Vacuolar and Plasma Membrane Proton Pumps Collaborate to Achieve Cytosolic pH Homeostasis in Yeast*

Vacuolar proton-translocating ATPases (V-ATPases) play a central role in organelle acidification in all eukaryotic cells. To address the role of the yeast V-ATPase in vacuolar and cytosolic pH homeostasis, ratiometric pH-sensitive fluorophores specific for the vacuole or cytosol were introduced into wild-type cells and vma mutants, which lack V-ATPase subunits. Transiently glucose-deprived wild-type cells respond to glucose addition with vacuolar acidification and cytosolic alkalinization, and subsequent addition of K+ ion increases the pH of both the vacuole and cytosol. In contrast, glucose addition results in an increase in vacuolar pH in both vma mutants and wild-type cells treated with the V-ATPase inhibitor concanamycin A. Cytosolic pH homeostasis is also significantly perturbed in the vma mutants. Even at extracellular pH 5, conditions optimal for their growth, cytosolic pH was much lower, and response to glucose was smaller in the mutants. In plasma membrane fractions from the vma mutants, activity of the plasma membrane proton pump, Pma1p, was 65–75% lower than in fractions from wild-type cells. Immunofluorescence microscopy confirmed decreased levels of plasma membrane Pma1p and increased Pma1p at the vacuole and other compartments in the mutants. Pma1p was not mislocalized in concanamycin-treated cells, but a significant reduction in cytosolic pH under all conditions was still observed. We propose that short-term, V-ATPase activity is essential for both vacuolar acidification in response to glucose metabolism and for efficient cytosolic pH homeostasis, and long-term, V-ATPases are important for stable localization of Pma1p at the plasma membrane.

The importance of V-ATPases 3 for acidification of the vacuole/lysosomes, Golgi apparatus, and endosomes of eukaryotic cells is well established (1,2). Multiple cellular processes, including secondary transport of ions and metabolites, maturation of iron transporters, endocytic and biosynthetic protein sorting, and zymogen activation depend on compartment acid-ification and have been linked to V-ATPase activity (1,3). In some cells such as macrophages, V-ATPases play specialized roles that clearly include regulation of cytosolic pH (4,5). However, although V-ATPases pump protons from the cytosol into organelles in all cells, they are not generally believed to play a major role in cytosolic pH regulation.
The yeast Saccharomyces cerevisiae has emerged as a major model system for eukaryotic V-ATPases. One reason for this is that yeast mutants lacking all V-ATPase activity (vma mutants) are viable, but loss of V-ATPase activity in eukaryotes other than fungi is lethal (6 -9). Yeast vma mutants do exhibit a set of distinctive phenotypes, however, that includes the inability to grow at pH values lower than 3 or higher than 7 and sensitivity to high extracellular calcium concentrations (2). This Vma Ϫ phenotype suggests a perturbation of pH homeostasis in these cells that is not fully understood. It has been suggested that vma mutants survive at low extracellular pH (pH 5) by endocytosis of acidic extracellular fluid and transport to the vacuole (6,10) or that they acidify the vacuole through diffusion of permeant acids (11). There have been few direct measurements of cytosolic or vacuolar pH in the vma mutants under different extracellular conditions, however (11).
pH homeostasis is critical for survival of yeast cells, as it is for all eukaryotic cells. V-ATPases function in tandem with Pma1p, an essential P-type proton pump localized to the plasma membrane, to help control pH (12,13). Glucose, the preferred carbon source for S. cerevisiae, is metabolized through fermentation and respiration. Enormous amounts of carbonic and organic acids are produced by energy metabolism, and these constitute the main source of protons in yeast (13,14). Despite major differences in localization and structure, there are a number of similarities in regulation and function between Pma1p and V-ATPases. Both Pma1p and the V-ATPase use energy from ATP hydrolysis to pump protons out of the cytosol (13). Both pumps are regulated by extracellular glucose, although their regulatory mechanisms are distinct. Glucose stimulates Pma1 H ϩ -ATPase activity by inducing phosphorylation (15), resulting in a decreased K m for ATP and an increased V max (16,17). V-ATPases are also activated by glucose, but they reversibly disassemble in response to glucose deprivation and readdition, resulting in higher levels of assembled pumps and ATPase activity at the vacuole in the presence of glucose (18,19). The activity of both pumps is sensitive to pH, indicating that they may be tuned to respond to changes in cytosolic pH in vivo (20,21). Pma1p, in particular, is activated in response to cytosolic acidification (21), and this mode of regulation is likely to be important for maintenance of cytosolic pH in a narrow range, relatively independent of extracellular pH. Taken together, these data suggest that in addition to their parallel roles in removal of protons from the cytosol, these two pumps are responsive to similar metabolic stimuli.
Both Pma1p and the V-ATPase are also electrogenic pumps. Each establishes a pH gradient and membrane potential (⌬⌿) critical for operation of other transporters in their respective membranes but also requires mechanisms for balancing the generation of membrane potential (22,23). Potassium ion is primarily responsible for balancing the plasma membrane potential in yeast (24). Transport of K ϩ ion into the cell results in a depolarization of the plasma membrane, allowing stimulation of Pma1p (22) and consequent cytosolic alkalinization (25,26), and pH-responsive regulation of the Trk1p potassium transporter via phosphorylation provides additional means of activating Pma1p when cytosolic pH drops. The exact mechanisms for balancing the membrane potential at the vacuolar membrane are less clear; many anion channels and other transporters are present in the vacuolar membrane that could contribute to the final potential, but their individual contributions to membrane potential are not well defined.
In addition to the primary proton pumps, multiple transporters and buffers combine to determine the final cytosolic and vacuolar pH under various conditions. Both the plasma membrane and organellar membranes contain alkali cation/H ϩ exchangers that contribute to overall pH homeostasis (27,28). In mammalian cells where there is no primary proton pump comparable with Pma1p; the plasma membrane NHE (Na ϩ / H ϩ ) antiporters play a central role in pH homeostasis (29). Yeast cells also have Na ϩ /H ϩ exchangers (Nha1p) at the plasma membrane, but their function in pH homeostasis is clearly secondary to that of Pma1p (27). In addition, both mammalian and yeast cells contain organellar K ϩ (Na ϩ )/H ϩ exchangers. The founding member of this class of exchangers, yeast Nhx1p (28), localizes to the yeast endosome and has clearly been shown to play a role in regulating both vacuolar and cytosolic pH (30,31). The existence of a distinct vacuolar K ϩ (Na ϩ )/H ϩ exchanger was inferred from physiological studies (32) and was recently proposed to be the VNX1 gene product (33). Both small molecule and protein buffers are also likely to contribute to pH homeostasis in the cytosol and vacuole. In addition to the potential role for permeant acids in the vma mutants, highlighted by Plant et al. (11), it is notable that the vacuole contains very high concentrations of polyphosphate that can provide buffering capacity (34,35). It has also been proposed that vacuolar proteins are well suited to buffer the vacuole to its typical pH range (36).
Several pieces of data suggest that the plasma membrane and vacuolar proton pumps Pma1 and V-ATPase are functionally interdependent. Bowman and Bowman (37) isolated mutants in Neurospora crassa that were resistant to the specific V-ATPase inhibitor concanamycin A and discovered that they contained mutations in Pma1p. Mislocalization of Pma1 from the plasma membrane to the endoplasmic reticulum (38) or the vacuole (39) in the vma mutants has been reported, but the physiological implications of the altered localization have not been clarified. To address a potential coordination of function between the V-ATPase and Pma1p, we examined pH homeostasis under varied extracellular conditions in two vma mutants, vma2⌬ and vma3⌬. Both mutants lack all V-ATPase activity, but the vma2⌬ mutation deletes a subunit of the peripheral V 1 sector of the enzyme, whereas the vma3⌬ mutation deletes the major proteolipid subunit of the proton-pore containing the V o sector (2). Measurements of cytosolic and vacuolar pH indicate that wild-type cells readjust pH in response to the addition of glucose and K ϩ ion, as expected but that the vma mutations have dramatically perturbed pH homeostasis in both the vacuole and the cytosol. Mislocalization of Pma1p may account for some of the defects in the vma mutants, but even an acute loss of V-ATPase activity in the presence of concanamycin A abolishes vacuolar pH responses and perturbs cytosolic pH homeostasis. These results suggest an unexpectedly large role for the V-ATPase in cellular pH homeostasis.

EXPERIMENTAL PROCEDURES
Materials-2Ј,7Ј-Bis(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester (BCECF-AM) was purchased from Invitrogen, and the yeast pHluorin plasmid was a generous gift from Dr. Rajini Rao (Johns Hopkins University). Polyclonal anti-Pma1p was a generous gift from Dr. Ramón Serrano. Monoclonal antibody against Pma1p (40B7) was obtained from Abcam, and monoclonal antibodies against Pep12p and alkaline phosphatase were purchased from Invitrogen. Alexa Fluor 488 goat anti-mouse IgG used for immunofluorescence was purchased from Invitrogen. Vanadate was purchased from Sigma and activated by heating at alkaline pH as described (40). Concanamycin A and other chemicals were purchased from Sigma.
Cytosolic and Vacuolar pH Measurements-Cytosolic pH was measured using a pH-sensitive green fluorescent protein, yeast pHLuorin, as described by Brett et al. (31). BY4741 wildtype and vma mutant cells were transformed with the yeast pHLuorin-containing plasmid (31), and transformants were selected in fully supplemented minimal medium lacking leucine (SDϪleucine) and confirmed by fluorescence microscopy (31). Cells containing the pHLuorin plasmid were grown to log phase (A 600 ϭ 0.5-0.6) in YEPD buffered to the desired pH. Cells were collected by centrifugation and washed 2-3 times in YEP (YEPD medium without glucose) buffered at pH 5 and suspended at 1 g of wet cell mass/ml in the same medium. For pH shift experiments, cells were grown to log phase as described above, but cells were harvested at an absorbance of 0.5, shifted to YEPD, pH 7.5, and incubated for 3-4 more hours. The cells were washed and resuspended in YEP pH 7.5. For pH measurements, 25 l of cell suspension was added to 2 ml of 1 mM MES/triethanolamine (TEA), pH 5, or MOPS/TEA, pH 7.5, and the mixture was stirred in a cuvette. Cytosolic pH responses were recorded at 30°C (a) after 5 min of stirring, (b) after the addition of glucose to 50 mM final concentration and an additional 5 min stirring, and (c) after the addition of KCl to 50 mM final and an additional 3 min of incubation. Fluorescence intensity at excitation wavelengths 405 and 485 nm was measured in triplicate for each sample at a constant emission wavelength of 508 nm in a SPEX Fluorolog-3-21 fluorometer. Calibration of fluorescence with pH was carried out for each strain in each experiment as described (31). Calibration curves were constructed for every strain and included buffers titrated to 5, 5.5, 6.0, 6.2, 6.5, 6.7, 7.0, and 7.5.
Vacuolar pH measurement used the pH-sensitive ratiometric dye BCECF-AM as described (11,30,31). Briefly, cells were grown to log phase in YEPD, pH 5 media, and collected by centrifugation, and 100 mg of cells mass were resuspended in 100 l of the same medium. For experiments at extracellular pH 7.5, cells were washed and resuspended in YEP pH 7.5. Cells were incubated in 50 M BCECF-AM for 30 min at 30°C with shaking. The cells were then washed 2-3 times with YEP, pH 5 or 7.5 (depending on the experiment), to remove the dye and resuspended at the same density in the same media. For vacuolar pH measurements, 25 l of cell suspension was added to 2 ml of 1 mM MES/TEA, pH 5, or MOPS/TEA, pH 7.5, and the mixture was stirred in a cuvette. Response of vacuolar pH was recorded at 30°C with stirring as described above, except that fluorescence intensity at excitation wavelengths 450 and 490 nm was measured in triplicate for each sample at a constant emission wavelength of 535 nm. Calibration of fluorescence with pH was carried out for each strain as described (30,31). Values are expressed as the mean Ϯ S.E. for each condition, and statistic significance was taken as p Յ 0.05 of identical means in the two-sample t test for paired means.
Proton Export-Proton pumping across the plasma membrane was measured by recording extracellular pH changes with a pH meter (Beckman Selection 2000) (43). Cells were grown to log phase in YEPD, pH 5 medium, washed, and resuspended as described above. 25 l of the cell suspension was incubated in 15 ml of buffer (1 mM MES/TEA, pH 5), and the extracellular pH was monitored for 16 min at 30°C with shaking. Extracellular pH was recorded manually every 30 s, and glucose (to 40 mM final concentration) and KCl (to 40 mM final) were added after 3 and 8 min of incubation, respectively.
Isolation of Plasma Membrane-Preparation of total extract and plasma membrane from yeast was done according to Panaretou and Piper (44) using wild-type and vma cells from the SF838-1D␣ strain background. One liter of cells in log phase (absorbance 0.6 -0.8) were harvested by centrifugation (5 min, 5000 ϫ g), then resuspended in 80 ml of 0.4 M sucrose in buffer A (25 mM imidazole-HCl, pH 7, containing 1 g/ml pepstatin A, 2 g/ml chymostatin, 1 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, and 1 g/ml leupeptin). The cell suspension was divided between two 50-ml centrifuge tubes and pelleted by centrifugation (5000 ϫ g for 10 min). Glass beads were added to twice the pellet volume of cells followed by enough 0.4 M sucrose in buffer A to cover the cells and glass beads. Cells were subjected to vortex mixing for three periods of 2.5 min each, with pauses of 30 s on ice. To pellet unbroken cells and glass beads, suspension was diluted three times in 0.4 M sucrose buffer A and centrifuged at 530 ϫ g for 20 min. Supernatant from this step was then centrifuged at 22,000 ϫ g for 30 min. The pellet was resuspended in 2 ml of buffer A by gentle vortexing (30 s), and 1-ml aliquots of resuspended membranes were loaded onto a discontinuous sucrose gradient (prepared by overlaying three 4-ml layers of 2.25, 1.65, and 1.1 M sucrose, all in buffer A) and centrifuged overnight (14 h) at 80,000 ϫ g in the Beckman SW40Ti rotor. Membranes banding at the 2.25/ 1.65 M are enriched for plasma membrane. This band was collected, diluted 4 times with buffer A, and pelleted at 30,000 ϫ g for 40 min. The pellet was resuspended in buffer B (25 mM imidazole-HCl, pH 7, 50% (v/v) glycerol) containing the same protease inhibitors and stored at Ϫ20°C.
Plasma membrane H ϩ -ATPase activity was assayed in membrane fractions with or without a 5-min preincubation with 50 M vanadate at room temperature with gentle shaking. ATPase activity was determined using the coupled enzyme assay (45), and vanadate-sensitive activity is reported. Typically 80% of the total ATPase activity in the plasma membrane fraction was inhibited by vanadate, and there was very little concanamycin A-sensitive ATPase activity, suggesting minimal contamination with vacuoles or endosomes. Protein concentrations were determined by the Lowry method (46).
For assessment of protein levels of Pma1p, Pep12p, and alkaline phosphatase, plasma membrane fractions were solubilized, separated by SDS-PAGE, and transferred to nitrocellulose as described (42), except that a portion of the samples to be blotted for Pma1p were solubilized in 100 mM Tris-HCl, pH 6.8, 4% SDS, 20% glycerol, and 10% ␤-mercaptoethanol rather than cracking buffer. Blots were probed with mouse monoclonal antibodies against Pep12p and alkaline phosphatase and rabbit polyclonal antibodies against Pma1p. Western blot signals were revealed with alkaline phosphatase-conjugated second antibody (anti-mouse or anti-rabbit as appropriate), and the signals were quantitated using ImageJ (National Institutes of Health). Equal protein concentrations were loaded for each strain; 40 g was loaded for detection of Pma1p, and 30 g was loaded for detection of alkaline phosphatase and Pep12p.
Immunolocalization-Immunofluorescence staining of cells was done essentially as described (47). For Pma1p staining, cells were grown to a density of 1 A 600 /ml, rapidly fixed by direct addition of formaldehyde for 30 min, and then harvested and fixed overnight with 4.4% formaldehyde in 0.1 M potassium phosphate, pH 6.5. Before staining, fixed cells were permeabilized with 2% SDS. They were then stained with mouse monoclonal anti-Pma1p (Abcam, 40B7 antibody) followed by incubation with goat anti-mouse IgG conjugated to Alexa Fluor 488 (Invitrogen). Stained cells were visualized under fluorescein fluorescence optics on a Zeiss Axioplan 2 fluorescence microscope.

RESULTS
vma Mutants Poorly Regulate Both Vacuolar and Cytosolic pH-One of the defining phenotypes of vma mutants is their sensitivity to external pH, but relatively little is known about how cytosolic and vacuolar pH are regulated in response to changes in extracellular pH, even in wild-type cells. Ratiometric

V-ATPases Contribute to Cytosolic pH Control in Yeast
fluorescent methods in living yeast cells have allowed measurements of vacuolar and cytosolic pH in cells grown under different conditions and in mutants lacking proteins implicated in pH homeostasis. We introduced the fluorescent dye BCECF-AM, which measures vacuolar pH in yeast (11,30), or yeast pHluorin, a green fluorescent protein analog that measures cytosolic pH (31), into wild-type and vma mutant cells and studied the dynamic adjustments of cytosolic and vacuolar pH in response to glucose and KCl addition. As described above, both glucose and potassium ion are strongly implicated in cytosolic and vacuolar pH homeostasis (25,26,32,48). Fig. 1A shows the pH responses of wild-type cells grown to log phase in buffered pH 5 medium, labeled with BCECF, and briefly starved for glucose in a weakly buffered solution at pH 5. Under these conditions, the vacuolar pH is initially 6.07 but drops to 5.61 within 5 min after the addition of glucose (to a final concentration of 50 mM). The addition of KCl (final concentration of 50 mM) to these cells consistently resulted in a small increase in pH. In the case of vma2⌬ and vma3⌬ mutants, which lack all V-ATPase activity, the vacuole was more alkaline than in the wild-type strain under all conditions. In glucosedeprived cells, vacuolar pH was 6.55. Surprisingly, the addition of glucose to the cells resulted in further alkalinization of the vacuole. As in wild-type cells, we consistently observed a small rise in pH upon KCl addition. Taken together, these results indicate that the addition of glucose to glucose-deprived cells acidifies the vacuole in wild-type cells, but this acidification is completely missing in cells lacking the V-ATPase, and glucose addition even results in further alkalinization.
We next measured the cytosolic pH in the wild-type and vma mutant cells under parallel conditions (Fig.  1B). Wild-type cells grown at pH 5 and briefly starved for glucose at extracellular pH 5 have a cytosolic pH of 6. The addition of glucose results in a rapid increase in cytosolic pH to almost 6.70, and the addition of potassium ion results in a small additional increase to a final pH of 6.98 (these final values of cytosolic and vacuolar pH are similar to those measured previously for wild-type cells by various techniques (26, 49 -52)). After a brief glucose deprivation, the cytosol of the vma mutants is much more acidic than that of the wild-type cells. There is a remarkably little response to glucose addition. There is, however, a larger change in cytosolic pH after KCl addition. Although the final cytosolic pH was still significantly lower than in wild-type cells, the change in pH in response to KCl addition in the mutant cells (0.72 pH units) approached the size of the response of wild-type cells to glucose addition. These experiments suggest that in the vma mutants potassium ion may partially compensate the lack of response to glucose in cytosolic pH regulation. They also suggest that the V-ATPase plays an unexpectedly prominent role in regulation, not only of vacuolar pH, but also of cytosolic pH.
All of the experiments done in Fig. 1, A and B, were performed in cells grown and measured at extracellular pH 5, optimal conditions for growth of the vma mutants. We were also interested in the pH responses of the wild-type and vma mutants at elevated extracellular pH. As shown in Fig. 1, C and D, wild-type cells grown at pH 5 then incubated for 3-4 h in pH 7.5 had a more alkaline pH in both the vacuole and the cytosol relative to cells maintained at pH 5. In contrast to the pH 5 measurements, the vma2⌬ and vma3⌬ showed a very similar vacuolar pH to the wild-type strain before glucose addition. The response of the wild-type and mutant cells to glucose addition was again opposite, however. Wild-type cells show a modest decrease in vacuolar pH in response to glucose, whereas FIGURE 1. vma mutants regulate vacuolar and cytosolic pH differently than wild-type cells. Vacuolar and cytosolic pH were measured in wild-type, vma2⌬, and vma3⌬ mutants. Cultures of isogenic wild-type (WT) or vma null mutants were grown in YEPD, pH 5, at 30°C to early log phase (A 600 0.6 -0.8). For A and B, vacuolar and cytosolic pH were measured directly; for C and D, cells were harvested at A 600 0.6 then shifted to pH 7.5 medium and grown for an additional 3-4 h. To measure vacuolar pH (A and C), cultures were loaded with BCECF-AM as described (30) (vacuolar labeling was confirmed by fluorescence microscopy). Fluorescence intensity values were collected at excitation wavelengths 450 and 490 nm and emission wavelength 535 nm, and the ratio of fluorescence signals from the two excitation wavelengths was calculated and converted to pH via a calibration curve (see "Experimental Procedures"). In the case of cytosolic pH (B and D), cells transformed with pHluorin (31) were used, and fluorescence intensity was measured at excitation wavelengths 405 and 485 nm and emission wavelength of 508 nm, then converted to cytosolic pH after calibration (see "Experimental Procedures"). Assays for both vacuolar and cytosolic pH were done by adding 25 l of yeast suspended at 1:2 (w/v) to 2 ml of 1 mM MES buffer adjusted to at pH 5 or in MOPS buffer at pH 7.5 (depending on final growth condition) with TEA. Fluorescence responses were recorded after 5 min of incubation without any further addition (black bars), 5 min after the addition of glucose to 50 mM final (dark gray bar), and 5 min after the addition of KCl to 50 mM final (light gray bar). Samples were stirred and maintained at 30°C during the measurements. The asterisk indicates a significant difference (p Ͻ 0.05) for the mutant sample relative to comparable wild-type sample.

V-ATPases Contribute to Cytosolic pH Control in Yeast
vma mutants show a large increase in vacuolar pH. The addition of KCl increased the vacuolar pH of the wild-type cells from 6.5 to 6.7 but had relatively little effect on the vma mutants. In wild-type cells cytosolic pH was somewhat higher at extracellular pH 7.5 than at extracellular pH 5 both before and after glucose addition, again consistent with previous measurements (26). Cytosolic pH in the vma mutants was higher at extracellular pH 7.5 than at extracellular pH 5 and showed less difference from wild type. The pattern of responses of both wild-type and the vma mutant strain to glucose and KCl was different at elevated extracellular pH. All three strains show less response to both glucose and KCl addition at extracellular pH 7.5.
Proton Export across the Plasma Membrane Is Slowed in vma Mutants-Although the dominant role of the V-ATPase in regulating vacuolar pH was expected, the effects on the vma mutations on cytosolic pH were larger than expected. The very modest response of cytosolic pH to glucose addition was particularly surprising, because the plasma membrane proton pump, Pma1, is expected to be activated by glucose and by cytosolic acidification and to be the predominant determinant of cytosolic pH (13). As an initial means of assessing Pma1p function in the mutant cells in vivo, we measured the rate of extracellular acidification, by wild-type cells placed in a weakly buffered solution, representative of export of metabolically generated protons, in response to addition of glucose (Fig. 2). A few minutes after glucose addition, the rate of export slows, but an additional burst of proton export is induced by the addition of 50 mM KCl.
After this burst, a new steady-state rate of proton export was established. In contrast, both the vma2⌬ and vma3⌬ mutants show a much slower rate of glucose-induced proton export, a smaller burst upon KCl addition, and a slower steady-state rate. These results were consistent with the measurements of cytosolic pH described above and suggest that Pma1p is not pumping protons to the exterior as well in the vma mutants as in the wild-type cells.
To better understand the reduced proton export in the vma mutants, we obtained a partially purified plasma membrane fraction from the wild-type and mutant strains by sucrose gradient fractionation. Vanadate-sensitive MgATPase activity, characteristic of Pma1p, was measured with the results shown in Table 1. Both the vma2 and vma3 mutants had significantly lower plasma membrane ATPase activity than the wild-type strain. These results suggest that reduced activity of Pma1 may be the primary source of the reduced rate of proton export in the mutants.
We also compared the levels of Pma1 protein in the partially purified plasma membrane fractions from wild-type and vma mutants. As shown in Fig. 3, the levels of Pma1p appear to be somewhat lower in the mutants relative to the wild-type cells. Protein levels estimated from the immunoblot indicated 76 and 59% as much Pma1p signal in vma2⌬ and vma3⌬ membranes as in wild-type membranes. Although there is very little concanamycin A-sensitive ATPase activity in the partially purified plasma membranes from wild-type cells, suggesting that con-  Immunoblots were prepared from wild-type (wt), vma2⌬, and vma3⌬ cells as described under "Experimental Procedures" and probed for Pma1p, Pep12p, and alkaline phosphatase (ALP). As described under "Experimental Procedures," equal amounts of protein from each strain were loaded to allow comparison and quantitation by NIH Image J.

TABLE 1 Vanadate-sensitive ATPase activity in plasma membrane fractions
Fractions enriched for plasma membrane were isolated from each strain (all three strains were in the SF838-1Da (pep4-3) background) as described under "Experimental Procedures" and assayed for ATPase activity in the presence and absence of 50 M sodium vanadate. Vanadate-sensitive ATPase activity is expressed as the mean Ϯ S.E. for three independent isolations. Mean values for the mutants were significantly different than the wild type (p Ͻ 0.05) but were not significantly different from each other.

Strain
Mean ATPase activity

V-ATPases Contribute to Cytosolic pH Control in Yeast
tamination with vacuoles and endosomes is minimal, loss of V-ATPase activity might well affect the behavior of vacuoles and endosomes during isolation. If contamination with these compartments is much higher in the vma mutants, then the specific activity of Pma1p and Pma1p protein levels would be artificially lowered. Because it is not possible to assess the level of contamination by these organelles via V-ATPase activity in the vma mutants, we compared the levels of the endosomal protein, Pep12p (53), and the vacuolar enzyme alkaline phosphatase in plasma membrane fractions. As shown in Fig. 3, there was ϳ50% more Pep12p in the vma2⌬ membranes than in wild type and 50% less in the vma3⌬ membranes. Alkaline phosphatase levels in the wild-type and vma2⌬ mutant membranes were very similar, and vma3⌬ mutant membranes have lower levels of alkaline phosphatase. These results indicate that lower levels of Pma1 ATPase activity in the mutant strains cannot be easily explained by vacuolar and endosomal contamination of the preparations. Instead, the data suggest that both the levels of Pma1p at the plasma membrane and activity of remaining Pma1p in the plasma membrane are reduced in the vma mutants. Pma1p Mislocalization in the vma Mutants-It has been reported previously that Pma1p is mislocalized in the vma mutants (38,39), but the localization of the pump in the mutants was not firmly established, and the implications of this mislocalization for overall pH homeostasis were not addressed. To address the question of localization, we visualized Pma1p by immunofluorescence microscopy using a monoclonal Pma1p-specific antibody. Fig. 4, upper panel, shows the characteristic bright staining of Pma1p at the plasma membrane of wild-type cells. In contrast, plasma membrane staining is reduced in both the vma2⌬ (Fig. 4, middle panel) and the vma3⌬ mutants (Fig. 4, lower panel). Both of the vma strains showed a much higher level of intracellular Pma1p in most cells. This intracellular fluorescence was distributed between the vacuole, which is readily identifiable under Normaski optics, and a number of small punctate structures located both adjacent to the vacuole and elsewhere in the cytosol. The punctate structures appeared to be more frequent in the vma3⌬ mutant than in the vma2⌬ mutant. The micrographs shown in Fig. 4 were obtained from a strain containing a pep4-3 mutant, which lacks activity of the major vacuolar protease, proteinase A (54), and poorly activates many of the other vacuolar proteases, resulting in an overall deficiency in vacuolar proteolysis. We also looked at Pma1p localization in a yeast strain containing wild-type proteinase A (Pep4ϩstrain). The Pep4ϩwild-type (Fig. 5, upper panel) strain showed virtually identical Pma1p staining to that shown in Fig. 4, upper panel. The vma strains containing wild-type proteinase A (Fig. 5, middle and lower panels) still had lower levels of plasma membrane staining but also had less intracellular fluorescence and fewer cells with clear vacuolar staining. Those results suggest that a larger proportion of Pma1p appears to be turned over by proteolysis in the Pep4ϩstrain, but it accumulates in the vacuole when vacuolar proteolysis is deficient. Loss of V-ATPase activity also slows vacuolar proteolysis but less effectively than the pep4-3 mutation (41). This may help to explain the appearance of some staining of Pma1p in the vacuole of the vma mutants containing Pep4p. Importantly, the dramatic decrease in plasma membrane levels of Pma1p in the vma mutants shown in Figs. 4 and 5 may well account for the decreases rate of proton export shown in Fig. 2.
Acute Inhibition of V-ATPase Activity Perturbs pH Homeostasis-The vma mutants experience a chronic loss of V-ATPase activity and may invoke compensatory mechanisms to cope with this loss. To observe the more immediate effects of loss of V-ATPase activity, we treated yeast cells with concanamycin A, a highly potent and specific inhibitor of V-ATPases . Pma1p is mislocalized to the vacuole in vma mutants lacking vacuolar protease activity. Pma1p was localized by immunofluorescence microscopy in both wild-type and vma mutants that contained a pep4-3 mutation, which inhibits most vacuolar protease activity. Nomarski images (left panels) and indirect immunofluorescence using anti-Pma1p antibody followed by Alexa Fluor 488-conjugated secondary antibody (right panels) are shown for each field. All fluorescence images were recorded with identical exposure times. (Fig. 6). As shown in Fig. 6A, a 30-min treatment of cells grown at pH 5 with 1 M concanamycin A abolishes all glucose-induced acidification of the vacuole. The initial pH after a brief glucose deprivation of concanamycin A-treated cells was lower than in the vma mutants cells, possibly because vacuolar acidification or buffering capacity established before inhibitor treatment was partially preserved, but there was no subsequent acidification with glucose. KCl addition allowed the normal small alkalinization. As in the vma mutant cells, the cytosolic pH was consistently lower in cells treated with concanamycin A (Fig. 6B). There is a somewhat more robust pH increase in response to glucose than in the vma mutant cells, but the final cytosolic pH in the presence of glucose and KCl is significantly lower than in wild type and comparable with the vma mutants. These results indicate that even a short term loss of V-ATPase activity affects both vacuolar and cytosolic pH homeostasis.
The more robust response of cytosolic pH to glucose addition in the concanamycin A-treated cells suggested that Pma1p is better able to export protons in these cells than in the vma cells. Pma1p localization was visualized by immunofluorescence microscopy in the presence of concanamycin A (Fig. 5,  bottom panel). Pma1p localization was indistinguishable in the presence and absence of concanamycin. This result is consistent with previous data suggesting that Pma1p has a very long (11-12 h) half-life at the plasma membrane (55), so major redistribution during a 30-min treatment with concanamycin A was unlikely. We also measured the rate of proton export in concanamycin-treated cells (Fig. 6C). Although the steady-state rate of proton export is slower in concanamycin-treated cells after both glucose and KCl addition, the effects are less dramatic than in the vma mutants. These results have several implications. First, redistribution of Pma1p away from the plasma membrane is a feature of chronic loss of V-ATPase activity that does not occur when the V-ATPase is acutely inhibited. This suggests that the effects of loss of Pma1p and loss of the V-ATPase on pH homeostasis can be partially dissociated by comparison on the vma deletion mutants and concanamycin-inhibited cells. Second, even in the concanamycininhibited cells, there is a pronounced perturbation of both cytosolic and vacuolar pH homeostasis, suggesting once again that the V-ATPase has an expectedly large role in overall pH homeostasis. Finally, there is some reduction in the rate of proton export in concanamycin A-treated cells. This suggests that even though Pma1p is present at the plasma membrane, it may be less efficient when the V-ATPase is inhibited, consistent with the lower Pma1p activity suggested by Table 1 and Fig. 3 (it is important to note that fungal Pma1p activity is not directly inhibited by concanamycin A at these relatively low concentrations (56). 4

DISCUSSION
How Much Do V-ATPases Contribute to Overall pH Homeostasis in Yeast?-The roles of Pma1p and the V-ATPase in pH homeostasis are highlighted in Fig. 7. The responses of wild-type cells shown in Figs. 1 and 2 are fairly easily explained by these mechanisms. At an extracellular pH of 5, wild-type cells had a low cytosolic pH in the absence of glucose, but the cytosol rapidly became more alkaline upon glucose addition, with an additional pH rise when KCl was added. Changes to the vacuolar pH were partially reciprocal; the vacuolar pH dropped in response to glucose (Fig. 1). A similar pattern of responses was seen at extracellular pH 7.5, except that both the cytosolic and vacuolar pH were higher than at pH 5 in the absence of glucose, and the changes with glucose and KCl addition were less pronounced. As in previous studies (26,52), we find that the cytosolic pH of cells supplied with glucose is relatively insensitive to extracellular pH, ranging from 6.7 at pH ext 5 to pH 7.1 at pH ext ϭ 7.5, and is even more constant if KCl is present (cytosolic pH 7.0 -7.1). Vacuolar pH appears to be more dependent on extracellular conditions (20,30) but is generally higher in yeast vacuoles than in mammalian lysosomes. We propose that the rise in cytosolic pH upon glucose addition reflects both the availability of metabolically generated ATP and H ϩ and glu-4 G. A. Martínez-Muñ oz and P. Kane, unpublished data.

FIGURE 5. Pma1p mislocalization occurs in vma strains containing vacuolar proteases but not in wild-type cells treated with concanamycin A.
Pma1p was localized as described in Fig. 4 in wild-type (wt), vma2⌬, and vma3⌬ mutant cells that are Pep4ϩ. The wild-type cells were also treated for 30 min with 1 M concanamycin A (wt ϩ conA) and then visualized. All fluorescence images were recorded with identical exposure times. JULY 18, 2008 • VOLUME 283 • NUMBER 29 cose activation of Pma1p and the V-ATPase (57,58). However, it should be noted that under similar conditions of brief (20 min) glucose deprivation in YEP, we have shown that there is a relatively modest decrease in cellular ATP levels during the deprivation and no overall increase in total ATP levels in the first 5 min after glucose readdition (19), so regulatory effects of glucose on the two pumps may be particularly important. The additional increase in cytosolic pH with KCl addition can be explained predominantly by the transport of K ϩ into the cell through the Trk1/Trk2 K ϩ -transporters, which balances the membrane potential generated by Pma1p-mediated H ϩ export (13,22,26). The data in Fig.  2 support this interpretation; proton export from the cell is initiated by glucose addition but accelerated by KCl addition. The drop in vacuolar pH upon glucose addition may well reflect specifically glucose-induced activation of the V-ATPase (58). In support of this, the acidification of the vacuole in response to glucose addition is completely lost in the vma mutants or in cells treated with concanamycin A (see below). We consistently observe a small rise in the pH of the vacuole in wild-type cells upon KCl addition; this may be attributed to the activity of Nhx1-type exchangers, but other mechanisms are also possible (28,32,33).

V-ATPases Contribute to Cytosolic pH Control in Yeast
In contrast, the vma mutants show multiple differences from wild-type cells in their management of internal pH. As expected from previous work, the vacuolar pH of the vma mutants was consistently higher than in the wild-type cells (41,52). Vacuolar acidification in response to glucose addition was completely absent, indicating that this is a V-ATPasedependent response. Instead, vacuolar pH increased upon glucose addition to the vma mutants at extracellular pH 5 or 7.5. This loss of vacuolar protons upon glucose addition could take place through some of the numerous H ϩ -coupled FIGURE 6. Acute inhibition of V-ATPase activity with concanamycin A affects pH homeostasis. Vacuolar (A) and cytosolic (B) pH responses were measured as described in Fig. 1 for wild-type cells grown at pH 5, labeled with BCECF in YEPD as in Fig. 1, and then incubated with 1 M concanamycin A or an equivalent concentration of DMSO (control) for 30 min at 30°C in YEP, pH 5. Assays for both vacuolar and cytosolic pH were done by adding 25 l of yeast suspended at 1:2 (w/v) to 2 ml of 1 mM MES buffer adjusted to pH 5 with TEA. Fluorescence responses were recorded after 5 min of incubation without any further addition (black bars), 5 min after the addition of glucose to 50 mM final (dark gray bar), and 5 min after the addition of KCl to 50 mM final (light gray bar). Samples were stirred and maintained at 30°C during the measurements. Concanamycin was present throughout the measurements for the concanamycin-treated cells. The asterisk indicates a significant difference (p Ͻ 0.5) for the concanamycin-treated cells relative to the untreated control under the same conditions. C, proton export from concanamycin-treated or control cells was measured as described in Fig. 2, except that wild-type cells were preincubated with 1 M concanamycin A (circles) or DMSO (triangles) before initiating the export assay.  antiporters present in the vacuolar membrane (3), but a loss of buffering power in the vacuole may also contribute. Vacuolar polyphosphates play an important buffering role (35), and vma mutants have very low levels of vacuolar polyphosphate (34,52). One significant implication of the vacuolar pH measurements is that the vma mutants do not appear to acidify of the vacuole through endocytic uptake of extracellular fluid (6,10). As shown in Fig. 1A, at external pH 5 in the presence of glucose and KCl, the vacuolar pH in the vma mutants is ϳ7.0. This result implies that the synthetic lethality between the vma and the end mutants (10) must have another explanation and is entirely consistent with the conclusions of Plant et al. (11), who found that in the absence of permeant acids the vma mutants had a near-neutral vacuolar pH.
The prominent effect of the vma mutations on cytosolic pH, particularly at pH ext 5, was initially surprising. Under conditions of cytosolic acidification, Pma1p activity is stimulated, and this stimulation is regarded as central to overall pH homeostasis (21,59). Remarkably, even in the presence of glucose and KCl, the cytosolic pH of the vma mutants was ϳ0.6 pH units lower than that of the wild-type cells. The failure of the vma mutants to respond to glucose addition with an alkalinization of the cytosol was particularly noteworthy. Coupled with the evidence that there is little activation of proton export upon glucose addition (Fig. 2), these results strongly suggested that Pma1p, rather than compensating for cytosolic acidification arising from loss of V-ATPase activity with increased activity, was actually less active in the mutants. This was confirmed biochemically (Table 1). In contrast to the glucose response, the cytosolic pH of the vma mutants was more responsive to KCl addition, and KCl addition also strongly stimulated proton export. This may reflect in part stimulation of Trk1 by acidic cytosolic pH (59) and the tight interdependence of pH and K ϩ homeostasis. Further experiments will be necessary to fully determine why pH homeostatic mechanisms fail to achieve control of cytosolic pH in the vma mutant cells.
After a shift to alkaline extracellular pH (Fig. 1B), vma mutants were able to achieve a cytosolic pH very similar to that of wild-type cells, along with a much more alkaline vacuolar pH (Fig. 1, C and D). The vma mutants do not grow at pH ext ϭ 7.5, but the source of their growth defect may not be directly related to pH homeostasis. It has been suggested that copper and iron limit growth of yeast cells at alkaline pH (60), and Davis-Kaplan et al. (61) have demonstrated that the vma mutants are constitutively iron-deprived. Transcriptional profiling of the vma mutants also suggests that they experience iron deprivation, even at pH 5 (62). Other factors such as iron deprivation may well be the direct cause of the pHdependent conditional lethality.
Mislocalization of Pma1p in the vma Mutant Cells-Our data indicate that Pma1 protein is present at lower levels at the plasma membrane of the vma mutants and in the absence of vacuolar protease activity accumulates in the vacuole. When normal levels of vacuolar protease are present, Pma1p appears to be rapidly turned over at the vacuole, although other Pma1p-containing intracellular membranes are visible in a population of the vma mutant cells (Figs. 3 and 4). We know neither how nor why Pma1p is mistargeted to the vacuole in the vma mutants. Pma1p is normally a long-lived protein, but a number of factors, including mutations in Pma1p itself, altered recruitment to rafts, or altered ubiquitination, can intervene to change its longevity or divert it away from the plasma membrane to the vacuole (63)(64)(65). There is also evidence of signal-induced alterations in Pma1p stability; specifically, activation of the calcium-activated protein phosphatase calcineurin results in lower Pma1p levels (66). The vma mutants rely on constitutive calcineurin activation for viability (67), so this may contribute to the lowering of Pma1p levels. Further experiments will be necessary to determine whether Pma1p is routed directly to the vacuole in the mutants or is retrieved from the plasma membrane by endocytosis. Perzov et al. (38) demonstrated that another plasma membrane protein, Gas1, is localized normally in the vma mutants, but Davis-Kaplan et al. (61) reported that Fet3/Ftr1, a plasma membrane transporter complex responsible for high affinity iron transport, is present at low levels in the plasma membrane of vma mutants, like Pma1p. Therefore, it is still unclear whether the mutants have a general defect in targeting and/or stability of a subclass of plasma membrane proteins or whether there is a specific mislocalization of Pma1p.
The question of why Pma1p is mistargeted is inextricably linked to the question of how it reaches the vacuole. Although the idea of populating the vacuole with an alternative proton pump in the vma mutants as a means of compensating for loss of V-ATPase activity is attractive, vacuoles from the vma mutant strain have almost no vanadate-sensitive ATPase activity and no ATP-driven proton pumping (41,68), and overexpression of Pma1p does not suppress the vma mutant phenotypes (11). It is still possible that Pma1p activity in earlier compartments, such as endosomes and/or Golgi, contributes to compartment acidification, and a need for populating endosomes with Pma1p could account for the synthetic lethality of the end and vma mutants (10). However, the cost of mislocalizing Pma1p, in terms of pH homeostasis and reduced activity of other plasma membrane transporters that rely on the pH gradient and membrane potential, would appear to be quite high. In fact, some of the wide range of phenotypes of the vma mutants (69) might arise indirectly from reduced Pma1p activity rather than directly from loss of V-ATPase activity. Nevertheless, In N. crassa, mutations that reduced Pma1p activity suppress concanamycin sensitivity (70), suggesting that reduced Pma1p activity may help cells to survive in the absence of a functional V-ATPase by a mechanism that is not yet clear.
Vacuolar and Plasma Membrane Proton Pumps Collaborate in pH Homeostasis-Reduced Pma1p function does not fully account for the perturbed pH homeostasis in the vma mutants, however. Although concanamycin-treated wildtype cells show normal plasma membrane staining of Pma1p, they still show pronounced perturbations in pH homeostasis. Not only is glucose-induced vacuolar acidification eliminated after 30 min of concanamycin A treatment, but cytosolic pH is significantly lower under all conditions. We propose that V-ATPases influence cytosolic pH homeostasis on at least two levels. First, they have a direct effect that is likely to involve sequestration of protons into the vacuole and other organelles and is lost in the concanamycin-treated cells. It is interesting, however, that the rate of proton export from the cell in response to glucose is actually lower in the concanamycin-treated cells. Thus loss of V-ATPase activity decreases Pma1p activity even when it is properly localized and could indicate a "cross-talk" or coordinate regulation between the V-ATPase and Pma1p. Both of these pumps are highly regulated at a number of levels, and their regulation is not completely understood (2,13,59). Second, V-ATPases appear to be critical for efficient localization of Pma1p to the plasma membrane. Taken together, the data indicate that Pma1p and V-ATPases are not independent players working toward a common goal of pH homeostasis but, rather, are highly interdependent pumps that are coordinately regulated on multiple levels.