The Ras/cAMP/Protein Kinase A Pathway Regulates Glucose-dependent Assembly of the Vacuolar (H+)-ATPase in Yeast*

Vacuolar (H+)-ATPases (V-ATPases) are ubiquitous, ATP-driven proton pumps that acidify organelles or the extracellular space. A rapid and effective mechanism for regulating V-ATPase activity involves reversible dissociation of the two functional domains of the pump, V1 and V0. This process is best characterized in yeast, where V-ATPases are reversibly disassembled in response to glucose depletion. To identify regulators that control this process in vivo, a genetic screen was performed in yeast to search for mutants that cannot disassemble their V-ATPases when grown in the absence of glucose. This screen identified IRA1 (inhibitory regulator of the Ras/cAMP pathway 1) and IRA2 as essential genes for regulating V-ATPase dissociation in vivo. IRA1 and IRA2 encode GTPase-activating proteins that negatively regulate Ras in nutrient-poor conditions. Down-regulation of Ras lowers cAMP levels by reducing adenylate cyclase activity. Decreased cAMP levels in turn lead to reduced activity of protein kinase A (PKA). Our results show that targeted deletion of IRA2 results in defective disassembly of the V-ATPase in response to glucose depletion, and reexpression of the gene rescues this phenotype. Glucose-dependent dissociation is also blocked in strains expressing the dominant active RAS2val19 allele or in strains deficient for the regulatory subunit of PKA, both of which lead to constitutively active PKA. These results reveal a role for PKA in controlling glucose-dependent V-ATPase assembly in yeast.

are important for a variety of physiological processes, including coupled transport of small molecules, such as neurotransmitters (3,4); activation of lysosomal/vacuolar acid hydrolases; processing of secreted peptides, such as insulin (5); and trafficking of endocytosed ligands or newly synthesized proteins to their appropriate cellular destinations (6). Entry of enveloped viruses, such as influenza virus, or bacterial toxins, such as diphtheria or anthrax toxin, is also dependent on acidification of intracellular compartments by V-ATPases (7). V-ATPases are also localized to the plasma membrane of certain cells, where they pump protons from the cytosol into the extracellular space. This activity is essential for normal physiological processes, such as acid/base balance by renal cells (8), pH regulation in macrophages and neutrophils (9,10), bone resorption by osteoclasts (11,12), and sperm maturation and storage in the epididymus (13). Certain types of cancer cells also target V-ATPases to the plasma membrane, and expression of surface V-ATPases correlates with a metastatic phenotype (14,15).
V-ATPases are composed of two functional domains known as V 1 and V 0 (1,2). The V 1 domain is located in the cytosol and carries out ATP hydrolysis, whereas the V 0 domain is membrane-bound and responsible for proton translocation across the lipid bilayer. The V 1 domain is a 650-kDa complex composed of eight different subunits (A-H) present in one or more copies. In mammals, the V 0 domain is a 260-kDa complex composed of five different subunits (a, c, cЉ, d, and e) present in single copies with the exception of c, which is present in multiple copies (1,16). Yeast express an additional subunit (cЈ) homologous to the c subunit (17), and mammalian V 0 domains sometimes contain an accessory subunit known as Ac45 (18).
V-ATPase activity is tightly controlled. Regulating V-ATPase activity allows cells to maintain organelles at a specific pH as well as increase or decrease proton flux across particular membranes in response to specific stimuli. One primary mechanism for regulating V-ATPase activity is reversible dissociation, which involves the physical separation or assembly of the V 1 and V 0 domains (2,19). This mechanism is a rapid and effective way to reduce or increase proton pumping activity and has been observed in several different organisms, including mammals, insects, and yeast. In dendritic cells of the immune system, V 1 domains will assemble with V 0 domains in response to maturation signals, allowing the pump to acidify organelles that function in antigen processing (20). Cultured kidney cells respond to increased glucose concentrations by assembling V 1 with V 0 in a manner that requires phosphatidylinositol 3-kinase activity (21). V-ATPase assembly is modulated during development in the tobacco hornworm gut (22), whereas in blowfly salivary gland epithelial cells, cAMP triggers assembly and proton pumping through a mechanism that may involve phosphorylation of the V 1 C subunit by protein kinase A (PKA) (23,24). V-ATPase regulation by reversible dissociation is best characterized in yeast (2). Here, V-ATPases are disassembled in response to glucose depletion and reassembled when glucose (or another rapidly fermented carbon source) is added back to the media (25). This process occurs within minutes, does not require new protein synthesis, and is not affected by disruption of the major signaling pathways known to respond to nutritional changes (26). Interestingly, the glycolytic enzyme aldolase binds V-ATPase subunits and may play a role in stabilizing the assembled pump (27,28). Point mutations in aldolase that disrupt its interaction with the V-ATPase without affecting glycolytic activity result in a disassembled pump, and overexpressing aldolase blocks V-ATPase dissociation upon glucose removal (29). It also appears that assembly and disassembly are independently controlled processes, because intact microtubules are required for disassembly but not reassembly (30), and a heterotrimeric complex known as RAVE functions in assembly but not disassembly (31,32). Although reversible dissociation has been well studied in yeast, it is not known how signals regarding nutritional state are transmitted to the V-ATPase to regulate its assembly.
In an effort to understand how V-ATPase activity is controlled in cells in response to stimuli, we performed a genetic screen in yeast to identify gene products required for downregulating V-ATPase activity through disassembly of the V 1 and V 0 domains. Because assembly or disassembly occurs rapidly and in response to specific stimuli, we hypothesized that a signaling molecule or pathway controls this process. This screen identified the Ras GTPase-activating proteins (GAPs) IRA1 and IRA2 as essential factors for regulating V-ATPase disassembly in yeast in response to glucose depletion. We show that expressing the dominant active RAS2 val19 allele or deleting the PKA regulatory subunit BCY1 results in a similar phenotype, suggesting that PKA mediates the effects of Ras on V-ATPase assembly.

EXPERIMENTAL PROCEDURES
Materials and Antibodies-G418, phenylmethylsulfonyl fluoride, dithiothreitol (DTT), raffinose, and Protein A-Sepharose were purchased from Sigma. Selection medium was purchased from Sunrise Science Products. Concanamycin A was purchased from Fluka Chemical Corp. 9-Amino-6-chloro-2-methoxyacridine was purchased from Invitrogen. Zymolyase 100T was purchased from Seikagaku America, Inc. Dithiobis(succinimidyl)propionate was purchased from Thermo Fisher Scientific, Inc. Aprotinin, leupeptin, and pepstatin were purchased from Roche Applied Science. Restriction enzymes and DNA ligase were purchased from New England Biolabs. Precast polyacrylamide Ready-Gels, Tween 20, SDS, nitrocellulose membranes (0.45-m pore size), and horseradish peroxidase-conjugated goat anti-mouse IgG were purchased from Bio-Rad. The chemiluminescence substrate for horseradish peroxidase was purchased from General Electric, and signal was detected using Eastman Kodak Co. BioMax Light film. Monoclonal antibodies 10D7 (anti-a) and 8B1 (anti-A) were purchased from Invitrogen. Peroxidase-conjugated monoclonal antibody 3F10 (anti-HA) was purchased from Roche Applied Science.
Genetic Screen-YPH500 was mutagenized using the mTn3 (LEU2 lacZ) transposon library (39), a kind gift from Michael Snyder. Yeast were transformed with NotI-digested library DNA, and the resulting mutants were selected on Ϫleucine, 2% raffinose plates and analyzed visually for red color. Approximately 30,000 mutants were screened, and 104 mutants that appeared red were selected, subcloned, and preserved as frozen stocks. To identify the location of the transposon in the selected mutants, yeast were transformed with pRSQ2-URA3, a kind gift from Michael Snyder, and selected on ϪleucineϪuracil plates. Yeast genomic DNA was isolated, digested with EcoRI, ligated, and transformed into Escherichia coli. Recovered plasmids were sequenced using a primer specific for a sequence in the insertion, 5Ј-GCTGCAAGGCGATTAAGTTGG-3Ј.
Glucose-dependent Dissociation-Yeast cells were grown in selection media to an absorbance at 600 nm of 0.75-0.9, and then two or three aliquots of yeast cells equivalent to 10 optical density units were taken for the assay. Yeast were converted to spheroplasts by incubating in Tris/DTT buffer (0.1 M Tris-HCl, pH 9.4, 10 mM DTT) for 10 min at 30°C, rinsing once in Spheroplast buffer (YPD with 0.1 M MES-Tris, pH 7.5, 0.7 M sorbitol, 2 mM DTT), and then incubating in Spheroplast buffer containing 0.3 mg of Zymolyase 100T for 40 min at 30°C. Yeast were then rinsed twice in YPD/0.7 M sorbitol (1% yeast extract, 2% bacto-peptone, 2% glucose, 0.7 M sorbitol) or YEP/1.2 M sorbitol (1% yeast extract, 2% bactopeptone, 1.2 M sorbitol) and then incubated in YPD/0.7 M sorbitol or YEP/1.2 M sorbitol for 30 min. For reassociation experiments, yeast were incubated in YEP/1.2 M sorbitol for 30 min followed by YPD/0.7 M sorbitol for 30 min. Yeast were then lysed in Lysis buffer (1ϫ PBS, pH 7.4, 1% polyoxyethylene 9-lauryl ether, 2 g/ml aprotinin, 5 g/ml leupeptin, 0.7 g/ml pepstatin, 1 mM phenylmethylsulfonyl fluoride), and cell lysate was then incubated for 1 h at 4°C in Lysis buffer containing 25 mM dithiobis(succinimidyl)propionate, precleared with Protein A-Sepharose for 30 min at 4°C, and then incubated with the monoclonal antibody 8B1 (anti-A) at 1:500 overnight at 4°C. Antibodies and bound protein were isolated with Protein A-Sepharose and separated by SDS-PAGE. After transferring proteins to nitrocellulose, Western blotting was performed using monoclonal antibodies 10D7 (anti-a) and 8B1 (anti-A). Antibodies were detected using horseradish peroxidase-conjugated secondary antibodies and chemiluminescent detection reagents.
Isolation of Vacuolar Membrane Vesicles-Vacuolar membrane vesicles were isolated as described in our previous study (40). Briefly, 1 liter of yeast cells was grown overnight at 30°C to an absorbance at 600 nm of 1.2-1.4 in selective media, rinsed once with water, and then converted to spheroplasts by incubating in Tris/DTT buffer for 20 min at 30°C, rinsing once in Spheroplast buffer, and then incubating in Spheroplast buffer containing 2 mg of Zymolyase 100T for 60 -90 min at 30°C. The resulting spheroplasts were osmotically lysed, and vacuolar membranes were isolated by centrifugation on two consecutive Ficoll gradients (12 and 8%) and diluted in transport buffer (15 mM MES-Tris, pH 7, 4.8% glycerol).

Yeast Genetic Screen to Identify V-ATPase Regulators-To
identify factors involved in V-ATPase dissociation in vivo, a genetic screen was performed in yeast to look for mutants that cannot dissociate their V-ATPases in response to glucose depletion. To perform the initial screen, we used the adenine-deficient Saccharomyces cerevisiae strain YPH500, which produces a red pigment derived from an intermediate in adenine biosynthesis. In cells containing a functional V-ATPase, this pigment accumulates in the acidic vacuole, turning the colony red. By contrast, in cells lacking functional V-ATPase, no pigment accumulates in the vacuole, and the colony appears white. Because glucose promotes V-ATPase assembly in yeast, YPH500 colonies normally appear red when plated on glucose. When V-ATPases are triggered to disassemble by plating on poorly fermented carbon sources, such as raffinose, the red pigment does not accumulate, and the colonies appear white (Fig. 1A). We used this property to screen for mutants that could not disassemble their V-ATPases in nutrient-poor conditions, because such mutants would appear red on raffinose plates. To perform the screen, a transposon library (39) was used to mutagenize the YPH500 genome, followed by plating of mutant cells on raffinose and selection of colonies that appeared red. From 30,000 colonies screened, 104 mutant colonies were isolated and confirmed as remaining red on raffinose plates.
Mutants that were positive by this initial screen were then tested using a secondary screen designed to directly measure V-ATPase assembly in response to glucose depletion. The yeast cell wall was removed, and the resulting spheroplasts were cultured in the presence or absence of glucose and then lysed. An antibody directed against the A subunit of the V 1 domain was used to immunoprecipitate both the V 1 domain and the V 1 V 0 holoenzyme (Fig. 1B). The immunoprecipitated proteins were separated by SDS-PAGE, and Western blotting was performed using antibodies directed against subunit a (as a measure of the amount of V 0 immunoprecipitated) and subunit A (as a measure of the amount of V 1 ). Dissociation appears as a reduction in the amount of V 0 immunoprecipitated with V 1 . Mutants that did not show a decrease in assembly of the V-ATPase in response to glucose removal were scored as being defective in in vivo dissociation. Of the 104 mutants isolated in the initial screen, 57 have been tested using the secondary screen. Of these, nine mutants showed defective dissociation in this assay.
IRA2 Is Required for V-ATPase Regulation by Disassembly-One mutant identified twice in the primary screen and confirmed using the secondary screen is disrupted in the IRA2 gene. Ira2p is a Ras GAP that stimulates the intrinsic GTPase activity of Ras proteins to convert them from the active, GTPbound form to the inactive, GDP-bound form (45)(46)(47). In S. cerevisiae, GTP-bound Ras activates adenylate cyclase, leading to increased cAMP levels and PKA activity (48). This pathway is stimulated by glucose and is inactive in nutrient-poor conditions, the same stimuli that affect V-ATPase activity. ira2⌬ mutants display a similar phenotype to mutants with overactive Ras, such as those expressing the dominant active RAS2 val19 allele (47).
To confirm that IRA2 is required for V-ATPase regulation, glucose-dependent dissociation was tested in a yeast strain in which the IRA2 gene had been replaced with the kanamycin resistance gene (34). Although glucose removal triggered normal disassembly of the V 1 domain from the V 0 domain in the parental strain expressing wild type IRA2, the ira2⌬ strain showed a dramatic reduction in V-ATPase regulation by disassembly (Fig. 2). Introduction of a plasmid containing the wild type IRA2 gene into the ira2⌬ stain restored the wild type phenotype. We conclude from these results that IRA2 is essential for V-ATPase disassembly in yeast in response to glucose depletion.
Loss of IRA2 Does Not Affect V-ATPase Stability or Activity-One possible way in which mutations could lead to reduced dissociation of the V-ATPase in response to glucose depletion is to increase the inherent stability of the V-ATPase complex. Certain point mutations in the G subunit of the yeast V-ATPase result in such increased stability and reduced dissociation following glucose removal (36). These G subunit mutations are characterized by increased levels of V 1 subunits on isolated vacuolar membranes (resulting from the increased stability of the complex). To test whether loss of IRA2 leads to a more stable V-ATPase complex, we performed Western blots on vacuolar membranes isolated from wild type, ira2⌬, or rescue strains maintained in the presence of glucose. Increased stability is reflected by more intense V 1 subunit staining on isolated vacuoles. No difference in the levels of V 1 or V 0 subunits was observed among these three strains (Fig. 3A), indicating that stability of the pump is not affected by loss of IRA2.
To verify that increased stability can be measured using this method, we analyzed yeast strains expressing an R25L point mutation in the V-ATPase G subunit. This point mutation has previously been reported to confer increased stability to the pump (36). Consistent with previous results, an increase in stability was observed in the R25L strain relative to wild type as evidenced by an increase in V 1 staining relative to V 0 (Fig. 3B). . Strategy for genetic screen to identify gene products involved in regulating glucose-dependent dissociation of the V-ATPase. The adenine-deficient yeast strain YPH500 was used for the screen. A, red pigment accumulates in the vacuole of these yeast in a manner that depends on V-ATPase activity. When grown on glucose, colonies appear red (F), indicating functional V-ATPase (left). Insertional mutagenesis was performed on YPH500, followed by plating on media containing raffinose, a poorly fermented carbon source known to trigger V-ATPase disassembly. When grown on raffinose, most colonies appeared white (E), indicating low V-ATPase activity due to dissociation of the pump (right). Mutants that appeared red on raffinose-containing media were selected as candidates for defective V-ATPase disassembly and were subject to a secondary screen designed to directly evaluate V-ATPase assembly state. B, a glucose-dependent dissociation assay was used for the secondary screen. Spheroplasts were incubated in the presence or absence of glucose followed by lysis in the presence of the detergent polyoxyethylene 9-lauryl ether to solubilize the V-ATPase. Antibodies that specifically recognize the A subunit of the V 1 domain were used to immunoprecipitate the V 1 domain and the V 1 V 0 holoenzyme. Assembly was measured by performing a Western blot on immunoprecipitated complexes using an antibody that specifically recognizes the a subunit of the V 0 domain. An antibody that recognizes subunit A of the V 1 domain was used as a loading control. Glucose-dependent dissociation is reflected as a decrease in the amount of a subunit immunoprecipitated with the A subunit antibody. Mutants that did not show such a decrease were identified as strains with defective V-ATPase regulation and were selected for further study. Wild type, ira2⌬, and the ira2⌬ yeast strain reexpressing wild type IRA2 on the plasmid pRS316 were tested for glucose-dependent dissociation, as described in the legend to Fig. 1B. The ira2⌬ strain showed similar levels of a subunit immunoprecipitated with the antibody directed against subunit A in the presence or absence of glucose, indicating defective dissociation of the pump in nutrient-limiting conditions. Reexpressing IRA2 in the ira2⌬ strains (rescue) restored the dissociation phenotype to that observed in the wild type strain.
We next tested whether loss of IRA2 leads to altered V-ATPase activity. In some cases, reduced activity can lead to defective dissociation (26), so a reduction in activity is another possible way in which in vivo dissociation could be inhibited. ATPase and proton pumping assays were performed on vacuolar membranes in the presence or absence of the specific V-ATPase inhibitor concanamycin A. No significant difference in concanamycin-sensitive ATPase or proton transport activity was observed between vacuolar membranes isolated from wild type, ira2⌬, or rescue strains (Fig. 4), suggesting that loss of IRA2 does not alter the enzymatic activity of the pump under normal conditions. IRA1 Is Required for V-ATPase Regulation-The IRA2 homolog in yeast, IRA1, also has GAP activity and negatively regulates the Ras/cAMP pathway in response to nutrient limitation (49). Deletion of either IRA1 or IRA2 leads to similar phenotypes, including increased levels of intracellular cAMP, increased sensitivity to heat shock and nitrogen starvation, and sporulation defects (47,49). IRA1 was also identified in the screen for V-ATPase regulators described above. As shown in Fig. 5, yeast with a disruption in IRA1 show defective V-ATPase dissociation upon glucose removal. These results suggest that endogenous levels of either protein cannot compensate for loss of the other.
Constitutively Active Ras2 Blocks V-ATPase Dissociation-Because Ira proteins are negative regulators of Ras, we next tested whether Ras plays a role in controlling V-ATPase assembly. To test the effects of overactive Ras on V-ATPase regulation, we analyzed glucose-dependent dissociation in a strain expressing RAS2 val19 , a dominant active mutant with greatly reduced GTPase activity (48). Interestingly, V-ATPase dissociation was blocked in strains expressing RAS2 val19 (Fig. 6), consistent with the reduced dissociation in the ira deletion strains.
We next wished to determine the effect of reducing Ras activity on V-ATPase dissociation. Ras is encoded by two genes in S.

FIGURE 3. Loss of IRA2 does not affect stability of assembled V-ATPases.
A, vacuolar membranes were isolated from wild type, ira2⌬, or rescue strains grown in the presence of glucose. Under these conditions, isolated membranes should contain maximal levels of assembled complexes, and no free V 1 domain should be present in the sample. Vacuolar membrane protein (175 ng) was separated by SDS-PAGE on 10% acrylamide gels, and Western blots were performed using antibodies that recognize the V 0 a subunit and V 1 A subunit as described under "Experimental Procedures." Increased stability of the assembled pump is reflected as an increase in the amount of A subunit relative to the a subunit. No difference in the relative amounts of A subunit was detected among the three strains, indicating that the phenotype observed in ira2⌬ mutants upon glucose depletion is not due to increased stability of the assembled pump. B, yeast deficient for the V-ATPase G subunit (vma10⌬) were transformed with the plasmid pRS315 containing N-Myctagged wild type G subunit or G subunit with a point mutation known to increase stability of the pump (R25L). Vacuoles were isolated from the resulting strains, and assembly was assayed by comparing relative amounts of subunit A as described in A. V-ATPases with the R25L mutation showed an increase in the amount of subunit A isolated with the vacuolar membranes, indicating increased stability. . V-ATPases from ira2⌬ yeast strains show wild type levels of enzymatic activity. Vacuolar membranes were isolated from wild type, ira2⌬, or rescue strains grown in the presence of glucose to ensure that membranes contained assembled complexes. Concanamycin-sensitive ATP hydrolysis and proton pumping were measured as described under "Experimental Procedures." The specific ATPase activity of vacuolar membranes from the wild type strain was 0.69 mol of ATP/min/mg of protein. For each strain, the assays were performed twice each for four independent vacuole preparations. The error bars indicate the S.E. between vacuole preparations. No significant difference was observed among the three strains, indicating that loss of IRA2 does not lead to an increase in the activity of assembled V-ATPase complexes. FIGURE 5. The IRA2 homolog IRA1 is required for V-ATPase disassembly upon glucose removal. A glucose-dependent dissociation assay was performed on strains containing wild type IRA1 or IRA1 with a transposon-mediated insertion as described in the legend to Fig. 1. Levels of subunit a do not decrease upon glucose removal in yeast with a disrupted IRA1 gene, indicating defective glucose-dependent V-ATPase dissociation.
cerevisiae, RAS1 and RAS2. Strains lacking either RAS1 or RAS2 are viable, but deletion of both isoforms in the same strain is lethal (48). When we tested glucose-dependent dissociation in yeast deficient for either RAS1 or RAS2, we found that V-ATPases dissociated normally in these strains (Fig. 6). Adding glucose back to the media led to reassembly of the pump in wild type, ras1⌬, and ras2⌬ strains but had no effect on yeast expressing RAS2 val19 (Fig. 6). We conclude from these results that Ras plays a role in regulating V-ATPase disassembly in yeast in response to glucose removal and that Ras1 and Ras2 can complement each other in this function.
PKA Regulates V-ATPase Assembly-In yeast, Ras stimulates adenylate cyclase, leading to increasing cAMP levels (48). Strains with activated Ras (through deletion of Ira proteins or expression of RAS2 val19 ) display elevated levels of cAMP and are phenotypically similar to strains with activated PKA. Therefore, we sought to determine whether PKA regulates V-ATPases by testing the effects of increased PKA activity on glucosedependent dissociation. PKA is a tetramer composed of two catalytic subunits and two regulatory subunits (48). cAMP activates PKA by binding the regulatory subunits, allowing the catalytic subunits to phosphorylate downstream substrates. Yeast with a deletion in the gene that encodes the regulatory subunits, BCY1, have constitutively active PKA that is no longer cAMPdependent (50). When we tested for glucose-dependent V-ATPase dissociation in a bcy1⌬ strain, we found that V-ATPases remain assembled upon glucose removal (Fig. 7). These data indicate that activated PKA causes V-ATPases to assemble and/or blocks V-ATPase disassembly, leading to increased V-ATPase activity. In addition, these data show that PKA mediates the effects of overactive Ras signaling (whether through decreased GAP activity or expression of dominant active Ras2) on V-ATPase assembly.
Constitutively Active Ras2 Affects the Aldolase/V-ATPase Interaction-To begin to explore the mechanisms by which PKA promotes V-ATPase assembly, we determined the effects of increased Ras signaling on the aldolase/V-ATPase interaction. Previous studies show that the glycolytic enzyme aldolase interacts with the V-ATPase and suggest that this interaction stabilizes the pump in its assembled state (27)(28)(29). Glucose removal disrupts the aldolase/V-ATPase interaction. We hypothesized that overactive Ras signaling would block the dissociation of aldolase upon glucose removal. Yeast expressing HA-tagged aldolase were transformed with RAS2 val19 , and the amount of aldolase immunoprecipitated with V 1 was measured. The amount of aldolase associated with the V-ATPase decreased upon glucose removal in the HA-aldolase strain, confirming previous reports (Fig. 8A) (28). Interestingly, the expression of RAS2 val19 resulted in decreased levels of aldolase bound to the V-ATPase even in the presence of glucose (Fig. 8A). Furthermore, no decrease in aldolase was observed upon glucose removal. The amount of bound aldolase in RAS2 val19 -expressing strains in either the presence or absence of glucose was about the same as in wild type strains in the absence of glucose. As before, V-ATPase assembly decreased upon glucose removal in the HA-aldolase strain but remained unchanged with the expression of RAS2 val19 (Fig. 8B). Thus, Ras signaling leads to increased V-ATPase assembly but decreased aldolase binding.

DISCUSSION
Regulation of V-ATPase activity by reversible dissociation is well characterized in yeast, but the signals that transmit information regarding nutritional status to the V-ATPase are unknown. By screening for mutants defective in V-ATPase disassembly upon glucose removal, we have identified the glucosesensitive Ras/cAMP/PKA pathway as playing an essential role in this process. We showed that V-ATPases remain assembled in the absence of glucose when PKA is stimulated through several methods, including reduction of Ras-GAP activity, expression of constitutively active Ras2, and deletion of the PKA regulatory subunit. This represents the first identification of signaling molecules involved in regulation of V-ATPase activity in yeast and marks a significant advance in our understanding of the factors that control V-ATPase assembly and activity in response to nutrient depletion.
PKA has been shown to play a role in V-ATPase assembly in other model systems. Exposure of insect epithelial cells to FIGURE 6. Effects of Ras mutations on reversible dissociation. Wild type, ras1⌬, ras2⌬, and wild type yeast expressing RAS2 val19 on the plasmid pRS415 were tested for glucose-dependent dissociation as described in the legend to Fig. 1B. Reassembly of the pump was tested by incubating the cells in the absence of glucose, followed by the presence of glucose, before lysing the cells for immunoprecipitation (indicated as Ϫ/ϩ). Reversible dissociation occurred normally in ras1⌬ and ras2⌬ strains but was blocked in yeast expressing RAS2 val19 . cAMP results in massive translocation of V 1 domains to the apical membrane and proton pumping across the cell surface (23,51). This effect is blocked by PKA inhibitors. In addition, the catalytic (but not regulatory) subunit of PKA translocates to the apical membrane when these cells are treated with the cAMP analog 8-CPT-cAMP (52). Interestingly, phosphorylation of the V-ATPase C subunit by PKA also occurs concomitant with these translocation events (24). Although the C subunit is not phosphorylated in the assembled holoenzyme, it is possible that transient phosphorylation of subunit C by PKA may serve as a signal for assembly in these cells. It will be important to identify the downstream PKA substrate in yeast cells and characterize how phosphorylation by PKA leads to V-ATPase assembly. Preliminary experiments in which the yeast V-ATPase was immunoprecipitated with antibodies against subunit A followed by immunoblot with an antibody directed against a phosphopeptide corresponding to the PKA recognition site have detected no difference in the presence or absence of glucose (data not shown). Because V-ATPase dissociation in yeast occurs only on specific membranes (i.e. vacuolar but not Golgi (53)), it will also be of interest to determine whether activated PKA localizes to specific membranes in yeast.
V-ATPases in dendritic cells of the immune system assemble and acidify digestive organelles in response to maturation signals (20), but the proteins that transmit such signals to the V-ATPase have not yet been identified. cAMP and PKA are known to play roles in the process of dendritic cell maturation (54,55), but whether they mediate V-ATPase assembly during this process is not known.
PKA can regulate V-ATPase activity through mechanisms that do not involve reversible dissociation but rather trafficking of fully assembled enzymes. cAMP-activated PKA stimulates V-ATPase trafficking in rat epididymal cells to increase V-ATPase activity at the plasma membrane, where proton extrusion by the V-ATPase aids in sperm maturation and storage (56). Treatment of macrophages with interleukin-1 leads to increased proton pumping by V-ATPases across the plasma membrane to assist in cytosolic pH regulation (57). PKA is involved in this process, but whether V-ATPase activity increases through trafficking or assembly is not known.
Previous studies have explored whether cAMP mediates the effects of glucose on V-ATPase assembly in yeast (26). The addition of exogenous cAMP to yeast deficient for PDE2, the primary phosphodiesterase that cleaves cAMP, was reported to have no effect on V-ATPase assembly. It is possible that the level of activation of PKA obtained by the addition of exogenous cAMP was not as high as that achieved in the present studies by disruption of IRA1 or IRA2, expression of RAS2 val19 , or deletion of the PKA regulatory subunit. Quantitation of the levels of activated PKA obtained under the previous experimental conditions may help to resolve this discrepancy.
Disruption of either IRA1 or IRA2 resulted in the same phenotype with respect to V-ATPase dissociation, suggesting that that endogenous levels of either protein are insufficient to mediate V-ATPase dissociation upon glucose removal. This observation is common for other ira⌬ phenotypes and may be due to the fact that deletion of either gene results in elevated Ras-GTP and cAMP levels (46,47,49,58).
Yeast harboring certain point mutations in the V-ATPase G subunit (V 1 ) display the same regulation defect as those with elevated Ras signaling (36). However, these G subunit mutations also lead to an increase in the inherent stability and activity of the pump, whereas the enzymatic properties of V-ATPases in ira2⌬ mutants are normal. The defects in V-ATPase disassembly in ira2⌬ mutants are therefore most likely due to dysfunctional regulation of the pump rather than a change in the enzyme itself.
The glycolytic enzyme aldolase binds assembled V-ATPase complexes and appears to play a role in stabilizing the pump (27,28). Yeast overexpressing aldolase exhibit the same phenotype as those with constitutively active PKA, with defective disassembly of the pump upon glucose depletion (29). The Ras/ cAMP/PKA pathway is known to mediate the response of yeast to glucose through activation of glycolytic enzymes (48). Therefore, it is not surprising that overactive Ras affected the aldolase/V-ATPase interaction, although the underlying mechanisms are unclear. Aldolase has one putative PKA recognition site, but there are no reports of PKA-dependent phosphorylation of the enzyme. As noted above, preliminary attempts to detect a change in phosphorylated proteins immunoprecipitated with an antibody against the V-ATPase A subunit showed FIGURE 8. Activated Ras2 affects the aldolase/V-ATPase interaction. A, yeast expressing HA-tagged aldolase and RAS2 val19 on the plasmid pRS316 were converted to spheroplasts, incubated in the presence or absence of glucose and then lysed. V-ATPases were immunoprecipitated using an antibody that recognizes the A subunit, and V-ATPase-associated aldolase was detected by performing a Western blot on immunoprecipitated proteins using an anti-HA antibody, with the anti-A antibody as a loading control. Glucose removal led to a decrease in associated aldolase in the wild type strain, consistent with previous reports. Expression of RAS2 val19 led to a decrease in associated aldolase in the presence of glucose, with no change in the amount of bound aldolase upon glucose removal. B, yeast expressing HA-tagged aldolase were assayed for glucosedependent V-ATPase dissociation, as described in the legend to Fig. 1B. Consistent with results shown in Fig. 6 for a different yeast strain, dissociation was normal in the HA-aldolase strain but was blocked upon expression of RAS2 val19 . no difference in the presence or absence of glucose. Other effectors may therefore be involved in transmitting the signals from PKA to the V-ATPase to affect assembly. Furthermore, in RAS2 val19 -expressing cells, V-ATPase assembly increased relative to wild type, whereas the levels of bound aldolase decreased, implying that Ras2 may promote V-ATPase assembly through a mechanism that does not depend on aldolase binding. Whether aldolase performs a V-ATPase stabilization role in mammalian cells has not been tested, although it is interesting to note that patients with certain aldolase mutations have the same clinical outcome as those expressing mutant forms of the renal-specific V-ATPase, namely renal tubular acidosis (59).
In conclusion, we have shown that regulation of V-ATPase assembly and activity in yeast in response to glucose is controlled by the Ras/cAMP/PKA pathway. This represents the first identification of signaling molecules responsible for controlling V-ATPase activity in yeast and allows for further work aimed at testing whether PKA regulates V-ATPase assembly in other systems and characterizing the mechanisms by which PKA accomplishes this. V-ATPases are key participants in a variety of normal and disease processes in cells, and a better understanding of the mechanisms that regulate V-ATPase activity in response to stimuli can aid in the development of effective and specific therapeutics.