling pathway. The finding that this effect was not complete is likely to be explained by the presence of a second low affinity binding site for Glc7p on the Reg1p molecule. This hypothesis is consistent with the partial recovery of PP‐1C in the LexA fusion protein pull‐down experiments shown in Figure 6 and data by Hartshorne and colleagues showing that other PP‐1 regulatory subunits may also contain more than one PP‐1C‐binding site (Hartshorne et al., 1998). Importantly, REG1 deletion or Reg1pF468D/R substitution had discrete effects on the phosphorylation state of a select few proteins and did not bring about a global cellular change in protein phosphorylation or alter the expression pattern of the entire yeast proteome. This observation is consistent with the general hypothesis of PP‐1 regulation in which distinct regulatory subunits target the catalytic subunit to discrete substrates. Reg1p is therefore the fourth example of a PP‐1 regulatory subunit that has been shown to act directly in a substrate targeting capacity both in vivo and in vitro. The finding that Hxk2p is phosphorylated in response to glucose starvation (also shown previously by Randez‐Gil et al., 1998) suggests that Reg1p–Glc7p activity may be regulated in the intact cell. We are currently investigating this hypothesis.
The finding that disruption of Reg1p function causes hyperphosphorylation of Hxk2p in vivo suggests that one of the mechanisms by which the phosphatase subunit mediates its effects on the pathways of glucose repression is by controlling the phosphorylation state of hexokinase II. Yeast strains containing deletion of the Hxk2 genes are defective in their glucose repression response (Gancedo, 1998). However, there are conflicting reports as to the functional significance of Hxk2p phosphorylation at Ser15 in vivo. Studies by Ma et al. (1990), in which the N‐terminal 15 amino acids of Hxk2p were deleted, concluded that Ser15 was not required for hexokinase activity or glucose repression in vivo. In contrast, more recent studies by Randez‐Gil et al. (1998), in which Ser15 was mutated to a non‐phosphorylatable alanine, produced mutants that were no longer able to glucose repress. Randez‐Gil et al. also showed glucose‐dependent dephosphorylation and dimerization of Hxk2p, and in vitro, the Hxk2p‐S15A mutant enzyme was unable to form monomers and exhibited a lower Km for glucose. These findings suggest that Ser15 phosphorylation may enable Hxk2p to scavenge glucose more efficiently under conditions when the sugar is limiting.
The mechanisms that lead to the increased phosphorylation of E1α and hsp60 following REG1 deletion or glucose starvation are likely to be indirect, resulting from changes in overall metabolic state rather than a loss of selective phosphatase activity mediated by Reg1p. E1α is encoded by the PDA1 gene and is an essential component of the pyruvate dehydrogenase (PDH) complex that converts pyruvate to acetyl‐CoA. Like its mammalian counterpart, yeast E1α catalysis is regulated by phosphorylation (Uhlinger et al., 1986; James et al., 1995). However, the increased phosphorylation of E1α observed in response to REG1 deletion is unlikely to be the result of a loss of PDH phosphatase activity mediated by Reg1p. First, E1α is a mitochondrial enzyme and Glc7p (PP‐1C) is not known to be a mitochondrial enzyme. Indeed, in Western blotting experiments of isolated yeast mitochondria, we found no evidence of PP‐1C in this organelle (data not shown). Secondly, dephosphorylation and activation of E1α are exclusively under the control of a specific PDH phosphatase related to the protein phosphatase 2C family (for a review see Reed et al., 1985). Although the increased phosphorylation of hsp60 was relatively weak (1.24 ± 0.29‐fold) compared with that of Hxk2p and E1α, this response was observed consistently in all experiments. The increased phosphorylation of hsp60 may be a non‐specific stress response because similar increases in phosphorylation of hsp60 (and other heat shock family members) have been observed following deletion of non‐related phosphatases and kinases in yeast (T.A.J.Haystead, unpublished data). Interestingly, as shown in Figure 5, the phosphorylation state of hsp60 was not affected by expression of the mutant fusion proteins, suggesting that increased phosphorylation of the protein observed in the reg1Δ strain is a non‐specific effect induced by the absence of the Reg1p protein itself.
It is likely that only changes in the most abundant phosphoproteins that are substrates for the Reg1p–Glc7p complex were detected with the two‐dimensional mapping protocols described in this study. Following two‐dimensional analysis, the Melanie program detected on average 450 ± 36 (SDM, n = 5) distinct [35S]methionine‐labeled proteins per gel, representing ∼6% of the 8300 genes in the entire S.cerevisiae genome. Therefore, the possibility exists that other proteins are also targets of Reg1p, but were not detected because they have a low level of expression or they did not enter the two‐dimensional gel. Indeed, genetic evidence suggests that although Reg1p has unique roles in the regulation of the glucose repression pathway, it may also act synergistically with its structural homolog Reg2p to control cell growth (Frederick and Tatchel, 1996). Furthermore, two‐hybrid analysis has demonstrated a direct interaction between Reg1p and Snf1, suggesting that the kinase may be a substrate for the Reg1p–Glc7p complex (Ludin et al., 1998). We are currently enriching the phosphoprotein content of our extracts prior to two‐dimensional SDS–PAGE in order to evaluate the effects of glucose starvation and REG1 deletion on the phosphorylation of lower abundance phosphoproteins.
The types of proteomic approaches that were utilized in this study to identify Hxk2p as a physiological target of the Reg1p–Glc7p complex herald a new age in which the functional consequences of genetic manipulations can be assessed directly and fully on the proteome as a whole. Because the microsequencing technology employed in this study gives an absolute identification of the protein sequenced, logical genetic and biochemical experiments can be carried out to probe the function and address the physiological relevance in the context of the intact cell. As the DDBJ/EMBL/GenBank databases are completed, this technology will have increasing relevance and become routinely applicable to studies in higher order eukaryotes, such as mammalian and plant cells, in addition to yeast.
Materials and methods
Strains and growth media
The S.cerevisiae strains used were MCY3278 (MATα reg1Δ::URA3 his3Δ200 leu2Δ1 trp1Δ63 ura3‐52 SUC2) and FY250 (MATα. his3Δ200 leu2Δ1 trp1Δ63 ura3‐52 SUC2). Both have the S288C genetic background. Cells were grown in the rich medium YPD [1% yeast extract (Difco), 2% bacto peptone (Difco), 2% dextrose (Sigma)] or in phosphate‐depleted YPD. Yeast were grown in 2% raffinose in those experiments involving glucose starvation.
Phosphate depletion of YPD (YPD‐PO4)
To 1 l of YPD liquid medium, 10 ml of 1 M MgSO4 and 10 ml of concentrated aqueous NH3 (27%) were added. The solution was stirred at room temperature for 30 min and filtered through Whatman No. 1 filter paper. After filtering, the solution was readjusted to pH 5.8 and autoclaved. This method produces medium containing low, 0.1 mM inorganic phosphate without decreasing organic phosphates.
A single yeast colony was used to inoculate 4 ml of YPD medium and was grown overnight at 37°C with vigorous shaking. In the morning, 50 ml of YPD was inoculated from the overnight culture to a final OD660 of 0.2. These cells were allowed to grow for two doublings and harvested by centrifugation at 800 g for 10 min. The harvested cells were washed with 10 ml of YPD‐PO4 and centrifuged at 800 g. The cells were resuspended in 15 ml of YPD‐PO4. YPD‐PO4 (50 ml) was inoculated with the resuspended cells to an OD660 of 0.2 and grown for two doublings (∼3 h). The cells were treated with 18.5 MBq of [32P]orthophosphate for 1 h and harvested by centrifugation (800 g for 10 min).
Preparation of yeast extracts
The harvested cells were resuspended into 10 ml of ice‐cold 5% trichloroacetic acid (TCA; J.T.Baker) and incubated on ice for 10 min. This mixture was centrifuged at 800 g for 10 min and the TCA decanted. The pellet was washed with 1 ml of acetone and centrifuged. After aspirating the acetone, the pellet was allowed to air dry overnight. Next, 0.5 ml of acid‐washed glass beads was added to the pellet with 480 μl of lysis buffer (0.1 M Tris–HCl pH 8, 0.3% SDS, 2.5% β‐mercaptoethanol, 150 μM sodium orthovanadate, 1 μM microcystin). The extract was vortexed for 10 min, alternating 1 min on ice, 1 min vortexing. This mixture was incubated at 100°C for exactly 10 s and allowed to recover on ice for 1 min. A 20 μl aliquot of RNase solution (0.5 M Tris–HCl pH 7.0, 0.05 M MgCl2, 200 U/ml RNase A) and 2 U of DNase I (Promega) were added and incubated on ice for 2 min. Samples used for two‐dimensional electrophoresis were saturated with urea by the addition of 500 mg of urea (Amresco) and 133 μl of buffer (4.75 M urea, 4% 3‐[(3‐cholamidopropyl)dimethylammonio]‐1‐propanesulfonate, 1% Bio‐Rad Biolytes, 5% β‐mercaptoethanol). The glass beads were removed by centrifugation at 13 000 r.p.m. and aspiration of the sample. The supernatant was brought to 25% TCA by the addition of ice‐cold 100% TCA and allowed to precipitate on ice for 10 min. The precipitate was pelleted by centrifugation at 14 000 g for 10 min. The resulting pellet was resuspended into 500 μl of two‐dimensional sample buffer (9.5 M urea, 2.0% Triton X‐100, 5% β‐mercaptoethanol). The protein concentration was measured using the method of Bradford (1976).
Determination of the specific activity of the γ‐phosphate of ATP
Following labeling with [32P]orthophosphate and homogenization, 100 μl of cell extract was treated with 10 μl of perchloric acid (55%) to precipitate the proteins. The extract was centrifuged (13 000 g for 5 min) and a nucleotide‐enriched fraction containing ATP, ADP and AMP prepared by extraction of the supernatant with a 10% excess (v/v) mixture of tri‐n‐octylamine/1,1,2,trichlorotrifluoro‐ethane (1:1 v/v). After vortexing, the extract was centrifuged (14 000 g for 3 min) and the neutralized upper aqueous layer removed for analysis by HPLC. The specific activity of the γ‐phosphate of ATP was determined following purification of ATP and ADP by anion‐exchange HPLC as described previously (Haystead and Hardie, 1988).
Two‐dimensional electrophoresis and mixed peptide sequencing
Protein maps were prepared by two‐dimensional electrophoresis. Isoelectric focusing was accomplished as the first dimension by preparation of a tube gel (4 cm×1 mm) comprised of 4% acrylamide, 0.8% 5/7 ampholyte, 0.8% 6/8 ampholyte, 0.4% 3/10 ampholyte and 2% Triton X‐100. NaOH (100 mM) was used as the basic buffer and 10 mM H3PO4 as the acidic buffer. The tube gel was pre‐focused using the following program: 200 V 15 min, 300 V 25 min, 400 V 25 min. The proteins were loaded (50 μg used for silver staining and 200 μg for transfer) onto the basic end of the gel and overlaid with 1 vol. of overlay buffer (9 M urea, 0.8% 5/7 ampholyte, 0.2% 3/10 ampholyte). Isoelectric focusing was performed by application of 500 V for 15 min followed by 750 V for 5.5 h. After completion of the program, the gel was expelled from the tube, laid upon a 12% polyacrylamide gel (6×10 cm×0.75 mm) and resolved at 150 V. The gels were equilibrated to transfer buffer (3 g/l Tris base, 14.25 g/l glycine, 200 ml/l methanol). The proteins were electroblotted to PVM pre‐wetted with methanol at 30 V overnight. The transferred proteins were stained with amido black (1 mg/ml in 10% acetic acid, 10% methanol) for 1 min, washed once with destain buffer (10% acetic acid, 10% methanol) and successively with water. After destaining, the membrane was air dried and applied to X‐ray film for autoradiography. The autoradiograph from each gel was compared using the Melanie software from Bio‐Rad. Spots of interest were aligned to the membrane and the corresponding stained bands excised. The excised pieces were treated with 200 μl of cyanogen bromide solution (500 mg/ml cyanogen bromide in 70% formic acid) for 90 min. The reaction was terminated by removal of the membrane and three washes with 1 ml of deionized water, then methanol followed by 50% methanol/water (v/v). The treated piece of PVM was placed in an Applied Biosystem 494 protein sequencer and 8–18 cycles of pulsed liquid chemistry carried out. The mixed peptide sequences generated were sorted and matched against the yeast protein databases by the FASTF algorithm (Damer et al., 1998).
Preparation of 32P‐labeled protein substrates
Purified bovine cAMP‐dependent PKA and phosphorylase b were gifts from John Lawrence (University of Virginia). Recombinant yeast hexokinase II was purchased from Sigma Chemical Co. For 32P‐labeled hexokinase II, 20 mg of the freeze‐dried protein was resuspended in HEPES buffer [25 mM HEPES pH 7.4, 1 mM dithiothreitol (DTT)]. The protein was phosphorylated to ∼1 mol/mol by the addition of 10 μl (4 μg/μl) of the native catalytic subunit of cAMP‐dependent PKA in the presence of 200 μM [γ‐32P]ATP and 5 mM MgCl2 for 30 min at room temperature. To determine the position of the phosphorylated amino acid, a sample (100 pmol) of the labeled protein was digested with CNBr (1 mg/ml for 16 h at room temperature). The major phosphopeptide was purified by reverse‐phase HPLC and sequenced. [32P]Phosphorylase a was prepared as described previously (Campos et al., 1996).
Yeast cell pellets were prepared from wild‐type and mutant strains grown in dextrose or raffinose by centrifugation of 50 ml cultures as described above. The pellets were resuspended in 10 ml of Tris–HCl buffer [50 mM Tris–HCl pH 8.0 at room temperature, 1 mM DTT, 1 mM EGTA, 1 mM EDTA, 1 mM benzamidine, 6 μg/ml leupeptin, 10 mM phenylmethylsulfonyl fluoride (PMSF)]. Glass beads were added and the cells homogenized for 8 min following the vortexing protocol outlined above. The extracts were centrifuged for 60 min at 100 000 g and the supernatant passed through a 0.2 μm filter. The supernatant was loaded onto a Pharmacia mono Q anion‐exchange smart column (0.1×1.0 cm) previously equilibrated in Tris–HCl buffer. The column was developed in this buffer over 45 min at a flow rate of 25 μl/min using a linear gradient to 600 mM NaCl. Hexokinase and phosphorylase phosphatase activity was measured in the collected fractions (25 μl). To correct for any apparent differences in phosphatase activity between samples due to protein loading, absorbance at 280 nm was monitored on‐line throughout anion‐exchange chromatography.
Phosphatase activity assay
Hexokinase and phosphorylase phosphatase activity in the collected fractions was measured simultaneously by automated robotic assay. A 10 μl aliquot of each column fraction was placed in a microtiter plate well. The robot added 50 μl of HEPES buffer containing 10 μM (final) of either 32P‐labeled hexokinase or 32P‐labeled phosphorylase a. This mixture was allowed to react for 10 min at room temperature and terminated by the addition of 100 μl of 25% TCA. The entire plate was centrifuged at 2500 r.p.m. for 10 min and 120 μl of the supernatant pipetted into 1.0 ml minitubes. Scintillation fluid (0.5 ml) was added and the amount of released phosphate determined in a Beckman LS6500 scintillation counter.
Site‐directed mutagenesis of the PP‐1C‐binding site in Reg1p
To introduce F468D and F468R point mutations into the PP‐1C‐binding site on Reg1p, the appropriate oligonucleotides were synthesized and used for plasmid mutation on the complementary strand. In order to introduce the F468R point mutation into Reg1p, site‐directed mutagenesis using the Morph kit (5 Prime→3 Prime, Inc., Boulder, CO) was performed on the pRJ65 plasmid (LexA–Reg1; Tu and Carlson, 1995) with the RegMut3 oligonucleotide. The DNA sequence of the plasmid was verified with an Applied Biosystems PRISM 377 DNA Sequencer (Foster City, CA). The mutated plasmids were introduced into strain MCY3278 carrying the genomic deletion of REG1 as described by Gietz et al. (1992).
We thank Perkin‐Elmer for the generous support in the purchase of the 492 cLc Procise sequenator. This work was support by grants (HL19242 and DK52378A) to T.A.J.H. and (GM354095) to M.C. from the National Institutes of Health. The FASTF and TFASTF programs can be accessed at http://fasta.bioch.virginia.ed. or by contacting T.A.J.H.
- Copyright © 1999 European Molecular Biology Organization