Characterization of yeast V-ATPase mutants lacking Vph1p or Stv1p and the effect on endocytosis

Vacuolar ATPases are multisubunit complexes, found in all eukaryotic cells,which are responsible for the acidification of intracellular compartments. These compartments include endosomes, lysosomes, Golgi membranes,clathrin-coated vesicles, several types of secretory granules and the central vacuoles of plants and yeast (Nelson and Harvey, 1999). Each of these intracellular compartments requires a specific internal pH that is generated by the V-ATPase function(Grabe and Oster, 2001). In a single cell, V-ATPases can be involved in a variety of essential cellular functions, including receptormediated endocytosis, post-translational modification, protein sorting along the secretory pathway, protein degradation and secondary transport (Stevens and Forgac, 1997). V-ATPase energizes the vacuolar system by driving the translocation of protons across the vacuolar membrane into the lumen at the expense of ATP hydrolysis.

In common with the evolutionarily related F-ATPase, whose primary function in eukaryotic cells is to form ATP, V-ATPase is composed of two functional domains. The peripheral V₁ domain, which is responsible for ATP hydrolysis, contains at least eight subunits, denoted as subunits A through H(Stevens and Forgac, 1997; Nelson and Harvey, 1999). The membrane V_o domain consists of at least five subunits (a, c,c′, c″ and d) and functions in proton translocation(Powell et al., 2000; Stevens and Forgac, 1997). In most of the eukaryotic organisms studied, a null mutation in the V-ATPase gene results in a lethal phenotype. In contrast, yeast cells lacking V-ATPase activity due to the same null mutation can survive, but only in a buffered medium at pH 5.5 (Nelson and Nelson,1990). These mutations exhibit a conditionally lethal phenotype in which cells cannot grow at a pH higher than 7.0 or various other medium conditions (Nelson and Nelson,1990). This feature makes yeast one of the favorite organisms for V-ATPase studies. Furthermore, the yeast genome has one gene encoding each subunit of the V-ATPase, except for VPH1 and STV1, which encode two isoforms of the 100 kDa a subunit (Vph1p and Stv1p). Disruption of both subunits is necessary to produce the null V-ATPase mutant phenotype(Manolson et al., 1992, 1994), which creates a unique opportunity to study the role of isoforms in the V-ATPase subunits. In other organisms, many of the V-ATPase subunit isoforms were demonstrated to be tissue-specific (Futai et al.,2000), yet some were found in specific locations in the same cells(Toyomura et al., 2000). In yeast, Vph1p was assigned to the vacuole, and a Golgi and/or endosome localization was proposed for Stv1p(Manolson et al., 1994). These observations suggest that subunit a may be responsible for the V-ATPase localization in specific subcellular compartments and/or regulation of enzymatic activity, which implies that there should be specific and distinct phenotypes for both mutants. The vph1Δ mutant is unable to accumulate Quinacrine in the vacuole, and therefore appears to be defective in vacuolar acidification, but no phenotype has been described for absence of Stv1p, and the stv1Δ mutant was reported to be identical to the wild type (Manolson et al.,1994).

Properties of V-ATPase complexes containing both disrupted Vph1 and Stv1 genes (double-deletion mutant) were recently studied with overexpression of each gene cloned into a yeast shuttle vector(Kawasaki-Nishi et al., 2001). It was demonstrated that a complex containing Stv1p showed lower assembly with the catalytic subunits and a lower ratio of proton transport to ATP hydrolysis than the V-ATPase complex containing Vph1p.

In this paper, we reinvestigate the subcellular locations of the Vph1p and Stv1p and also define the characteristics of the separate vph1Δand stv1Δ phenotypes in comparison to the wild type and double-deletion mutant. We used a new pH-sensitive LysoSensor Green DND-189 dye to determine the extent of acidification within the intracellular compartments in both wild type and null mutants for each of the two isoforms of subunit a. The V-ATPase was suggested to be directly involved in endocytosis and membrane fusion (Wendland et al., 1998; Ungermann et al., 1999; D'Hondt et al.,2000; Peters et al.,2001). Using the endocytic marker FM 4-64 we show that the endocytic process is inhibited whenever there is a decrease in the cellular V-ATPase activity.

The `wild type' strain used is a haploid S. cerevisiae W303(MATa/α trp1 ade2 his3 leu2 ura3). The other haploid strains used in this work are: vma3Δ (MATa ade2 trp1 ura3 his3 VMA3::LEU2); stv1Δ (MATa ade2 ura3 his3 leu2 STV1::TRP1); vph1Δ/stv1Δ (MATa ade2 trp1 ura3 his3 VPH1::LEU2 STV1::URA3); vph1Δ(MATα ade2 ura3 his3 leu2 trp1 VPH1::LEU2); vma8Δ (MATα ade2 ura3 his3 leu2 trp1 VMA8::LEU2); vtc1Δ (MATa ade2 trp1 ura3 his3 leu2 VTC1::URA3). The cells were grown in a YPD medium containing 1 % yeast extract, 2 % bactopeptone and 2 % dextrose. The medium was buffered as previously described (Noumi et al.,1991).

All the yeast strains used in this work were prepared as previously described (vma3Δ, Nelson and Nelson, 1990; vma8Δ and vtc1Δ, Cohen et al., 1999). Constructs for disruption of the VPH1 and STV1 genes were prepared as follows. Each gene was amplified from yeast genomic DNA by polymerase chain reaction (PCR) with specific primers, cloned into pGEM-T easy or pBluescript SK, respectively. For VPH1, a fragment of 610 base pairs (bp) was deleted by digestion with HincII and EcoRV, and a blunt-end LEU2 marker was inserted in its place. For STV1 (which was cloned into the KpnI-BamHI sites of pBluescript) a 2360 bp fragment was deleted by digestion with EcoR1 and XhoI,filled in and a blunt-end TRYP1 or URA3 marker inserted. The constructs were excised using appropriate enzymes at either end and used for yeast W303 or vph1Δ (haploid strains) transformation. Homologous recombination and null V-ATPase phenotypes were checked as previously described (Cohen et al.,1999).

Polyclonal antibody against Stv1p was obtained by injecting rabbits with a chimeric protein containing the maltose-binding protein and hydrophilic sequence of amino acid residues 231-436 of Stv1p. The DNA fragment encoding these amino acids was amplified by PCR with introduced EcoRI and HindIII restriction sites. The amplified DNA fragment was cloned in-frame to the maltose-binding protein in the plasmid PMAL-C (New England Biolab). Following sequence verification, 500 ml of bacterial culture was grown to A₆₀₀=0.5, induced with isopropyl-β-D-thiogalactoside (IPTG) for 3 h and harvested by centrifugation at 4000g. The cells were broken using a French press and the protein purified on a column containing maltose—agarose. The fractions containing the chimeric protein were dissociated by SDS, loaded on a preparative gel and electrophoresed. The gel was briefly stained by Coomassie Blue, the identified protein band cut out and the fusion protein was electroeluted. About 0.25 mg fusion protein was injected into rabbits as previously described (Nelson,1983). Polyclonal antibody against Sec14p was prepared as a chimeric protein containing the maltose-binding protein and the whole open-reading frame sequence of the Sec14p was prepared using a procedure similar to that described for the Stv1p antibody. Monoclonal antibodies that recognize Vph1p (10D7-A7-B2), CPY (10A5-B5) and Pep12p (2C3G4) were from Molecular Probes, Inc. Polyclonal antibody against the purified Pma1p was raised in rabbits and used at a dilution of 1:10,000. Antibody against Sed5p was a generous gift from Dr Randy Schekman (University of Berkeley, USA). Polyclonal antibodies against Vma8p and Vma5p were raised in guinea pigs and used at a dilution of 1:1000. The antibody detection system (ECL) was from Amersham. Western blots were performed as previously described(Cohen et al., 1999).

Whole-cell extracts from the strains were prepared as described by Lyons and Nelson (1984) with several modifications. Briefly, yeast cells (20 ml) were grown to late logarithmic phase, centrifuged and washed once with water. Cells were resuspended in 0.45 ml water. Sodium hydroxide and 2-mercaptoethanol were added to a final concentration of 0.2 mol l^-1 and 0.5 %, respectively. The suspension was placed on ice for 20 min, then 0.125 ml of dissociation buffer(2 % SDS, 80 mmol l^-1 Tris-HCl, pH 6.8, 1 % 2-mercaptoethanol, 10 %glycerol and 0.01 % Bromophenol Blue) was added and adjusted immediately with HCl to pH 7.5-8. The proteins were extracted for 1 h at room temperature and insoluble material was removed by centrifugation in an Eppendorf centrifuge(18,000 g) at room temperature for 10 min.

For membrane preparation, yeast cells were grown in 500 ml YPD medium (pH 5.5) to A₆₀₀=1.0. The suspension was centrifuged at 3000 g for 5 min and the pellet was washed once with water and again with 1 mol l^-1 sorbitol. The cell wall was digested by 2.5 U zymolyase in 10ml solution containing 10 mmol l^-1 Hepes, pH 7.5, and 1 mol l^-1 sorbitol. After 30 min incubation at 30 °C, the suspension was centrifuged in 15 ml Corex tubes at 3000 g for 5 min. Glass beads(1 ml) were added to the pellet, together with 1 ml of a solution containing 30 mmol l^-1 Mops, pH 7.0, 10 μl protease inhibitor cocktail(Sigma), 1 mmol l^-1 phenylmethylsulfonyl fluoride (PMSF), 1 mmol l^-1 EDTA and 1 mmol l^-1 EGTA. The suspension was vortexed five times for 30 s each with incubation on ice for 30 s inbetween. The solution was removed from the glass beads and placed in a fresh Corex tube for centrifugation at 1000 g for 5 min, forming a pellet containing the cell debris and nuclei. The supernatant was centrifuged at 115,000 g for 30 min and the pellet was suspended in 0.3-0.5 ml of solution containing 10 mmol l^-1 Tris-Cl, pH 7.5, 1 mmol l^-1 EDTA,2 mmol l^-1 dithiothreitol (DTT), 25 % glycerol, and stored as the membrane fraction at -80 °C. Sucrose gradients were also used to estimate the relative density of various membrane fractions. The gradients (20 %-60 %sucrose) were made as described by Lupashin et al.(1997), and were centrifuged in a Beckman SW-40 rotor for 14h at 150,000 g.

Differential centrifugation analysis and sucrose gradient fractionation of Golgi membranes were performed as previously described(Graham et al., 1994; Graham and Krasnov, 1995). Briefly, spheroplasts (300 A₆₀₀ units) obtained by zymolyase treatment (0.125 mg ml^-1 in 50 mmol l^-1Tris-HCl, pH 7.5, 1.4 mol l^-1 sorbitol, 40 mmol l^-12-mercaptoethanol) were lysed by sevenfold dilution in hypoosmotic buffer (0.1 mol l^-1 sorbitol, 10 mmol l^-1 TEA, pH 7.5, 1 mmol l^-1 EDTA) followed by Dounce homogenisation (25 strokes). The lysate was centrifuged at 1000 g for 6 min to generate P1 (pellet)and S1 (supernatant) fractions, and the latter was centrifuged at 13,000 g to generate P13 and S13 fractions. The S13 fraction was layered onto a two-step sucrose cushion consisting of 1 ml of 66 % sucrose and 1 ml of 20 % sucrose, then centrifuged in a Beckman SW-40 rotor at 120,000 gfor 2h at 4 °C. The membranes present at the 20 %-66 % sucrose interface(P120) were collected in a volume of approx. 1.2 ml and adjusted to approx. 48% sucrose using a 66 % sucrose solution. The membrane sample (1.5 ml) was layered on top of a 60 % sucrose cushion (0.5 ml) and a sucrose step gradient consisting of 47.5 % (1.5 ml), 45 % (1.0 ml), 42 % (2.0 ml), 40 % (2.0 ml), 38% (1.0 ml), 36 % (1.0 ml) and 32 % (1.5 ml) sucrose was layered on top of the sample. All sucrose solutions were prepared in 10 mmol l^-1Na-Hepes, pH 7.5. The gradients were centrifuged in an SW-40 rotor at 31,000 g for 17 h at 4 °C. Fifteen fractions (approx. 0.7 ml) were collected, starting from the bottom of the gradient. The proteins from each fraction were precipitated by addition of 0.13 ml of 5× SDS-sample buffer and 1.0 ml of ice-cold ethanol. After incubation on ice for 30 min, the samples were centrifuged at 18,000 g for 30 min at 4 °C. After aspiration of the supernatants, the pellets were resuspended in 0.05 ml of SDS-sample buffer and stored at -20 °C.

For preparation of vacuoles, cells were grown in YPD medium adjusted to pH 5.5 with HCl and harvested at a cell density of about 0.8 A₆₀₀ units. Vacuolar membranes were prepared according to the method of Uchida et al.(1985), except that the 8 %Ficoll gradient purification step was omitted, the homogenization buffer contained no magnesium and the vacuoles were washed only once with the EDTA buffer (10 mmol l^-1 Tris-HCl, pH 7.5, 1 mmol l^-1 EDTA, 2 mmol l^-1 DTT). ATP-dependent proton uptake activity was assayed by following the absorbency changes of Acridine Orange at 490-540 nm as previously described (Supek et al.,1994). The reaction mixture (1 ml) contained 20 mmol l^-1 Mops-Tris, pH7, 15 mmol l^-1 KCl, 135 mmol l^-1 NaCl and 15 μmol l^-1 Acridine Orange. Isolated yeast vacuoles (10-30 μg) were added to the reaction mixture followed by 10μl of 0.1 mol l^-1 MgATP. The reaction was terminated by the addition of 1 μl of 1 mmol l^-1 carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone (FCCP).

For FM4-64 vacuolar staining, cells were grown in YPD to A₆₀₀=0.8-1.6. Cells (20-40 A₆₀₀ units ml^-1 in YPD medium) were incubated on ice for 30 min with 30μmol l^-1 FM4-64 dye (Molecular Probes Inc.), washed once with YPD and incubated for 60 min for steady-state experiments as described(Vida and Emr, 1995).

For Quinacrine staining, yeast cells were grown in YPD to A₆₀₀=0.8. The cells were cooled on ice for 5 min and 1 ml of cell suspension was sedimented by centrifugation and resuspended in 100μl YPD containing 100 mmol l^-1 Hepes, pH 7.6, and 200 μmol l^-1 freshly prepared Quinacrine. The suspension was incubated for 5-10 min at 30 °C and cooled on ice for 5 min. The cells were sedimented by centrifugation and resuspended in 1 ml 100 mmol l^-1 Hepes, pH 7.6, 2 % glucose. The cells were washed twice with the same cold buffer and resuspended in 0.1 ml of the same solution. 4 μl of the cell suspension was mixed on the microscope slide with 4 μl of 1 % low-melting agarose kept at about 45 °C and covered with a glass cover. The cells were examined using a Zeiss LSM510 confocal laser microscope. Fluorescence profiles were generated using a Zeiss LSM Image browser.

For LysoSensor Green DND-189 staining (Molecular Probes Inc), all strains were grown at 30 °C to A₆₀₀=0.7-1. Cells were harvested and washed with uptake buffer (YPD containing 100 mmol l^-1 Hepes, pH 7.6). The cell pellets were resuspended at a concentration of 15 A₆₀₀ units ml^-1 in uptake buffer and dye was added to a final concentration of 4 μmol l^-1from a stock solution of 1 mmol l^-1 in DMSO. Cells were then incubated for 5 min at 30 °C and washed once with the same buffer. Cell pellets were resuspended in fresh YPD, pH 7.6, at 15 A₆₀₀units ml^-1, placed on standard slides with low-melting agarose and photographed immediately after staining. The samples were viewed using a confocal laser scanning microscope (Zeiss) equipped with an Argon 458 nm laser and a C-Apochromat 63× water immersion objective. A LP505 filter was used for LysoSensor Green DND- 189 fluorescence. The images were recorded,merged and processed using the Zeiss LSM Image browser.

To analyze the intracellular localization of Stv1p, we prepared a rabbit polyclonal antiserum against a bacterially expressed fragment of Stv1p (amino acid residues 231-436; see Materials and methods). Fig. 1A shows that affinity-purified anti-Stv1p antibodies recognized a protein of approx. 110 kDa on an immunoblot of a wild-type cell lysate, as was previously reported(Manolson et al., 1994). This protein was not present in the stv1▵ cell lysate and was found in greater abundance in the vph1▵ strain than in wild type(Fig. 1B). These results suggest that Stv1p compensates for the loss of Vph1p in the vph1▵ strain by increasing the stability or overproduction of Stv1p in that strain, since there are no differences in the mRNA levels for either subunit a isoform following disruption of one of them(Manolson et al., 1994).

Immunolocalization studies of Vph1p and Stv1p in yeast suggested that Vph1p is localized in the central vacuole, whereas Stv1p is localized in some other intracellular compartments, possibly the Golgi or endosomes(Manolson et al., 1994). Several studies showed that V-ATPase is present along the Golgi complex(Moriyama and Nelson, 1989; Ying et al., 2000; Grunow et al., 1999). Subcellular fractionation of the Golgi membranes using the technique of Graham and Krasnov (1995) was used to determine the location of the Stv1p subunit more precisely. This technique separates the vacuoles more efficiently and allows better detection of the V-ATPase complex in the different Golgi and endosomal fractions. The cell membranes were applied to the bottom of a sucrose gradient and centrifuged to equilibrium. Fractions were collected from the bottom and probed with antibodies against Stv1p, Vph1p, Pep12p, Sec14p and Sed5p. Sed5p is the marker for early Golgi compartments, Sec14p for the late Golgi vesicles and Pep12p for endosomes. In the fractionation profile shown in Fig. 2, Vph1p and Stv1p were broadly distributed throughout the gradient. Furthermore, the Vph1p profile matches that of the Pep12p profile, suggesting an endosomal distribution on the gradient. ALP (vacuolar membrane protein) marks the vacuolar membrane contamination. Stv1p is located in two peaks along the gradient. The first peak (fractions 3-5) coincides slightly better than Vph1p with the Sec 14p(trans-Golgi) distribution, and the second peak (fractions 12,13)matches the broad profiles of Pep12p and Sed5p(Fig. 2). We also found that the V₁ subunit, Vma5p, peaked in the first four fractions. The peak of Vma5p overlapped mostly with that of Sec14p, raising the possibility that the assembled V-ATPase complex present in a late Golgi compartment is more stable. A similar distribution was also observed for Vma8p, another V₁ subunit (not shown). Assuming that the amount of assembled V-ATPase is an indicator of its ATPase activity, these results support previous reports that acidification develops along the Golgi complex and is maximal in the trans-Golgi compartment(Llopis et al., 1998; Schapiro and Grinstein, 2000). Fractionation in yeast does not yield clean separation of compartments, so it is difficult to draw exact conclusions about localization. Assuming that the two subunit a isoforms are in separate complexes, and that this western profile represents assembled and unassembled proteins, we can cautiously say that from the relative amounts, at least in fraction 3, Stv1p is the V-ATPase isoform in the Golgi complex.

It was proposed that Stv1p does not normally reside on the vacuolar membrane in wild-type cells, and only overexpressed Stv1p-HA in vph1Δ/stv1Δ and vph1Δ strains could be found in the vacuolar membrane (Manolson et al., 1994; Kawasaki-Nishi et al., 2001). On the other hand, detectable levels of Vma1p and Vma2p were found to be associated with vacuoles from the vph1Δstrain (Manolson et al.,1992). Our antibody revealed that purified vacuoles from wild-type cells also contain Stv1p. Fig. 3A shows a western analysis of purified vacuoles from wild-type, stv1Δ and vph1Δ probed with antibodies against Stv1p, Vph1p, Vma5p and Vma8p. Stv1p is detected in vacuolar membranes from the wild-type and vph1Δ cells. These data also demonstrate that the level of Stv1p in the vph1Δ mutant strain increased in comparison to the wild-type strain, as was found for Vph1 in stv1Δ mutant (Fig. 1B). On the other hand, the amount of V-ATPase complex in the vph1Δ was reduced significantly (based on amounts of the V₁ subunits of V-ATPase, Vma5p and Vma8p present in those vacuolar membranes) as compared to the wild type.

Although vacuolar membranes isolated from the vph1Δ strain contained an evidently assembled V-ATPase complex (see Fig. 3A), it was still possible that other V-ATPase-containing membranes also contaminated the vacuoles of vph1Δ strain and wild-type. We could detect Golgi and endosomal proteins in isolated vacuoles with antibodies against Sec14p and Pep12p,respectively (data not shown). Fig. 3B depicts two fractions of the gradient (8 and 13) showing,respectively, endosomal and intact vacuolar fractions, which were obtained from a sucrose gradient fractionation of vph1Δ strain,decorated with antibody against CPY (a soluble vacuolar marker), Pep12p antibody (an endosomal marker), and Vma8p and Stv1p antibodies. Fraction 8 represents the peak of the endosomal fractions and fraction 13 the peak of the vacuolar fractions. CPY is a vacuolar-soluble protein, so we also probed both fractions with antibody against ALP, a resident of the vacuolar membrane, that was present throughout the gradient (not shown). These data are in agreement with the assumption that an assembled V-ATPase complex is present in the vacuole in vph1Δ strain with no detectable endosomal contamination.

Previous work has demonstrated that vacuoles of a vph1Δmutant with neutral lumenal pH do not accumulate Quinacrine and have no detectable bafilomycin A₁-sensitive V-ATPase activity(Manolson et al., 1994). Because we showed that small amounts of assembled V-ATPase complex are present in vacuoles from our vph1Δ strain, it was of interest to assess their proton pumping activity. Fig. 4 shows the ATP-dependent proton uptake activity of vacuoles isolated from wild-type, stv1Δ and vph1Δstrains. The ATP-dependent proton uptake activity from the vph1Δ mutant (26 nmol min^-1 mg^-1) was 30-fold less than the specific activity of wild-type (817 nmol min^-1 mg^-1), but nevertheless measurable. This explains the observation that Quinacrine fluorescence, which is dependent on higher proton concentrations, was not observed in the vph1Δ strain. Strains that overexpress Stv1p showed a marked reduction in the assembled V-ATPase complex, even though there were appreciable amounts of Stv1p in the vacuolar membrane (Kawasaki-Nishi et al.,2001). It appears that deletion of the VPH1 gene results in naturally overexpressed Stv1p in vacuolar membranes (Figs 1B, 3A). These results also suggest that the compensation of a vph1Δ mutant by Stv1p is only partial.

As a control for V-ATPase activity we isolated vacuoles from a stv1Δ strain. Surprisingly, the initial rate of V-ATPase activity in these vacuoles was diminished by 40-50 % (see Fig. 4). This highly reproducible result led us to do a more careful analysis of the amounts of V₁ subunits upon the vacuolar membranes in this strain, which are more indicative of the enzyme's activity. A reduction in Vma5p and Vma8p is apparent on western blot analysis (Fig. 3A), whereas the Vph1p level is slightly higher than in wild type. This pattern suggests that in wild type the Stv1p in its post-Golgi location plays a role in facilitating the assembly process of the holoenzyme in the vacuolar membrane.

LysoSensor Green DND-189 is an acidotropic probe that accumulates in the membranes of acidic organelles as a result of protonation. It was used to investigate the acidic compartments in cultured cerebellar granule cells and plant tissue (Cousin and Nicholls,1997; Guttenberger,2000). We used this dye to stain the acidic compartments of yeast(see Materials and methods). The dye labeled the vacuolar membranes of wild-type cells in less than 5 min, and the signal remained stable for more than 1h. The staining is more efficient than with Quinacrine, as shown in Fig. 5A, which shows parallel labeling of wild-type cells with LysoSensor Green DND-189 and Quinacrine (a vital dye for the vacuolar lumen). LysoSensor Green detects acidic vacuoles in almost 100 % of the stained cells. The vacuolar staining was confirmed by double labeling with FM4-64 lipophilic styryl dye, which selectively stains endocytic compartments and yeast vacuolar membranes. Fig. 5B shows double-staining images at different intervals. After 25 min the green staining of the vacuolar membrane is clearly visible, and the red staining by FM 4-64 is concentrated mainly in cytoplasmic vesicles, which may correspond to endosomes(Vida and Emr, 1995). Some of these vesicles are definitely not stained by LysoSensor Green; conversely some vesicles are stained by LysoSensor Green but not by FM4-64. After 45 min the staining of the vacuolar membrane by the FM4-64 is almost complete. Double staining of the vacuolar ring (Fig. 5B) also demonstrates that LysoSensor Green DND-189 is an efficient dye for the staining of acidic compartments in yeast.

LysoSensor Green was used to examine vacuolar membrane staining of wild type, vph1Δ and stv1Δ mutants and the double mutant vph1Δ/stv1Δ cells. The vacuolar membrane was stained normally in wild-type cells, but no staining was observed in the double mutant vph1Δ/stv1Δ, other V-ATPase null mutants(Fig. 6), or in wild-type cells treated with the specific V-ATPase inhibitor concanamycin A (not shown). LysoSensor Green staining and V-ATPase activity are correlated in the vtc1Δ mutant, which was shown to have diminished V-ATPase activity in vitro (Cohen et al.,1999), while its staining with the vital dye was much less intense than that of the wild-type cells (not shown). Our results obtained with the stv1Δ mutant, which grows well in YPD medium, pH 7.5, showed that the intensity of the vacuolar staining was lower than in the wild type(Fig. 6). This observation is in agreement with the data in Fig. 4, which demonstrate reduced V-ATPase activity in the stv1Δ strain. Interesting results were observed in vph1Δ cells. Half of the cells were weakly stained and the others, when the signal was intensified, showed a clear difference between vph1Δ and the double mutant in vacuolar staining(Fig. 6). This observation suggests that, in vph1Δ cells, Stv1p might reach the vacuolar membrane, as also indicated by western analysis(Fig. 3). Moreover, when grown on a medium buffered at pH 7.5, the staining of vacuoles in the vph1Δ cells is intensified(Fig. 6).

We recently investigated the effect of vacuolar ATPase null mutations on the targeting of the plasma membrane H⁺-ATPase (Pma1p) through the secretory pathway (Perzov et al.,2000). We showed that the amounts of Pma1p in the plasma membranes of V-ATPase-depleted mutants were markedly reduced, and a large amount of the protein was accumulated in the ER-Golgi in a non-active form. If Stv1p and Vph1p are part of separate V-ATPase complexes in distinct compartments, we might expect differential effects on Pma1p distribution in both mutants. We therefore performed sucrose gradient fractionation on total membranes prepared from the wild-type, vph1Δ, stv1Δ and vph1Δ/stv1Δ strains. As shown in Fig. 7A, the distribution of Pma1p in the sucrose gradient fractions of the membrane preparation was not altered in stv1Δ and vph1Δ strains, but was altered in the vph1Δ/stv1Δ null mutant, as is usually observed in other null V-ATPase mutants. These results indicate that only complete inactivation of the V-ATPase in all cellular compartments leads to redistribution of Pma1p. In contrast, we observed defects in processing of vacuolar hydrolyses in vph1Δ cells, while in the stv1Δ strain the vacuolar proteins mature as in the wild type(Fig. 7B).

FM4-64 is a fluorescent lipophilic dye, which serves as a marker for endocytosis and has been used for this purpose in yeast(Vida and Emr, 1995). The dye is initially incorporated into the plasma membrane and, through endocytosis,reaches and accumulates in the vacuolar membrane. It has been suggested that the acidification of endocytic organelles by V-ATPase is essential for endosomal trafficking in mammalian cells(Clague, 1998). We used FM4-64 dye to investigate the role of acidification in endosomal trafficking in yeast. The wild-type and null V-ATPase strains were stained with FM4-64 and the localization of the dye recorded after 20, 40, 60 and 90 min. Fig. 8A shows that, in the wild-type strain, the dye reaches the vacuole after 20 min and stains it strongly after 40 min. Vacuolar staining in the vma3Δ mutant is inhibited, and after 60 min the cells do not even show the vacuolar staining which the wild-type cells do after 20 min. After 60 min of internalization,the dye stained vacuolar membranes only in the wild type, whereas in the vma3Δ mutant, vph1Δ/stv1Δ cells or other null V-ATPase mutants, much of the internalized dye only reaches vesicular intermediates in the cytoplasm and levels do not change much during this time. Only after 90 min does some of the dye reach the vacuolar membrane. Fig. 8B summarizes the staining data obtained with vph1Δ and stv1Δ cells. The transport of the dye to the vacuole in vph1Δ cells resembles that in null mutants, but slight staining of the vacuolar membranes is observed in some cells (Fig. 8B). In stv1Δ cells, FM4-64 reached the vacuolar membrane with kinetics similar to the wild type, but a mix of fragmented vacuoles was observed very frequently (a similar vacuolar pattern was obtained by staining the stv1Δ strain with LysoSensor Green dye, see Fig. 6). Fragmented vacuoles are a component of the phenotype in the stv1Δ strain.

Fragmented vacuoles and no defect in endocytic delivery of the dye to the vacuole were also detected in vps1Δ cells(Raymond et al., 1992; Nothwehr et al., 1995; Vida and Emr, 1995). Vps1p is a dynamin-like protein required for formation of endosome-bound vesicles from the Golgi, and in vps1Δ mutants endosomal-bound transport is diverted to the cell surface (Nothwehr et al., 1995).

These data indicate that the null V-ATPase mutants, as well as the vph1Δ mutant, markedly slow down the endocytic process in the yeast cells. To test this further, we stained the wild-type strain in the presence of 3 μmol l^-1 concanamycin A, a specific inhibitor of V-ATPase. Fig. 9 shows that the drug inhibited the internalization of FM4-64, making the wild type similar to the stained V-ATPase null mutant: endosomal-like structures near the vacuolar membrane are stained, but the stain does not reach the vacuole membrane.

Vacuolar H⁺-ATPase generates the energy and appropriate pH for the various compartments of the vacuolar system of the cell(Stevens and Forgac, 1997; Nelson and Harvey, 1999). In many plant and animal species, the various subunits of V-ATPase have numerous isoforms (Futai et al., 2000). It is believed that each isoform resides in a different compartment and may even play a role in targeting the whole complex into this compartment. In yeast, all subunits except for subunit a are encoded by a single gene, so the study of Vph1p and Stv1p is interesting because subunit a may have a role in targeting the complex to specific organelles, as well as in modulating the activity of the enzyme. We set out to describe the phenotype of the STV1 null mutant. Previously Stv1p was assigned to Golgi and/or endosome compartments by immunofluorescent staining(Manolson et al., 1994),correlated with its Golgi retention motive FXFXD; however, no specific features were found for its null mutant. This could be explained by redundancy or by lack of methods to discover the phenotype. First, we have to establish that Stv1p and Vph1p subunits are part of different V-ATPase complexes. This was proposed following work in which Stv1p was first described(Manolson et al., 1994) but no direct proof was available. We tried to immunoprecipitate the V-ATPase membrane complex (V_o) from isolated yeast membranes with Vph1p commercial monoclonal antibody 10D7 (which recognizes its epitope only after V₁ dissociation). We detected Vma3p in the immunoprecipitate, but did not detect any Stv1p with our polyclonal antibody for the Stv1p that was found in the supernatant (data not shown).

V-ATPase is present along the secretory pathway and plays a central role in its proper function, especially in the Golgi complex(Moriyama and Nelson, 1989; Grunow et al., 1999; Ying et al., 2000; Schoonderwoert et al., 2000). To demonstrate the subcellular location of Stv1p we used a Golgi fractionation method (Graham and Krasnov,1995), which separates between cis- and trans-Golgi. Using this gradient we found that Stv1p and Vph1p in the wild type showed different profiles: Vph1p exhibited mainly an endosomal pattern, and was probably on its way to the vacuole, whereas Stv1p peaked nearer to trans-Golgi. V₁ subunits Vma5p and Vma8p showed a main peak in trans-Golgi, which might indicate that this location has the most stable V-ATPase complex. The pH along the secretory pathway falls from the ER through Golgi and reaches maximum acidity in lysosomes(Grabe and Oster, 2001; Llopis et al., 1998). In the Golgi complex the trans-Golgi is the most acidic compartment, which is in good agreement with the high levels of V-ATPase amounts described above.

By disrupting one of the a subunits we can demonstrate that the other partially compensates for the lack of its counterpart. Therefore, both mutants of one isoform lack some features of the null V-ATPase mutant growth profile. However, it may be that the ability of Stv1p to compensate for the lack of Vph1p in the vph1Δ mutant is at much lower efficiency (Figs 3, 4) because it is less stable in the vacuolar membrane (Kawasaki-Nishi et al., 2001). This is seen in the higher rate of growth at pH 7.5,as well as in the processing of the vacuolar proteins ALP and CPY (see Fig. 7B) when compared to vph1Δ. We do not see the same results for the reciprocal mutant(stv1Δ), which might indicate that Vph1p functions normally in place of Stv1p (Fig. 7B). It was therefore interesting to see that the specific activity of proton pumping by V-ATPase in vacuoles isolated from the stv1Δ mutant was diminished in comparison to the wild-type strain. Even though Vph1p itself is overexpressed in the vacuolar membrane in the stv1Δ mutant(Figs 3A, 4), there is less holoenzyme present in it, hence the decrease in activity. Stv1p is therefore not redundant in the wild type. The complex containing it has a role in its proper functioning, as does the complex containing the Vph1p.

The same results were obtained by in vivo staining of these yeast strains with a fluorescent pH indicator, LysoSensor Green. Examination of stained yeast cells revealed residual staining in vacuolar membranes, in the Vph1p null mutant (but not in the double mutant), which correlates with the small but measurable H⁺ translocating activity of V-ATPase in the vacuoles of the vph1Δ mutant(Fig. 6).

The relationship between V-ATPase function and the endocytosis processes is intriguing. On the one hand, endocytosis is responsible for the viability of yeast V-ATPase null mutants, while on the other, some of the endocytosis mutants were described to have a very similar phenotype to the V-ATPase null mutant (Munn and Riezman,1994; D'Hondt et al.,2000; Yoshida and Anraku,2000). The acidification of endosomal compartments due to V-ATPase activity and its relation to endocytic processes, e.g. ligandreceptor dissociation during receptor-mediated endocytosis, is well-established in mammalian cells (Mellman et al.,1986). Other studies showed that endosomal carrier vesicle formation, as well as transfer between late endosomes and lysosomes, are pH-sensitive processes inhibited by the specific V-ATPase inhibitor bafilomycin A (Clague et al.,1994; van Weert et al.,1995; van Deurs et al.,1996). In yeast, the same relationship has yet to be demonstrated. The FM4-64 dye staining of yeast cells was used to visualize endocytosis in vivo (Vida and Emr,1995), and to stain the V-ATPase null mutant in comparison to the wild-type strain. We found inhibition of FM 4-64 internalization in the V-ATPase null mutants. After treatment for 60 min, during which time all the dye concentrated in the vacuolar membrane in the wild type, much of the internalized dye in the mutants only reached endosomes surrounding the vacuolar membrane. We could mimic the mutant in the wild-type cells by addition of concanamycin A, which specifically inhibits V-ATPase activity(Fig. 9).

Since the staining kinetics of this dye are used as a measure for endocytosis, we conclude that V-ATPase mutants inhibit endocytosis in yeast as well. We see that the staining of the plasma membrane (not shown) and of the cytoplasmic vesicles, presumably endosomes, is similar for both the wild type and null V-ATPase mutants. However, the final step of internalization of the dye in the vacuolar membrane is markedly dependent on proper V-ATPase functioning. FM4-64 was used to examine whether the stv1Δ or vph1Δ might have a phenotype in endocytosis. Fig. 8 shows that the Vph1p null mutant displays the double-mutant phenotype, while the Stv1p null mutant has similar staining kinetics to wild type but exhibits prevalent vacuolar localization.

As expected for such a fundamental enzyme, its null mutation, if viable,should affect a vital pathway or pathways of the cell. We conclude that the V-ATPase function is to determine the pH conditions, not only in the lumen of their respective organelles, but also in processes involved in intervesicular activities connected with membrane fusion. A similar conclusion was reached by Ungermann et al. (1999), who showed that V-ATPase activity is needed for in vitro homotypic fusion processes in yeast. This may relate to the recently published data on the participation of the Vma3p of V-ATPase's V_o sector in membrane fusion processes (Peters et al.,2001). Numerous protein complexes are involved in membrane fusion,a process that is not yet completely resolved(Wickner and Haas, 2000; Pelham, 2001). The fact that V-ATPase is suggested as taking part in various stages of this process, and that there are multiple spontaneous revertants of V-ATPase null mutants(Cohen et al., 1999), hint that the suppressor mutations come from the vast pool of the fusion machinery.

We thank Dr A. Bardul for his assistance with the confocal microscopy analysis, and Professor G. Schatz for the gift of FM4-64 and LysoSensor Green DND-189. This project has been funded by the BMBF and supported by BMBF's International Bureau at the DLR.