Yeast Phosphofructokinase-1 Subunit Pfk2p Is Necessary for pH Homeostasis and Glucose-dependent Vacuolar ATPase Reassembly*

Background: V-ATPase proton pumps are regulated by glucose. The glycolytic enzyme phosphofructokinase-1 binds to V-ATPase. Results: Mutant phosphofructokinase-1 (pfk1Δ and pfk2Δ) cannot acidify the vacuoles, although V-ATPases are competent to transport protons. Conclusion: Phosphofructokinase-1 is necessary to have fully active V-ATPase in vivo. Significance: Understanding how glucose controls V-ATPase is critical for understanding cellular adaptations to nutritional changes. V-ATPases are conserved ATP-driven proton pumps that acidify organelles. Yeast V-ATPase assembly and activity are glucose-dependent. Glucose depletion causes V-ATPase disassembly and its inactivation. Glucose readdition triggers reassembly and resumes proton transport and organelle acidification. We investigated the roles of the yeast phosphofructokinase-1 subunits Pfk1p and Pfk2p for V-ATPase function. The pfk1Δ and pfk2Δ mutants grew on glucose and assembled wild-type levels of V-ATPase pumps at the membrane. Both phosphofructokinase-1 subunits co-immunoprecipitated with V-ATPase in wild-type cells; upon deletion of one subunit, the other subunit retained binding to V-ATPase. The pfk2Δ cells exhibited a partial vma growth phenotype. In vitro ATP hydrolysis and proton transport were reduced by 35% in pfk2Δ membrane fractions; they were normal in pfk1Δ. In vivo, the pfk1Δ and pfk2Δ vacuoles were alkalinized and the cytosol acidified, suggestive of impaired V-ATPase proton transport. Overall the pH alterations were more dramatic in pfk2Δ than pfk1Δ at steady state and after readdition of glucose to glucose-deprived cells. Glucose-dependent reassembly was 50% reduced in pfk2Δ, and the vacuolar lumen was not acidified after reassembly. RAVE-assisted glucose-dependent reassembly and/or glucose signals were disturbed in pfk2Δ. Binding of disassembled V-ATPase (V1 domain) to its assembly factor RAVE (subunit Rav1p) was 5-fold enhanced, indicating that Pfk2p is necessary for V-ATPase regulation by glucose. Because Pfk1p and Pfk2p are necessary for V-ATPase proton transport at the vacuole in vivo, a role for glycolysis at regulating V-ATPase proton transport is discussed.

imately 70% of the pumps are assembled into V 1 V o complexes. The ratio of assembled to disassembled V-ATPases is dynamic and responds to variations in glucose concentration. Glucose removal leads to V 1 V o disassembly, which helps reserve energy when glucose is limiting. Once energy is restored, after glucose readdition, V 1 V o reassembles (2).
It has been proposed that V-ATPase pumps are structurally primed to disassemble so that disassembly occurs easily and promptly when energy is low (17,18). However, V 1 V o reassembly may require some form of energy (e.g. ATP) to introduce structural changes necessary to reform the subunit-subunit interactions in V 1 V o complexes.
Reassembly is facilitated by a V-ATPase exclusive assembly factor in yeast, the regulator of ATPase of vacuoles and endosomes (RAVE) complex (19,20). The RAVE complex consists of three subunits, Skp1p, Rav1p, and Rav2p; Rav1p aids in the reassociation of cytosolic V 1 with V o at the membrane (21)(22)(23). Although the mechanisms involved in yeast reversible disassembly remain elusive, components of the glycolytic pathway and Ras/cAMP/PKA pathway are involved (15,24). The cytosol and extracellular pH also have been shown to affect yeast V-ATPase reassembly in response to glucose (25,26).
Glycolytic enzymes interact with V-ATPase and may functionally couple glycolytic ATP production and V-ATPase proton transport (27)(28)(29)(30)(31). There is evidence that glycolysis itself could regulate yeast V 1 V o reassembly and/or V-ATPase activity (15). First, reassembly can be triggered by fructose and mannose that are rapidly fermentable sugars like glucose. Second, glucose oxidation beyond glucose 6-phosphate formation is required for reassembly. Third, the ratio of assembled to disassembled V-ATPase pumps gradually increases as the glucose concentration increases, indicating that reversible disassembly is not an all-or-none response.
The interplay between glycolysis and V-ATPase is conserved; it has been described in renal epithelial cells (32), viral infections (33), and the metabolic switch in cancers (34,35). Two glycolytic enzymes can modulate the V-ATPase function, aldolase and phosphofructokinase-1. Aldolase associates with yeast (27,28,36), plant (37), and mammalian (38) V-ATPases. The interaction with aldolase is glucose-dependent in yeast and necessary for stable V 1 V o assembly. Phosphofructokinase-1 interacts with V-ATPase in vacuolar membranes and directly binds to V-ATPase V o subunit a in vitro (29,30). Phosphofructokinase-1 also co-localizes with the V-ATPase V o subunit a isoform a4 (V o a4) in the ␣-intercalated cells of the cortical collecting duct. This interaction may be physiologically relevant. A genetic mutation in the human subunit V o a4 that causes hered-itary distal renal tubular acidosis also prevents V o a4 binding to phosphofructokinase-1 (30).
The yeast ortholog of human phosphofructokinase-1 consists of two tetramers, each made of two subunits, Pfk1p (␣ subunit) and Pfk2p (␤ subunit) (39). Deletion of both subunits prevents yeast growth on glucose, but the single deletion strains metabolize glucose (40). They are therefore suitable to study the interplay between phosphofructokinase-1 and V-ATPase; we anticipated V 1 V o complexes to be assembled in pfk1⌬ and pfk2⌬ mutants.
This study examined V-ATPase functions at steady state and under disassembly and reassembly conditions in the pfk1⌬ and pfk2⌬ mutants. Each mutant failed to acidify vacuoles, even though V-ATPases were catalytically active in vitro. Overall V-ATPase function was significantly more affected in pfk2⌬ than pfk1⌬. The pfk2⌬ cells exhibited a partial vma growth phenotype, enhanced Rav1p-V 1 binding, and abnormal V 1 V o reassembly after readdition of glucose to cells briefly deprived of glucose.

EXPERIMENTAL PROCEDURES
Materials and Strains-Zymolase 100T was purchased from Seikagaku (Tokyo, Japan), concanamycin A from Wako Biochemicals (Richmond, VA), and Ficoll from United Stated Biologicals. The antibody to the Myc antigen was from Invitrogen. dithiobis(succinimidyl) propionate was purchased from Pierce and Tran 35 S-label from MP Biomedicals (Santa Ana, CA). Alkaline phosphatase-conjugated secondary antibodies were from Promega and horseradish peroxidase secondary antibodies from Invitrogen. All other reagents were from Sigma. The Saccharomyces cerevisiae strains referred to throughout are listed in Table 1. The mutant strains were verified by PCR. The primers used in this study are listed in Table 2.
Construction of the pfk2⌬ Rav1p-Myc Strain-The pfk2⌬ Rav1p-Myc mutant was made by disrupting the PFK2 gene using PCR-based homologous recombination, in which LEU2containing cassettes were generated pfk2⌬-5Ј and pfk⌬-3Ј primers ( Table 2) and pRS315 as template. The Rav1p-Myc strain (21) was transformed directly with the PCR product using the lithium acetate method. Transformants were selected on fully supplemented synthetic complete (SC) medium lacking leucine (SCϪLeu) plates. The mutant was tested for integration by PCR.
Growth Phenotype-Overnight cell cultures were grown to stationary phase, diluted to 0.1 A 600 /ml in fresh YEPD or selective medium buffered to pH 5.0, and cells grew for 4 -6 h. Cultures were washed twice with sterile ddH 2 O and 2.5 A 600 cells  (42,43). Cells were maintained in SCϪUra medium buffered to pH 5.0. Fluorescence (ex 405, ex 485; em 508) was monitored and the cytosol pH measured as described before using a FluoroMax 4 spectrofluorometer (Horiba Jobin Yvon) (43). The pHLuorin plasmid was created by Dr. Rajini Rao (Department of Physiology, Johns Hopkins University) and was a generous gift from Patricia Kane (SUNY Upstate Medical University, Syracuse).
Vacuolar pH was measured using the ratiometric fluorescent dye BCECF-AM (44 -46). Glucose-dependent vacuolar pH changes were monitored continuously for 20 min after the addition of 2% glucose (final concentration) to cells briefly (10 min) depleted of glucose. The fluorescence intensity (490 nm/450 nm) was measured in a FluoroMax 4 spectrofluorometer (Horiba Jobin Yvon).
Immunoprecipitations-Nondenaturing immunoprecipitations from whole cell lysates were performed as described previously (47), except that the monoclonal antibodies 13D11 (anti-V 1 subunit B) and 10D7 (anti-V o subunit a, Vph1p isoform) were used. To conduct immunoprecipitations from the cytosol, cytosolic fractions were isolated by high speed centrifugation (100,000 ϫ g for 1 h in a Beckman optima L-100 XP), and 4 mg of total protein was used per immunoprecipitation. Protein was separated in 10% SDS-PAGE and analyzed by Western blotting. The nitrocellulose membranes were scanned using a Bio-Rad ChemiDoc XRSϩ and the intensity of protein bands quantified using the Multi Gauge and GraphPad Prism 5 software. Protocols previously described were followed (15) to conduct the immunoprecipitations from whole cell lysates and biosynthetically radiolabeled cells, except that the chases were performed at the indicated times and the antibodies 13D11 and 10D7 were used. The SDS-polyacrylamide gels (13% acrylamide) were dried, scanned in a Fuji Scanner FLA-5100, and analyzed using the Multi Gauge and GraphPad Prism 5 software.

RESULTS
Studying V-ATPase functions in a phosphofructokinase-1null mutant strain (pfk1⌬ pfk2⌬) is not straightforward because pfk1⌬ pfk2⌬ cells cannot metabolize glucose (40), which causes V 1 V o disassembly (15). The single deletion strains pfk1⌬ and pfk2⌬ are suitable for these studies. They grow on glucose as the only carbon source because each phosphofructokinase-1 subunit Pfk1p and Pfk2p (␣ and ␤, respectively) has catalytic and regulatory functions (40,51). Thus, upon deletion of one subunit, the other subunit retains catalytic activity sufficient to support glycolysis. We examined the roles of individual phosphofructokinase-1 subunits for V-ATPase activity, assembly, and regulation.
Phosphofructokinase-1 Subunits Pfk1p and Pfk2p Co-precipitate with V-ATPase-First, we asked whether the individual phosphofructokinase-1 subunits retain binding to V-ATPase in the single deletion mutants. We immunoprecipitated V-ATPase and conducted immunoblotting using a polyclonal antibody that recognizes both phosphofructokinase-1 subunits. Yeast pfk1⌬ and pfk2⌬ cells and an isogenic wild-type strain were grown in rich media in the presence of 2% glucose (YEPD, normal yeast growth medium). The cells were lysed and the V-ATPase complex immunoprecipitated under nondenaturing conditions using the monoclonal antibody 13D11 to the V-ATPase V 1 subunit B (anti-B) (47).
As expected, both phosphofructokinase-1 subunits co-precipitated with V-ATPase in wild-type cells (Fig. 1A). So did each individual subunit in pfk1⌬ and pfk2⌬. Deletion of one phosphofructokinase-1 subunit did not prevent interaction of the other subunit with V-ATPase. The subunit expressed in pfk1⌬ cells (subunit Pfk2p) and pfk2⌬ cells (subunit Pfk1p) can associate with V-ATPase.
Next, we asked whether the subunit expressed in the pfk1⌬ and pfk2⌬ mutants was present at the vacuolar membrane. We purified vacuolar membrane fractions from pfk1⌬, pfk2⌬, and wild-type cells by density gradient centrifugation and conducted Western blotting (Fig. 1B). Both subunits were detected in wild-type membranes. The subunit Pfk1p was in pfk2⌬ membranes and Pfk2p in pfk1⌬ membranes. Given that individual phosphofructokinase subunits co-precipitate with V-ATPase and that pfk1⌬ and pfk2⌬ cells metabolize glucose (40), we asked whether V-ATPase was functional in pfk1⌬ and pfk2⌬.
The pfk2⌬ Mutants Exhibit Partial vma Growth Phenotype-V-ATPase inactivation leads to a conditionally lethal growth phenotype that is pH-dependent in yeast, the vacuolar membrane ATPase (vma) phenotype. The vma mutants grow at pH Phosphofructokinase-1 Subunits and V-ATPase Function 5.0, but cannot grow at either pH 7.5 or at pH 7.5 in the presence of high concentrations of calcium chloride (2). We plated pfk1⌬ and pfk2⌬ cells on YEPD media buffered to pH 5.0, pH 7.5, and pH 7.5 plus calcium chloride to determine whether deletion of one subunit altered normal V-ATPase function (Fig. 2). For reference, we compared pfk1⌬ and pfk2⌬ growth with V-ATPase mutants (vma2⌬ and vma6⌬). The vma2⌬ and vma6⌬ mutants lack all V-ATPase activity because the structural genes VMA2 and VMA6 that encode subunits B of V 1 and d of V o , respectively, were deleted (52,53). As expected, vma2⌬ ( Fig. 2A) and vma6⌬ (Fig. 2B) did not grow at pH 7.5 and pH 7.5 plus calcium chloride.
The pfk2⌬ cells exhibited a partial vma growth phenotype, in which the growth defect was evident at pH 7.5 plus calcium chloride and accentuated at 37°C ( Fig. 2A, right panel). This growth defect was much subtler in the pfk1⌬ mutant. At the less stringent growth condition, pH 7.5, the pfk2⌬ cells grew normally, so did pfk1⌬ and the isogenic wild-type strain. To address whether vma growth defects in pfk2⌬ were specific to PFK2, we expressed the PFK2 gene from a CEN plasmid in pfk2⌬ cells. As expected, the empty vector did not rescue growth on pH 7.5 plus calcium chloride plates (Fig. 2B). Exogenously expressed PFK2 rescued pfk2⌬ growth (pfk2⌬ (PFK2), Fig. 2B), indicating that the vma growth phenotype of pfk2⌬ was caused by lack of PFK2 expression. These results suggest that the phosphofructokinase-1 subunit Pfk2p is necessary for normal V-ATPase function.
Vacuolar and Cytosolic pH Are Significantly Altered in pfk2⌬ Cells-Having shown that lack of PFK2 leads to growth defects typical of cells with impaired V-ATPase activity (vma phenotype), we asked whether pH homeostasis was altered. V-ATPase mutants have alkalinized vacuoles and acidified cytosol because V-ATPase proton transport is necessary to sustain yeast vacuolar and cytosolic pH homeostasis (45).
We used fluorometric assays and pH-sensitive fluorescent dyes to measure vacuolar (BCECF) and cytosolic (pHLuorin) pH (42)(43)(44) in pfk1⌬ and pfk2⌬ cells in vivo (Fig. 3). As expected, pH homeostasis was aberrant in a control strain that lacks all V-ATPase function (vma2⌬). The vacuolar pH of vma2⌬ cells was considerably more alkaline and its cytosol more acidic than wild-type cells.
The fact that pH alterations were milder in pfk1⌬ than pfk2⌬ and more severe in the vma2⌬ mutant is consistent with the extent of the growth defects observed in these strains (Fig. 2). Together these phenotypes suggest that subunit Pfk2p is more critical than Pfk1p to sustain optimal V-ATPase proton transport at steady state in vivo.
V-ATPase Is Catalytically Competent in pfk2⌬-To directly address the effect that deletion of PFK1 and PFK2 has on V-FIGURE 1. V-ATPase associates with phosphofructokinase-1 subunits Pfk1p and Pfk2p. A, V-ATPase co-immunoprecipitates (IP) with Pfk1p and Pfk2p from whole cell lysates. Isogenic wild-type, pfk1⌬, and pfk2⌬ cells were grown overnight to mid-log phase (0.8 -1.0 A 600 /ml). Cells were converted to sheroplast by zymolase treatment and V-ATPase immunoprecipitated under nondenaturing conditions using the monoclonal antibody 13D11 to subunit B of V 1 and protein A-Sepharose. The immunoprecipitated protein was separated by 10% SDS-PAGE and immunoblotted with antibodies to phosphofructokinase and V 1 subunits A and B using horseradish peroxidase secondary antibodies. HC, antibody heavy chain; LC, antibody light chain. B, Pfk1p and Pfk2p subunits co-purify with vacuolar membrane fractions. Vacuolar membrane vesicles (0.25 g of total membrane protein) were purified from pfk1⌬, pfk2⌬, and wild-type cells by density gradient centrifugation. Membranes were immunoblotted as described above. FIGURE 2. The pfk2⌬ mutant exhibits partial vma growth phenotype. A, growth of the pfk2⌬ strain is sensitive to pH 7.5 with calcium chloride. Cell cultures were grown overnight to mid-log phase and 10-fold serial dilutions stamped onto YEPD plates adjusted to pH 5.0, pH 7.5, and pH 7.5 plus 100 mM CaCl 2 . Cell growth was monitored for 3 days at 30°C and 37°C. B, PFK2 rescues the vma growth phenotype of pfk2⌬. The wild-type strain and the mutant strains pfk1⌬, pfk2⌬, and vma6⌬ were transformed with the empty CEN plasmid pRS316. The pfk2⌬ cells were also transformed with the gene PFK2 expressed from the same plasmid under control of its endogenous promoter (pfk2⌬ (PFK2)). Serial dilutions of the cells were stamped onto SCϪUra plates adjusted to pH 7.5 plus 100 mM CaCl 2 and allowed to grow for 3 days at 30°C and 37°C. Shown are representative plates of triplicates.

Phosphofructokinase-1 Subunits and V-ATPase Function
ATPase catalytic activity, we purified vacuolar membrane fractions by density gradient centrifugation. We measured ATP hydrolysis and proton transport in vitro in the presence and absence of the V-ATPase inhibitor conanamycin A (43).
The V-ATPase pumps at pfk1⌬ and pfk2⌬ membranes were significantly active. The conanamycin A-sensitive ATP hydrolysis and proton transport were partially reduced by approximately 35% in the pfk2⌬ membranes (Fig. 4A). They were normal in pfk1⌬. Comparative membrane protein titrations in Western blots showed equivalent amounts of V 1 (subunits A and B) and V o (subunit a) subunits in wild-type, pfk1⌬, and pfk2⌬ membranes (Fig. 4B), suggesting that V 1 V o assembly is normal at the vacuolar membrane.
Subtle alterations of the V 1 V o assembly level may not be detected by Western blotting. We further examined whether loss of Pfk1p or Pfk2p affected V 1 V o assembly by using a more sensitive approach. We biosynthetically radiolabeled the cells with 35 S to estimate the total fraction of V 1 V o complexes assembled in pfk1⌬ and pfk2⌬. The radiolabeled V-ATPase pumps were immunoprecipitated from whole cell lysates with the antibodies 13D11 (anti-B, recognizes V 1 and V 1 V o ) and 10D7 (anti-a, recognizes V o ) and the total fraction of assembled V-ATPase complexes calculated as described before (15).
Consistent with previous studies (2,15), approximately 60% of the V-ATPase pumps were assembled in wild-type cells at steady state (Fig. 4C). A comparable fraction of assembled V-ATPases was detected in pfk2⌬ and pfk1⌬ cells, consistent with the Western blots (Fig. 4B) and further suggesting that the absence of either phosphosfructokinase-1 subunit does not disturb biosynthetic V 1 V o assembly. Next, we asked whether V-ATPase reversible disassembly was normal.
Glucose-dependent V-ATPase Reassembly Is Defective in pfk2⌬ Mutants-Until now our studies have been conducted at steady state, in the presence of abundant glucose (2% glucose). We determined whether the phosphofructokinase-1 subunits were necessary for V 1 V o reversible disassembly in response to glucose depletion and readdition. It is known that under disassembly and reassembly conditions the equilibrium [ (2); lack of glucose favors disassembly, glu-cose readdition promotes reassembly and restores steady-state equilibrium.
The wild-type pfk1⌬ and pfk2⌬ cells were biosynthetically radiolabeled and chased in the presence (assembly condition) and absence (disassembly condition) of 2% glucose and after readdition of 2% glucose following a brief glucose depletion period (reassembly condition). We estimated the fraction of assembled V 1 V o by nondenaturing immunoprecipitation experiments using the anti-B and anti-a antibodies (15).
The pfk1⌬ cells disassembled and reassembled V 1 V o normally, as wild-type cells (Fig. 5A). Approximately 70 -80% of the total V 1 V o complexes disassembled upon glucose depletion. An equivalent proportion of V 1 V o complexes reassembled after glucose readdition to pfk1⌬. Notably, disassembly was normal in the pfk2⌬ strain, but reassembly was significantly reduced (Fig. 5B). Only 50% reassembly was detected relative to wildtype cells. Maximum reassembly is achieved within 5 min in wild-type cells; an increase of the incubation time from 5 min to up to 15 min did not lead to an increase of pfk2⌬ assembly after glucose readdition. These results suggest that the mechanism(s) of reassembly are defective in pfk2⌬, whereas the kinetics of reassembly remains similar to wild-type cells.
V-ATPase-Rav1p Interaction Is Enhanced in the pfk2⌬ Strain-Reassembly requires the RAVE complex, particularly the RAVE subunit Rav1p that connects RAVE with V 1 , subunit C, and V o (21)(22)(23). Because these interactions are necessary for reassembly of V-ATPase pumps at the vacuolar membrane, we asked whether the Rav1p-V 1 binding was affected in pfk2⌬. We deleted PFK2 in cells expressing Myc epitope-tagged genomic RAV1 (Rav1p-Myc). It has been shown that wildtype Rav1p-Myc cells retain normal growth and assemble functional V 1 V o complexes at vacuolar membranes (21). Thus, the Myc tag does not interfere with steady-state V-ATPase assembly and activity.
To address whether Rav1p binding to V 1 was affected in pfk2⌬ we immunoprecipitated V 1 from a cytosolic fraction isolated from the pfk2⌬ Rav1p-Myc strain. We estimated approximately 5-fold more Rav1p bound to V 1 subunits A and B in pfk2⌬ than wild-type cells (Fig. 6A). This increase in Rav1p-V 1 FIGURE 3. pfk1⌬ and pfk2⌬ mutants have altered vacuolar and cytosol pH homeostasis at steady state. A, the vacuolar lumen is alkalinized in live pfk1⌬ and pfk2⌬ cells. Overnight mid-log phase cultures from wild-type cells, mutant cells (pfk1⌬, pfk2⌬, vma2⌬), and the mutant pfk2⌬ expressing exogenous PFK2 from the CEN plasmid pRS316 (pfk2⌬ (PFK2)) were stained with 50 M BCECF-AM for 30 min at 30°C. The ratio of fluorescent emission (535 nm) excited at 490 and 450 nm was measured in a fluorometer to quantitatively assess vacuolar pH. The average fluorescence over 6 min at 1-min intervals was compared with a standard curve to generate absolute pH values. B, the cytosol pH is acidified in live pfk1⌬ and pfk2⌬ cells. The wild-type, pfk1⌬, pfk2⌬, and vma2⌬ cells expressing cytosolic pHLuorin were grown overnight to mid-log phase (0.4 -0.6 A 600 /ml). The cells were transferred to 1 mM HEPES/MES buffer, pH 5.0, containing 2% glucose at a cell density of 5.0 A 600 /ml. The ratio of fluorescent emission (508 nm) excited at 405 nm and 485 nm was measured for 6 min at 10-min intervals. The average fluorescence was estimated and calibration curves made in parallel used to calculate pH values in a fluorometer. Vacuolar and cytosol data are presented as average pH values from three independent experiments, error bars ϭ Ϯ S.D. *, p Ͻ 0.05; ***, p Ͻ 0.001 compared with wild-type control as measured by a two-tailed unpaired t test.

Phosphofructokinase-1 Subunits and V-ATPase Function
binding was not caused by enhanced expression of cytosolic Rav1p and/or V 1 subunits because these proteins were comparable in whole cell lysates from pfk2⌬ and wild-type cells (Input, Fig. 6A). Rather, Rav1p-V 1 binding was increased in the cytosol of pfk2⌬, suggesting an enhanced Rav1p-V 1 affinity at steady state. . V-ATPase activity is partially reduced at pfk2⌬ vacuolar membrane vesicles. A, ATP hydrolysis and proton transport are differentially affected in pfk1⌬ and pfk2⌬ cells. Vacuolar membrane fractions were purified from the isogenic wild-type, pfk1⌬, pfk2⌬, and vma2⌬ cells. ATP hydrolysis (left) was assayed spectrophotometrically in the presence and absence of the V-ATPase inhibitor concanamycin A by using an enzymatic coupled assay that measures NADH oxidation at 340 nm. The wild-type specific activity of the concanamycin A-sensitive ATP hydrolysis was 2.5-4 mol of ATP/min/mg of protein. The strains showing significant activity (wild-type, pfk1⌬, and pfk2⌬) were equally inhibited (ϳ80%) by 100 nM concanamycin A. ATP-dependent proton transport (right) was measured via fluorescence quenching of 1 M 9-amino-6-chloro-2-methoxyacridin (ex 410 nm; em 490 nm) upon the addition of 0.5 mM ATP/1 mM MgSO 4 to 5 g of total protein in vacuolar membranes vesicles. Initial velocities were calculated for 15 s following MgATP addition. The average wild-type slope was Ϫ1320.32 fluorescence units/15 s. Data represent three independent vacuolar preparations. *, p Ͻ 0.05; ***, p Ͻ 0.001 decreased ATPase activity compared with wild-type membranes as measured by two-tailed unpaired t test. B, wild-type levels of V o V 1 complexes are assembled at the vacuolar membrane of pfk1⌬ and pfk2⌬ mutants. The purified vacuolar membrane vesicles were analyzed by quantitative immunoblotting using antibodies against V 1 subunits B and A and the V o subunit a (Vph1p isoform) and alkaline phosphatase-conjugated secondary antibodies. Serial dilutions of wild-type and mutant membrane fractions were prepared and the indicated amounts of vacuolar protein loaded per well and separated on 10% SDS-polyacrylamide gels. A representative gel of three independent vacuolar preps is shown. C, pfk1⌬ and pfk2⌬ assemble wild-type levels of V 1 V o complexes at steady state. Isogenic wild-type, pfk1⌬, and pfk2⌬ cells were biosynthetically radiolabeled with Tran 35 S for 60 min and V-ATPase immunoprecipitated from whole cell lysates under nondenaturing conditions using the antibodies anti-B (recognizes V 1 and V 1 V o ) and anti-a (recognizes V o ). The protein was separated in 13% SDS-polyacrylamide gels and assembled V 1 V o estimated as the fraction of V o immunoprecipitated with anti-B relative to the total immunoprecipitated with both antibodies. Gels from three independent experiments were analyzed in a Fuji Scanner FLA-5100 and analyzed using the Multi Gauge and GraphPad Prism 5 software.

FIGURE 5. Glucose-dependent V 1 V o reassembly is defective in pfk2⌬.
A, glucose-dependent disassembly and reassembly are normal in pfk1⌬ cells. Biosynthetically radiolabeled wild-type and pfk1⌬ cells were chased in YEP medium containing 2% glucose (YEPD) for 20 min (1), lacking glucose for 10 min (2), and in medium lacking glucose for 10 min followed by an additional 10-min chase after readdition of glucose to a final concentration of 2% glucose (3). The cells were lysed and immunoprecipitated, and gels from four independent experiments were analyzed as described for Fig. 4C. A representative gel is shown (left). Analyzed results are expressed as the average Ϯ S.D. (error bars) relative to wild-type (right). B, glucose-dependent reassembly is defective in pfk2⌬ cells. Wild-type and pfk2⌬ cells were radiolabeled, chased, and immunoprecipitated, and V 1 V o assembly levels were analyzed as described for Figs. 4C and 5A. Sample 1 was chased in YEPD for 25 min, 2 in YEP for 10 min, and 3-5 were chased respectively for 5, 10, and 15 min after glucose readdition (2% final concentration). The data are expressed as the average Ϯ S.D. relative to wild-type (right). *, p Ͻ 0.05 compared with wild-type as measured by two-tailed unpaired t test. JULY 11, 2014 • VOLUME 289 • NUMBER 28

Phosphofructokinase-1 Subunits and V-ATPase Function
To eliminate the possibility that the Myc tag itself affected Rav1p-V 1 binding, we also immunoprecipitated V-ATPase complexes from whole cell lysates after brief glucose depletion and readdition. As expected for wild-type Rav1p-Myc cells, approximately 70% of the V 1 V o disassembled and reassembled after glucose removal and its readdition, respectively (Fig. 6B). Only half of the V 1 V o complexes reassembled after readdition of glucose to pfk2⌬ Rav1p-Myc cells. We concluded that tagged Myc did not interfere with V-ATPase reversible disassembly. Thus, the increased level of RAVE-V 1 in the cytosol of pfk2⌬ is a phenotypic trait of the phosphofructokinase-1 mutant pfk2⌬.
Rav1p also interacts with V o subunit a at the membrane, specifically the subunit a isoform Vph1p (vacuolar isoform) (23). We isolated membrane fractions to determine whether binding of Rav1p to V o subunit a was also enhanced in pfk2⌬. We detected similar levels of Rav1p in pfk2⌬ and wild-type vacuolar membranes (data not shown). These results suggest that the binding of RAVE to V o was not altered in pfk2⌬. A greater affinity of Rav1p-V 1 binding could lead to reassembly defects of pfk2⌬.
Glucose-dependent V 1 V o Reassembly Fails to Acidify pfk2⌬ Vacuoles-In wild-type cells, glucose depletion alkalinizes the vacuoles, and glucose readdition acidifies the vacuolar lumen (45). These pH changes are V-ATPase-dependent because V 1 V o reassembly restores ATP hydrolysis and proton transport (2,45). We monitored vacuolar pH in the phosphofructokinase-1 mutants under reassembly conditions. The vacuoles were loaded with the pH-sensitive fluorophore BCECF-AM and deprived of glucose for 10 min, after which glucose was readded and the pH continuously measured (Fig. 7).
The wild-type and pfk1⌬ cells acidified the vacuolar lumen after glucose readdition, indicating that reassembled V 1 V o resumed proton transport. Glucose readdition to pfk2⌬ did not result in vacuolar acidification. Instead, the vacuolar pH gradually increased until reaching approximately pH 6.5, the same pH measured at steady state (Fig. 3). A similar alkalinization of the vacuolar lumen is obvious in the V-ATPase mutant vma2⌬ (Fig. 7) and other yeast mutants lacking V-ATPase activity. These results indicate that reassembled V-ATPase pumps are inactive in the pfk2⌬ strain, although the vacuolar pH increased abruptly in vma2⌬ and gradually in pfk2⌬.

DISCUSSION
We studied V-ATPase assembly, activity, and regulation in phosphofructokinase-1 single deletion mutants lacking the structural genes PFK1 and PFK2. Whereas both phosphofructokinase-1 subunits make contributions to V-ATPase activity, the most critical subunit is Pfk2p.
The pfk1⌬ and pfk2⌬ Mutants Are Suitable to Study V-ATPase-V-ATPase studies using pfk1⌬pfk2⌬ cells are difficult to interpret because the double deletion mutant cannot metabolize glucose. Although pfk1⌬pfk2⌬ tolerates 0.2% glucose (29), suppression of glycolysis and glucose-dependent sig-FIGURE 6. The interaction between Rav1p and cytosolic V 1 is increased in pfk2⌬ cells. A, cytosolic fractions from pfk2⌬ cells contain more Rav1p-V 1 than wild-type cells. Overnight mid-log phase cultures (0.8ϳ1.0 A 600 /ml) of wild-type and pfk2⌬ cells expressing Rav1p-Myc were lysed and the cytosolic fraction prepared by centrifugation (100,000 ϫ g for 1 h). Cytosolic V 1 complexes were immunoprecipitated with anti-B antibody. Immunoprecipitated protein (IP) and total cytosolic protein (Input) were loaded on 10% SDS-polyacrylamide gels. Rav1p and V 1 (subunits A and B) were detected by immunoblotting with anti-Myc, anti-B, and anti-A monoclonal antibodies, respectively. A representative gel is shown (left). Gels from three independent experiments were scanned using a Bio-Rad ChemiDoc XRSϩ, and data were analyzed using Multi Gauge and GraphPad Prism software. Data are expressed as -fold increase Rav1p:V 1 subunit ratio Ϯ S.D. (error bars) relative to wild-type (right). B, glucose-dependent V 1 V o reassembly is defective in Rav1p-Myc pfk2⌬ cells. Isogenic wild-type and pfk2⌬ mutant cells expressing genomic Rav1p-Myc were grown overnight to mid-log phase. Cells were converted to sheroplast incubated in YEPD media (2% glucose) for 20 min (ϩG), deprived of glucose for 10 min (ϪG), and deprived of glucose for 10 min followed by an additional 10 min after glucose readdition to a final concentration of 2% glucose (Ϫ/ϩG). V-ATPase was immunoprecipitated (200 A 600 /immunoprecipitation) as described for Fig. 1A, and immunoblots were analyzed with antibodies to V o subunit a (vacuolar isoform Vph1p) and V 1 subunits A and B. One representative gel is shown (left). Gels from three independent experiments were analyzed, and total V 1 V o was estimated as for Fig. 4C. The data are expressed as the average Ϯ S.D. relative to wild-type. **, p Ͻ 0.01 compared with wild-type as measured by two-tailed unpaired t test.

Phosphofructokinase-1 Subunits and V-ATPase Function
nals promotes V 1 V o disassembly. Only 30 -40% of the V 1 V o complexes are assembled at 0.2% glucose in wild-type cells (15).
Because there is an equal contribution of Pfk1p (subunit ␣) and Pfk2p (subunit ␤) in the wild-type phosphofructokinase-1 heteromeric complex, each individual subunit can assemble to an unknown structure that is partially active in vivo (51,54). Each single deletion strain grows on glucose, indicating that pfk1⌬ and pfk2⌬ have phosphofructokinase-1 activity above a threshold level that supports sufficient glycolytic flow.
The Phosphofructokinase-1 Subunit Pfk2p Is Necessary for Cellular pH Homeostasis and V-ATPase Regulation-Heteromeric (wild-type cells) and homomeric (pfk1⌬ and pfk2⌬ cells) phosphofructokinase-1 complexes co-precipitate with V-ATPase. Our study suggests that phosphofructokinase-1 interacts with V 1 V o because Pfk1p and Pfk2 were detected in vacuolar membrane fractions and co-immunoprecipitated with the anti-B antibody (13D11). However, we cannot exclude the possibility that it binds to cytosolic V 1 as well because the antibody 13D11 recognizes both V 1 V o and V 1 . Additional studies will be necessary to determine whether the interactions between Pfk2p and V-ATPase are direct or indirect, for example, involving other glycolytic enzymes.
Pfk2p could bridge interactions involving phosphofructokinase-1 and V-ATPase. Yeast phosphofructokinase-1 forms stable ␣ 4 ␤ 4 complexes in which ␤ subunits (Pfk2p) are at the periphery (39), more readily available to form intermolecular interactions with other proteins, including V-ATPase.
The subunit ␣ is not critical for V-ATPase function, although the vacuole and cytosol pH homeostasis are altered in vivo in pfk1⌬. The pfk1⌬ cell exhibits wild-type levels of V-ATPase activity in membrane fractions and V 1 V o assembly at steady state. They also disassemble and reassemble V 1 V o normally and resume vacuolar acidification after glucose readdition.
At steady state, V-ATPase proton transport is somewhat inhibited in vivo; pfk1⌬ has more alkaline vacuoles and acidic cytosol than wild-type cells. Together with the fact that pfk1⌬ growth is slightly reduced on plates buffered to pH 7.5 containing calcium chloride, these results indicate that V-ATPase proton transport is suppressed in pfk1⌬ vacuolar membranes in vivo at steady state.
The subunit ␤ is critical for V-ATPase function; V-ATPase activity and its regulation are more severely impaired in the pfk2⌬ mutant. Vacuolar and cytosol pH alterations are more pronounced in pfk2⌬ than pfk1⌬ at steady state. The vacuolar and cytosol pH, respectively, increased and decreased by 0.75 pH unit each. For reference, complete lack of V-ATPase activity in vma2⌬ changes the vacuole and cytosol pH by about 1.0 pH unit.
V-ATPase reversible disassembly is also defective in pfk2⌬. Only half of the V-ATPase complexes reassemble after glucose readdition compared with wild-type cells. Notably, pfk2⌬ does not acidify the vacuolar lumen after glucose readdition. Because glucose-triggered vacuolar acidification is a direct outcome of V-ATPase reassembly and reactivation (45), these results suggest that V 1 V o complexes do not pump protons after reassembly occurs in pfk2⌬.
One explanation is that pfk2⌬ does not acidify the vacuoles because 50% of V 1 V o reassembles and those pumps are ϳ35% less active. However, wild-type yeast vacuoles are acidified when 30 -40% of the V-ATPase pumps reassemble after addition of 0.2% glucose. 3 Another explanation is that RAVE-mediated reassembly is defective in pfk2⌬.
Enhanced Rav1p-V 1 Interaction May Impair pfk2⌬ Glucosedependent Reassembly-The chaperone activity of RAVE directly involves its subunit Rav1p, which binds to cytosolic V 1 and subunit C and connects RAVE and V-ATPase subunits (21)(22)(23). We estimated approximately 5-fold more Rav1p-V 1 complexes in the cytosol of pfk2⌬ than wild-type cells (Fig. 6). Because these experiments were conducted at steady state, excess Rav1p-V 1 complexes are not a byproduct of failure to reassemble V 1 V o .
RAVE likely facilitates the structural changes necessary to properly reform V 1 V o subunit interactions (7,17). Reassembled V 1 V o complexes can be inactive if reassociation of V 1 and/or subunit C with V o is structurally flawed or "loose." Structurally loose complexes will not acidify vacuoles; they could also break apart V 1 and V o during the immunoprecipitation experiments (Fig. 5).
RAVE exclusively controls assembly of Vph1p-containing V-ATPases, which reside in the vacuole (23). Thus, Vph1pcontaining V-ATPases are defective in pfk2⌬. In agreement with this notion, vacuolar pH and glucose-dependent reassembly are altered in pfk2⌬. In addition, pfk2⌬ and rav1⌬ mutants share common growth characteristics, a partial vma growth phenotype (23). This phenotype suggests that the nonvacuolar V-ATPase pumps probably are functional. Like rav1⌬, pfk2⌬ may retain normal V-ATPase function at the Golgi and endosomes, which house the second V o subunit a isoform Stv1p (23).
A difference between pfk2⌬ and rav1⌬ is that biosynthetic V 1 V o assembly is normal in pfk2⌬, but the glucose-triggered reassembly is partially reduced. Both mechanisms, biosynthetic assembly and glucose-dependent reassembly, are severely defective in rav1⌬, which also lacks V-ATPase activity at the vacuolar membrane (21). Our interpretation of these results is that RAVE activity is partially compromised in pfk2⌬. Only its function at mediating glucose-triggered reassembly is defective.  . Glucose-dependent vacuolar acidification is impaired in pfk2⌬ cells. The pfk1⌬, pfk2⌬, and vma2⌬ mutants and a wild-type strain were stained with 50 M BCECF-AM for 30 min at 30°C. The cells were depleted of glucose for 10 min, after which glucose (2% final concentration) was added (arrow) and fluorescence monitored for an additional 20 min. The ratio of fluorescent emission (535 nm) excited at 490 and 450 nm was measured as for Fig. 3A. The vacuolar pH was estimated using calibration curves made in parallel.

Phosphofructokinase-1 Subunits and V-ATPase Function
Are V-ATPase and Glycolysis Directly Coupled in Vivo?-Glycolysis is necessary to reassemble V 1 V o complexes (15). This study suggests that glycolysis may also modulate V-AT-Pase activity under steady-state conditions. In vivo vacuolar acidification is defective in pfk1⌬ and pfk2⌬, although the V-ATPase pumps are active in vitro. Reactivation in vitro could be explained if the mechanism that inhibits proton transport in vivo is lost during the purification of membrane vesicles.
Recent crystallographic data of the yeast V-ATPase complex suggest that V 1 V o reassembly may require some energy input in addition to RAVE (7,17). It is conceivable that reduced glycolytic flow and therefore lower energy (e.g. ATP) production contribute to the V 1 V o reassembly defects in pfk2⌬.
There is evidence that the glycolytic flow is reduced in the pfk1⌬ and pfk2⌬ mutants compared with wild-type cells (40). Homomeric complexes consisting of subunit ␤ (pfk1⌬ cells) are more active than the subunit ␣-containing complexes (pfk2⌬ cells) (40). The concentration of phosphofructokinase-1 reaction product is 10 -15-fold lower in the pfk2⌬ mutant compared with wild-type cells; it is only 2-fold lower in pfk1⌬ (40). In addition, the NADH spike triggered by readdition of glucose is reduced or absent in pfk2⌬, but not in pfk1⌬ (55,56).
Additional studies will be necessary to understand the interplay between V-ATPase and glycolysis. It likely involves additional glycolytic enzymes. A supercomplex consisting of V-ATPase and glycolytic enzymes has been proposed before (28 -30, 36, 37); it could functionally and structurally couple V-ATPase activity and energy metabolism. It also could directly provide glycolytic ATP to drive proton transport. Other rapid energy-consuming processes rely on glycolysis-derived ATP locally made by glycolytic enzymes at the site where the energy supply is needed. These processes include, glutamate uptake in synaptic vesicles (57), regulation of ATP-sensitive K ϩ channels in cardiac myocytes (58,59), and phototransduction at rod and cone cells (16).
In summary, deletion of the structural genes encoding subunits of the glycolytic enzyme phosphofructokinase-1 interferes with yeast V-ATPase function and regulation in vivo, despite that V-ATPases are catalytically competent and the cells metabolize glucose. V-ATPase proton transport at the vacuole is inhibited by unknown cellular mechanisms; a reduction of the glycolytic flow is an attractive candidate. Given that RAVE-V 1 binding is enhanced and V 1 V o reassembly reduced in pfk2⌬, we concluded that Pfk2p is necessary for normal RAVE functions.