The Amino-terminal Domain of the Vacuolar Proton-translocating ATPase a Subunit Controls Targeting and in Vivo Dissociation, and the Carboxyl-terminal Domain Affects Coupling of Proton Transport and ATP Hydrolysis*

The 100-kDa “a” subunit of the vacuolar proton-trans-locating ATPase (V-ATPase) is encoded by two genes in yeast, VPH1 and STV1 . The Vph1p-containing complex localizes to the vacuole, whereas the Stv1p-containing complex resides in some other intracellular compartment, suggesting that the a subunit contains information necessary for the correct targeting of the V-ATPase. We show that Stv1p localizes to a late Golgi compartment at steady state and cycles continuously via a prevacuolar endosome back to the Golgi. V-ATPase complexes containing Vph1p and Stv1p also differ in their assembly properties, coupling of proton transport to ATP hydrolysis, and dissociation in response to glucose depletion. To identify the regions of the a subunit that specify these different properties, chimeras were constructed containing the cytosolic amino-terminal domain of one isoform and the integral membrane, carbox-yl-terminal domain from the other isoform. Like the Stv1p-containing complex, the V-ATPase complex containing the chimera with the amino-terminal domain of Stv1p localized to the Golgi and the complex did not dissociate in response to glucose depletion. Like the

The V-ATPases 1 are a family of ATP-dependent proton pumps responsible for acidification of intracellular compartments in eukaryotic cells (1)(2)(3)(4)(5)(6)(7)(8). Acidification of these compartments is crucial for such processes as receptor-mediated endocytosis, intracellular trafficking, the processing and degradation of macromolecules, and the coupled transport of small molecules. In addition, V-ATPases in the plasma membrane of specialized cells function in such processes as pH homeostasis (9), bone resorption (10), renal acidification (11), potassium transport (12), and tumor metastasis (13). In yeast, the V-ATPase functions to create the driving force for uptake of small molecules and ions into the vacuole (14) and is important for post-Golgi protein trafficking (15)(16)(17).
The 100-kDa "a" subunit of the V 0 domain has a bipartite structure containing a hydrophilic amino-terminal domain of ϳ50 kDa and a hydrophobic carboxyl-terminal domain containing multiple transmembrane helices (27)(28)(29). Topological studies employing cysteine mutagenesis and chemical labeling have led to a model for the a subunit containing nine transmembrane helices, with the amino-terminal domain on the cytoplasmic side of the membrane and the carboxyl terminus on the luminal side of the membrane (30), although data suggesting a cytoplasmic orientation of the carboxyl terminus have also been reported (31). Mutagenesis studies have identified several buried charged residues in the last two transmembrane helices that appear to play some role in proton transport (32,33), suggesting that the 100-kDa subunit functions analogously to subunit a of the F-ATPases. It is the interaction between the F-ATPase a subunit and the ring of proteolipid c subunits that is thought to facilitate proton movement through the F 0 domain (19,34,35).
In yeast the a subunit is encoded by two genes (VPH1 and STV1), which show ϳ50% identity at the amino acid level (28,29). Vph1p has been shown to associate with V-ATPases targeted to the vacuole, whereas Stv1p is present in V-ATPase complexes localized to some other intracellular compartment (29). The identity of the signal responsible for this differential localization is not known, but the role of the a subunit in targeting is supported by data from mammalian cells. Thus, in mouse, the a subunit is also encoded by multiple genes that are expressed in a tissue-specific manner (36,37). The a3 isoform has been shown to localize to the plasma membrane in osteoclasts (37), whereas the a4 isoform localizes to the plasma membrane of renal intercalated cells (38).
In addition to differences in localization, the a subunit isoforms in yeast have also been shown to differ in their ability to assemble with the V 1 domain, in the coupling of proton transport and ATP hydrolysis and in their ability to undergo in vivo dissociation in response to glucose depletion (39). This latter process has been shown to play an important role in regulation of V-ATPase activity in yeast cells (40) and has been suggested to function in regulation in higher eukaryotes as well (41,42). To identify the regions of the a subunit responsible for these differences, chimeric proteins have been constructed containing the amino-terminal domain of one isoform and the carboxyl-terminal domain of the other. Our results suggest that intracellular targeting and in vivo dissociation of the V 1 domain are controlled by signals residing within the amino-terminal domain. In contrast, the coupling of ATP hydrolysis to proton translocation is controlled by the carboxyl-terminal domain.

EXPERIMENTAL PROCEDURES
Materials-Zymolyase 100T was obtained from Seikagaku America (Ijamsville, MD). Concanamycin A was purchased from Fluka Chemical Corp. (Milwaukee, WI). Protease inhibitors were from Roche Molecular Biochemicals. The monoclonal antibodies against the HA epitope, 3F10 and 12CA5, were also from Roche Molecular Biochemicals. Monoclonal antibodies against ALP (1D3-A10), Pep12p (24 -2C3G4), the V-ATPase A subunit (8B1-F3), the B subunit (13D11-B2), and the Alexa (A594)conjugated goat anti-rabbit antibody were from Molecular Probes (Eugene, OR). The affinity-purified polyclonal antibody against the HA epitope has been described previously (17). The purified anti-HA monoclonal antibody used for immunofluorescence (HA.11) was purchased from Covance Research Products (Richmond, CA). Biotin-conjugated goat anti-mouse and streptavidin-conjugated fluorescein isothiocyanate (FITC) antibodies were obtained from Jackson ImmunoResearch (West Grove, PA). Escherichia coli and yeast culture media were purchased from Difco. Restriction endonucleases, T4 DNA ligase, and other molecular biology reagents were from New England Biolabs (Beverly, MA), Life Technologies Inc., or Promega (Madison, WI). Oligonucleotides were synthesized by Keystone Laboratories (Camarillo, CA). ATP, phenylmethylsulfonyl fluoride, and most other chemicals were purchased from Sigma.
Plasmid Construction-The plasmids used in this study are shown in Table I. pDJ48 was generated by the insertion of a triple HA epitope tag sequence into the unique BglII site in pMM1 (position 675 of the STV1 ORF). STV1-HA was removed from pDJ48 following digestion with KpnI and SacI and subcloned into the same sites of pRS306 to produce pDJ74. pDJ48 was digested with HpaI and EcoRI, and the ends were filled in using Vent DNA polymerase, removing base pairs 163-2341 of the STV1 ORF. This was ligated with a 1-kilobase pair SalI-EcoRV fragment from pFA6-kanMX2 containing the kanamycin resistance gene, which had been filled in as above (43), generating pKEB11.
Plasmid pSKN11 (STV1-2xHA in pRS316) and pSKN12 (VPH1-2xHA in pRS316) (39) were used to generate constructs expressing Vph1p under the control of the STV1 promoter (pSKN13), Vph1p and Stv1p chimeras (pSKN23, pSKN24), and the Stv1p-(⌬165-208) deletion (pSKN34). pSKN13 was generated by recombinant polymerase chain reaction using pSKN11 or pSKN12 as a template. pSKN23 and pSKN24 were generated by swapping the region encoding the amino-terminal and carboxyl-terminal domains of each gene at a unique restriction enzyme site, MluI, which was generated by site-directed mutagenesis at the position of Leu 455 in STV1-2xHA and at Leu 409 of VPH1-2xHA. The plasmids pSKN14 and pSKN25 were generated by digestion of pSKN11 and pSKN23 with XbaI and BamHI followed by subcloning of the insert into the 2-m plasmid YEp352. pSKN34 was generated by deletion of the region corresponding to Pro 165 -Pro 208 of Vph1p.
Yeast Strains-Yeast strains used in this study are shown in Table I. Strains were constructed by standard genetic techniques and grown in  MAT␣ ura3-52 leu2-3,112 his4-519 ade6 gal2 pep4-3  66  KEBY2  MAT␣ ura3-52 leu2-3,112 his4-519 ade6 gal2 pep4-3  STV1ϻ3XHA   This study   KEBY4  MAT␣ ura3-52 leu2-3,112 his4-519 ade6 gal2 pep4-3  . YEPD buffered to pH 5.0, 5.5, or 7.5 and YEPD with 100 mM CaCl 2 were also used for the selection and growth of Vma Ϫ strains (16). Strains KEBY2, KEBY4, and KEBY9 were derived from SF838-1D, and all other strains were from SF838Ϫ9D. Strain KEBY4 was generated by transforming KEBY2 with SalI-and NotI-cut pKEB11. Kanamycin-resistant colonies were selected on YEPD with 200 g/ml geneticin sulfate (G418). Insertion of the Kan r gene into the STV1 ORF was confirmed by polymerase chain reaction from genomic DNA using oligonucleotides complementary to the STV1 sequence and to a sequence within Kan r . Loss of Stv1p-HA was assessed by Western blot using the 12CA5 monoclonal antibody. KEBY4 (stv1⌬::Kan r ) was then transformed with ApaI/BamHI-digested pBJ6714, and Leu ϩ colonies were selected for lack of growth on YEPD with 100 mM CaCl 2 , producing KEBY9. Strain KEBY44 was created by transformation of NBY72 with EcoRI-digested pDJ74. Colonies were selected on medium lacking uracil and were then plated on 5-fluoroorotic acid-containing minimal medium to select for Ura Ϫ loop outs. The presence of Stv1p-HA was confirmed by Western blot. Strain KEBY45 was derived from KEBY44 by transformation with BamHI-and PstI-digested pKJH2. Colonies selected on medium lacking leucine were screened for deletion of VPS27 by colony overlay assay to detect CPY secretion (45) and by immunofluorescence with anti-Pep12p to screen for class E Vps Ϫ vacuolar morphology (46). Immunofluorescence Microscopy-Indirect immunofluorescence microscopy was performed as described previously (45). Cells were grown in YEPD or YEPD, pH 5.0, at 30°C to 1 A 600 /ml, fixed by the addition of 3% formaldehyde, incubated for 1 h, and then followed by a 16-h incubation at room temperature in 4% paraformaldehyde, 50 mM KPO 4 , pH 6.5. Cells were converted to spheroplasts using 250 g/ml Zymolyase 100T and permeabilized by treatment with 5% SDS for 5 min or 1% SDS for 2 min, as described in the figure legends. Cells were allowed to adhere to poly-L-lysine-coated multiwell slides. Nonspecific antibody binding was blocked by incubation of the cells in phosphate-buffered saline with 5 mg/ml bovine serum albumin and 1% normal goat serum. All antibodies were diluted in phosphate-buffered saline with 5 mg/ml bovine serum albumin, which was also used for all washes. Antibodies against ALP, Pep12p, and HA were pre-absorbed to yeast proteins (to remove nonspecific binding) by incubation with pho8⌬,pep12⌬ or cells without an HA-tagged protein, respectively. Antibody incubations were performed at room temperature as follows: 2 h for primary antibodies (except 4 h for anti-ALP) and 1 h for secondary and tertiary antibodies. The affinity-purified polyclonal antibody against the HA epitope was used at a dilution of 1/200 and the anti-HA monoclonal antibody (HA.11) at 1/500. Monoclonal antibodies against ALP and Pep12p were used at dilutions of 1/3 for anti-ALP tissue culture supernatant and 1/1000 for purified anti-Pep12p. Images were generated using a Bio-Rad MRC 1024 confocal microscope.

SF838-1D
Analysis of 100-kDa a Subunit Expression and V-ATPase Assembly-Yeast cells transformed with pSKN11 (Stv1p-HA), pSKN13 (Vph1p-HA), pSKN23 (Stv1-Vph1p-HA), or pSKN24 (Vph1-Stv1p-HA) were grown to log phase at 30°C in selective medium; whole cell lysates were prepared as described previously (39), and the proteins were separated by SDS-PAGE on 8% acrylamide gels. The expression of the 100-kDa subunits was detected by Western blotting using the horseradish peroxidase-conjugated monoclonal antibody 3F10 against HA. To evaluate assembly of the V-ATPase, the complex was detergent-solubilized from whole cell lysates and immunoprecipitated using the monoclonal antibody 8B1-F3 against subunit A of the V 1 domain, as described previously (39). Following separation of proteins by SDS-PAGE on 8% acrylamide gels, Western blotting was performed using antibodies against both subunit A and the HA epitope. Blots were developed using a chemiluminescent detection method obtained from Kirkegaard & Perry Laboratories (Gaithersburg, MD). The assembly competence of the chimeric proteins was reflected by their ability to be immunoprecipitated using the anti-A subunit antibody.

Detection of V-ATPase Subunits Present on Isolated Vacuolar
Membranes-Vacuolar membranes isolated from each of the mutants were subjected to SDS-PAGE, and Western blots were probed with the mouse monoclonal antibodies 3F10 against HA, 8B1-F3 against the A subunit (Vma1p), and 13D11-B2 against the B subunit (Vma2p). Blots were also probed with rabbit polyclonal antibodies against Vma6p, Vma7p, Vma8p, Vma10p, and Vma13p, as well as Vma4p (a generous gift of Dr. Daniel J. Klionsky), and a mouse monoclonal antibody against Vma5p (a generous gift of Dr. Patricia M. Kane). Following removal of unbound FIG. 1. Stv1p-HA colocalizes with the late Golgi marker A-ALP but shows little overlap with the PVC marker Pep12p. KEBY44 cells (pho8⌬ STV1::HA) were transformed with the A-ALP plasmid pSN55. Immunofluorescence was performed as described under "Experimental Procedures," and cells were permeabilized with 5% SDS for 5 min (upper panels) or 2% SDS for 1 min (lower panels). Stv1p-HA was visualized using a rabbit polyclonal antibody against the HA epitope followed by a secondary Alexa anti-rabbit antibody. The same cells were also stained for A-ALP or Pep12p using monoclonal antibodies against ALP and Pep12p, respectively, followed by anti-mouse biotin and streptavidin-FITC. Confocal micrographs were taken simultaneously of the red and green fluorescence channels, with an 8% bleed through correction, and overlapped to produce the merged image. primary antibodies by washing, blots were incubated with horseradish peroxidase-conjugated secondary antibodies (Bio-Rad) and developed using chemiluminescence as above.
In Vivo Dissociation of the V-ATPase in Response to Glucose Deprivation-Dissociation of the V-ATPase in response to glucose depletion was measured as described previously (39). The vph1⌬ stv1⌬ strain KEBY9 expressing Vph1p-HA, Stv1p-HA, Vph1-Stv1p-HA, or Stv1-Vph1p-HA from the single copy plasmid pRS316 was grown in selective medium overnight to an absorbance at 600 nm of Ͻ1.0. The cells were converted to spheroplasts by treatment with Zymolase 100T and incubated in YEP media with or without 2% glucose for 40 min at 30°C. Spheroplasts were pelleted and lysed in phosphate-buffered saline containing 1% C 12 E 9 , protease inhibitors, and 1 mM dithiobis(succinimidyl propionate). An aliquot (corresponding to ϳ6 ϫ 10 5 cells) was removed to allow analysis of proteins present in the whole cell lysate. The V-ATPase complexes were immunoprecipitated from the remainder of the lysate (corresponding to ϳ3 ϫ 10 6 cells) using 8B1-F3 against the A subunit and protein G-agarose followed by separation on 8% acrylamide gels and transfer to nitrocellulose. Western blotting was then performed separately using the horseradish peroxidase-conjugated monoclonal antibodies 3F10 against HA to detect the V 0 domain or antibody 8B1-F3 against the A subunit to detect the V 1 domain followed by a horseradish peroxidase-conjugated secondary antibody. Dissociation of the V-ATPase complex is reflected as a reduction in the amount of the 100-kDa subunit immunoprecipitated using the antibody directed against subunit A (located in the V 1 domain). Blots were developed as described above.
Other Procedures-ATPase activities were measured using a coupled spectrophotometric assay with the modification of using 0.35 mM NADH instead of 0.5 mM NADH (39). ATP-dependent proton transport was measured by fluorescence quenching using the fluorescence probe ACMA in transport buffer (25 mM Mes-Tris, pH 7.2, 5 mM MgCl 2 , and 1 mM ATP) as described previously (39) in the presence or absence of 1 M concanamycin A, a specific inhibitor of the V-ATPase (47). SDS-polyacrylamide gel electrophoresis was carried out as described by Laemmli (48).

RESULTS
Stv1p Is Localized at Steady State to the Late Golgi Compartment-The two isoforms of the yeast 100-kDa subunit of the V-ATPase are localized to distinct intracellular compartments. A previous report (29) showed that although Vph1p is localized to the vacuolar membrane at steady state, Stv1p is localized to punctate structures that are non-vacuolar. We sought to identify the intracellular compartment to which Stv1p is targeted by colocalization with known marker proteins. We constructed a version of STV1 with a triple HA epitope tag sequence in the same position as described previously (29). STV1-HA was then integrated into the yeast genome, thus creating a strain (KEBY44) in which the sole genomic copy of STV1 is epitopetagged. This tagged version of Stv1p is fully functional as measured by complementation of growth of a vph1⌬ stv1⌬ strain on 100 mM CaCl 2 . 2 To compare the localization of Stv1p-HA with the late Golgi marker protein A-ALP, we used a yeast strain in which the gene encoding wild type alkaline phosphatase (ALP/Pho8p) had been deleted. A-ALP is a chi-  were transformed with empty pRS316 plasmid (vector), with pRS316 expressing an HA-tagged version of Vph1p from the VPH1 promoter, or with plasmids expressing HA-tagged versions Vph1p or Stv1p from the STV1 promoter. Whole cell lysates were prepared, and proteins were separated by SDS-PAGE on 8% acrylamide gels. The lysate from 5 ϫ 10 6 cells was loaded in each lane. Western blotting was then performed using the anti-HA antibody as described under "Experimental Procedures." b, KEBY9 cells (vph1⌬ stv1⌬) were transformed with empty pRS316 plasmid (vector) or with pRS316 expressing HA-tagged versions of Vph1p, Stv1-Vph1p, or Vph1-Stv1p, all from the STV1 promoter. Whole cell lysates were prepared, and proteins were analyzed as described in a. meric protein with the luminal and transmembrane domains of ALP fused to the cytoplasmic domain of dipeptidyl aminopeptidase A (DPAP A, also known as Ste13p). This protein can be detected using a monoclonal antibody raised against the luminal domain of ALP and has been shown to reside in the late Golgi at steady state (49,50). As shown in Fig. 1, Stv1p-HA showed significant overlap with A-ALP by indirect immunofluorescence visualized with confocal microscopy. Nearly all the structures stained for Stv1p-HA also stained for A-ALP. Stv1p-HA also showed colocalization with another late Golgi marker, Sec7p-GFPx3 (51). 3 In contrast, Pep12p, a marker of the prevacuolar compartment (PVC, 52,53), shows weak colocalization with Stv1p-HA, with only a small percentage of structures staining for both proteins (Fig. 1, lower panels). These data suggest that Stv1p is predominantly localized to the late Golgi compartment at steady state.
Many yeast late Golgi membrane proteins such as DPAP A, Kex2p (proteases that processes ␣-factor), and Vps10p/Pep1p (the receptor for carboxypeptidase Y) cycle via the PVC and are retrieved from this compartment back to the late Golgi (49, 50, 54 -58). Thus, these proteins are constantly cycling between the late Golgi and PVC, with the majority concentrated in the late Golgi in a wild type cell. To determine whether Stv1p also cycles via the PVC, we deleted VPS27 in the KEBY44 cells used in the previous experiment. VPS27 is a class E VPS gene and is thus required for the exit of proteins from the PVC (46,59). Cells lacking Vps27p display a distinct morphological phenotype, with a large aberrant PVC (the "class E compartment") typically adjacent to the vacuole. Late Golgi proteins that cycle continuously via the PVC, such as Vps10p, Kex2p, and A-ALP as well as the PVC marker Pep12p and proteins en route to the vacuole such as Vph1p, are concentrated in the class E compartment in vps27⌬ cells (17,59). As shown in Fig. 2, Stv1p-HA colocalized with both A-ALP and Pep12p in the class E compartment in vps27⌬ cells (large arrows), suggesting that it does cycle via the PVC and back to the late Golgi. Similar to A-ALP, but in contrast to Pep12p, Stv1p-HA also exhibited additional punctate staining in the vps27⌬ cells, which is similar to the pattern seen for Stv1p-HA in wild type cells (Fig. 1). This result is consistent with some proportion of Stv1p-HA remaining in a late Golgi compartment. The population of Stv1p-HA that does not accumulate in the class E compartment partially overlaps with A-ALP (small arrows, Fig. 2). Our results show that Stv1p-HA is predominantly localized to the late Golgi compartment but cycles via the PVC and back to the late Golgi in a manner similar to many other late Golgi membrane proteins.
Construction and Expression of Vph1p/Stv1p Chimeras-V-ATPase complexes containing Vph1p and Stv1p differ both in intracellular localization (see above and Ref. 29) and in other properties, including activity, coupling efficiency, and in vivo dissociation in response to glucose depletion (39). To determine which regions of the a subunit are responsible for these differences, chimeras were constructed containing the hydrophilic amino-terminal domain of one isoform and the hydrophobic carboxyl-terminal domain of the other isoform. Fig. 3 shows a schematic illustration of the HA-tagged Vph1p, Stv1p, the chimeric proteins, and a deletion construct of Stv1p in which amino acid residues 165-208 were removed. Chimera Vph1-Stv1p-HA contained residues 1-409 of Vph1p-HA and residues 456 -890 of Stv1p, whereas chimera Stv1-Vph1p-HA contained residues 1-455 of Stv1p-HA and residues 410 -840 of Vph1p (the difference in residue number for the junction point between Vph1p and Stv1p is due to a 46-residue insertion in the amino-terminal domain of Stv1p). The Vph1-Stv1p-HA chi-mera contained a tandem HA tag after residue Asn 185 , whereas the Stv1-Vph1p-HA chimera contained a tandem HA tag after residue Leu 227 . These HA tags, located in the amino-terminal domain, were shown previously not to interfere with the stability or assembly of Vph1p and Stv1p, respectively (39). Both chimeras were expressed using the STV1 promoter, thus ensuring more uniform levels of expression of the chimeric proteins. Fig. 4 shows the expression levels of the Vph1-Stv1p-HA and Stv1-Vph1p-HA chimeras, Vph1p-HA and Stv1p-HA using the STV1 promoter, and Vph1p-HA using the VPH1 promoter. 4 As can be seen, both chimeras were expressed at levels comparable with Vph1p-HA driven from the STV1 promoter (Fig. 4b), which in turn was lower than either Vph1p-HA or Stv1p-HA driven from their own promoters (Fig. 4a). The lower level of Vph1p-HA expressed from the STV1 promoter than from the VPH1 promoter likely reflects the higher transcriptional efficiency of the latter promoter (29). The fact that Stv1p-HA is nevertheless expressed at nearly the same steady state level as Vph1p-HA, when each is driven from its own promoter (Fig.  4a), may be due to a greater stability of Stv1p-HA relative to Vph1p-HA. The presence of two bands differing by 5-6 kDa for the Stv1-Vph1p-HA chimera (Fig. 4b) may reflect a partial proteolysis of this chimera. Nevertheless, each chimeric gene (STV1-VPH1-HA and VPH1-STV1-HA) is capable of complementing the growth defect of the vph1⌬ stv1⌬ strain, indicating that the V-ATPase complexes containing these chimeric proteins possess a substantial level of activity.
The Amino Termini of Vph1p and Stv1p Contain the Information Necessary for Their Intracellular Localization-To investigate the localization of Vph1p-HA expressed from the STV1 promoter, and the Stv1-Vph1p-HA and Vph1-Stv1p-HA chimeras, we performed immunofluorescence using an anti-HA antibody. As shown in Fig. 5, Vph1p-HA expressed under the control of the STV1 promoter was properly localized to the vacuole membrane. This shows that the differential localization of Vph1p-HA and Stv1p-HA was not due to differences in the expression level of the two proteins but rather was likely due to distinct targeting information within the proteins themselves. As shown with Stv1p-HA expressed from integrated STV1::HA (Fig. 1), Stv1p-HA expressed from the CEN plasmid showed typical Golgi staining (Fig. 5). However, a small amount of Stv1p-HA was seen on the vacuole membrane in cells expressing Stv1p-HA from the low copy, centromere-based plasmid. This is consistent with previous results showing that overexpression of Stv1p-HA causes some vacuolar localization (29,39), probably due to saturation of the Golgi retention or retrieval machinery.
As shown in Fig. 5, the chimera with the amino terminus of Vph1p-HA and the transmembrane carboxyl-terminal region of Stv1p (Vph1-Stv1p-HA) was localized to the vacuole mem-3 K. Bowers and T. H. Stevens, unpublished results. 4 It should be noted that the Vph1-Stv1p chimera migrates faster on SDS-PAGE than Vph1p (Fig. 4), despite the fact that they contain (including the HA epitope tags) 862 and 858 amino acid residues, respectively (Fig. 3). This aberrant migration appears to be due to differences in amino acid composition of the amino-and carboxyl-terminal domains of Vph1p and Stv1p such that the chimeras do not migrate as expected based upon the number of amino acid residues. This can also be seen by comparing the migration of Stv1p and the Stv1-Vph1p chimera in Figs. 6 -8. Thus, although Stv1p and the Stv1-Vph1p chimera have 908 and 904 amino acid residues, respectively, and would therefore be predicted to comigrate on SDS-PAGE, the Stv1-Vph1p chimera migrates more slowly than Stv1p. It thus appears as though the amino-terminal domain of Vph1p causes the chimera to migrate faster than expected, whereas the amino-terminal domain of Stv1p causes the chimera to migrate more slowly, although it is also possible that the observed differences are due to the carboxyl-terminal domains.
brane. The corresponding chimera with the amino-terminal region of Stv1p-HA and the carboxyl-terminal region of Vph1p localized predominantly to the Golgi. This suggests that the information for the differential localization of Vph1p and Stv1p resides within the amino-terminal, cytoplasmic domains of the proteins.
Golgi Localization of Stv1p Is Not Determined by an FXFXD Motif-Interestingly, the amino acid sequence of Stv1p has an FXFXD motif within the amino terminus of the protein that is not conserved in the sequence of Vph1p (Fig. 6a). The FXFXD motif is a previously characterized sequence required for the Golgi retrieval of Ste13p/DPAP A and A-ALP (50). An in-frame deletion of amino acids 165-208 of Stv1p (Stv1p-(⌬165-208)-HA) removes the FXFXD motif (FSFDD) as well as additional amino acids that are not conserved in Vph1p by sequence alignment. As can be seen from Fig. 6b, the Stv1p-(⌬165-208)-HA protein was expressed at approximately the same level as Vph1p-HA expressed from the STV1 promoter in whole cell lysates. Stv1p-(⌬165-208)-HA assembles into a functional V-ATPase complex, because expression of this protein suppresses the growth defects of a vph1⌬ stv1⌬ yeast strain. Furthermore, as shown in Fig. 6c, the Stv1p-(⌬165-208)-HA staining pattern was indistinguishable from that of Stv1p-HA (i.e. it localized to the late Golgi). Mutation of FXFXD to AXAXA by site-directed mutagenesis also had no effect on the steady state localization of Stv1p. 2 These data show that deletion of residues 165-208 of Stv1p, including the FXFXD motif, does not grossly affect the steady state localization of Stv1p.
Vph1p/Stv1p Chimeras Are Competent for Assembly into V-ATPase Complexes-As an initial measure of the assembly competence of the chimeric proteins, detergent-solubilized extracts of whole cell lysates were prepared from cells expressing the chimeric proteins, and immunoprecipitation was performed using the monoclonal antibody 8B1-F3 against the A subunit (Vma1p/Tfp1p) of the V 1 domain. Following separation of the immunoprecipitated proteins by SDS-PAGE, Western blotting was performed using antibodies against both subunit A and the HA epitope tags introduced into the chimeric a subunits. The degree of assembly of the V-ATPase is indicated by the amount of subunit a immunoprecipitated using the antibody against subunit A. As can be seen from Fig. 7, the Vph1-Stv1p-HA chimera showed somewhat lower assembly than the Stv1-Vph1p-HA chimera, which in turn showed almost the same level of assembly as Stv1p-HA itself. These data indicate that both chimeric forms of the a subunit are competent to assemble with the other subunits of the V-ATPase complex.
As a further measure of the assembly competence of the chimeric proteins, vacuoles were isolated from the strains expressing each of the a subunit constructs, and Western blotting was performed using antibodies against subunits A-H of the V 1 domain as well as subunits a and d of the V 0 domain. It has been shown previously that disruption of assembly of the V-ATPase complex results in the disappearance from the vacuolar membrane of multiple V-ATPase subunits (60). To ensure the presence of Stv1p-HA and the Stv1-Vph1p-HA chimera in the vacuolar membrane, these two proteins were overexpressed using the 2-m plasmid YEp352 as described previously (39). As can be seen in Fig. 8, the chimeras show reduced assembly relative to the native 100-kDa subunits, with the Vph1-Stv1p-HA chimera showing lower assembly than the Stv1-Vph1p-HA chimera. Stv1p-HA also showed reduced assembly relative to Vph1p-HA, consistent with our previous observations (39). Although vacuoles from cells expressing the Stv1-Vph1p-HA chimera showed lower levels of V 1 subunits than vacuoles from cells expressing Stv1p-HA, this does not appear to reflect reduced assembly of the chimera because the chimeric protein itself is also present at reduced levels in the vacuole. It is also notable that subunit B (Vma2p) association with the vacuole appears to be less affected by the switch in 100-kDa subunits than other V 1 subunits. Nevertheless, the data in Fig.  8 confirm that both chimeric forms of the a subunit are competent to assemble with the remaining V-ATPase subunits, although assembly is reduced relative to the native a subunit isoforms.
Vph1p/Stv1p Chimeras Differ in the Coupling Efficiency of Proton Transport and ATP Hydrolysis-We have shown previously that V-ATPase complexes containing Stv1p-HA have a 4 -5-fold lower ratio of coupling of proton transport to ATP hydrolysis than complexes containing Vph1p-HA (39). Table II shows proton transport (as measured by ACMA quenching) and concanamycin-sensitive ATPase activities measured for vacuoles isolated from strains expressing Vph1p-HA or the Vph1-Stv1p-HA chimera in pRS316 or Stv1p-HA or the Stv1-Vph1p-HA chimera in YEp352. As can be seen, vacuoles isolated from the strain expressing the Stv1-Vph1p-HA chimera are more active for both proton transport and ATPase activity than vacuoles isolated from the strain expressing the Vph1-Stv1p-HA chimera, consistent with the lower degree of assembly observed for the Vph1-Stv1p-HA chimera (Figs. 7 and  8). Comparison of the ratio of proton transport to ATP hydrolysis (Table II, 3rd column) reveals that the 3-4-fold difference in coupling ratio between Vph1p-HA and Stv1p-HA is also detected in comparison with the Stv1-Vph1p-HA and Vph1-Stv1p-HA chimeras. Thus, the tighter coupling of Vph1p-HA relative to Stv1p-HA appears to be due to differences in the carboxyl-terminal domain. It should be noted that because of the low proton transport activity of complexes containing the Vph1-Stv1p-HA chimera relative to the error of these measurements, there is considerable uncertainty in the calculation of the coupling ratio. Nevertheless, even the highest ratio calculated from the extreme values of the fluorescence quenching and ATPase activities permissible from the error bars for the Vph1-Stv1p-HA chimera (0.45/0.02 ϭ 22.5) is lower than the ratio observed for the Stv1-Vph1p-HA chimera.
The Amino-terminal Domain of the a Subunit Controls in FIG. 5. Immunolocalization of Vph1p/Stv1p chimeras. KEBY9 cells (vph1⌬ stv1⌬) were transformed with pRS316 expressing HA tagged versions of Vph1p, Stv1p, Stv1-Vph1p, or Vph1-Stv1p, all expressed from the STV1 promoter (plasmids pSKN13, pSKN11, pSKN23, and pSKN24, respectively). Immunofluorescence was performed as described under "Experimental Procedures," and cells were permeabilized with 2% SDS for 1 min. The tagged proteins were visualized using a monoclonal anti-HA antibody followed by anti-mouse biotin and streptavidin-FITC. Each panel represents a confocal micrograph of representative cells.
Vivo Dissociation of the V-ATPase in Response to Glucose Depletion-We have shown previously that V-ATPase complexes containing Vph1p-HA that are localized to the vacuole show in vivo dissociation in response to glucose depletion, whereas complexes containing Stv1p-HA that are localized to its normal compartment do not (39). Stv1p-containing complexes that have been targeted to the vacuole through overexpression, on the other hand, do show dissociation in response to glucose withdrawal. These results suggest that in vivo dissociation of the V-ATPase is primarily controlled through the membrane environment in which the V-ATPase resides. However, we have shown previously (39) that V-ATPase complexes containing Vph1p that have been blocked in transport to the vacuole by disruption of the VPS21 or the VPS27 gene still show glucosedependent dissociation, although to a somewhat reduced degree. Thus, V-ATPase complexes present in compartments other than the central vacuole can still undergo dissociation in response to glucose withdrawal.
To begin to address whether the signals that control dissociation of the V-ATPase are distinct from those that control targeting of the complex, the in vivo dissociation behavior of complexes containing the two chimeric forms of the a subunit were compared. Spheroplasts were incubated in media with or without glucose for 40 min followed by detergent solubilization and immunoprecipitation using the anti-A subunit antibody 8B1-F3. Western blotting was then performed using antibodies against both subunit A and the HA epitope tag. Dissociation of the V-ATPase is reflected as a decrease in the amount of subunit a immunoprecipitated using the antibody against subunit A. Consistent with our previous observations (39), V-ATPase complexes containing wild type Vph1p-HA showed dissociation in response to glucose withdrawal (Fig. 9b), whereas those containing Stv1p-HA did not (Fig. 9a). Interestingly, the Vph1p-Stv1p-HA chimera, like Vph1p-HA, also exhibited glucose-dependent dissociation, whereas the Stv1p-Vph1p-HA chimera (like Stv1p-HA), did not (Fig. 9a). Thus, the dissociation  pSKN13 and pSKN34, respectively). Levels of the proteins in whole cell lysates (5 ϫ 10 6 cells per lane) were determined by Western blot using the anti-HA antibody as described in Fig. 4. c, cells were transformed with pRS316 expressing HA-tagged versions of Stv1p or Stv1p-(⌬165-208) (plasmids pSKN11 and pSKN34), and immunofluorescence was performed as described in Fig. 5. behavior of the chimeras correlates with the source of the amino-terminal domain.

Targeting Information for the Yeast V-ATPase Complexes Is
Contained within the Amino-terminal, Cytosolic Domains of the Subunit a Isoforms-Differential targeting of subunit a isoforms of the V-ATPase may provide a means of localizing V-ATPase complexes to different subcellular sites. In this study, we show that the V-ATPase complex containing the 100-kDa subunit a isoform Stv1p resides in the late Golgi at steady state, where it colocalizes with the chimeric protein A-ALP (see Fig. 1). Amino acid sequence comparison has shown that Stv1p is 54% identical and 71% similar to the yeast vacuolar a subunit isoform Vph1p (29). Given this significant amount of se-quence similarity, these proteins provide interesting tools for the study of targeting signals in yeast and for the analysis of different subunit a isoforms. The greatest sequence divergence is within the amino-terminal domains of the proteins, before the first putative transmembrane domain, suggesting that information for their differential targeting might be contained within this domain. In fact, constructing chimeric proteins in which the amino-terminal domains of Vph1p and Stv1p have been exchanged does indeed change the localization of these proteins (see Fig. 5) but still results in assembled and functional V-ATPase complexes (Figs. 7 and 8). This demonstrates that the localization signals are contained within the aminoterminal, cytosolic domains of the a subunits. Many cytosolic domains of membrane proteins have been shown to contain targeting signals, often short amino acid sequences required for interaction with the cytosolic sorting machinery (for review see Ref. 61). Because previous evidence (54,62) suggests that targeting to the vacuole in yeast may require no specific signals, it seems more likely that the Stv1p isoform contains Golgi retention or retrieval motifs within its amino-terminal domain.
Although Stv1p-HA is localized to the late Golgi at steady state, it accumulates in the large, aberrant PVC of vps27⌬ cells (Fig. 2). These cells have defects in the retrieval of proteins of the PVC and back to the Golgi, as well as protein transport out of the PVC to the vacuole (46,59). In addition to proteins en FIG. 7. Analysis of assembly competence of chimeric proteins by immunoprecipitation and Western blot. Whole cell lysates were prepared from KEBY9 (vph1⌬ stv1⌬) cells expressing HA-tagged versions Vph1p, Stv1p, or the Stv1-Vph1p or Vph1-Stv1p chimeras using the plasmids pSKN13, pSKN11, pSKN23, and pSKN24, respectively. Lysates were detergent-solubilized, and V-ATPase complexes were immunoprecipitated using the monoclonal antibody 8B1-F3 against the A subunit (Vma1p) as described under "Experimental Procedures." Following separation of the proteins by SDS-PAGE on 8% acrylamide gels, Western blot analysis was performed using antibodies against both HA and subunit A as described.  a ATPase activities were measured on isolated vacuolar membranes (5 g of membrane protein) prepared as described under "Experimental Procedures" from cells transformed with HA-tagged versions of Vph1p or Vph1-Stv1p in pRS316 or Stv1p or Stv1-Vph1p in YEp352. Activities were measured in the presence of 1 mM ATP and the absence or presence of 1 M concanamycin A. The results shown represent the concanamycin-sensitive portion of the activity.
b ATP-dependent proton transport activities were estimated from the initial rate of ATP-dependent fluorescence quenching in the presence of 1 mM ATP using the fluorescence dye ACMA in the absence or presence of 1 M concanamycin A.
FIG. 9. In vivo dissociation of V-ATPase complexes in response to glucose depletion in cells expressing Vph1p, Stv1p, or the chimeric proteins. a, spheroplasts were prepared from KEBY9 (vph1⌬ stv1⌬) cells expressing HA-tagged versions of Vph1p, Stv1p, the Stv1-Vph1p chimera, or the Vph1-Stv1p chimera (all expressed from the STV1 promoter using pRS316). The spheroplasts were incubated in YEP media in the absence or presence of 2% glucose (as indicated) for 40 min followed by detergent solubilization and immunoprecipitation using the monoclonal antibody 8B1-F3 against the A subunit (Vma1p) as described under "Experimental Procedures." Following separation of the proteins by SDS-PAGE on 8% acrylamide gels, Western blot analysis was performed using antibodies against both HA and subunit A as described. b, a shorter exposure of the gel shown in a to allow the glucose-dependent decrease in Vph1p immunoprecipitated with the anti-A subunit antibody to be more readily visualized.
route to the vacuole, vps27⌬ cells concentrate late Golgi proteins that cycle via the PVC, such as Vps10p and A-ALP (49,59). Our data suggest that Stv1p, like Vps10p and A-ALP, exits the late Golgi, travels to the PVC, and is retrieved, resulting in steady state Golgi localization. Specific signals for the Golgi retrieval of Vps10p, A-ALP, and Kex2p have been identified and are contained within the cytoplasmic domains of these membrane proteins (50,56,57,54). Each of these signals contains an aromatic amino acid that is important to maintain the Golgi localization of the protein. Interestingly, the aromatic sorting motif found in A-ALP, FXFXD, is also found within the amino-terminal domain of Stv1p but not in Vph1p (see Fig. 6). However, deletion of this motif from Stv1p did not significantly alter its steady state Golgi localization (Fig. 6). This suggests that some other signal is required for the retrieval of Stv1p to the Golgi or that there are redundant signals. There is evidence that proteins are retained in the late Golgi, in addition to being retrieved from the PVC, as a means of Golgi localization. Wild type A-ALP and Kex2p, for example, show slowed exit from the Golgi, compared with an A-ALP mutant lacking residues 2-11 (49) and a mutant Kex2p truncated at residue 772, respectively (63). This suggested that the regions deleted in these mutants contained retention signals for A-ALP and Kex2p. However, the amino terminus of Stv1p does not particularly resemble these retention sequences, suggesting that Stv1p has a novel targeting signal.
Comparison of the amino acid sequences of Stv1p and Vph1p reveals a stretch of amino acids in Stv1p that are not conserved in Vph1p and thus may contain targeting information (Stv1p residues 148 -197, see Fig. 6). The Stv1p-(⌬165-208)-HA mutant protein has most of this region removed, and our results suggest that the Golgi localization signal is not contained in these residues (see Fig. 6). Residues 148 -164 of Stv1p may therefore contain targeting information. We analyzed the importance of this region in targeting by constructing truncated Stv1 proteins. However, Stv1p-HA lacking residues 148 -197 or residues 129 -164 of Stv1p could not be detected by Western blot and failed to complement the growth phenotypes of a vph1⌬ stv1⌬ yeast strain. 5 Thus large deletions of this region of Stv1p result in unstable protein. In addition to this region of Stv1p that does not align with Vph1p, there are many other residues in the Stv1p cytosolic domain that are not similar to residues in the Vph1p sequence. The Stv1p amino-terminal domain is predicted to be 450 amino acids long (29), and there are, for example, 16 aromatic amino acids in Stv1p that do not correspond to aromatic amino acids in Vph1p by sequence alignment. None of these aromatic amino acids, however, matches known retrieval motifs except the FXFXD motif discussed above.
The Hydrophobic, Carboxyl-terminal Domains of Vph1p and Stv1p Control V-ATPase Assembly and the Ratio of ATP Hydrolysis to Proton Translocation-In addition to showing distinct patterns of intracellular localization, V-ATPase complexes containing Vph1p and Stv1p have also been shown to differ in assembly, coupling, and in vivo regulation (39). Thus, Vph1p-containing complexes show ϳ10-fold greater assembly with the V 1 domain than do Stv1p-containing complexes, giving rise to an ϳ10-fold higher V max of ATPase activity for the former class (39). In addition, Vph1p-containing complexes show a 5-fold higher ratio of proton transport to ATP hydrolysis than Stv1p-containing complexes (39), suggesting a tighter coupling of proton transport and ATP hydrolysis with Vph1p. Although both the Vph1-Stv1p-HA and Stv1p-Vph1p-HA chimeras show reduced assembly with the V 1 domain relative to that observed for Vph1p-HA, nevertheless, the Stv1-Vph1p-HA chimera shows better assembly than does the Vph1-Stv1p-HA chimera (Fig. 8), suggesting that the steady state difference in assembly between Vph1p and Stv1p may be attributable to differences in the carboxyl-terminal domain. Similarly, complexes containing the Stv1-Vph1p-HA chimera show a 3-fold higher ratio of proton transport to ATP hydrolysis than complexes containing the Vph1-Stv1p-HA chimera (Table II), suggesting that the tightness of coupling of proton transport and ATP hydrolysis is also affected by the carboxyl-terminal domain of the protein. These results are consistent with our previous identification of mutations in the carboxyl-terminal domain that affect both assembly and activity of the V-ATPase (32,33). It should be noted, however, that because of the reduced assembly of both chimeric forms of the a subunit (Fig. 8), the observed differences in coupling efficiency may be due to the loss of one or more V 1 subunits from the corresponding complex.
In Vivo Dissociation of the V-ATPase in Response to Glucose Deprivation Is Controlled by the Amino-terminal Domains of Vph1p and Stv1p-By contrast with the assembly and coupling properties of the V-ATPases, the in vivo dissociation, like the targeting information, appears to reside in the amino-terminal domain of the a subunit. The Kane laboratory has shown (40) that dissociation of the V-ATPase into separate V 1 and V 0 domains represents an important mechanism for regulating the yeast V-ATPase in vivo, and a similar mechanism has been proposed to control V-ATPase activity in insects during molting (41). We have shown previously that whereas Vph1p-containing complexes present in their normal intracellular site (the vacuole) undergo dissociation in response to glucose depletion, Stv1p-containing complexes present in the Golgi do not (39). When Stv1p-containing complexes are forced (through overexpression of Stv1p) to localize to the vacuole, in vivo dissociation is now observed (39). These results suggest that the cellular localization of the V-ATPase is the primary determinant in controlling dissociation in response to glucose depletion. We would then predict that chimeras localized to the vacuole should show glucose-dependent dissociation, whereas those that localized to the Golgi would not. In fact, this prediction is supported by the observation that complexes containing the Vph1-Stv1p-HA chimera dissociated in response to glucose depletion, whereas those containing the Stv1-Vph1p-HA chimera did not (Fig. 9). These results also indicate that there is no signal controlling in vivo dissociation of the V-ATPase located in the carboxyl-terminal domain of the a subunit that overrides the targeting information located in the amino terminus.
The amino-terminal domain of the a subunit has been shown previously to interact with both subunit A and subunit H of the V 1 domain (64) and has been suggested to form part of the peripheral stator connecting V 1 and V 0 (64,65). Based upon these results, it might have been predicted that the aminoterminal domain would play a more important role in controlling interactions between the V 1 and V 0 domains, including assembly, coupling, and regulation of dissociation. The fact that only the latter process appears to be dependent upon signals in the amino-terminal domain suggests that the a subunit may also make contact with the V 1 domain through the carboxyl-terminal region. Alternatively, conformational differences between the carboxyl-terminal regions may be transmitted to the rest of the molecule through the amino terminus of subunit a. It is interesting, in this regard, that a subunit mutations affecting assembly of the V-ATPase appear to cluster on the luminal side of the membrane (30, 32, 33), suggesting a luminal domain that may regulate assembly of the complex. With respect to the tightness of coupling of proton transport and ATP hydrolysis, the dependence on the hydrophobic carboxyl-terminal domain is easier to understand because this coupling is likely to be critically dependent upon the interaction of the a subunit with the proteolipid subunits, as has been shown to be true for the F-ATPases (19,34,35). Additional work will be required to identify residues within each of the a subunit domains that directly participate in the multitude of functions served by this subunit.