ABSTRACT
Eukaryotic cells have evolved mechanisms for ensuring growth and survival in the face of stress caused by a fluctuating environment. Saccharomyces cerevisiae has two homologous glycerol-3-phosphate dehydrogenases, Gpd1 and Gpd2, that are required to endure various stresses, including hyperosmotic shock and hypoxia. These enzymes are only partially redundant, and their unique functions were attributed previously to differential transcriptional regulation and localization. We find that Gpd1 and Gpd2 are negatively regulated through phosphorylation by distinct kinases under reciprocal conditions. Gpd2 is phosphorylated by the AMP-activated protein kinase Snf1 to curtail glycerol production when nutrients are limiting. Gpd1, in contrast, is a target of TORC2-dependent kinases Ypk1 and Ypk2. Inactivation of Ypk1 by hyperosmotic shock results in dephosphorylation and activation of Gpd1, accelerating recovery through increased glycerol production. Gpd1 dephosphorylation acts synergistically with its transcriptional upregulation, enabling long-term growth at high osmolarity. Phosphorylation of Gpd1 and Gpd2 by distinct kinases thereby enables rapid adaptation to specific stress conditions. Introduction of phosphorylation motifs targeted by distinct kinases provides a general mechanism for functional specialization of duplicated genes during evolution.
INTRODUCTION
Cytoplasmic glycerol-3-phosphate dehydrogenases (GPDs) catalyze the NADH-dependent reduction of dihydroxyacetone phosphate (DHAP) to glycerol-3-phosphate. This reaction is the first step in a biosynthetic pathway leading to glycerol production that is conserved throughout eukaryotes (Fig. 1A). In budding yeast, this pathway is important for cell growth in a number of contexts. During exponential growth, GPD activity maintains cellular redox balance by reoxidizing NADH produced from glycolysis and is required for anaerobic growth in synthetic medium (6, 12). Under conditions of high osmolarity, glycerol produced by this pathway accumulates intracellularly and plays an essential osmoprotective role (2). Independent of glycerol production, GPDs participate in a mitochondrial NADH shuttle that is dispensable for aerobic growth but may promote an increased life span caused by caloric restriction (26, 44). GPDs also serve to reduce the accumulation of the toxic metabolic by-product methylglyoxal, presumably through disposal of its precursor DHAP (1). Lastly, glycerol-3-phosphate is a precursor in phospholipid biosynthesis, though an alternate route via acylation of DHAP compensates in the absence of GPDs (30).
Gpd1 and Gpd2 are reciprocally phosphorylated at a conserved site in response to glucose. (A) The NADH-consuming biosynthetic pathway leading to glycerol production in yeast involving Gpd1 and Gpd2. (B) Schematic depiction of primary sequence features of Gpd1 and Gpd2, i.e., the N-terminal localization sequence (blue), a conserved phosphorylation site (red), and the catalytic domain (gray). Inset, sequence context of the conserved phosphorylation site (underlined). (C) Reciprocal phosphorylation of Gpd1 and Gpd2 in response to glucose. Cells expressing WT Gpd1-His6-GFP (left) or Gpd2-His6-GFP (right) or the indicated phosphorylation site mutant proteins from their own promoters on low-copy-number plasmids were propagated to mid-exponential phase in medium containing a high glucose concentration (2%) (H), and then a portion of each culture was shifted to the same medium a containing limiting glucose concentration (0.05%) (L). After 90 min, samples of each culture were harvested and lysed and the His6-tagged proteins in the extract were enriched by immobilized metal affinity chromatography and then resolved by Phos-tag PAGE before (−) or after (+) treatment with λ protein phosphatase (λPP) and analysis by immunoblotting with anti-GFP antibodies. (D) Analysis of Gpd1 using a phospho-specific antibody. The experiment was performed as in the left side of panel C, except that samples were analyzed by standard SDS-PAGE, followed by immunoblotting with anti-phospho-RXRXXS to detect Gpd1 phosphorylation and anti-GFP antibody to detect total Gpd1.
Despite catalyzing the same chemical reaction with similar kinetics (3, 6), Gpd1 and Gpd2 have only partially overlapping function in vivo. For example, single deletion of GPD2, but not GPD1, leads to slow growth under anaerobic conditions (6, 12). Conversely, adaptation to high osmolarity is more dependent on GPD1 (2). The distinct functions of the two GPDs has been attributed in part to differential protein compartmentalization: Gpd2 localizes to both the cytosol and mitochondria, while Gpd1 is found both in the cytosol and in peroxisomes (36, 69). In addition, GPD1 and GPD2 are differentially regulated at the transcriptional level. GPD1, for example, is transcriptionally upregulated in response to hyperosmotic stress and is generally regarded as a key transcriptional target of the Hog1 mitogen-activated protein kinase (MAPK) pathway (2). Likewise, transcription of the GPD2 gene increases under anaerobic conditions (6). Direct evidence that transcriptional induction of GPDs is functionally significant, however, is lacking. Indeed, recent studies suggest that the Hog1-dependent transcriptional response is dispensable for adaptation to high osmolarity (13, 73).
Collectively, these observations suggest that there may be alternative modes of regulation for Gpd1 and Gpd2. Here we show that yeast GPDs are subject to posttranslational regulation through phosphorylation at a conserved site near the catalytic domain. Revealingly, we find that the two isozymes are reciprocally phosphorylated in vivo by different protein kinases. Phosphorylation curtails GPD activity in vitro and in vivo, and we show that Gpd1 dephosphorylation is an important component of the rapid pretranscriptional response to hyperosmotic stress. Thus, differential phosphorylation of Gpd1 and Gpd2 serves as a mechanism that contributes to their functional specialization, thereby facilitating the adaptation of yeast to diverse environments.
MATERIALS AND METHODS
Yeast strains and medium.All of the single-deletion strains used in this study (gpd1Δ, gpd2Δ, ypk1Δ, ypk2Δ, sch9Δ, and snf1Δ strains) were derived from strain BY4741 (MATα his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) of the S288C lineage by replacement of the corresponding gene with the KanMX cassette (75) and were purchased from Open Biosystems. The ypk1ts ypk2Δ mutant, generated in another S288C-derived strain (YPH499), was previously described (16). The ypk1Δ ypk2Δ mutant strain (W303 genetic background) harboring a YPK1 or ypk1-as low-copy-number expression plasmid was previously reported (11) and kindly provided by Tobias Walther (Yale University). The elm1Δ tos3Δ sak1Δ mutant strain (63) was generously provided by Martin Schmidt (University of Pittsburgh). The gpd1Δ gpd2Δ mutant strain was generated by PCR-based gene deletion as follows. The HIS3 gene was amplified from a nonreplicating plasmid template using primers G2KO-F and G2KO-R (Table 1). The PCR product was used for transformation of the BY4741 gpd1Δ mutant strain, followed by selection on medium lacking histidine. Disruption of the entire GPD2 open reading frame (ORF) was confirmed by PCR using the G2KO-A/HIS3-B and HIS3-C/G2KO-D primer pairs (Table 1).
Primers used in this study
Unless otherwise noted, yeast strains were cultured at 30°C in a synthetic complete medium containing 2% glucose and lacking leucine (SC-Leu), uracil (SC-Ura), or both (SC-Ura-Leu), as appropriate for plasmid maintenance. Powdered components for yeast medium were obtained from MP Biomedical (CSM dropout formula) and BD Biosciences (yeast nitrogen base, yeast extract, and peptone). Glucose and myriocin were obtained from Sigma, galactose was from USB, and 1-(1,1-dimethylethyl)-3-(1-naphthalenylmethyl)-1H-pyrazolo[3,4-d]pyrimidin-4-amine (1NM-PP1) was from EMD Millipore.
Vector construction and mutagenesis.All of the primers used in vector construction and mutagenesis are shown in Table 1, and the plasmids used are listed in Table 2. To generate the low-copy-number Gpd1 expression vector p415-Gpd1, we amplified the entire Gpd1 ORF region and 1 kb of the upstream sequence by PCR using primers G1P-F and G1-R with BY4741 genomic DNA as the template. The PCR product was cloned into the XhoI and NotI sites of pRS415-GFP. The low-copy-number Gpd2 expression vector p415-Gpd2, encoding the entire ORF plus 1 kb of the upstream sequence, was constructed in a similar way using primers G2P-F and G2-R. For expression of Gpd1 under the control of the GPD2 promoter, the GPD2 promoter region and the Gpd1 coding sequence were amplified individually, fused by overlap extension PCR, and subcloned into p415-GFP. For expression in bacteria, the full Gpd1 ORF and the Snf1 catalytic domain coding sequence (residues 41 to 315) were amplified by PCR using the primers listed in Table 1 and introduced into pGEX-4T2 at the BamHI and NotI sites.
Yeast plasmids used in this study
Plasmids bearing the point mutations listed in Table 2 were generated by QuikChange mutagenesis (Stratagene) according to the manufacturer's instructions. All plasmids were confirmed by sequencing through the entire coding sequence.
Protein expression and purification.Wild-type (WT) and mutant proteins used for in vitro enzyme assays (Gpd1, Gpd2, Ypk1, Ypk2, Sch9, and Yck1) were expressed and purified in yeast from the galactose-inducible tandem affinity purification tag plasmid pBG1805 using YSC3880 (MATa pep4-3 his4-580 ura3-53 leu2-3,112) as the host strain according to the following protocol originally described by Gelperin et al. (32). Yeast cells transformed with the appropriate expression plasmid were grown in 1 ml of SC-Ura with 2% dextrose overnight, washed with SC-Ura plus 2% raffinose, and diluted 1,000-fold into 400 ml of the same medium. After growth at 30°C for 15 h, fusion protein expression was induced by the addition of 200 ml of 3× YEP-Gal (3% yeast extract, 6% peptone, 6% galactose). After 6 h of incubation, cells were pelleted, washed with ice-cold water, and stored at −80°C.
Lysates from harvested yeast cells were obtained by resuspension in 250 μl of yeast lysis buffer (50 mM Tris-HCl at pH 7.5, 150 mM NaCl, 1 mM EGTA, 10% glycerol, 0.1% Triton X-100, 0.5 mM DTT, 1 mM phenylmethylsulfonyl fluoride [PMSF], 1× complete protease inhibitor cocktail [Roche]) with 150 mM NaCl and vortexing for 2 × 3 min at 4°C with 250 μl of acid-washed glass beads (0.5 mm), followed by centrifugation for 1 min at 13,000 rpm. The supernatant solution was transferred to a fresh tube, and the residual pellet was resuspended in yeast lysis buffer with 650 mM NaCl and vortexed and pelleted as before. The combined lysates were mixed with 25 μl of IgG-Sepharose beads (GE Healthcare) and rotated for 2 h at 4°C. Beads were pelleted (2,000 rpm, 5 min, 4°C) washed (5 min, 4°C) twice with wash buffer (50 mM Tris-HCl at pH 7.5, 150 mM NaCl, 10% glycerol, 0.1% Triton X-100) and twice with elution buffer (50 mM Tris-HCl at pH 7.5, 150 mM NaCl, 25% glycerol, 0.1% Triton X-100). Beads were resuspended in 200 μl elution buffer containing 40 μg/ml glutathione S-transferase (GST)–3C protease and rotated overnight at 4°C. After centrifugation, the supernatant was mixed with glutathione-Sepharose beads for 2 h to remove the 3C protease. Following centrifugation, the supernatant was snap-frozen in aliquots at −80°C.
For purification of GST-tagged Gpd1 and Snf1 from Escherichia coli BL21(DE3), cells transformed with the appropriate expression plasmid were grown in LB medium to an optical density at 600 nm (OD600) of 0.6 and expression was induced by adding isopropyl-β-d-thiogalactopyranoside to 0.8 mM for 4 h at 30°C. Cells were pelleted, washed once with cold water, and snap-frozen. The frozen pellets were resuspended in 5 ml ice-cold resuspension buffer (20 mM Tris, 140 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol [DTT], 10 μg/ml leupeptin, pH 7.5). Cells were treated with 100 μg/ml lysozyme for 10 min on ice and lysed by adding sodium deoxycholate to 0.05%. After the addition of 25 μl 200 mM PMSF, the suspension was incubated for 15 min at room temperature, 30 U/ml DNase I and 50 mM MgCl2 were added, and incubation was continued for an additional 15 min. Lysates were cleared by centrifugation at 20,000 × g for 15 min at 4°C. Proteins were isolated by incubating the cleared lysate with 500 μl of prewashed glutathione-Sepharose 4B (GE Healthcare) at 4°C for 2 h with mixing. Resin was washed twice for 5 min with ice-cold GST wash buffer (50 mM Tris, 50 mM NaCl, 1 mM DTT, 0.01% NP-40, 10% glycerol, pH 8.0). Proteins were eluted from the resin by incubation for 10 min at 4°C with 500 μl GST wash buffer containing 10 mg/ml reduced glutathione. The eluate was dialyzed overnight at 4°C into wash buffer. Dialyzed protein was snap-frozen in aliquots and stored at −80°C.
For use in kinase assays, the purified GST-Snf1 catalytic domain was activated in vitro by Elm1. Purified, tandem affinity purification (TAP)-tagged Elm1 (27) (0.1 μg, obtained from Martin Schmidt) was mixed with 1 μg of the Snf1 catalytic domain in 25 μl of reaction buffer (20 mM HEPES at pH 7.4, 10 mM MgCl2, 1 mM DTT, 0.1% Tween 20, 100 μM ATP) and incubated for 1 h at 30°C.
Analysis of protein phosphorylation in yeast.For analysis of GPD phosphorylation in vivo, 10 μl of a saturated overnight culture of yeast harboring the appropriate low-copy-number expression plasmid was inoculated into 10 ml of SC-Leu medium plus 2% dextrose. After growth at 30°C for 12 h, a portion of the culture was left in the same medium, while the remainder was harvested by centrifugation, washed with prewarmed water, and resuspended in the indicated prewarmed medium (SC-Leu plus 0.05% dextrose or SC-Leu plus 2% dextrose containing either 1 M NaCl, 1 M sorbitol, or 500 nM 1NM-PP1) and incubation was continued for the indicated time. Cells were harvested by centrifugation and stored at −80°C.
Yeast cells were lysed as described above for purification of overexpressed proteins, using 100 μl buffer for each round of lysis. The combined samples were mixed with 20 μl of Talon resin (Clontech), incubated for 1 h with rotation, and quickly washed with wash buffer. Proteins were eluted with 100 μl of imidazole buffer (20 mM Tris-HCl at pH 7.5, 100 mM NaCl, 250 mM imidazole, 1 mM DTT, 10 μg/ml leupeptin) for 10 min at 4°C. Samples (10 μl) were resolved by Phos-tag SDS-PAGE using 7.5% acrylamide, 100 μM Phos-tag acrylamide (Wako Pure Chemical Industries, Ltd.), and 100 μM MnCl2. Immediately following electrophoresis, Phos-tag gels were washed sequentially with transfer buffer containing 50 mM EDTA and fresh transfer buffer for 10 min each, transferred to PVDF membrane, and probed with anti-green fluorescent protein (anti-GFP) antibody (1:5,000; Clontech).
Enzyme assays.For radiolabel assays, purified kinase (0.1 μg) and substrate (0.1 μg) were incubated in 25 μl kinase assay buffer (20 mM HEPES at pH 7.4, 10 mM MgCl2, 1 mM DTT, 0.1% Tween 20, 100 μM ATP, 0.25 μCi of [γ-33P]ATP) and incubated for 30 min at 30°C. Reactions were terminated by the addition of SDS-PAGE sample buffer and 5 min of boiling. Proteins were resolved by SDS-PAGE and analyzed using a phosphorimager (Bio-Rad). For immunoblot analysis, kinase assays were conducted in the absence of radiolabeled ATP, proteins were resolved by standard 10% SDS-PAGE or Phos-tag SDS-PAGE using 7.5% acrylamide (37.5:1 monoacrylamide/bisacrylamide ratio), 50 μM Phos-tag acrylamide, and 50 μM MnCl2. Gels were washed as described above, transferred to PVDF membrane, and immunoblotted with rabbit anti-Akt phosphosubstrate (anti-RXRXXpS, 1:1,000; Cell Signaling Technology, Inc.) or antihemagglutinin (anti-HA; 1:2,000; Covance) polyclonal antibodies.
For the immunoprecipitation-kinase assay (see Fig. 3G), yeast cells harboring pBG1805-Ypk1 or a control plasmid were lysed after galactose induction and challenge with a high salt concentration, as indicated. Ypk1 was immunoprecipitated using IgG-Sepharose (GE Healthcare), and washed beads were incubated with bacterially expressed Gpd1 in the presence of ATP. Gpd1 phosphorylation was assessed by immunoblotting using the Akt phosphosubstrate antibody.
Enzymatic activity of Gpd1 and Gpd2 was determined by spectrophotometric analysis of NADH reduction by the following protocol as previously described (15). Purified GPD (0.25 μg WT or mutant) was incubated with Ypk1 (0.25 μg), Snf1 (0.25 μg), or control buffer, as appropriate, in 25 μl kinase assay buffer (20 mM HEPES at pH 7.4, 10 mM MgCl2, 1 mM DTT, 0.1% Tween 20, 100 μM ATP) for 30 min at 30°C. Kinase reactions were stopped by adding EDTA (10 mM). Aliquots of the kinase reaction mixture (5 μl containing 50 ng GPD) were incubated with 95 μl reaction buffer (9.2 mM triethylamine at pH 7.2, 0.227 mM NADH, 1.93 mM DHAP, 9.2 mM EDTA, 0.184 mM β-mercaptoethanol) in a UV-transparent 96-well plate (Corning). Enzyme activity was determined in a plate-reading spectrophotometer (Tecan Safire) by monitoring the decrease in NADH absorbance at 340 nm for 30 min at room temperature. The activity was calculated from the linear rate during the initial reaction velocity period.
To measure glycerol production in untreated cultures, the indicated strains transformed with the indicated plasmids were grown in SC-Leu medium with 2% glucose. When cultures reached an OD600 value of 1.0 (exponential phase) or 11 (early stationary phase), aliquots (1 ml) were snap-frozen and processed as described below. To assay glycerol production following various treatments, cultures of the indicated yeast strains transformed with the indicated plasmids were grown to an OD600 value of approximately 1.0 in SC-Leu plus 2% glucose. Cells were pelleted and resuspended to an OD600 value of 1.0 in fresh medium (SC-Leu plus 0.05% glucose or SC-Leu plus 2% glucose containing an additional 1 M NaCl, 1 M sorbitol, or 500 nM 1NM-PP1 as indicated). Aliquots (1 ml) were withdrawn after 15 min, 30 min, or 1 h as indicated, snap-frozen, and processed as follows. To determine the glycerol concentration, frozen cultures were heated to 100°C for 10 min. After centrifugation at 13,000 rpm for 1 min, supernatants were stored at −20°C until analyzed. The glycerol concentration was determined using a glycerol detection kit (BioVision Research) as directed by the manufacturer. The glycerol concentration was calculated from a standard curve.
Yeast growth assays.To assay growth in liquid culture following hypertonic stress, cells were grown to exponential phase (OD600 of 1.0) in a selective medium (SC-Leu) with 2% glucose. Cells from 20 ml culture were harvested by centrifugation (4,000 rpm, 5 min) at room temperature. Cell pellets were washed once with fresh prewarmed (30°C) medium, and cells were harvested again by centrifugation. Harvested cells suspended with SC-Leu medium with 2% glucose and 1 M NaCl or 1 M sorbitol and were cultured in a shaking incubator at 30°C. Cell densities (OD600) were measured at 15-min intervals.
For growth assays on solid medium, overnight cultures in selective medium were diluted to an OD600 of 1.0. A series of 10-fold dilutions was spotted onto agar plates containing standard selective medium with or without an additional 1 M NaCl or 1 M sorbitol, and the plates were incubated at 30°C for 3 days.
RESULTS
Gpd1 and Gpd2 are reciprocally phosphorylated in response to glucose.We recently reported a large-scale analysis of protein kinase phosphorylation site motifs in budding yeast (50). On the basis of this work, potential substrates for 62 yeast kinases were predicted through a combination of peptide library screening and bioinformatic analysis (42). Among these predictions, Gpd1 and Gpd2 were identified as candidate substrates for Snf1, the yeast ortholog of the AMP-activated protein kinase catalytic subunit. Gpd1 and Gpd2 each have a single predicted Snf1 phosphorylation site at an analogous position amino terminal to its catalytic domain (Fig. 1B). Phosphopeptides corresponding to this site from both Gpd1 and Gpd2 have been frequently identified by mass spectrometry (MS) in phosphoproteomic studies, suggesting that they were authentic phosphorylation sites in vivo (4, 10, 19, 28, 45, 61). Interestingly, closer inspection of the surrounding sequences revealed that the phosphorylation sites for Gpd1 and Gpd2 fell within distinct sequence motifs. The Gpd2 phosphorylation site (Ser72) conforms to the canonical Snf1 consensus sequence (L/I/M-X-R-X-X-S-X-X-X-L/I) (21) confirmed by our peptide library screening. In contrast, the analogous site in Gpd1 (Ser24) diverges from this motif in that it has an Arg situated five residues upstream of the phosphoacceptor site (the P −5 position). These observations suggested that Gpd1 and Gpd2 might be phosphorylated by distinct kinases in vivo, perhaps under different conditions.
To assess the phosphorylation of these sites in vivo, we expressed epitope-tagged versions of these proteins in yeast cells and analyzed their migration by using phosphate affinity (Phos-tag) PAGE, in which the mobility of phosphoproteins is retarded (40). To determine whether Gpd1 and Gpd2 undergo phosphorylation under distinct conditions, we first compared cultures growing exponentially in excess (2%) glucose with parallel cultures shifted briefly into limiting (0.05%) glucose. The Gpd1 species observed in high-glucose medium was converted to a higher-mobility species in low-glucose medium, equivalent to that produced in vitro by phosphatase treatment (Fig. 1C, left). Thus, under high-glucose conditions, the population of Gpd1 molecules was almost quantitatively in a phosphorylated state. A Gpd1-S24A mutant exhibited mobility equivalent to that of dephosphorylated WT Gpd1 in high- or low-glucose medium, indicating that the observed mobility shift required phosphorylation at the expected site. To confirm that the mobility shift was due to specific phosphorylation at Ser24, we performed a similar experiment in which Gpd1 phosphorylation was assessed by immunoblotting with a phosphospecific antibody that recognizes R-X-R-X-X-phospho-S sites (Fig. 1D). We observed that Gpd1 is reactive with this antibody in cells grown in high-glucose medium, and the signal decreases when cells are shifted into limiting glucose. Furthermore, the reactivity with the antibody is completely abolished by either phosphatase treatment or mutation of Ser24 to Ala. These data indicate that Gpd1 is phosphorylated at Ser24 in a glucose-sensitive manner.
Importantly, Gpd2 displayed the reciprocal behavior. In high-glucose medium, Gpd2 migrated as a single fast-mobility species but a substantial fraction was converted to two lower-mobility species in low-glucose medium that were removed by in vitro phosphatase treatment (Fig. 1C, right). Thus, under high-glucose conditions, the population of Gpd2 molecules was quantitatively in a dephosphorylated state. A Gpd2-S72A mutant exhibited a mobility equivalent to that of dephosphorylated WT Gpd2 in high- or low-glucose medium, indicating that the observed mobility shift required phosphorylation at the expected site.
Gpd2 is a Snf1 substrate in vitro and in vivo.Snf1 is the catalytic subunit of a heterotrimeric complex required for yeast cells to reprogram their metabolism in response to carbon source limitation (33). The complex is activated under conditions of energy stress as a consequence of rising ADP and falling ATP levels (17, 48, 74). Our observation that Gpd2 was phosphorylated upon glucose limitation at a site matching the Snf1 consensus suggested that it is likely to be a direct substrate of this kinase. Purified recombinant Gpd2 was phosphorylated by the Snf1 catalytic domain, as judged by radiolabel incorporation from [γ-33P]ATP, which was completely abolished by the S72A mutation (Fig. 2A). Mobility shift on Phos-tag PAGE demonstrated that Snf1 could phosphorylate Gpd2 to nearly complete stoichiometry in vitro (Fig. 2B). In yeast lacking either Snf1 itself or all of its three upstream activating kinases (Elm1, Sak1, and Tos3 [63]), the Gpd2 mobility shift induced by glucose limitation did not occur (Fig. 2C). Furthermore, we could rescue the Gpd2 mobility shift defect of a snf1Δ mutant strain by the expression of Snf1 from a low-copy-number plasmid (Fig. 2D). The same behavior was observed in the same cells for a well-characterized Snf1 substrate, Mig1, which also undergoes a phosphorylation-dependent mobility shift under energy stress conditions (67). Collectively, these results strongly suggest that Gpd2 is a direct substrate of Snf1 in vivo.
Phosphorylation of Gpd2 by Snf1. (A) Snf1 phosphorylates Gpd2 in a radiolabel kinase assay. TAP-tagged Gpd2 (WT or S72A mutant) purified from yeast was incubated with the bacterially expressed Snf1 catalytic domain and radiolabeled ATP, resolved by SDS-PAGE, and analyzed by autoradiography. (B) Mobility shift analysis indicates that Snf1 phosphorylates Gpd2 to high stoichiometry. TAP-tagged Gpd2 (WT or S72A mutant) was incubated alone, with λ protein phosphatase (λPP), or with the Snf1 catalytic domain in the presence of unlabeled ATP, resolved by Phos-tag PAGE, and analyzed by immunoblotting. (C) Gpd2 phosphorylation depends on Snf1 in vivo. Cultures of WT and snf1Δ and elm1Δ tos3Δ sak1Δ mutant yeast cells expressing Gpd2-His6-GFP from its own promoter on a low-copy-number vector were grown to mid-exponential phase in SC-Leu medium containing 2% glucose (H). The cultures were then split and either left untreated or transferred into SC-Leu containing 0.05% glucose (L). After 90 min, cells were harvested and lysed and Gpd2 was partially purified and analyzed as described in the legend to Fig. 1C. (D) Reexpression of Snf1 rescues the Gpd2 phosphorylation defect of a snf1Δ mutant strain. The WT and snf1Δ mutant strains, as indicated, were transformed with an empty vector (−) or the same plasmid expressing SNF1 and then cotransformed with a plasmid expressing either Gpd2-His6-GFP (top) or HA-tagged Mig1 (bottom). Cells were grown, treated, and lysed as described for panel C. Partially purified Gpd2 was subjected to Phos-tag PAGE (top), and total cell lysates containing Mig1 were subjected to standard SDS-PAGE (bottom) before immunoblotting with the indicated antibodies. (E) Phosphorylation of Gpd2 at Ser72 by Snf1 primes for phosphorylation at Ser75 by Yck1. Purified, HA-tagged Gpd2 was incubated with ATP and the indicated kinases, resolved by Phos-tag SDS-PAGE, and analyzed by immunoblotting. For the far right lane, after phosphorylation, the sample was treated with λPP. The arrow indicates the slowest-migrating dually phosphorylated species. (F) Priming-dependent phosphorylation of Ser75 in vivo. Cultures of either a gpd2Δ single mutant (top) or a snf1Δ mutant (bottom) expressing WT Gpd2-(His)6-GFP or its S72A or S75A mutant form were treated and analyzed as described for panel C.
We noted in some experiments that the slower-migrating Gpd2 species observed in low-glucose medium resolved into a doublet (Fig. 1C and 2F). Since mutation of Ser72 completely prevented Gpd2 phosphorylation in vivo and in vitro, a likely explanation is that subsequent to phosphorylation at Ser72 by Snf1, Gpd2 is further phosphorylated by another protein kinase. MS studies indicate that Ser75 is an additional phosphorylation site on Gpd2 in vivo (4, 19, 28, 45, 61). Members of the casein kinase I (CKI) family strongly prefer substrates that have been “primed” by prior phosphorylation at the P −3 position (29, 50), making Ser75 a likely CKI target after Ser72 phosphorylation by Snf1 (Fig. 2E). To test this hypothesis, we incubated recombinant Gpd2 with Yck1, a yeast CKI homolog (55), either alone or in combination with Snf1, and analyzed the reaction products by Phos-tag PAGE. Yck1 alone could not phosphorylate Gpd2. However, as anticipated, the combination of Snf1 and Yck1 retarded mobility even further than Snf1 alone did (Fig. 2E). In vivo, a Gpd2-S75A mutant exhibited the expected Snf1-dependent mobility shift in response to glucose limitation, but the slowest-migrating species was eliminated (Fig. 2F). We conclude that Snf1 phosphorylation at Ser72 primes Gpd2 for subsequent phosphorylation by Yck1 (or a related kinase) at Ser75.
Gpd1 is a substrate of Ypk1 and Ypk2 in vivo and in vitro.The observation that Gpd1 phosphorylation was promoted by glucose suggested that it may be downstream of the target of rapamycin (TOR) complexes, which are active under conditions of nutrient sufficiency (46). In yeast and multicellular eukaryotes, TOR kinases are found in two distinct complexes, TORC1 and TORC2, which differ in their regulation and downstream effectors (46, 77). Our initial assumption was that Gpd1 phosphorylation was dependent on the TORC1 complex, because it is essential for many cellular responses to glucose in yeast. Furthermore, TORC1 phosphorylates and is necessary for full activation of the protein kinase Sch9 (68), whose preferred phosphorylation site motif includes Arg at the P −5 and P −3 positions, as found upstream of Ser24 in Gpd1 (50). However, purified Sch9 was unable to phosphorylate Gpd1 in vitro (Fig. 3A), and Gpd1 phosphorylation in high-glucose medium was not decreased by deletion of SCH9, deletion of TOR1 or by treatment with the TORC1-specific inhibitor rapamycin (data not shown). Because they share a phosphorylation motif (R-X-R-X-X-S) with Sch9 (16, 50), we also tested the TORC2-dependent kinases Ypk1 and Ypk2. In vitro, Ypk1 (but not Sch9 or Ypk2) phosphorylated Gpd1 (which was overexpressed and purified from yeast) well above the background, as judged by reactivity with the anti-R-X-R-X-X-phospho-S phospho-specific antibody (Fig. 3A and B), by incorporation of radioactivity from [γ-33P]ATP (Fig. 3B), or by mobility shift on Phos-tag gels (Fig. 3C). Phosphorylation by Ypk1, as well as phosphospecific antibody reactivity, was completely abolished in the Gpd1-S24A mutant, indicating that this site is the sole in vitro Ypk1 phosphorylation site in Gpd1 (Fig. 3B and C).
Phosphorylation of Gpd1 by Ypk1 and Ypk2 is sensitive to osmotic conditions. (A) TAP-tagged Gpd1 purified from yeast was incubated with ATP alone; in the presence Ypk1, Ypk2, or Sch9; or in the presence of λPP, as indicated. After incubation, samples were resolved by SDS-PAGE and analyzed by immunoblotting with anti-phospho-RXRXXS antibody to detect Gpd1 phosphorylation and with anti-HA antibody to detect total Gpd1. (B) Purified WT Gpd1 or Gpd1-S24A was incubated with [γ-33P]ATP in the absence (−) or presence (+) of purified Ypk1. After incubation, samples were resolved by SDS-PAGE and analyzed by autoradiography (top), by staining with Coomassie blue to confirm the presence of the substrate (middle), and with anti-phospho-RXRXXS antibody as an independent confirmation of phosphorylation (bottom). (C) Purified WT Gpd1 and Gpd1-S24A were incubated with either Ypk1 or λPP in the presence of ATP and resolved by Phos-tag PAGE and analyzed by immunoblotting with anti-HA antibody. (D) Cultures of otherwise isogenic WT and ypk1ts ypk2Δ mutant strains expressing Gpd1-His6-GFP from its own promoter on a low-copy-number vector were grown to mid-exponential phase at 30°C and then shifted to 37°C. At the indicated times, samples were withdrawn and the harvested cells were lysed and analyzed as described in the legend to Fig. 1C. (E) Cultures of ypk1Δ ypk2Δ mutant cells expressing Gpd1-His6-GFP and either WT Ypk1 (ypk2Δ, left) or Ypk1-as (ypk2Δ ypk1-as, right) were grown to mid-exponential phase and then treated with 1NM-PP1 or a vehicle control. After 30 min, cultures were analyzed for Gpd1 phosphorylation as described in the legend to Fig. 1D. (F) A culture of gpd1Δ mutant cells expressing Gpd1-His6-GFP was grown to mid-exponential phase, and then samples were treated for 30 min with 1 M NaCl with or without 1.25 μM myriocin, as indicated. The cells were then analyzed as described for panel D. (G) Cells transformed with either an empty vector or the same vector expressing TAP-tagged Ypk1 from the GAL1 promoter were grown to mid-exponential phase and either not induced or induced with galactose. Cultures were then left untreated or treated with 1 M NaCl for 15 min prior to harvesting, cell lysis, and immunoprecipitation of Ypk1. Immobilized Ypk1 was incubated with ATP and purified, bacterially expressed GST-Gpd1. After incubation, samples were resolved by SDS-PAGE and immunoblotted with anti-phospho-RXRXXS to detect GST-Gpd1 phosphorylation, with anti-HA antibodies to determine the level of Ypk1, and with anti-GST antibodies to confirm that equivalent amounts of GST-Gpd1 were present. (H) Hyperosmotic shock blocks TORC2-dependent phosphorylation of Ypk1. Cultures of WT cells (strain BY4741) and an otherwise isogenic hog1Δ mutant, each expressing Ypk111A-myc as a reporter for TORC2-dependent phosphorylation at Thr622, as described in detail elsewhere (57), were grown to mid-exponential phase and split, and then one-half of each culture was exposed to 1 M sorbitol for 10 min. Cells were harvested and lysed, and the resulting extracts were resolved by Phos-tag PAGE and analyzed by immunoblotting with anti-c-myc MAb 9E10 to detect all Ypk1 species and with anti-phospho-p38 antibodies to detect activated, dually phosphorylated Hog1, as described in detail elsewhere (73). (I) Cultures of WT and hog1Δ mutant cells expressing tagged Gpd1-His6-GFP (WT or S24A mutant) were grown to mid-exponential phase and then either left untreated or treated with 1 M NaCl for 30 min. Samples were then processed as described in the legend to Fig. 1C.
Although Ypk1 was the only robust Gpd1 kinase in vitro, we found that the ypk1Δ single mutation did not prevent Gpd1 phosphorylation in vivo (data not shown). Because Ypk1 and Ypk2 are closely related and have partially redundant functions (16, 18, 58), it seemed likely that they both contribute to Gpd1 phosphorylation in vivo. Because a ypk1Δ ypk2Δ double mutant is inviable, we first examined Gpd1 phosphorylation in a ypk2Δ mutant strain carrying a temperature-sensitive YPK1 allele (16). After a shift to the restrictive temperature, Gpd1 phosphorylation decreased in the ypk1ts ypk2Δ mutant strain but not in the otherwise isogenic WT strain (Fig. 3D), suggesting that Gpd1 is indeed phosphorylated at Ser24 by the combined action of the two kinases. Second, we used a ypk1Δ ypk2Δ mutant strain expressing analog-sensitive mutant protein Ypk1 (Ypk1-L424G, designated Ypk1-as) that is susceptible to specific inhibition by the compound 1NM-PP1 (11). Treatment of this strain with 1NM-PP1 induced dephosphorylation of Gpd1, as judged by elimination of phospho-specific antibody reactivity (Fig. 3E), confirming that Gpd1 phosphorylation requires Ypk1 and Ypk2 in vivo. The sequence surrounding Ser24 has multiple additional Ser residues that have been identified by MS to be sites of phosphorylation in vivo, including the P +3 residue Ser27 (4, 10, 28, 45, 61). As seen for Gpd2, Ser24 phosphorylation of Gpd1 may serve to prime Gpd1 for further phosphorylation at nearby sites.
Because Gpd1 has an important role in resistance to hyperosmotic stress, we examined whether its phosphorylation state was osmosensitive. Indeed, we found that addition of a high concentration of NaCl (Fig. 3F) caused Gpd1 to become rapidly and stoichiometrically dephosphorylated (an effect that was also induced by sorbitol [see Fig. 7A]). Gpd1 dephosphorylation was partially reversed by treatment with the drug myriocin, an inhibitor of sphingolipid biosynthesis that stimulates Ypk1 activation (11, 57). These results suggested that Ypk1 activity might be downregulated upon hyperosmotic stress. To test this possibility directly, we overexpressed epitope-tagged Ypk1 in yeast cells, immunoprecipitated the enzyme from control cells and cells subjected to a brief shift to high salinity, and then performed an immune complex kinase assay with recombinant Gpd1 as the substrate (Fig. 3G). We found that hyperosmotic treatment reduced the ability of Ypk1 to phosphorylate Gpd1 to an undetectable level. Indeed, we found that TORC2-dependent phosphorylation of Ypk1 at Thr662, a modification required for maximal activation of Ypk1 (57), is abrogated when cells are subjected to hyperosmotic shock (Fig. 3H). Taken together, our data indicate that Ypk1 (and/or Ypk2) are sufficiently active under normal growth conditions to support a high level of Gpd1 phosphorylation, but hyperosmotic shock markedly reduces the activity of these enzymes, permitting rapid Gpd1 dephosphorylation.
Many cellular responses to high osmolarity are mediated by the MAPK Hog1, which is activated by hyperosmotic conditions (34). However, both inactivation of Ypk1 (Fig. 3H) and Gpd1 dephosphorylation (Fig. 3I) occurred in hog1Δ mutant cells, indicating that downregulation of TORC2-dependent phosphorylation of Ypk1 does not require Hog1 action.
Phosphorylation negatively regulates GPDs.The phosphorylation sites on Gpd1 and Gpd2 described here are located very close to their respective catalytic domains (Fig. 1B). To examine whether phosphorylation affects catalytic activity, purified Gpd1 and Gpd2 were phosphorylated to full stoichiometry by the appropriate kinases and then assayed spectrophotometrically by following the conversion of DHAP and NADH to glycerol-3-P and NAD+. For both isozymes, we found that phosphorylation caused a roughly 2-fold decrease in activity (Fig. 4). The activity of Gpd2 was further reduced when it was prephosphorylated by both Snf1 and Yck1 (Fig. 4C), indicating that multisite phosphorylation has an additive inhibitory effect. Demonstrating that the observed effects are due to phosphorylation at the physiologically relevant sites, neither Gpd1-S24A nor Gpd2-S72A exhibited any decrease in activity after incubation with the cognate kinase (Fig. 4). In fact, Gpd2-S72A, but not Gpd1-S24A, displayed somewhat higher activity than the corresponding WT enzyme. Mimicking phosphorylation by replacement of two Ser residues (the initial phosphoacceptor and the P +3 residue) with Glu reduced the activity of both enzymes to the level observed after treatment with the appropriate protein kinase (Fig. 4A and B).
Phosphorylation decreases the catalytic activity of Gpd1 and Gpd2 and curtails glycerol production in vivo. (A) Gpd1 activity following phosphorylation by Ypk1. Gpd1-TAP (WT or S24A or S24E,S27E [SSEE] mutant form) purified from yeast was preincubated with ATP in the presence or absence of Ypk1 and then catalytic activity with DHAP as the substrate was measured by monitoring NADH consumption spectrophotometrically (top). In parallel, mobility on Phos-tag PAGE was used to assess the extent of phosphorylation (bottom). (B) Gpd2 activity following phosphorylation by Snf1. The catalytic activities and electrophoretic mobilities of Gpd2, Gpd2-S72A, and Gpd2-S72E,S75E were assessed as described for panel A following phosphorylation by Snf1. (C) Multisite phosphorylation has an additive inhibitory effect on Gpd2 activity. Purified Gpd2 was incubated with ATP in the absence (−) or presence of Snf1 alone, Yck1 alone, or both enzymes, and then enzymatic activity was assessed as described for panel B. In all of the panels, bars show average initial reaction rates (n = 3) and error bars show standard deviations.
We also examined the effect of phosphorylation site substitutions on glycerol production in vivo. In agreement with the prior findings of others (6), we found that gpd2Δ mutant cells displayed glycerol production substantially lower than that of WT cells under standard growth conditions (Fig. 5). Although a gpd1Δ mutation alone had no effect on glycerol production in the absence of hyperosmotic stress, a gpd1Δ gpd2Δ mutant strain produced no detectable glycerol, demonstrating that Gpd1 does contribute to glycerol formation. Reexpression of WT Gpd2 in either gpd2Δ or gpd1Δ gpd2Δ mutant cells restored glycerol production to the level seen in the WT strain. In exponential-phase cultures, we found that cells expressing the phosphomimetic mutant protein Gpd2-S72E,S75E produced significantly less glycerol than WT cells, while the Gpd2-S72A mutant supported levels of glycerol production equivalent to WT (Fig. 5A and B), consistent with Gpd2 being largely dephosphorylated under these conditions. In cultures grown to stationary phase (a condition in which Snf1 is activated after glucose depletion, Fig. 5C and 5D), we also observed that cells expressing Gpd2-S72E,S75E produced less glycerol, but under these circumstances cells expressing phosphorylation-resistant Gpd2-S72A consistently overproduced glycerol in comparison to the WT strain. Lastly, cells subjected to acute glucose limitation (Fig. 5E and F) overproduced glycerol when expressing Gpd2-S72A, but the phosphomimetic S72E,S75E mutant had no significant effect. The lack of effect of phosphomimetic mutation under these conditions is likely due in part to increased abundance of Gpd2 protein observed upon acute glucose limitation (evident in Fig. 2D for example), which may be related to transcriptional upregulation of GPD2 reported to occur upon glucose depletion (72). We note that because Gpd2-S72A has higher activity than WT Gpd2 in vitro (Fig. 4B), the increased glycerol production supported by this mutant could not be unequivocally attributed to lack of inhibitory phosphorylation. Nonetheless, the decreased glycerol production observed in cells expressing Gpd2-S72E,S75E under multiple conditions is consistent with the importance of phosphorylation in negatively regulating Gpd2 activity. Thus, Snf1-mediated phosphorylation of Gpd2 presumably serves to curtail glycerol production. In keeping with the predominant role of Gpd2 in the absence of hyperosmotic stress, we found that reexpressing either Gpd1-S24A or Gpd1-S24E,S27E had no detectable effect on glycerol production in either gpd1Δ or gpd1Δ gpd2Δ mutant cells during stationary phase (Fig. 5G and 5H).
Effect of Gpd2 mutation on glycerol production in vivo. Cultures of WT and gpd1Δ, gpd2Δ, or gpd1Δ gpd2Δ mutant yeast cells were transformed with either an empty low-copy-number vector (−) or the same plasmid expressing Gpd2, Gpd2-S72A, or Gpd2-S72E,S75E, Gpd1, Gpd1-S24A, or Gpd1-S24E,S27E (SSEE) as indicated. Cells were grown to exponential phase (A and B) or early stationary phase (C, D, G, and H), and the total glycerol concentration was determined as described in Materials and Methods. (E and F) Exponential-phase cultures (in medium containing 2% glucose) were transferred to fresh medium containing 0.05% glucose, and glycerol production was assayed 1 h later. Bars show the average glycerol concentrations across three separate cultures grown in parallel; error bars show standard deviations. *, P < 0.03 (compared to the identical strain expressing the WT Gpd isozyme, by unpaired one-tailed t test).
Transcriptional upregulation and dephosphorylation both promote Gpd1 function to support growth at high osmolarity.Because phosphorylation inhibits Gpd1 catalytic activity, we hypothesized that its dephosphorylation facilitates growth under high osmolarity conditions. Although the effects were very modest, we consistently found that Gpd1-S24A supported somewhat better growth at high salinity than did Gpd1-S24E,S27E, especially in gpd1Δ gpd2Δ mutant cells, in long term agar plate assays (Fig. 6, top and middle panels). We considered the possibility that the elevated expression of Gpd1 that occurs under hyperosmotic conditions masks the effects of Gpd1 phosphorylation. To test this possibility, we constructed low-copy-number plasmids in which GPD1 was placed under the control of the GPD2 promoter, which is not activated by hyperosmotic stress (6). We found that WT Gpd1 expressed from the GPD2 promoter was incapable of rescuing growth of a gpd1Δ gpd2Δ mutant strain at high salinity, and provided slower growth in the presence of high sorbitol (Fig. 6, bottom panel). Strikingly, however, Gpd1-S24A expressed from the GPD2 promoter did support growth of the gpd1Δ gpd2Δ mutant cells at high salinity (albeit not as robust growth as a GPD1+ GPD2+ strain). In addition Gpd1-S24A supported faster growth on sorbitol than WT Gpd1 expressed from the GPD2 promoter. These results suggest that dephosphorylation and transcriptional upregulation both contribute to the control of Gpd1 activity during hyperosmotic stress.
Both Gpd1 dephosphorylation and transcriptional induction are essential mechanisms for survival and long-term growth at high osmolarity. Ten-fold serial dilutions of WT or gpd1Δ or gpd1Δ gpd2Δ mutant cells, as indicated, expressing WT Gpd1, Gpd1-S24A (GPD1 SA), or Gpd1-S24E,S27E (GPD1 EE) from either the native GPD1 promoter (top and middle panels) or the GPD2 promoter (bottom panel) were spotted onto agar plates containing SC-Leu with or without 1 M NaCl or 1 M sorbitol and incubated at 30°C for 3 days.
Gpd1 dephosphorylation is required for rapid adaptation to osmotic stress.Based on our observations, it seemed likely that dephosphorylation of Gpd1 in response to hyperosmotic stress may facilitate short-term adaptation through increased glycerol production. Therefore, we performed a time course in which both Gpd1 protein level and phosphorylation state were monitored after NaCl or sorbitol addition (Fig. 7A). Gpd1 was almost completely dephosphorylated within 15 min, well before higher levels of Gpd1 accumulated. Over time, Gpd1 levels increased, commensurate with the known transcriptional induction of GPD1, reaching a new steady-state level by 1 h. By this point, approximately 50% of the Gpd1 was rephosphorylated. To correlate its phosphorylation state with its activity, we measured glycerol produced in the first 15 min following NaCl or sorbitol treatment in a gpd1Δ mutant strain expressing WT Gpd1, Gpd1-S24A or Gpd1-S24E,S27E (Fig. 7B). Consistent with the importance of Gpd1 in cells coping with hyperosmotic stress, gpd1Δ mutant cells produced substantially less glycerol than the WT strain in response to NaCl or sorbitol, with the residual glycerol being produced by Gpd2. Consistent with the observed Gpd1 dephosphorylation in response to hyperosmolarity contributing to optimal short-term glycerol production, gpd1Δ mutant cells expressing Gpd1-S24A produced somewhat more glycerol after challenge with NaCl or sorbitol, whereas gpd1Δ mutant cells expressing Gpd1-S24E,S27E produced substantially less glycerol (Fig. 7B). Furthermore, we found that in the absence of osmotic stress, treatment of ypk1Δ ypk2Δ mutant cells expressing Ypk1-as with 1NM-PP1 led to glycerol production greater than that of an untreated or a control strain (Fig. 7C), confirming that the TORC2-Ypk1/Ypk2 pathway restrains glycerol production, presumably acting through Gpd1 phosphorylation.
Gpd1 dephosphorylation is essential for rapid adaptation to hyperosmotic stress. (A) Gpd1 is rapidly dephosphorylated in response to hyperosmolarity. Cultures of gpd1Δ mutant cells expressing Gpd1-His6-GFP from a low-copy-number plasmid were grown to mid-exponential phase and then transferred into medium containing 1 M NaCl (left panel). At the indicated times, samples were withdrawn and lysates were prepared. Gpd1-His6-GFP was recovered on immobilized metal affinity resin and either resolved by standard SDS-PAGE and analyzed by immunoblotting with anti-phospho-RXRXXS (top) or resolved by Phos-tag PAGE and analyzed by immunoblotting with anti-GFP (bottom). In a separate experiment, identical cultures were processed in the same manner except that they were treated with 1 M sorbitol rather than NaCl (right panel). (B) Gpd1 dephosphorylation promotes glycerol production following hyperosmotic shock. Cultures of WT and gpd1Δ mutant cells transformed with either an empty plasmid (−) or the same vector expressing WT Gpd1, Gpd1-S24A, or Gpd1-S24E,S27E (SSEE) were harvested 15 min after treatment with 1 M NaCl or 1 M sorbitol. The total glycerol concentration was determined as described in the legend to Fig. 4C. Bars show average glycerol concentrations (n = 3 for separate cultures incubated in parallel); error bars show standard deviations. *, P < 0.005 (compared to gpd1Δ mutant cells expressing WT Gpd1 by unpaired one-tailed t test). (C) Acute inhibition of Ypk1 leads to enhanced glycerol production. A ypk1Δ ypk2Δ mutant strain harboring a plasmid expressing either WT Ypk1 (ypk2Δ) or Ypk1-as (ypk2Δ ypk1-as) was grown to exponential phase and transferred to fresh medium in the presence or absence of 1NM-PP1. After 30 min, the glycerol concentration was determined. (D and E) Gpd1 dephosphorylation accelerates adaptation to hyperosmolarity. The transformed strains used in panel B or similarly transformed gpd1Δ gpd2Δ mutant strains were grown to mid-exponential phase and transferred to into medium containing 1 M NaCl (D) or 1 M sorbitol (E) to an OD600 of 1.0. Cell density (OD600) was determined at 15-min intervals.
To test whether Gpd1 dephosphorylation facilitates growth after a hyperosmotic challenge, we exposed cells to high salinity and monitored culture growth over time. As observed before (34), cells expressing WT Gpd1 exposed to high salinity underwent a temporary growth arrest but resumed growth within 90 min, whereas the gpd1Δ mutant strain carrying the empty vector underwent a significantly longer delay (3 h) (Fig. 7D, left panel). Most significantly, gpd1Δ mutant cells expressing Gpd1-S24A resumed growth slightly faster than gpd1Δ mutant cells expressing WT Gpd1 and, conversely, gpd1Δ mutant cells expressing Gpd1-S24E,S27E exhibited a somewhat longer growth delay than gpd1Δ mutant cells carrying the empty vector. These effects were even more pronounced when gpd1Δ gpd2Δ double mutant cells were used (Fig. 7D, right panel). In response to sorbitol treatment, resumption of growth was not delayed in cells expressing Gpd1-S24E,S27E but did occur at a lower rate (Fig. 7E). On the basis of these results, we conclude that lack of Gpd1 phosphorylation under hypertonic stress is important for optimal glycerol production and that glycerol synthesis is the primary adaptation process required for the resumption of cell growth.
DISCUSSION
Regulation of Gpd1 and the kinases Ypk1 and Ypk2 by hyperosmotic stress.Control of Gpd1 activity has been ascribed primarily to transcriptional regulation of the GPD1 gene. Like many responses to hyperosmotic stress, transcriptional induction of GPD1 is dependent on the MAPK Hog1 (2). Transcription of such genes involves recruitment of Hog1 to promoter regions, where it phosphorylates transcription factors, chromatin remodeling enzymes, and the basal transcriptional machinery (24). In addition to long-term transcriptional responses, adaptation to stress also involves rapid responses that occur prior to new protein synthesis. For example, following exposure to a high salt concentration, Hog1 phosphorylates the potassium channel Tok1 and the Na+/H+ antiporter Nha1 at the plasma membrane to promote restoration of proper ion balance (54). Also, accumulation of intracellular glycerol prior to transcriptional induction of Gpd1 has been attributed to increased flux through glycolysis (25), as well as rapid closing of the plasma membrane glycerol channel Fps1 (47, 65). While direct phosphorylation of Fps1 by Hog1 has been demonstrated under certain conditions (51), in response to hypertonicity, Hog1 appears to regulate Fps1 indirectly through phosphorylation and inhibition of the Fps1-activating proteins Rgc1 and Rgc2 (9). System level modeling of the hyperosmotic stress response has suggested that these early events are more critical for initial adaptation to stress, whereas new gene transcription assists cells in coping with repeated stress (49). Furthermore, it was recently shown that a Hog1 mutant incapable of nuclear entry does not support transcriptional upregulation of Hog1-induced genes yet allows growth at high osmolarity (73). Importantly, growth of cells expressing nonnuclear Hog1 under hyperosmotic stress conditions still required GPD1, suggesting that Gpd1 may be regulated by mechanisms other than transcription. Indeed, the findings presented here demonstrate that hyperosmotic stress prevents Ypk1-dependent inhibitory phosphorylation of Gpd1, causing its dephosphorylation and activation to enhance glycerol production, which is required for the rapid pretranscriptional adaptation to hyperosmolarity.
Our data indicate that Ypk1 and Ypk2, two TORC2-dependent kinases, contribute to Gpd1 phosphorylation in vivo and that hyperosmotic stress downregulates their activity by preventing their TORC2-mediated activation. The TORC2-Ypk1/Ypk2 pathway is involved in multiple cellular functions, including cytoskeletal organization, endocytosis, and cell wall integrity (46). Substantial evidence suggests a critical role for this pathway in sphingolipid homeostasis (7, 11, 57, 62, 64). Ypk1 promotes sphingolipid biosynthesis through phosphorylation of the l-serine:palmitoyl coenzyme A transferase inhibitors Orm1 and Orm2 (57) and influences other aspects of plasma membrane lipid homeostasis through phosphorylation of the aminophospholipid flippase-activating kinase Fpk1 (56). Our results show that TORC2-dependent phosphorylation of Ypk1 and presumably Ypk2 is impaired under hyperosmotic conditions. Ypk1/Ypk2 activity absolutely requires activation loop phosphorylation at Thr504 (in Ypk1) by the eisosome-associated protein kinases Pkh1 and Pkh2, which is sufficient for Ypk1 (and Ypk2) function under normal growth conditions (56, 58, 59, 71). In addition, Ypk1/Ypk2 activity requires subsequent TORC2-dependent phosphorylation at the so-called turn motif (Ser644 in Ypk1), and full activation also requires TORC2-dependent phosphorylation at the so-called hydrophobic motif (Thr662 in Ypk1), the latter of which is necessary to support growth when sphingolipid biosynthesis is compromised (38, 57, 59). Recruitment and activation of Ypk1 (and Ypk2) to TORC2 apparently involve the adaptor proteins Slm1 and Slm2 (11, 22, 53, 64). Blockade of sphingolipid biosynthesis reportedly releases Slm1 and Slm2 from eisosomes, permitting their translocation to TORC2 complexes (11). Slm protein relocalization and enhancement of TORC2-dependent Ypk1/2 phosphorylation can also be induced by decreasing external osmolarity (11). Thus, one mechanism consistent with our observation that hyperosmotic stress prevents TORC2-mediated phosphorylation of Ypk1 is that under this condition, Slm proteins remain eisosome associated. Another potential mechanism for Ypk1/Ypk2 inactivation under hyperosmotic conditions could involve dephosphorylation by another Slm1- and Slm2-interacting protein, the stress-activated Ca2+-calmodulin-dependent phosphoprotein phosphatase calcineurin, which can negatively regulate TORC2 and Slm1/2 (14, 22, 52). However, activation of Ypk1 by sphingolipid depletion appears to be calcineurin independent (11). Interestingly, the effector kinase Pkc1 of the cell wall integrity pathway, which is also controlled by TORC2, is also activated by hypotonic shock and inhibited by hyperosmolarity (23, 39). Collectively, our results and the findings recounted above show that TORC2, Slm1/2, and Ypk1/2 constitute a novel osmosensing pathway, independent of the Hog1 MAPK, to coordinate cell growth with adaptation to changing environmental conditions.
Regulation of Gpd2 by Snf1.While dephosphorylation of Gpd1 is important for adaptation to high osmolarity, it seems most likely that Gpd2 phosphorylation by Snf1 provides a mechanism for fine-tuning of cellular carbon flow. Like mammalian cells (70), rapidly dividing yeast cells produce energy primarily from glycolysis when glucose is abundant, even under aerobic conditions (30). In yeast, production of ethanol consumes most of the NADH produced through glycolysis. However, to maintain redox balance, yeast cells also reoxidize a portion of the NADH generated during glycolysis via reduction of DHAP to glycerol-3-P. Glycerol-3-P can be either dephosphorylated to glycerol or fatty acid esterified to form phosphatidic acid (30). Upon glucose depletion, however, continued growth requires that yeast cell metabolism change from primarily fermentative to oxidative, a process known as the diauxic shift. This adaptation involves significant reprogramming of cellular transcription, largely to facilitate utilization of the acetate, ethanol, lactate, and glycerol generated during fermentation and reoxidation of the NADH generated by mitochondrial respiration. Snf1 phosphorylation and inhibition of Gpd2 presumably serve to impede wasteful glycerol production during the diauxic shift. Gpd2 phosphorylation could also contribute to a recently reported function of Snf1 in decreasing metabolic flux from free fatty acids to phospholipids (20). Although our results show that Snf1 negatively regulates Gpd2 activity in vivo, we found no detectable effect on cell growth in either liquid medium or on agar plates upon the expression of either mutant protein Gpd2-S72E,S75E or Gpd2-S72A (unpublished observations). The levels of both Gpd1 and Gpd2 are highly elevated in stationary-phase cells, when they are thought to have a role in life span extension by participating in a mitochondrial NADH shuttle (26) or via increased glycerol production (72). Thus, the impact of Gpd2 phosphorylation on cell growth may be evident only during the diauxic transition when Snf1 is activated.
Functional differentiation of GPD isozymes.In addition to having differential transcriptional regulation, Gpd1 and Gpd2 also have distinct patterns of subcellular localization. Although both are mainly cytoplasmic, Gpd2 is partly mitochondrial and Gpd1 is partly peroxisomal (36, 41, 60, 69). Like yeast cells, mammalian cells also have a peroxisomal pool of GPD thought to provide NAD+ necessary for fatty acid oxidation (66). However, Gpd1 is not required for the growth of yeast cells when oleic acid is the sole carbon source, suggesting that other sources of NAD+ are sufficient for this process (36). The regulatory phosphorylation sites we have identified are located adjacent to the peroxisomal localization sequence of Gpd1. A previous study using phosphomimetic and unphosphorylatable mutants indicated that phosphorylation was required for localization to peroxisomes (36) and reported that hypertonic stress caused Gpd1 to relocalize to the nucleus, consistent with our observation that Gpd1 becomes dephosphorylated under these conditions. Theoretically, sequestration in peroxisomes could contribute to the negative regulation of Gpd1 by phosphorylation. However, neither deletion of the peroxisomal localization signal nor constitutive anchoring of Gpd1 to peroxisomes affects cell growth under conditions of high osmolarity (36), suggesting that the primary effect of phosphorylation with respect to hyperosmotic stress is to decrease Gpd1 catalytic activity.
Gpd2 compartmentalization to mitochondria has been proposed to underlie its specific requirement in anaerobic growth. In this location, Gpd2 can contribute to mitochondrial redox balance in the absence of respiration (69). In contrast to the localization sequence of Gpd1, that of Gpd2 is situated much further away from its phosphorylation site(s) (Fig. 1B). Indeed, we observed no obvious perturbation of the localization of Gpd2-GFP upon the manipulation of its phosphorylation state (unpublished observations), suggesting that, as for Gpd1, Gpd2 phosphorylation affects primarily its catalytic activity.
Gpd2 is the primary GPD isozyme responsible for glycerol production during exponential growth (6) (Fig. 5C). Under conditions of hyperosmotic stress, Gpd1 becomes the predominant isozyme, even prior to the onset of its transcriptional induction (Fig. 7B). Our data suggest that reciprocal phosphorylation of the two GPDs contributes to such “isozyme switching” in response to a changing environment. GPD1 and GPD2 likely arose from gene duplication and divergence (76), a process that allows genes to acquire new or more specialized functions through subsequent mutation. Introduction, modification, or loss of phosphorylation sites can occur rapidly in evolution and provide a simple mechanism for functional specialization or divergence (5, 8, 10, 35, 37, 43). Computational analysis of the budding yeast phosphoproteome has predicted that pairs of phosphorylation sites positionally conserved between paralogs tend to be phosphorylated by distinct protein kinases (31), and the present study provides experimental evidence in support of such predictions. Differential phosphorylation of yeast GPDs is thus likely to be an example of a general mechanism by which duplicated genes acquire distinct functions.
ACKNOWLEDGMENTS
We are grateful to Martin Schmidt and Tobias Walther for helpful comments on the manuscript. We also thank Martin Schmidt for sharing plasmids, strains, and purified Elm1 protein; Tobias Walther for the strain expressing analog-sensitive Ypk1; Michael Snyder for movable ORF expression constructs; and Mark Hochstrasser for general yeast reagents.
Y.J.L. was supported by a Rudolph J. Anderson Postdoctoral Fellowship through Yale University. This research was supported by NIH R01 research grants GM079498 and GM102262 (to B.E.T.) and GM021841 (to J.T.).
G.R.J. prepared kinase expression constructs and recombinant kinases. F.M.R. performed analysis of Ypk1 phosphorylation in vivo. All other experiments were performed by Y.J.L. J.T. and B.E.T. supervised research, and the paper was written by Y.J.L., J.T., and B.E.T.
We have no conflicts of interest to declare.
FOOTNOTES
- Received 3 July 2012.
- Returned for modification 7 August 2012.
- Accepted 12 September 2012.
- Accepted manuscript posted online 17 September 2012.
- Copyright © 2012, American Society for Microbiology. All Rights Reserved.
