Regulation of Vacuolar H+-ATPase (V-ATPase) Reassembly by Glycolysis Flow in 6-Phosphofructo-1-kinase (PFK-1)-deficient Yeast Cells*

Yeast 6-phosphofructo-1-kinase (PFK-1) has two subunits, Pfk1p and Pfk2p. Deletion of Pfk2p alters glucose-dependent V-ATPase reassembly and vacuolar acidification (Chan, C. Y., and Parra, K. J. (2014) Yeast phosphofructokinase-1 subunit Pfk2p is necessary for pH homeostasis and glucose-dependent vacuolar ATPase reassembly. J. Biol. Chem. 289, 19448–19457). This study capitalized on the mechanisms suppressing vacuolar H+-ATPase (V-ATPase) in pfk2Δ to gain new knowledge of the mechanisms underlying glucose-dependent V-ATPase regulation. Because V-ATPase is fully assembled in pfk2Δ, and glycolysis partially suppressed at steady state, we manipulated glycolysis and assessed its direct involvement on V-ATPase function. At steady state, the ratio of proton transport to ATP hydrolysis increased 24% after increasing the glucose concentration from 2% to 4% to enhance the glycolysis flow in pfk2Δ. Tighter coupling restored vacuolar pH when glucose was abundant and glycolysis operated below capacity. After readdition of glucose to glucose-deprived cells, glucose-dependent V1Vo reassembly was proportional to the glycolysis flow. Readdition of 2% glucose to pfk2Δ cells, which restored 62% of ethanol concentration, led to equivalent 60% V1Vo reassembly levels. Steady-state level of assembly (100% reassembly) was reached at 4% glucose when glycolysis reached a threshold in pfk2Δ (≥40% the wild-type flow). At 4% glucose, the level of Pfk1p co-immunoprecipitated with V-ATPase decreased 58% in pfk2Δ, suggesting that Pfk1p binding to V-ATPase may be inhibitory in the mutant. We concluded that V-ATPase activity at steady state and V-ATPase reassembly after readdition of glucose to glucose-deprived cells are controlled by the glycolysis flow. We propose a new mechanism by which glucose regulates V-ATPase catalytic activity that occurs at steady state without changing V1Vo assembly.

Vacuolar H ϩ -ATPase (V-ATPase) 3 is a highly conserved ATP-driven proton pump distributed throughout the endo-membrane system. Intracellular V-ATPase generates and maintains the acidic organelle luminal pH necessary for membrane trafficking, protein sorting and processing, and zymogen activation (1)(2)(3)(4)(5). V-ATPase is also present at the plasma membrane of cells specialized for active proton transport. Kidney intercalated cells (6), bone osteoclasts (7), and epididymis clear cells (6) express V-ATPases at the plasma membrane, which are necessary for systemic acid-base balance, bone resorption, and sperm maturation, respectively.
V-ATPase consists of two multisubunit domains, V 1 and V o , that are responsible for ATP hydrolysis and proton transport, respectively (8,9). Peripheral subunits form V 1 , the catalytic domain attached to the cytosolic side of the membrane. V 1 is bound to V o , the integral membrane domain that forms the path for proton transport. During catalysis, ATP hydrolysis within V 1 drives active transport of protons from the cytosol to the other side of the membrane via rotation of a proteolipid ring structure in V o . Detachment of V 1 from V o is an important mechanism that inhibits V-ATPase proton transport (3,10,11). Disassembly and reassembly of V 1 V o has been observed in yeast, insects, and mammalian cells (3,(11)(12)(13). However, the cellular mechanisms governing V-ATPase regulation by reversible disassembly are not well understood. In yeast, V-ATPase reversible disassembly is intertwined with cytosol pH changes (14,15) the RAS/cAMP/PKA pathway (16), and glycolysis (3).
Disassembly of the yeast V 1 V o complex occurs when glucose is not available. Upon glucose depletion, V 1 subunit C is released into the cytoplasm, causing the rest of the V 1 domain to dissociate from V o (10,17,18). Disassembly completely inactivates V-ATPase pumps because V 1 without V o cannot hydrolyze ATP, and V o without V 1 cannot transport protons (2). Consequently, protons cannot not be redistributed from the cytosol into the vacuole, which alkalinizes the vacuolar lumen and acidifies the cytosol (19,20). Disassembly is reversed by readdition of glucose to yeast cells (10,21). After reassembly, V 1 V o resumes ATP-driven proton transport, which restores vacuolar and cytosol pH homeostasis (20).
Complete glucose depletion causes about 75% of the V 1 V o complexes to disassemble (10,21); the remaining 25% of the pumps maintain the V-ATPase activity necessary to sustain basal cellular functions. Presumably yeast V 1 V o disassembly preserves energy when glucose, the main energy source, is lim-ited. This hypothesis is supported indirectly by observations that rapidly fermentable carbon sources like glucose, mannose, and fructose trigger reassembly (21), whereas less efficiently fermentable carbon sources like raffinose and galactose or nonfermentable carbon sources like glycerol and ethanol cannot trigger reassembly (10,21). These studies indirectly linked the V 1 V o assembly levels with glycolysis. Additional evidence suggesting that reassembly correlates with glycolysis comes from studies using the nonmetabolizable glucose analogue 2-deoxyglucose, which cannot substitute for glucose in triggering reassembly (21), and also from studies using a phosphoglucose isomerase (pgi1) glycolytic mutant in which fructose, but not glucose, induces V 1 V o reassembly (21).
V-ATPase function is defective in both PFK-1 deletion mutants, but it is more severely impaired in pfk2⌬ (24). Glucosedependent V 1 V o reassembly is normal in pfk1⌬, but it is reduced by 40 -50% in pfk2⌬. In addition, V-ATPase proton transport is partially suppressed at steady state. The cells have alkalinized vacuoles and display pH and calcium growth sensitivity, which is characteristic of yeast cells with partially defective V-ATPase activity. The mechanisms by which V-ATPase is suppressed in pfk2⌬ cells are elusive. Subunit Pfk1p may regulate V-ATPase through its interaction with V 1 V o in the pfk2⌬ mutant. In addition, a reduction in PFK-1 function and the glycolysis flow may inhibit V-ATPase in pfk2⌬ cells.
We have now investigated the interrelation between glycolysis and V-ATPase function to gain new knowledge of the mechanisms underlying glucose-dependent V-ATPase regulation. After readdition of glucose to glucose-deprived cells, V 1 V o reassembly is proportional to the glycolysis flow until pfk2⌬ reaches a metabolic threshold necessary to complete V 1 V o reassembly. At steady state, V-ATPase also communicates with the glycolysis flow. This communication modulates proton transport to restore pH homeostasis when glucose is abundant in pfk2⌬. This is a new mechanism by which glucose regulates V-ATPase. It occurs at steady state and does not involve disassembly/reassembly.

Stimulation of Glycolysis Rescues V-ATPase Function in pfk2⌬ at Steady
State-To test the hypothesis that V-ATPase function is adjusted in response to changes in the glycolysis flow, we manipulated glycolysis in pfk2⌬ cells and measured V-ATPase functions. First, we measured PFK-1 activity in pfk2⌬ cells to determine the extent of PFK-1 inhibition. As shown in Fig. 1A, pfk2⌬ cells only express subunit Pfk1p. The enzymatic activity of PFK-1 was 42% lower in pfk2⌬ than wildtype cells. This result is consistent with previous reports indicating that PFK-1 homomeric complexes consisting of subunit ␣ (pfk2⌬ cells) are less active than wild-type native PFK-1 hetero-octameric complexes (␣ 4 ␤ 4 ) (27).
In yeast, the final product of glycolysis, pyruvate, is fermented into ethanol, the levels of which were measured in both strains at steady state (Fig. 1B). In the presence of 2% glucose, which is the standard glucose concentration in yeast growth FIGURE 1. PFK-1 activity and glycolysis are defective in pfk2⌬ cells. A, PFK-1 enzymatic activity is reduced in pfk2⌬ cells. Wild-type and pfk2⌬ cells were grown overnight to mid-log phase (optical density of 0.6 -0.8 A 600 /ml) in YEP medium containing 2% glucose. The cells were converted to spheroplast by zymolase treatment, and protein was separated on a 10% SDS-PAGE gel. The gel was immunoblotted with primary polyclonal antibodies to PFK-1 and secondary antibodies conjugated to alkaline phosphate. Protein markers (right) are 150, 100, and 75 kDa. PFK-1 enzymatic activity was measured for an optical density of 20.0 A 600 obtained from cells grown overnight to mid-log phase in medium containing 2% glucose. Data are expressed as the average Ϯ S.D. (n ϭ 3 separate experiments). Relative averaged values were used to express percentage activity. The wild-type PFK-1 activity in the whole cell lysate was 0.014 -0.017 mol of fructose 1,6-bisphosphate/min/optical density A 600 . B, ethanol concentration is lower in pfk2⌬ than in wild-type cells. The cells were cultured to an optical density of 0.6 -0.8 A 600 /ml in YEP medium containing 2% or 4% glucose, converted to spheroplasts by zymolase treatment, and resuspended in fresh medium containing 1.2 M sorbitol plus 2% or 4% glucose. After 20 min at 30°C, the ethanol concentration was measured. Data represent three independent experiments. Error bars are standard deviation. Statistically significant differences (*, p Ͻ 0.05; **, p Ͻ 0.01) were determined by two-tailed unpaired t test. medium, the ethanol concentration was observed to be 6.6 Ϯ 0.9 mM and 3.8 Ϯ 0.2 mM in the wild-type and pfk2⌬ strains, respectively. This result indicates that glycolysis was defective in pfk2⌬. The 42% ethanol concentration reduction is consistent with the 42% reduction of PFK-1 activity measured in pfk2⌬ cells and with independent studies that reported reduced fructose-1,6-biphosphate, the 6-phosphofructo-1-kinase reaction product (27)(28)(29).
Previous characterizations of glycolysis (27) and V-ATPase pumps (24) in pfk2⌬ cells were conducted in 2% glucose. Although 2% glucose is optimal for wild-type cells, it might not be optimal for PFK-1 mutants. We reasoned that the pfk2⌬ glycolytic mutant would require larger concentrations of glucose to maintain a substantial glycolytic flow. The pfk2⌬ cells were grown in medium containing 2% or 4% glucose, and the ethanol concentration was compared with wild-type cells (Fig.  1B). Although lower ethanol levels were detected in 2% glucose, the ethanol concentration increased in 4% glucose in pfk2⌬ cells. It reached 83% of the ethanol concentration in the wildtype cells, indicating that glycolysis was stimulated.
Because the vacuolar pH is regulated by V-ATPase, vacuolar alkalization is a direct consequence of inhibiting V-ATPase activity. The effects of stimulating glycolysis on V-ATPase proton transport were determined by measuring the vacuolar pH in vivo. We used the pH-sensitive fluorophore BCECF, which accumulates in yeast vacuoles. The cells were loaded with BCECF-AM, and the pH was measured fluorometrically according to standard calibration curves (20,24,31). The vacuolar pH was measured in wild-type cells and pfk2⌬ after growing the cells in 2% versus 4% glucose. vma3⌬ cells were used as a control because vma3⌬ completely lacks V-ATPase function (32). In wild-type cells, where the V-ATPase pumps are fully active, the vacuolar lumen was acidic (pH ϭ 5.2-5.9) at both glucose concentrations ( Fig. 2A). In 2% glucose, the pfk2⌬ vacuoles were alkalinized (pH 6.6), similar to the inactive V-ATPase mutant vma3⌬ (pH 6.8).
In 4% glucose, the acidic pH of the vacuoles was rescued. The pfk2⌬ vacuolar lumen pH in 4% glucose (pH 5.8) was almost identical to the wild-type vacuolar lumen pH in 2% glucose (pH 5.9). These results show that stimulation of glycolysis with 4% glucose stimulated V-ATPase proton transport in pfk2⌬ cells at steady state. Thus, they indicate that V-ATPase proton transport is coupled to the flow of glycolysis at steady state.
Sensitivity to elevated pH and calcium at all temperatures is a signature of mutants that lack all V-ATPase activity (vacuolar and Golgi V-ATPase). It is known as the Vma Ϫ (vacuolar membrane ATPase) growth phenotype (33) and is only shown when over ϳ70% of V-ATPase function is compromised (34 -36). The vma mutants grow at pH 5.0, but growth is reduced or abolished at neutral pH in the presence of high concentrations of calcium (32). The physiological basis of Vma Ϫ growth is not fully understood because the downstream consequences of inhibiting V-ATPase are extensive (33). Vacuolar alkalization alone is not sufficient to cause the Vma Ϫ phenotype (37). However, pfk2⌬ growth was fairly reduced with 2% glucose under non-permissive growth conditions at 37°C (Fig. 2B) (24). This modest growth defect was indicative of defective vacuolar but not Golgi V-ATPase function in pfk2⌬ (37).
To determine whether the Vma Ϫ growth is rescued after stimulation of glycolysis, pfk2⌬ cells were plated on medium containing 100 mM CaCl 2 buffered to pH 7.5 ( Fig. 2B) that was supplemented with 2% glucose (left panels) versus 4% glucose (right panels). The Vma Ϫ growth defect was rescued in 4% glucose, indicating that the vacuolar V-ATPase function was restored. As expected, the wild-type strain grew under all the conditions, and vma3⌬, which completely lacks V-ATPase function (32), failed to grow under non-permissive conditions. The pfk1⌬ mutant strain was examined for reference. The pfk1⌬ strain lacks the PFK-1 subunit Pfk1p but has greater glycolysis flow than the pfk2⌬ strain (27). The pfk1⌬ cells closely mimicked the wild-type cells, as reported before (24). A normal pfk1⌬ growth was consistent with the concept that vacuolar V-ATPase function is coupled to the glycolysis flow.
Metabolic Reactivation Is Defective in pfk2⌬ Cells-Readdition of glucose to cells deprived of glucose rapidly reactivates glycolysis (38 -40). We measured ethanol under V-ATPase disassembly (glucose depletion) and reassembly (glucose readdition) conditions to determine whether the glycolysis flow dictates the V-ATPase reassembly level. As described below, glucose-dependent metabolic reactivation was defective in pfk2⌬, which cannot sufficiently reassemble V 1 V o after readdition of glucose.
Under disassembly conditions (10 min after glucose removal), the ethanol concentration dropped to about 2.5 Ϯ 0.2 mM in the wild-type and pfk2⌬ strains (Fig. 3A). As a control, FIGURE 2. Metabolic reactivation is defective in pfk2⌬ cells. A, the acidic vacuolar pH is restored in 4% glucose. Wild type, pfk2⌬, and vma3⌬ cells were cultured to mid-log phase in YEP containing 2% or 4% glucose. The cells were harvested and 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 to quantitatively assess vacuolar pH. The average fluorescence over 6 min was compared with a standard curve to generate absolute pH values. Data are presented as average pH values from three independent experiments, and error bars are standard deviation. Statistically significant differences (*, p Ͻ 0.05) were determined by two-tailed unpaired t test. B, the vma Ϫ growth phenotype is rescued in 4% glucose. Cells were cultured to mid-log phase in 2% or 4% glucose, and 10-fold serial dilutions were stamped onto YEP plates containing 2% or 4% glucose adjusted to pH 5.0 and pH 7.5 plus 100 mM CaCl 2 . Cell growth was monitored for 72 h at 37°C. Shown are representative plates of triplicates.
the ethanol concentration was determined in cells treated with 2-deoxyglucose after glucose was removed. Because of its inability to activate glycolysis beyond 2-deoxyglucose-6-phosphate formation (41), the observed ethanol concentration of 2.5 mM was comparable with that of the glucose-deprived wild type. Because 2-deoxyglucose does not support ethanol production, this indicates that glycolysis had stalled upon removal of glucose.
Under reassembly conditions (10 min after glucose readdition), the concentration of ethanol increased, indicating that glycolysis resumed. Although addition of 2% glucose increased ethanol concentrations to 4.1 Ϯ 0.2 mM and 2.9 Ϯ 0.2 mM in the wild-type and pfk2⌬ strains, respectively, steady state levels were not reached. These results also showed that 2% glucose resumed glycolysis in the two strains, but to a lesser extent in pfk2⌬. The pfk2⌬ mutant, when 2% glucose was readded, mimicked the wild-type cells deprived of glucose, suggesting that insufficient glycolytic flow upon reactivation is responsible for the V 1 V o reassembly defects in pfk2⌬.
Addition of glucose after nutrient limitation is known to trigger a rapid increase of NADH/NAD ϩ that inhibits the glycolytic enzyme triose phosphate dehydrogenase (39) and results in a transient peak in NADH levels (42). To monitor the rate of metabolic reactivation when glucose was readded, NADH was measured by autofluorescence. Readdition of 2% glucose stimulated NADH synthesis in wild-type and pfk2⌬ cells (Fig. 3B). However, the NADH synthesis rate was reduced by 70% in pfk2⌬ cells. These results are consistent with the ethanol measurements, indicating that glucose-mediated metabolic reactivation was defective in pfk2⌬.
We measured metabolic reactivation after adding 4% glucose. Fig. 3, A and B, shows the level of ethanol and NADH formation that resulted from the addition of 4% glucose to pfk2⌬ after a brief glucose depletion period. The concentration of ethanol reached a steady-state level of 6 mM Ϯ 0.6 mM and 3.8 Ϯ 0.7 mM for the wild type and pfk2⌬, respectively. The rate of NADH synthesis in wild-type and pfk2⌬ cells increased by 15%. It reached up to 46% in pfk2⌬ cells, which was a larger increase than that observed upon addition of 2% glucose. Thus, doubling the concentration of glucose from 2% to 4% stimulated glucose-dependent metabolic reactivation by about 50% in pfk2⌬, as indicated by ethanol levels and the NADH formation rate. Next, we exposed pfk2⌬ to 2% and 4% glucose to manipulate glycolysis and directly establish the role of glycolysis in V-ATPase assembly and activity.
4% Glucose Rescues Glucose-dependent Reassembly and Vacuolar Acidification-If V 1 V o reassembly is governed by the glycolysis flow, we anticipated reassembly levels to increase after stimulating glycolysis in pfk2⌬ cells. Fig. 4A shows the extent of V 1 V o reassembly in pfk2⌬ after addition of 4% glucose. Reassembly levels were measured using biosynthetically 35 S-radiolabeled cells in pulse-chase experiments as described under "Experimental Procedures." The radiolabeled cells were chased in YEP medium containing 2% glucose for 20 min (steady-state condition, ϩ G), YEP medium without glucose for 10 min (disassembly condition, Ϫ G), and after readdition of varied concentrations of glucose (0.1-4%) for an additional 10 min (reassembly conditions, Ϯ G). After the chases, the V-ATPase complexes were immunoprecipitated under nondenaturing conditions with anti-V 1 subunit B to immunoprecipitate V 1 and V 1 V o versus anti-V o subunit a, which can only immunoprecipitate V o when it is disassembled from V 1 .
About 80% of the V 1 V o complexes disassembled upon glucose depletion (Fig. 4A). After glucose readdition, reassembly was proportional to the concentration of glucose added. However, pfk2⌬ required greater concentrations of glucose to reach The wild-type and pfk2⌬ strains were cultured to an optical density of 0.6 -0.8 A 600 /ml in YEP medium containing 2% glucose and converted to spheroplast by zymolase treatment. Spheroplasts were resuspended in fresh medium containing 1.2 M sorbitol and incubated at 30°C as follows: in the presence of 2% glucose (ϩ G, 20 min), without glucose (Ϫ G, 10 min), and after readdition of 2% glucose, 4% glucose, or 2% 2-deoxyglucose to glucose-deprived cells (Ϯ G, 10 min). The ethanol concentration was measured using a colorimetric assay. Data represent three independent experiments. Error bars are standard deviation. The dashed line represents the Ն40% reassembly threshold. B, the rate of NADH formation after glucose re-addition is decreased in pfk2⌬ cells. Wild-type and pfk2⌬ cells were cultured overnight in medium containing 2% glucose, and then cells were resuspended in 50 mM potassium phosphate buffer (pH 6.8) for 3 h, after which NADH autofluorescence was monitored (emission (Em), 450 nm; excitation (Ex) 366 nm) for 60 s in the absence of glucose and for another 120 s after 2% or 4% glucose final concentration was added. Velocities were calculated for the initial 15 s following glucose addition and expressed relative to wild-type cells. Data represent two independent experiments. Error bars are standard deviation. Statistically significant differences (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001) are as compared with the wild type in the presence of 2% glucose and were determined by two-tailed unpaired t test.

Yeast V-ATPase Regulation by Glycolysis
wild-type levels of reassembly. In the wild-type cells, 50% reassembly was detected after addition of 0.3% glucose; 100% reassembly was reached with 0.5% glucose. In contrast, addition of 0.3% glucose did not trigger significant reassembly in pfk2⌬ cells, and 2% glucose led to only 60% reassembly. Approaching steady-state levels of assembly (100%) required 4% glucose.
Readdition of 2% glucose, which restored 62% of the steadystate ethanol concentration, triggered comparable levels of V 1 V o reassembly (60%). Readdition of 4% glucose, which restored steady-state ethanol levels led to steady-state reassembly levels (100%). Collectively, these results showed that V 1 V o reassembly was proportional to the glycolysis flow in pfk2⌬ until the cells reached Ն40% of the steady-state wild-type ethanol concentration and NADH formation rate.
The V 1 V o complexes reactivate after reassembly, which acidifies the vacuolar lumen and restores membrane gradients and secondary transport systems (20). To determine whether reassembly restored V-ATPase proton transport, we measured the vacuolar pH after readdition of 4% glucose to pfk2⌬. Vacuolar acidification was detected within 1-2 min of glucose readdition to the wild-type cells (Fig. 4B), which is within the timescale when reassembly occurs (3,21). The V 1 V o reassembly level was somewhat higher in 4% glucose (Fig. 4A), and the wild-type vacuole pH was slightly more acidic.
Glucose-induced acidification was defective in pfk2⌬ in response to 2% glucose even though 60% of the V 1 V o complexes reassembled in the glycolytic mutant (Fig. 4A). However, V-ATPase proton transport resumed after 4% glucose readdition. Reactivation of V-ATPase proton transport was indicated by the rapid acidification of the vacuolar lumen. The vma2⌬ strain that cannot assemble V 1 V o complexes (43) was used as a negative control because vma2⌬ completely lacks V-ATPase activity and vacuolar buffering capacity (2). As expected, the net vacuole pH was considerably more alkaline in vma2⌬ than in wild-type and pfk2⌬ cells. A gradual pH drop was detected in vma2⌬ at 3 min after V-ATPase reactivation (acidification) was complete in wild-type and pfk2⌬ cells (Fig. 4B). This observation further indicated that glucose-mediated vacuolar acidification was V-ATPase-dependent in pfk2⌬ cells.
Proton Transport Is Increased at Vacuolar Membranes-The rates of proton transport and ATP hydrolysis were measured in purified vacuolar membranes from pfk2⌬ cells. V-ATPase activity was determined in the presence and absence of the specific V-ATPase inhibitor concanamycin A, and the ratio of proton pumping to ATP hydrolysis was used as a means of estimating the coupling efficiency of the enzyme when glycolysis was stimulated in pfk2⌬ cells.
In wild-type cells cultured in 4% glucose, ATP hydrolysis and proton transport activities increased by 21% and 17%, respectively, compared with those grown in 2% glucose (Fig. 5A). In contrast, ATP hydrolysis was unchanged in pfk2⌬ membranes from cells cultured in 2% versus 4% glucose, which was 64 -67% of the wild-type ATP hydrolysis. However, proton transport significantly increased (by 30%) when pfk2⌬ cells were grown in 4% glucose from that observed in 2% glucose. As a result, the proton transport/ATP hydrolysis ratio increased 24% in the pfk2⌬ mutant (from 0.63 to 0.83). Western blotting analyses of vacuolar membrane fractions showed comparable levels of V 1 subunit A and subunit B in 2% glucose and 4% glucose (Fig. 5B). Based on these results, we concluded that the V-ATPase pumps fine-tuned their catalytic activity in 4% glucose, which enabled pfk2⌬ cells to restore vacuolar pH homeostasis.
Pfk1p Subunit Binding to V-ATPase Decreases in 4% Glucose-The individual PFK-1 subunits expressed in pfk1⌬ (subunit Pfk2p) and pfk2⌬ (subunit Pfk1p) cells co-immunoprecipitate with V-ATPase and co-purify with vacuolar membrane fractions (24). Thus, deletion of one PFK-1 subunit does not prevent interaction of the other subunit with V-ATPase.
We asked whether binding of the subunit Pfk1p to V-ATPase was affected in 4% glucose in the pfk2⌬ mutant. The level of Pfk1p subunit co-immunoprecipitated with the anti-V 1 subunit A monoclonal antibodies that recognize V 1 and V 1 V o complexes (21) was analyzed. About 60% fewer Pfk1p co-immunoprecipitated with V 1 subunits in pfk2⌬ cells in 4% glucose (Fig.  6). This decrease in Pfk1p-V-ATPase binding was not caused by change in expression of Pfk1p and/or V-ATPase subunits because these protein expression levels were comparable in whole cell lysates from 2% and 4% glucose (Fig. 6, Input). Rather, Pfk1p-V-ATPase binding was decreased in 4% glucose Results are presented as average from three independent experiments. Error bars are standard deviation. Statistically significant differences (*, p Ͻ 0.05; ***, p Ͻ 0.001) as compared with steady state (2% glucose) were determined by two-tailed unpaired t test. Error bars show mean Ϯ S.D. B, readdition 4% glucose restores pfk2⌬ vacuolar acidification after reassembly. Cells were stained with BCECF-AM and deprived of glucose for 10 min in 1 mM HEPE-MES (pH 5.0) buffer, and glucose was readded to a final concentration of 2% or 4% (arrow). The ratio of fluorescent emission (535 nm) excited at 490 nm and 450 nm was calculated, and the vacuolar pH was measured using calibration curves. Data represent three independent experiments. Error bars are standard deviation.
in pfk2⌬, which was suggestive of lower Pfk1p-V-ATPase binding affinity in this mutant.

Discussion
This study capitalized on the cellular mechanisms suppressing V-ATPase function in pfk2⌬ to gain new knowledge of the mechanisms underlying glucose-dependent V-ATPase regulation. We took advantage of the fact that V-ATPase is fully assembled in pfk2⌬ cells at steady state and glycolysis partially suppressed to manipulate the glycolysis flow and assess its direct involvement on V-ATPase function.
At steady state, the ratio of proton transport to ATP hydrolysis increases in response to high glucose levels in pfk2⌬. Enhanced V-ATPase proton transport restores vacuolar pH homeostasis. It likely allows cells to preserve energy when glycolysis is suboptimal and glucose abundant (4% glucose). One importance of these findings is that they revealed V-ATPase elasticity of coupling as a new mechanism of how glucose regulates V-ATPase pumps without changing the V 1 V o assembly state. This study also showed that, under V 1 V o reassembly conditions, the level of glucose-induced reassembly directly corresponds to the glycolysis flow in pfk2⌬ cells. V 1 V o reassembly is complete after the rate of glycolysis reaches a threshold (40 -46% of the wild-type rate in 2% glucose) when metabolism resumes after glucose is readded to glucose-deprived cells.
The Glycolysis Flow Communicates with V-ATPase and Regulates Its Activity at Steady State-At steady state, in 2% glucose, the ethanol concentration is significantly reduced (by 42%) in pfk2⌬ cells. This level of glycolytic reduction mimics the level of PFK-1 activity reduction (42%) (Fig. 1A), as expected, because PFK-1 catalyzes the second limiting step reaction of the pathway.
V-ATPase proton transport is suppressed in 2% glucose in pfk2⌬. Consequently, the vacuolar lumen is alkalinized ( Fig.  2A). This alteration in V-ATPase function is not due to V 1 V o disassembly or catalytic defects because pfk2⌬ cells cultured in 2% glucose display wild-type levels of V 1 V o complexes at the membrane and express catalytically competent wild-type V-ATPase pumps (24). Rather, vacuolar V-ATPase function is inhibited in vivo in pfk2⌬ cells cultured in 2% glucose. This study indicates that V-ATPase activity is suppressed in pfk2⌬ because glycolysis is suppressed. Regulation of V-ATPase activity at steady state by the glycolysis flow has not been reported before.
V-ATPase activity is remarkably sensitive to changes in the glycolysis flow. Addition of 4% glucose restores 83% of wildtype ethanol concentration at steady state (Fig. 1B). This signif-FIGURE 5. Proton transport is increased at vacuolar membranes from pfk2⌬ cells cultured in 4% glucose. A, V-ATPase coupling efficiency is increased in 4% glucose. Vacuolar membrane fractions from wild-type and pfk2⌬ cells were purified by density centrifugation. ATP hydrolysis (left panel) was assayed spectrophotometrically in the presence and absence of the V-ATPase inhibitor concanamycin A (100 nM) by using an enzymatic coupled assay that measures NADH oxidation at 340 nm. The average wild-type specific activity in 2% glucose for the concanamycin A-sensitive ATP hydrolysis was 3.0 mol of ATP/min/mg of total vacuolar protein. ATP-dependent proton transport (right panel) was measured by fluorescence quenching of 1 M 9-amino-6-chloro-2-methoxyacridin (excitation, 410 nm; emission, 490 nm) upon addition of 0.5 mM ATP/1 mM MgSO 4 to 5 g of total protein in vacuolar membrane vesicles. Initial velocities were calculated for 15 s following MgATP addition. The average wild-type slope was Ϫ1117.47 fluorescence units/15 s. Data represent six independent vacuolar preps. Statistically significant differences (*, p Ͻ 0.05; **, p Ͻ 0.01; ns, not significant) were as compared with the wild type in the presence of 2% glucose and determined by two-tailed unpaired t test. B, V-ATPase assembly is comparable in 4% glucose and 2% glucose. Vacuolar membrane vesicles were purified from pfk2⌬ and wild-type cells cultured overnight in 2% glucose or 4% glucose. Membrane protein (1 g total membrane protein/well) was separated by SDS-PAGE in 10% gels. Gels were immunoblotted with primary monoclonal antibodies to V 1 subunit A and subunit B, and secondary antibodies conjugated to alkaline phosphate. Protein markers (left) are 77 and 50 kDa. This gel was modified to excise a lane containing pfk1⌬ membranes.

FIGURE 6. Pfk1p subunit binding to V-ATPase decreases in 4% glucose.
Overnight mid-log phase cultures (optical density of 0.8 -1.0 A 600 /ml) were lysed, and V-ATPase complexes were immunoprecipitated with anti-A monoclonal antibody. Immunoprecipitated protein (IP) and total lysate protein (Input) were loaded on 10% SDS-PAGE gels. Pfk1p and V-ATPase (V 1 subunits A and B) were detected by immunoblots using, respectively, anti-PFK-1 polyclonal antibodies and anti-B and anti-A monoclonal antibodies and horseradish peroxidase secondary antibodies. Ab, antibody alone; HC, antibody heavy chain. Protein markers are 150, 100, 75, and 50 kDa. A representative gel is shown (top panel). Gels from two independent experiments were scanned using a Bio-Rad ChemiDoc XRSϩ, and data were analyzed using Multi Gauge V3.0 and GraphPad Prism 5 software. Data were expressed as -fold increase Pfk1p:V 1 subunit ratio Ϯ S.D. relative to the wild type (bottom panel).
icant stimulation of glycolysis also restores the vacuolar acidic pH ( Fig. 2A). This result is indicative of a causal correlation between the glycolysis flow and V-ATPase activity. It is evidence that glucose regulates V-ATPase activity at steady state. Accordingly, the PFK-1 subunit deletion mutant pfk1⌬, which has milder glycolysis defects than pfk2⌬ (27), has milder V-ATPase alterations (24). Thus, V-ATPase proton transport is adjusted in response to the glycolytic flow at steady state.
V-ATPase Adjusts Catalytic Activity in Response to Glucose When Glycolysis Is Suboptimal-V-ATPase catalytic coupling is tighter in pfk2⌬ cells in 4% glucose (Fig. 5A) when glycolysis operates below capacity at steady state. The proton transport/ ATPase ratio increases by 24%, indicating that V-ATPase is more efficient (transports more protons per ATP hydrolyzed), probably to help cells to preserve energy. Importantly, these findings indicate that ATP hydrolysis and proton transport by V-ATPase pumps may not always be optimally coupled in vivo.
Using pfk2⌬ cells, this study showed that glucose can control V-ATPase activity without changing the V-ATPase assembly state. Until now, glucose-dependent regulation of yeast V-ATPase has been via V 1 V o disassembly and reassembly in response to glucose depletion and readdition, respectively (3). Although the ability of V-ATPase to change coupling of ATP hydrolysis and proton pumping has been described before (44), the involvement of glucose and glycolysis has not been reported. Increased efficiency of coupling was observed in some genetic mutants (45) and when V-ATPase pumps were exposed to low ATP concentrations in vitro (46). In yeast, the N-end domain of V o subunit a (47, 48) and a non-homologous region of the V 1 subunit A (45) have been implicated in this type of regulation.
Changes in the proton transport to ATPase activity ratio may be linked to "loose" conformations required for disassembly and reassembly because, during catalysis, relative rotation of V 1 and V o subunits requires stable V 1 V o assembly. Accordingly, catalytic slip does not occur in the molecular motors that do not reversibly disassemble, such as the F 1 F o ATP synthase. One possibility is that, in 4% glucose, V 1 V o intermediary structures that slip during catalysis are stabilized in pfk2⌬ to restore the membrane pH gradients necessary for an array of other transporters when glucose concentration is high (20).
Complete V 1 V o Reassembly after Glucose Readdition Requires the Glycolysis Flow to Reach a Threshold-Glucose depletion, which inhibits glycolysis, leads to V 1 dissociation from V o , and its readdition, which resumes glycolysis, causes V 1 to reassociate with V o (3). Thus, yeast V-ATPase assembly is intimately linked to the glycolysis flow.
We reasoned that if a metabolic input drives V 1 V o reassembly, then it would reach a certain threshold within the first minutes after glucose readdition to glucose-deprived cells. Below this threshold, reassembly is defective. We found the threshold to be 40 -46% of the wild-type glycolytic capacity. We have chosen to monitor the glycolytic intermediate NADH to establish the metabolic threshold for reassembly after glucose readdition. The NADH formation rate corresponds to the rate of glycolysis. In addition, NADH autofluorescence is measureable in intact cells in real time. The wild-type NADH formation rate in 2% glucose is used as a reference for these studies because 2% is the standard glucose concentration in yeast growth medium.
When 2% glucose is readded, the NADH formation rate is only 31% of the wild-type rate. Under these conditions, the pfk2⌬ cells cannot sufficiently reassemble V 1 V o . Addition of 4% glucose, which stimulates the rate of glycolysis to 46% (the proposed NADH threshold), prompts wild-type levels of V 1 V o reassembly in pfk2⌬ (Fig. 4A). Likewise, V 1 V o reassembly is complete when the cells produce Ն40% the wild-type ethanol levels (Fig. 3A). With reassembly, V 1 V o proton transport resumes, and the vacuoles are acidified after 1-2 min of 4% glucose readdition to pfk2⌬.
The finding that V 1 V o remains silenced in pfk2⌬ after readdition of 2% glucose even though 60% of the V-ATPase complexes are reassembled is intriguing. It can be explained if inhibitory MgADP is trapped in the catalytic sites (49) because glycolysis is significantly suppressed. Alternatively, a regulatory subunit such as V 1 subunit H, which silences V 1 after disassembly (49) and activates V 1 V o complexes at the membrane (50), could have retained an inhibitory conformation preventing proton transport in 2% glucose (51,52).
Although glycolytic ATP formation is an attractive possibility to signal reassembly and/or reactivation in 4% glucose, the levels of ATP during glucose deprivation and readdition do not appear to correlate to the V-ATPase assembly state (21). Wildtype yeast cells recover steady-state ATP concentrations in the absence of glucose (21).
How Does PFK-1 Regulate V-ATPase Activity?-The finding that V-ATPase function is fully rescued in 4% glucose in pfk2⌬ cells that lack the PFK2 gene argues against direct regulation of V-ATPase through its physical interaction with the PFK-1 subunit Pfk2p. However, the PFK-1 subunit Pfk1p may play an inhibitory role in pfk2⌬ cells. The level of Pfk1p that co-immunoprecipitates with V-ATPase subunits is significantly reduced in pfk2⌬ cells in 4% glucose (Fig. 6). This decrease can be explained if Pfk1p is recruited to support greater glycolytic demands in 4% glucose, after which V-ATPase proton transport is enhanced. Whether binding of subunit Pfk1p to V-ATPase is inhibitory is an interesting possibility that will be addressed in future studies.
This study showed that at least one mechanism by which PFK-1 modulates V-ATPase is via the glycolysis flow. The RAS/cAMP/PKA signaling pathway, which has been linked to glucose-dependent V 1 V o disassembly and reassembly, is also intertwined with PFK-1. Activation of the RAS/cAMP/PKA pathway enhances the glycolysis flow because PKA stimulates formation of fructose-2,6-bisphosphate, the most potent activator of PFK-1. Thus, RAS/cAMP/PKA could control V-AT-Pase assembly via glycolysis (16).
We cannot eliminate the possibility that glucose-dependent pH changes could also regulate V-ATPase. Cytosol alkalization is glucose concentration-dependent and correlates to the level of V 1 V o assembly (14,53). A larger ⌬pH between the cytosolic and the luminal sides of the vacuolar membrane may stimulate V 1 V o activity. Similar to the activating effect ⌬pH has on the F o subunit a of the evolutionary related F-ATP synthase (54), binding of protons to the membrane-bound V o subunit a of the V-ATPase could prime V 1 V o proton transport. In Arabidopsis, the V-ATPase coupling ratios are sensitive to the cytosol and vacuolarpH (55).Inthelemonfruit,variablecouplingandpH-dependent slip regulate the V-ATPase pump (56).
Clearly, the scope of glucose-dependent V-ATPase regulation is more complex than initially anticipated. It extends beyond V 1 V o disassembly and/or reassembly and is intimately linked to the metabolic state of a cell, specifically the glycolysis flow. The finding that V-ATPase changes catalytic efficiency when the cellular demands for membrane transport increase (4% glucose) and glycolytic ATP production operates below capacity can have implications in human health, particularly for distal renal tubular acidosis (25), viral infections (57), and the metabolic switch in cancers (58,59), where glucose-dependent regulation of V-ATPase is conserved.
Growth Phenotype-Cells were grown overnight to 0.6 -1.0 A 600 /ml in YEPD medium buffered to pH 5.0 with 50 mM succinic acid/50 mM sodium phosphate. Cultures were washed twice with sterile double-distilled H 2 O and cells with an optical density of 2.5 A 600 were resuspended into 1 ml of sterile doubledistilled H 2 O. 10-fold serial dilutions were stamped onto YEP containing 2% glucose and 4% glucose. Plates were buffered to pH 5.0 with 50 mM succinic acid/50 mM sodium phosphate (pH 7.5) with 50 mM MES/50 mM MOPS or pH 7.5 with 100 mM calcium chloride added. The plates were incubated for 72 h at 30°C and 37°C.
Immunoprecipitations-Pulse-chase experiments were conducted at the indicated glucose concentrations and times following protocols described before (21). The V-ATPase complexes were immunoprecipitated from whole cell lysates with the monoclonal antibodies 13D11 (anti-V 1 subunit B) and 10D7 (anti-V o subunit a, Vph1p isoform) and the protein separated by SDS-PAGE (13% acrylamide gels). The gels were dried, scanned in a Fuji scanner (FLA-5100), and analyzed using Multi Gauge and GraphPad Prism 5 software as described previously (24). The proportion of V o assembled into V 1 V o , determined by comparing the amount of V o subunit a immunoprecipitated with 13D11 with the total amount of V o subunit a immunoprecipitated with both antibodies. For non-radiolabeled immunoprecipitations, the 13D11 antibody was used to immunoprecipitate V-ATPase from whole cell lysates (61). Protein was separated by SDS-PAGE in 10% gels and analyzed by Western blotting using the monoclonal antibodies 8B1 (anti-V 1 subunit A) and 13D77 (anti-V 1 subunit B) and yeast PFK-1 polyclonal antibodies. The nitrocellulose membranes were blotted with horseradish peroxidase secondary antibody and scanned using a Bio-Rad ChemiDoc XRSϩ, and then the intensity of protein bands was quantified using Multi Gauge and GraphPad Prism 5 software.
NADH Autofluorescence-NADH was monitored as described by Poulsen et al. (42) with the following modifications. The cells were grown overnight to an optical density of 0.6 -1.0 A 600 /ml in YEPD, harvested, and cells with an optical density of 100 A 600 were resuspended in 50 mM potassium phosphate buffer (pH 6.8) up to a density of 10% by weight. The cells were starved of glucose by incubation in the same phosphate buffer for 3 h on a rotary shaker at 30°C and placed on ice for 10 min. The NADH fluorescence intensity (excitation at 366 nm, emission at 450 nm) was monitored without glucose for 60 s and then continuously recorded for an additional 90 s with readdition of 2% glucose or 4% glucose at 30°C in a FluoroMax 4 spectrofluorometer (Horiba Jobin Yvon Inc.).
Ethanol Concentration-For steady-state analyses, the cells were cultured overnight to an optical density of 0.6 -1.0 A 600 /ml in YEP (pH 5.0) medium (wild-type and pfk2⌬) containing 2% or 4% final glucose concentration. Cells with a total of 2.0 optical density A 600 /ml per condition were harvested and then converted to spheroplasts by zymolase treatment (35). The spheroplasts were incubated at 30°C for 10 min in YEP adjusted to pH 5.0 with 50 mM succinic acid/50 mM sodium phosphate containing 2% or 4% glucose plus 1.2 M sorbitol. The ethanol concentration was measured using an ethanol assay kit (ab65343, Abcam) according to the instructions of the manufacturer. For glucose-depletion and readdition analyses, the cells were grown in medium containing 2% glucose overnight and converted to spheroplasts as described above, and then the spheroplasts were incubated in YEP medium with 2% or 4% glucose for 20 min, in YEP for 10 min, and in YEP for 10 min, followed by addition of 2% glucose, 4% glucose, or 2% 2-deoxyglucose for 10 min.
PFK-1 Enzymatic Activity-Wild-type and pfk2⌬ cells were cultured overnight to an optical density of 0.6 -1.0 A 600 /ml in YEPD (pH 5.0) medium. Cells were converted to spheroplasts and lysed in 15 mM MES-Tris (pH 6.9) containing 5% glycerol at a final concentration of 1.0 optical density A 600 /l. PFK-1 activity was measured spectrophotometrically at 37°C using the coupled enzyme assay of Lotscher et al. (62). Whole cell lysates (20.0 optical density A 600 ) was added to the enzymatic assay mixture (25 mM Tris (pH 6.9), 2 mM ATP, 5 mM MgCl 2 , 2 mM phosphoenol pyruvate, 30 units/ml pyruvate kinase, 30 units/ml L-lactic dehydrogenase, and 0.5 mM NADH), and the reaction was started by addition of 5 mM fructose-6-phospate. NADH oxidation was monitored spectrophotometrically at 340 nm for 5 min. One unit of PFK-1 activity is defined as 1 mol fructose 1,6-bisphosphate formed/ min in an optical density of 20 A 600 .
Other Methods-Vacuolar membranes fractions were purified by Ficoll density gradient centrifugation as described before (35,48,61). Protein concentration was measured by the Bradford assay (63). ATP hydrolysis was measured by monitoring NADH oxidation spectrophotometrically (62) using 5 g of total vacuolar membrane protein in the presence and absence of 100 nM concanamycin-A. Proton transport was measured by monitoring 9-amino-6-chloro-2-methoxyacridin quenching after addition of MgATP (48).
Author Contributions-K. J. P. conceived and coordinated the study, analyzed the results, and wrote the paper. C. Y. C. designed, performed, and analyzed the experiments. D. D. performed the PFK1 activity measurements.