The Golgi Apparatus Plays a Significant Role in the Maintenance of Ca2+ Homeostasis in the vps33Δ Vacuolar Biogenesis Mutant of Saccharomyces cerevisiae*

The vacuole is the major site of intracellular Ca2+ storage in yeast and functions to maintain cytosolic Ca2+ levels within a narrow physiological range. In this study, we examined how cellular Ca2+ homeostasis is maintained in a vps33Δ vacuolar biogenesis mutant. We found that growth of the vps33Δ strain was sensitive to high or low extracellular Ca2+. This strain could not properly regulate cytosolic Ca2+ levels and was able to retain only a small fraction of its total cellular Ca2+ in a nonexchangeable intracellular pool. Surprisingly, thevps33Δ strain contained more total cellular Ca2+ than the wild type strain. Because most cellular Ca2+ is normally found within the vacuole, this suggested that other intracellular compartments compensated for the reduced capacity to store Ca2+ within the vacuole of this strain. To test this hypothesis, we examined the contribution of the Golgi-localized Ca2+ ATPase Pmr1p in the maintenance of cellular Ca2+ homeostasis. We found that avps33Δ/pmr1Δ strain was hypersensitive to high extracellular Ca2+. In addition, certain combinations of mutations effecting both vacuolar and Golgi Ca2+ transport resulted in synthetic lethality. These results indicate that the Golgi apparatus plays a significant role in maintaining Ca2+homeostasis when vacuolar biogenesis is compromised.

the mammalian PMCA plasma membrane family of Ca 2ϩ ATPases. The loss of Pmc1p results in an inability to grow in the presence of high environmental Ca 2ϩ (7). The second protein known to be involved in vacuolar Ca 2ϩ transport is the H ϩ /Ca 2ϩ exchanger encoded by the VCX1 (HUM1) gene (14,15). Although mutants that do not express Vcx1p show little or no decrease in Ca 2ϩ tolerance, the combination of pmc1⌬ and vcx1⌬ mutations leads to a more severe Ca 2ϩ -sensitive phenotype than the loss of either transporter alone. Both the expression and function of these two vacuolar Ca 2ϩ transporters are regulated by calcineurin, a highly conserved protein phosphatase that is activated by Ca 2ϩ /calmodulin. As in mammalian cells, the activation of yeast calcineurin can be blocked by the immunosupressant drugs cyclosporin A (CsA) 1 and FK506 (16,17). Although the functional relationship between these two vacuolar Ca 2ϩ transporters is complex, it has been reported that calcineurin activation stimulates Pmc1p function and inhibits Vcx1p function (14,15).
Several other genes encoding potential Ca 2ϩ ATPases have been identified within the yeast genome (18); however, the only member of this group demonstrated to play a role in Ca 2ϩ transport is encoded by the PMR1 gene. Pmr1p is related to the SERCA family of Ca 2ϩ ATPases and has been shown to reside in the Golgi apparatus of S. cerevisiae (19 -22). Although Pmr1p and Pmc1p both act to partition Ca 2ϩ into distinct cellular compartments, their roles in Ca 2ϩ homeostasis do not appear to be equivalent. First, cells lacking Pmc1p are sensitive to high environmental Ca 2ϩ , whereas cells lacking Pmr1p cannot grow under low Ca 2ϩ conditions. In addition, the total cellular Ca 2ϩ level in a pmc1⌬ strain is 2-3-fold lower than normal, but the total cellular Ca 2ϩ level in the pmr1⌬ mutant is 4 -5-fold higher than normal. These different phenotypes suggest that the vacuole and the Golgi apparatus normally carry out distinct roles in Ca 2ϩ homeostasis.
Genetic screens have identified at least 60 different genes involved in vacuolar protein localization (23). Among these, the class C vacuolar protein sorting mutants (which include the vps11, vps16, vps18, and vps33 mutants) result in the most severe defects in vacuolar biogenesis. For example, strains carrying the vps33⌬ mutation lack a morphologically distinguishable vacuole but instead accumulate small vesicular and Golgi-like structures (24 -26). These anomalous compartments may result from the inability to dock and/or fuse late transport vesicles from the biosynthetic, endocytic, and autophagic pathways with the vacuole (27). A vps33⌬ strain was also found to secrete Ͼ90% of soluble vacuolar proteins such as carboxypeptidase Y and to mislocalize nearly 50% of the vacuolar membrane protein ␣-mannosidase to the cell surface (24).
In this study we asked how the severe defects in vacuolar biogenesis associated with the vps33⌬ mutation affect cellular Ca 2ϩ homeostasis. We found that the vps33⌬ strain was sensitive to both high and low levels of environmental Ca 2ϩ and was unable to regulate cytosolic Ca 2ϩ levels properly when exposed to a sudden, large increase in environmental Ca 2ϩ . Despite its defect in vacuolar biogenesis, we found that the vps33⌬ strain contains more total cellular Ca 2ϩ than a wild type strain. To determine whether other intracellular compartments compensate for reduced vacuolar Ca 2ϩ storage, we examined whether the Golgi-localized Ca 2ϩ ATPase Pmr1p plays a significant role in Ca 2ϩ homeostasis in the vps33⌬ strain. We found that PMR1 expression is elevated in the vps33⌬ strain. We also found that a vps33⌬/pmr1⌬ strain is hypersensitive to high extracellular Ca 2ϩ , and the combination of certain mutations effecting both vacuolar and Golgi Ca 2ϩ transport results in synthetic lethality. These results indicate that the Golgi apparatus plays a significant role in maintaining Ca 2ϩ homeostasis when vacuolar biogenesis is compromised.

MATERIALS AND METHODS
Strains Used-Strains used in this study are listed in Table I. The PMC1 and VCX1 genes were disrupted using the one-step gene replacement method (28). A 1.62-kb fragment of the PMC1 gene was generated by PCR using wild type yeast genomic DNA as template. The forward primer used was 5Ј-ATCGGTACCA CTTGGATTGC AT-3Ј, and the reverse primer was 5Ј-CATGGATCCT GCCATCCTCA-3Ј. These primers contained KpnI and BamHI restriction endonuclease sites respectively (underlined). The PCR product was digested with KpnI and BamHI and cloned into a pBluescript II KS (ϩ) plasmid. The 1.06-kb segment of the PMC1 gene was then removed by digestion with AflIII and EcoRI and replaced by the TRP1 gene taken from pJJ280 plasmid (29). A KpnI/NotI fragment containing the disrupted pmc1⌬::TRP1 fragment was then used to transform yeast. Trp ϩ colonies were selected, and the correct gene replacement was confirmed by PCR.
Similarly, a 2.04-kb fragment of the VCX1 gene was generated by PCR using genomic DNA as template. The forward primer used was 5Ј-CGTGGTACCT TGTCATCCTCAC-3Ј, and the reverse primer was 5Ј-GCTAGGATCC GCTAAAATAG G-3Ј. Again, these primers contained KpnI and BamHI restriction endonuclease sites, respectively (underlined). The fragment was digested with these enzymes and cloned into a pBluescript II KS (ϩ) plasmid. A 1.56-kb fragment was removed from the VCX1 DNA by digestion with HincII and HindIII endonucleases and replaced with a fragment containing the URA3 gene obtained from pJJ244 (29). A KpnI/BamHI fragment containing the disrupted vcx1⌬::URA3 fragment from this plasmid was used to transform yeast. The replacement of wild type VCX1 was confirmed by PCR analysis. Other genetic manipulations were carried out by standard methods (30).
Culture Media-Bacterial strains were grown on standard media (31). Yeast strains were maintained on YP medium containing 2% D-glucose (YPD) or synthetic minimal medium containing 2% D-glucose (SMD) and other supplements as required (30). Growth media were routinely buffered with 40 mM MES-Tris, pH 5.5.
Determination of Ca 2ϩ Concentration in Media-EGTA was used to reduce the Ca 2ϩ concentration of buffered media. Because YPD and SMD media contain divalent cations other than Ca 2ϩ , the effective concentrations of Mg 2ϩ , Mn 2ϩ , Fe 2ϩ , K ϩ , and Na ϩ were considered when calculating free Ca 2ϩ concentrations. Known quantities of CaCl 2 stock solutions were added, and the resulting free Ca 2ϩ concentrations were calculated based on the total concentration of Ca 2ϩ as well as other cations, pH, and temperature of the medium. These calculations were done using the Maxchelator 1.2 program.
Measurements of Total Cellular Ca 2ϩ , Mg 2ϩ , Na ϩ , K ϩ , and Phosphate Levels-50 -100 A 600 units of yeast growing in YPD supplemented with CaCl 2 or EGTA were harvested by centrifugation at 5,000 ϫ g for 5 min. The cell pellets were washed with fresh YP and transferred to microcentrifuge tubes whose mass had previously been determined gravimetrically to an accuracy of 0.1 mg on an analytical balance. The tubes were centrifuged at 15,000 ϫ g for 5 min, and the supernatants were removed carefully. The tubes were then respun, and any remaining supernatant was again removed. The tubes containing the pellets were weighed to determine the wet weight of the pellet, and the pellets were then dried to completion in a Savant SpeedVac system. The tubes were then weighed again to determine the dry weight of the pellet. 1 M HCl was added to the dry pellets, and the capped microcentrifuge tubes were vortexed and incubated on a rocker for at least 24 h. Thereafter each sample was centrifuged briefly in a microcentrifuge, and multiple aliquots of each supernatant were taken for ion measurements. Ca 2ϩ , Na ϩ , and K ϩ measurements of aliquots were carried out with an Eppendorf EFOX-5070 flame photometer; Mg 2ϩ levels of aliquots were determined using a Varian AA-20 atomic absorption spectrophotometer. Cellular ion concentrations were then calculated based on the dry weight of the samples and dilution factors. Total combined orthophosphate and polyphosphate levels (referred to as total inorganic phosphate) were determined in the 1 M HCl hydrolysate described above using an acid molybdate-based diagnostic kit (Sigma). The phosphorus levels measured represent the sum of the acid-hydrolyzed polyphosphate and the inorganic phosphate present (32).
Measurement of Cytosolic Free Ca 2ϩ Concentration-A pEVP11based plasmid containing a functional apoaequorin gene (pAEQ) was transformed into yeast using the LEU2 gene as selectable marker (1). This plasmid was a gift from Patrick Masson. Cells containing the pAEQ plasmid were grown in SMD medium containing other necessary supplements and were harvested in the logarithmic growth phase. 10 A 600 units of cells were resuspended in 0.2 ml of aequorin test medium, which consists of SMD medium (which contains 1 mM Ca 2ϩ ) supplemented with 2 mM EGTA and 20 mM MES-Tris, pH 6.5. The free Ca 2ϩ concentration of this medium was calculated to be 6 M. To convert the apoaequorin to aequorin, 10 l of 590 M coelenterazine (dissolved in methanol) was added, and the cells were incubated for 20 min at room temperature. They were then centrifuged briefly in a microcentrifuge, and the supernatant containing excess coelenterazine was removed. The cells were washed again in 0.5 ml of aequorin test medium, and the cells were then resuspended in test medium and incubated at room temperature for 20 min before initiating the experiment. A Berthold Lumat 9050 luminometer was used to collect aequorin light emission data at 200-ms intervals. The data were downloaded directly to a computer using the MS Windows Terminal software and transferred to Microsoft Excel 5.0 for analysis.
To determine the concentration of cytosolic Ca 2ϩ using the aequorin reporter system, it was necessary to determine: 1) the total amount of reconstituted aequorin available for light emission and 2) the relationship between Ca 2ϩ concentration and light emission (33). The total amount of reconstituted aequorin was determined routinely in a crude extract of each strain by measuring the maximum light emission (L max ) value in the presence of a saturating concentration of Ca 2ϩ . To prepare This study the crude extract, 2 A 600 units of cells in 0.2 ml of aequorin standard buffer (100 mM MES-Tris, pH 6.5; 150 mM KCl; 20 mM NaCl; 5 mM MgCl 2 ; and 2 mM phenylmethylsulfonyl fluoride) were lysed by agitation with glass beads at 4°C. A 25-l aliquot was placed in the luminometer, and the L max of this sample was induced by injecting 25 l of a 50 mM CaCl 2 solution. The L max value was generally between 0.5 and 1.0 ϫ 10 7 relative light units/s. The protein concentration of cell lysates was also measured using a Bio-Rad protein assay kit. A correction factor based upon the L max value/unit of protein was determined for each strain, and this value was used to correct for minor differences in the concentration of aequorin in different strains.
To determine the relationship between the free Ca 2ϩ concentration and aequorin-based light emission, a standard curve was prepared using a cell lysate as described (33). Briefly, increasing concentrations of CaCl 2 were added to a crude extract of wild type cells prepared in aequorin standard buffer. To determine the cytosolic Ca 2ϩ concentration within intact cells, both the L observed in intact cells and the L max emission observed in a crude extract of the same cells were determined. The ratio between these values (L:L max ) was then used to estimate the cytosolic free Ca 2ϩ concentration from our standard curve. In no case was the L value in an experiment greater than 2-3% of the L max value. Thus, the absolute amount of reconstituted aequorin was not limiting in any of these experiments. 45 Ca 2ϩ Uptake and Release-To determine the rate of Ca 2ϩ uptake by different mutant strains, cells were grown in SMD medium to approximately 1.0 A 600 /ml. Cells were harvested and resuspended in a buffer containing 40 mM MES-Tris, pH 6.5, and 20 mM D-glucose. An aliquot of 45 Ca 2ϩ (NEN Life Science Products) was then added, and aliquots were filtered through 0.45-m Millipore filters on a 12-position Millipore vacuum manifold at the indicated times. The filtered cells were washed immediately with two 5-ml aliquots of ice-cold blocking solution (150 mM NaCl, 20 mM MgCl 2 , and 2 mM LaCl 3 ). The cell-associated counts on the filter were then determined by scintillation counting. To calculate absolute Ca 2ϩ levels, cpm were converted to mmol of Ca 2ϩ /kg dry mass based upon total cellular Ca 2ϩ measurements as determined by flame photometry under identical growth conditions. Cells for Ca 2ϩ exchange experiments were grown in YPD medium to a density of 0.05 A 600 . The medium was then supplemented with 45 Ca 2ϩ , and the cells were grown to a cell density of 0.5-1 A 600 /ml. The cells were then harvested by centrifugation at 4,000 ϫ g for 5 min, washed, and resuspended in fresh YPD supplemented with 50 mM CaCl 2 . At the indicated times, aliquots were removed, filtered, washed, and processed for scintillation counting as described above.
Northern Analysis-RNA extraction and Northern analysis were carried out as described previously (34). Strains were grown in YPD medium in the presence of 1 mM EGTA (estimated to result in 0.01 mM free Ca 2ϩ ) or 50 mM calcium to 1 A 600 /ml. A 0.56-kb region of the PMR1 gene was amplified by PCR using the primers DB-483 (5Ј-GGC-CCCAATGAAATAACCGT AG-3Ј) and DB-484 (5Ј-CCTGTTCCTAC GACGATACCC T-3Ј). The ACT1 probe was prepared by PCR amplification using the primers DB-154 (5Ј-GCGCG GAATT CAACG TTCCA GCCTT CTAC-3Ј) and DB-155 (5Ј-GGATG GAACA AAGCT TCTGG-3Ј). All probes were labeled with [␣-32 P]dATP using the random hexamer method. Radioactivity in specific hybrids was quantitated using a Phos-phorImager (Molecular Dynamics). After quantitating the radioactivity associated with PMR1 mRNA, the membranes were hybridized with the ACT1 probe. After background correction, the PMR1 signal of each sample was corrected with the ACT1 mRNA control. These corrected values were then normalized to the wild type strain grown under low Ca 2ϩ conditions.

Sensitivity of Yeast Vacuolar Mutants to Different Environmental Ca 2ϩ
Concentrations-We initially compared the Ca 2ϩ tolerance of yeast strains containing knockouts of genes involved in vacuolar Ca 2ϩ transport (pmc1⌬, vcx1⌬, or pmc1⌬/ vcx1⌬) vacuolar biogenesis (vps33⌬) or a combination of both classes (vps33⌬/pmc1⌬/vcx1⌬). Each strain was streaked onto YPD plates supplemented with increasing concentrations of CaCl 2 or with 10 mM EGTA and incubated at 30°C for 48 h. The wild type, pmc1⌬, vcx1⌬, and pmc1⌬/vcx1⌬ strains grew similarly on standard YPD plates (buffered to pH 5.5) containing 0.3 mM Ca 2ϩ (Fig. 1A), whereas the colony size of the vps33⌬ and vps33⌬/pmc1⌬/vcx1⌬ strains was slightly smaller. The wild type, pmc1⌬, and vcx1⌬ strains also grew similarly on YPD medium supplemented with 100 mM CaCl 2 , whereas the growth rate of the pmc1⌬/vcx1⌬ double mutant was reduced significantly on this medium (Fig. 1B). In contrast, neither the vps33⌬ strain nor the vps33⌬/pmc1⌬/vcx1⌬ strain was able to form visible colonies under these growth conditions during the 48-h incubation period.
When the YPD plates were supplemented with 200 mM CaCl 2 , both the wild type and vcx1⌬ strains grew somewhat more slowly than on YPD plates supplemented with 100 mM CaCl 2 . The pmc1⌬/vcx1⌬ double mutant was unable to grow under these conditions, whereas the pmc1⌬ strain grew much more slowly than the wild type strain (Fig. 1C). A further doubling of the Ca 2ϩ concentration in the YPD plate to 400 mM completely inhibited growth of the pmc1⌬ strain but not the growth of the wild type and vcx1⌬ strains (not shown). None of these strains was inhibited by the addition of either 400 mM NaCl or 400 mM KCl to the YPD plates, indicating that the increased osmolarity associated with 200 mM CaCl 2 did not cause the growth sensitivity described above. We conclude that strains harboring the vps33⌬ mutation show greater sensitivity to high extracellular Ca 2ϩ than strains carrying the pmc1⌬ mutation, the vcx1⌬ mutation, or both mutations together. Overall, the rank order of Ca 2ϩ sensitivity observed for these strains was: vps33⌬/pmc1⌬/vcx1⌬ and vps33⌬ strains Ͼ pmc1⌬/vcx1⌬ strain Ͼ pmc1⌬ strain Ͼ vcx1⌬ and wild type strains.
We also examined whether the growth of these strains was sensitive to inhibition by the chelating agent EGTA. We found that pmc1⌬, vcx1⌬, and pmc1⌬/vcx1⌬ strains grew similarly to the wild type strain on YPD plates buffered to pH 5.5 and supplemented with 10 mM EGTA (Fig. 1D). In contrast, the growth of the vps33⌬ and vps33⌬/pmc1⌬/vcx1⌬ strains was severely inhibited under these conditions, suggesting that they require a higher minimal level of environmental Ca 2ϩ for efficient growth than the other strains. However, not only Ca 2ϩ but other cations such as Zn 2ϩ , Fe 2ϩ , and Mn 2ϩ are also com- plexed effectively by EGTA. To confirm that low environmental Ca 2ϩ was responsible for EGTA sensitivity, we supplemented EGTA-pretreated media with different divalent cations to determine the component(s) required for growth of the vps33⌬ strain. We found that the addition of Ca 2ϩ could restore a significant amount of growth in YPD medium treated with EGTA, whereas several other cations (Mg 2ϩ , Mn 2ϩ , Fe 2ϩ , Zn 2ϩ , and Cu 2ϩ ) could not (data not shown). These results lead us to conclude that the vps33⌬ and vps33⌬/pmc1⌬/vcx1⌬ strains are more sensitive to either high or low levels of environmental Ca 2ϩ than the wild type, pmc1⌬, vcx1⌬, and pmc1⌬/vcx1⌬ strains.
Measurement of Rapid Changes in the Cytosolic Ca 2ϩ Concentration upon External Ca 2ϩ Challenge-Yeast cells, like mammalian cells, have been reported to maintain cytosolic free Ca 2ϩ levels in the range of 50 -200 nM (1-3). To determine how the above mutations affect the ability of yeast to maintain cytosolic Ca 2ϩ homeostasis, we introduced a plasmid encoding a cytosolic form of apoaequorin into each strain (1). Apoaequorin can be converted to aequorin by incubating the strains with the membrane-permeant cofactor coelenterazine. Once active aequorin is generated, it is capable of emitting light as a function of the free Ca 2ϩ concentration present in the cytosol (33). In the experiments described here, the aequorin-dependent light emission of each strain was sampled throughout the experiment at 200-ms intervals. To determine the cytosolic Ca 2ϩ concentration as a function of light emission, a standard curve was prepared using crude extracts from the wild type strain where the light emission at each Ca 2ϩ concentration was correlated to the L max each sample was capable of discharging ( Fig. 2A). Using this method, the relative light units/s emitted from the wild type strain routinely corresponded to a resting free cytosolic Ca 2ϩ concentration of ϳ75 nM when cells were incubated in a medium containing low (ϳ6 M) free Ca 2ϩ (for further details, see "Materials and Methods").
To determine how various mutations affect the ability of these strains to respond to a sudden increase in extracellular Ca 2ϩ , 50 mM CaCl 2 was injected rapidly into the cell suspension while the cytosolic aequorin-dependent light emission was continuously monitored. We found that the light emission of the wild type strain increased rapidly and reached a peak level corresponding to ϳ300 nM cytosolic Ca 2ϩ within 5 s (Fig. 2B). The Ca 2ϩ concentration decreased rapidly thereafter and returned to a new steady-state free cytosolic Ca 2ϩ concentration of ϳ80 -85 nM within 90 s.
The light emission measured in the pmc1⌬/vcx1⌬ strain corresponded to a basal cytosolic Ca 2ϩ concentration of 75-80 nM. When 50 mM CaCl 2 was injected, the light emission reached a peak value corresponding to ϳ385 nM cytosolic free Ca 2ϩ , which was somewhat higher than the peak observed with the wild type strain. The recovery phase of the pmc1⌬/vcx1⌬ strain was also much weaker than the wild type control. The postshock steady-state cytosolic Ca 2ϩ concentration was ϳ310 nM, which was 4-fold higher than the steady-state cytosolic Ca 2ϩ concentration observed in the wild type strain after the same Ca 2ϩ shock. This suggests that the loss of the Pmc1p and Vcx1p vacuolar Ca 2ϩ transporters severely compromises the ability of this strain to return its cytosolic Ca 2ϩ concentration to a low resting level after exposure to elevated extracellular Ca 2ϩ .
We next examined the response of strains carrying the vps33⌬ mutation to Ca 2ϩ shock. We found that the initial resting cytosolic Ca 2ϩ concentration was ϳ165 nM, which was 2-fold higher than the wild type strain. The basal cytosolic Ca 2ϩ level measured in the vps33⌬/pmc1⌬/vcx1⌬ strain was ϳ210 nM, which was almost 3-fold higher than the wild type strain. When the vps33⌬ strain was exposed to Ca 2ϩ shock, the maximum light emission was nearly 100-fold higher than observed with the wild type strain and corresponded to a peak cytosolic Ca 2ϩ concentration of ϳ1.75 M (Fig. 2B). This level was 5-fold higher than the peak observed with the wild type strain. Like the pmc1⌬/vcx1⌬ strain, the recovery of the vps33⌬ strain from the peak cytosolic Ca 2ϩ level was much weaker than the wild type control and reached a new steadystate level at ϳ470 nM (6-fold higher than the wild type strain). The vps33⌬/pmc1⌬/vcx1⌬ strain exhibited a high peak of cytosolic Ca 2ϩ which corresponded to ϳ1.5 M, which was somewhat lower than was observed with the vps33⌬ strain. However, the recovery of this strain from the peak level was even weaker than the vps33⌬ strain and reached a new steady-state level of ϳ660 nM (more than 8-fold higher than the wild type strain). The weaker recovery of this strain may indicate that a low level of residual function of the Pmc1p and/or the Vcx1p transporters remains within the vesicles that accumulate in the vps33⌬ strain. When taken together, these results indicate that strains carrying the vps33⌬ mutation are severely compromised in their ability to regulate basal cytosolic Ca 2ϩ levels and are unable to sequester efficiently the cytosolic Ca 2ϩ that enters the cell after an acute Ca 2ϩ shock.
FIG. 2. Measurement of rapid changes in cytosolic free Ca 2؉ levels after a Ca 2ϩ shock. Panel A, standard curve correlating free Ca 2ϩ concentration to aequorin-dependent light emission as measured in crude extracts. Panel B, changes in the cytosolic free Ca 2ϩ concentration were measured in strains containing aequorin after the addition of 50 mM CaCl 2 to the medium. Light emission was monitored over a 2-min period from wild type, pmc1⌬/vcx1⌬, vps33⌬, and vps33⌬/ pmc1⌬/vcx1⌬ strains. The Ca 2ϩ shock was initiated by injecting 50 mM CaCl 2 into the test medium after measuring the basal light emission for 10 s (for further details, see "Materials and Methods").

Measurement of Total Cellular Ca 2ϩ , Mg 2ϩ , and Phosphate Levels in Yeast Vacuolar
Mutants-A large fraction of total cellular Ca 2ϩ , Mg 2ϩ , and polyphosphate normally resides within the vacuole (6,8,35,36). To determine how these various vacuolar mutations affect the capacity to store these compounds within the vacuole, we measured their total cellular levels (Fig. 3). We did not detect a significant change in the level of Mg 2ϩ in the pmc1⌬/vcx1⌬ strain, although a small (22%) decrease in total cellular inorganic phosphate (orthophosphate and polyphosphate) was observed. In contrast, the total cellular Ca 2ϩ level was reduced nearly 2-fold in the pmc1⌬/vcx1⌬ strain.
The cellular levels of these three compounds were significantly different in strains carrying the vps33⌬ mutation. We found that the total amount of cellular Mg 2ϩ was 3-fold lower in both the vps33⌬ and vps33⌬/pmc1⌬/vcx1⌬ strains. Similarly, the total inorganic phosphate level was reduced more than 4-fold in the vps33⌬ strain and 6-fold in the vps33⌬/pmc1⌬/ vcx1⌬/ strain. Thus, strains carrying the vps33⌬ mutation exhibited a severe reduction in the cellular content of Mg 2ϩ and inorganic phosphate, consistent with a reduced capacity to store these ions within the vacuole of strains carrying the vps33⌬ mutation.
Because Ͼ90% of total cellular Ca 2ϩ is normally stored within the vacuole (5, 6), we expected the vps33⌬ strain also to contain a much lower level of total Ca 2ϩ . However, we found that both the vps33⌬ and vps33⌬/pmc1⌬/vcx1⌬/ strains contained 15-20% more cellular Ca 2ϩ than the wild type strain. Thus, the vacuolar biogenesis defect associated with the vps33⌬ mutation resulted in a net increase in total cellular Ca 2ϩ , and this phenotype was epistatic to the decrease in total cellular Ca 2ϩ observed in the pmc1⌬/vcx1⌬ mutant. When taken in conjunction with the observation that the vacuolar storage of Mg 2ϩ and inorganic phosphate is compromised in strains carrying the vps33⌬ mutation, these results suggest that another intracellular compartment is capable of compensating for the defects in Ca 2ϩ storage and homeostasis in strains carrying the vps33⌬ mutation.
Membrane Permeability of the Vacuolar Mutants-The results described above indicate that the vps33⌬ mutation has effects on Ca 2ϩ homeostasis which differ significantly from the combined loss of the Pmc1p and Vcx1p vacuolar Ca 2ϩ transporters. One possible explanation for the higher level of Ca 2ϩ observed is that the rate of Ca 2ϩ uptake in the vps33⌬ strain is increased. To test this possibility, we measured the rate of 45 Ca 2ϩ uptake in each strain (Fig. 4). A CaCl 2 solution contain-ing the radionuclide was added to cells at a final concentration of 1 mM. Aliquots were then collected at intervals over a period of 90 s to determine the rate of Ca 2ϩ uptake. All four strains (wild type, pmc1⌬/vcx1⌬, vps33⌬, and vps33⌬/pmc1⌬/vcx1⌬) showed a similar rate of Ca 2ϩ uptake, indicating that the vps33⌬ mutation does not significantly alter the rate of Ca 2ϩ uptake under the conditions examined (1 mM extracellular Ca 2ϩ ). If the vps33⌬ mutation altered the plasma membrane permeability in a more general, nonspecific way, it is likely that the concentration of other intracellular ions would also be altered. To examine this possibility, we measured the steadystate concentrations of the monovalent K ϩ and Na ϩ ions in each strain when grown in YPD medium (Fig. 5). We found that none of the strains had any significant differences in the total cellular concentrations of either cation. Taken together, these results indicate that the membrane permeabilities of Ca 2ϩ , K ϩ , and Na ϩ are not altered significantly in strains carrying the vps33⌬ mutation under the conditions examined. lar Ca 2ϩ found in yeast cells exists in two distinct forms, termed the exchangeable and nonexchangeable pools (6,8). The exchangeable pool represents Ca 2ϩ that can readily leave the cell, whereas the nonexchangeable pool is thought to represent a more stable pool of Ca 2ϩ located primarily within the vacuole in a complex with polyphosphate. To determine the partitioning of cellular Ca 2ϩ between the exchangeable and nonexchangeable pools in the vps33⌬ strains, we measured 45 Ca 2ϩ efflux. Strains were grown in YPD medium containing 45 Ca 2ϩ for four generations. After washing and resuspending the cells in fresh YPD medium containing 50 mM CaCl 2 , the amount of 45 Ca 2ϩ that remained associated with cells from each strain was determined at various times (Fig. 6A). We found that the wild type and pmc1⌬/vcx1⌬ strains quickly exchanged a small portion of the total cellular Ca 2ϩ during the first 15 min and subsequently exchanged Ca 2ϩ at a much slower rate. In contrast, both strains carrying the vps33⌬ mutation exhibited a much longer period of Ca 2ϩ exchange which extended for 90 min for the vps33⌬/pmc1⌬/vcx1⌬ strain and 210 min for the vps33⌬ strain. This indicates that most of the Ca 2ϩ within strains carrying the vps33⌬ mutation does not reside within a nonexchangeable pool. Although the larger size of the exchangeable Ca 2ϩ pool may partially account for the increased period of time required to release the exchangeable Ca 2ϩ pool in these strains, other factors may also be involved.
Because the strains carrying the vps33⌬ mutation exhibited a prolonged time of release of their exchangeable pools, we compared the nonexchangeable and exchangeable Ca 2ϩ pools in each strain after Ca 2ϩ efflux was allowed to proceed for 210 min (Fig. 6, B and C). Under these conditions, we found that the wild type strain contained 4.9 mmol of Ca 2ϩ /kg dry mass in its nonexchangeable pool. In contrast, the pmc1⌬/vcx1⌬ strain retained only 0.5 mmol of Ca 2ϩ /kg of dry mass after 210 min of efflux. Similarly, the vps33⌬ strain held 1.0 mmol of Ca 2ϩ /kg dry mass, and the vps33⌬/pmc1⌬/vcx1⌬strain held 0.6 mmol of Ca 2ϩ /kg dry mass in their nonexchangeable pools. Thus, the nonexchangeable pool in each of these mutant strains is 5-10fold smaller than in the wild type strain, indicating that all three mutant strains are severely compromised in their ability to store Ca 2ϩ within the vacuolar nonexchangeable pool.
When we calculated the amount of Ca 2ϩ that was readily mobilized during 210 min of efflux, we found that the exchangeable pool in the wild type strain contained 1.2 mmol of Ca 2ϩ /kg dry mass. This pool held 6.9 mmol of Ca 2ϩ /kg dry mass in the vps33⌬ strain and 8.1 mmol of Ca 2ϩ /kg dry mass in the vps33⌬/pmc1⌬/vcx1⌬ strain. Thus, the absolute amount of Ca 2ϩ in the exchangeable pool in these strains was 6 -7-fold larger than in the wild type strain. In contrast, the amount of Ca 2ϩ in the exchangeable pool in the pmc1⌬/vcx1⌬ strain was 1.8 mmol of Ca 2ϩ /kg dry mass, which was only 1.5-fold higher FIG. 5. Measurement of total cellular Na ؉ and K ؉ levels. Cultures of the indicated strains were grown in standard YPD medium, and the relative amounts of K ϩ (panel A) and Na ϩ (panel B) were determined as described under "Materials and Methods." WT, wild type.
FIG. 6. Measurement of 45 Ca 2؉ efflux. The indicated strains were grown for four generations in YPD medium supplemented with 45 Ca 2ϩ . To initiate Ca 2ϩ release, the strains were harvested, washed, and resuspended in YPD supplemented with 50 mM CaCl 2 . The amount of 45 Ca 2ϩ that remained cell-associated was determined at the indicated times and converted to total cellular Ca 2ϩ as described under "Materials and Methods." Panel A, absolute amounts of cell-associated Ca 2ϩ . Squares, wild type; diamonds, pmc1⌬/vcx1⌬; circles, vps33⌬; and triangles, vps33⌬/pmc1⌬/ vcx1⌬. Panel B, nonexchangeable Ca 2ϩ pools. Panel C, exchangeable Ca 2ϩ pools. WT, wild type. than the wild type strain. These results indicate that although the nonexchangeable Ca 2ϩ pools within the three vacuolar mutant strains are similar, the exchangeable pools found in the vps33⌬ and vps33⌬/pmc1⌬/vcx1⌬ strains are roughly 4-fold larger than those found in the pmc1⌬/vcx1⌬ strain.
Sensitivity of Vacuolar Mutants to Cyclosporin A-Several studies have found that the loss of calcineurin function leads to a significant increase in the steady-state level of cellular Ca 2ϩ (7,14,(37)(38)(39). To determine how the vps33⌬ strain responds to such an increase in intracellular Ca 2ϩ , we compared the growth of these strains on YPD plates (pH 5.5) with and without 20 g/ml CsA (Fig. 7). In the absence of CsA, the vps33⌬ and the vps33⌬/pmc1⌬/vcx1⌬ strains again had a slightly slower growth rate than the other strains. However, when CsA was added to the plates the growth of the vps33⌬/pmc1⌬/ vcx1⌬ strain was completely blocked, whereas the vps33⌬ strain showed a severe growth defect compared with plates lacking CsA. In contrast, growth of the wild type and pmc1⌬/ vcx1⌬ strains was unaffected by the presence of CsA. These results are consistent with the possibility that the vps33⌬ mutation reduces the ability of these strains to sequester adequately the increased intracellular Ca 2ϩ that accumulates upon the inhibition of calcineurin function.
The Golgi Ca 2ϩ ATPase Pmr1p Participates in the Maintenance of Cellular Ca 2ϩ Homeostasis during Ca 2ϩ Stress-The results presented above clearly demonstrate that the vps33⌬ mutation severely disrupts intracellular Ca 2ϩ homeostasis. Despite the severe defects in vacuolar structure and Ca 2ϩ sequestration which result from this mutation, they remain viable and accumulate a normal amount of total cellular Ca 2ϩ . This raised the possibility that other intracellular organelles may compensate for the loss of vacuolar Ca 2ϩ storage in these strains. Besides the vacuole, two compartments within the secretory pathway have also been implicated in Ca 2ϩ storage in yeast. The PMR1 gene encodes a Ca 2ϩ ATPase that has been localized to the Golgi apparatus (19,21,22) and was also recently reported to influence the rate of degradation of proteins within the endoplasmic reticulum (40). Given this well defined role of Pmr1p as a Ca 2ϩ ATPase within a non-vacuolar compartment, we next tested whether Pmr1p may be involved in maintaining Ca 2ϩ homeostasis in strains defective in vacuolar biogenesis.
First, we examined PMR1 mRNA levels to determine whether its expression changes in response to either the concentration of environmental Ca 2ϩ or mutations that effect Ca 2ϩ homeostasis. To provide the broadest range of environmental Ca 2ϩ concentrations during this experiment, strains were grown in YPD containing 1 mM EGTA (calculated to reduce the free Ca 2ϩ concentration to approximately 0.01 mM) or in YPD supplemented with 50 mM CaCl 2 . RNA was extracted from each strain, and the level of PMR1 mRNA was determined (relative to an ACT1 control). In the wild type strain, we found that the relative level of PMR1 mRNA increased 1.4-fold as extracellular Ca 2ϩ increased (Fig. 8). In the pmr1⌬/vcx1⌬ strain, we found that the PMR1 mRNA level was slightly elevated in the low Ca 2ϩ medium and was increased to 1.6-fold above the wild type control when the environmental Ca 2ϩ was increased. Finally, the PMR1 mRNA level in the vps33⌬ strain was 1.6-fold higher than the wild type strain when grown in the presence of low Ca 2ϩ and was increased to 2.2-fold higher than the wild type control when grown in the presence of 50 mM Ca 2ϩ . These results indicate that PMR1 gene expression increases moderately as a function of the Ca 2ϩ stress on a wild type strain or as a consequence of mutations that effect the maintenance of intracellular Ca 2ϩ homeostasis.
To address further the role of Pmr1p in maintaining Ca 2ϩ homeostasis in strains carrying the vps33⌬ mutation, we examined the progeny of a cross between a pmr1⌬ strain and a vps33⌬/pmc1⌬/vcx1⌬ strain. A total of 36 tetrads was dissected, and the genotype of the 107 viable spores was determined. We found that all but three possible combinations of mutations were obtained. The nonviable combinations, which all contained the pmr1⌬ mutation, were: pmr1⌬/pmc1⌬/ vcx1⌬, pmr1⌬/vps33⌬/pmc1⌬, and pmr1⌬/vps33⌬/pmc1⌬/ vcx1⌬. This indicates that the loss of both Ca 2ϩ transporters located in the vacuole (pmc1⌬ and vcx1⌬) in conjunction with the Golgi apparatus Ca 2ϩ transporter (pmr1⌬) is lethal. Although strains lacking both the vacuolar Ca 2ϩ ATPase (Pmc1p) and Golgi apparatus Ca 2ϩ ATPase (Pmr1p) were viable, the introduction of mutations that further compromised Ca 2ϩ homeostasis (either the vps33⌬ or the vcx1⌬ mutation) apparently resulted in a lethal imbalance in Ca 2ϩ homeostasis. These results indicate that specific combinations of both vacuolar and Golgi mutations lead to insurmountable defects in Ca 2ϩ homeostasis.
Previous studies reported that disruption of the PMR1 gene does not confer sensitivity to elevated levels of environmental Ca 2ϩ (19 -21). This led to the conclusion that the Golgi apparatus does not play a significant role in maintaining cellular Ca 2ϩ homeostasis under conditions of Ca 2ϩ stress. To determine whether the Golgi apparatus plays a more significant role in this process when vacuolar Ca 2ϩ storage is compromised, we next examined the ability of the pmr1⌬/vps33⌬ strain to grow in the presence of elevated environmental Ca 2ϩ (Fig. 9). This strain grew somewhat slower than the vps33⌬ strain on stand- ard YPD medium. Although the vps33⌬ and vps33⌬/pmc1⌬/ vcx1⌬ strains were capable of growth on YPD plates containing 50 mM CaCl 2 , growth of the pmr1⌬/vps33⌬ strain was completely inhibited. These results indicate that the Golgi apparatus acts to compensate for the defective Ca 2ϩ homeostasis associated with the vps33⌬ strain.
Finally, we examined whether strains carrying the pmr1⌬ mutation alone also exhibited a growth defect in the presence of high environmental Ca 2ϩ (Fig. 10). We found that each of the four strains examined (wild type, pmr1⌬, pmc1⌬, and vcx1⌬) grew with similar rates on plates containing 100 mM CaCl 2 . However, we found that the pmr1⌬ and pmc1⌬ strains were unable to grow on plates containing 500 mM CaCl 2 , whereas the wild type and vcx1⌬ strains did grow under these conditions. These results indicate that in cells with intact vacuolar function, Pmr1p plays a more important role in maintaining Ca 2ϩ homeostasis upon exposure to extreme Ca 2ϩ stress than the vacuolar Vcx1p transporter.

DISCUSSION
Wild type strains of S. cerevisiae are capable of maintaining intracellular Ca 2ϩ levels within a narrow range when faced with extracellular Ca 2ϩ concentrations ranging from Ͻ1 M to Ͼ100 mM. Consistent with the fact that the yeast vacuole normally contains Ͼ90% of the total cellular Ca 2ϩ , mutations in Ca 2ϩ transporters which limit vacuolar Ca 2ϩ uptake have been shown to cause a 2-3-fold reduction in the total cellular Ca 2ϩ levels (7,14,15). Similarly, we observed a 2-fold decrease in total cellular Ca 2ϩ in the pmc1⌬/vcx1⌬ strain. In contrast, we found that strains carrying the vps33⌬ vacuolar biogenesis mutation have total cellular Ca 2ϩ levels that are slightly higher than the wild type strain. This result was surprising based upon the severe defects in vacuolar biogenesis caused by mutations in this gene (24 -26) in conjunction with our finding that the steady-state levels of two other substances normally stored primarily within the vacuole (Mg 2ϩ and polyphosphate) were greatly reduced. Because the vps33⌬/pmc1⌬/vcx1⌬ strain (which lacks both known vacuolar Ca 2ϩ transporters) also had this high level of total cellular Ca 2ϩ , the increased accumulation of Ca 2ϩ cannot be attributed to the residual function of these transporters in a prevacuolar compartment. Instead, our results suggest that the loss of most (or all) vacuolar Ca 2ϩ storage in strains carrying the vps33⌬ mutation leads to the redistribution of a significant portion of intracellular Ca 2ϩ into the Golgi apparatus and possibly other intracellular compartments as well.
The Golgi apparatus contains the only non-vacuolar Ca 2ϩ ATPase (Pmr1p) that has been characterized in yeast (19 -22). Because the pmr1⌬ strain was not previously found to be sensitive to elevated extracellular Ca 2ϩ , it was not thought to play a significant role in maintaining cellular Ca 2ϩ homeostasis. However, we found that a pmr1⌬/vps33⌬ strain is more sensitive to elevated extracellular Ca 2ϩ than the vps33⌬ strain alone, and PMR1 gene expression is elevated in the vps33⌬ strain. In addition, we found that certain combinations of mutations affecting both vacuolar and Golgi Ca 2ϩ transport (pmr1⌬/pmc1⌬/vcx1⌬, pmr1⌬/vps33⌬/pmc1⌬, and pmr1⌬/ vps33⌬/pmc1⌬/vcx1⌬) resulted in synthetic lethality. Taken together, these results indicate that the Golgi apparatus of yeast plays a significant role in cellular Ca 2ϩ homeostasis through a Pmr1p-dependent mechanism when vacuolar Ca 2ϩ storage is compromised. We also found that a pmr1⌬ strain with normal vacuolar function is sensitive to high levels of Ca 2ϩ in the growth medium. Given the fact that the Golgi has not previously been observed to play a role in Ca 2ϩ homeostasis under other growth conditions, Golgi Ca 2ϩ sequestration may only play a significant role in cellular Ca 2ϩ homeostasis when the cytosolic Ca 2ϩ load exceeds the capacity of the vacuolar Ca 2ϩ storage system.
Although our study clearly implicates Pmr1p in the maintenance of Ca 2ϩ homeostasis in vps33⌬ strains, we observed only a 2-fold increase in PMR1 transcription. Although a larger increase might have been expected, it is possible that PMR1 expression is regulated primarily at a post-transcriptional level. In this way a significant increase in Pmr1p activity could occur without a concomitant increase in mRNA abundance (or protein abundance if the regulation is exerted at a post-translational level). Alternatively, Pmr1p may be present and active under all conditions, but the vacuolar Ca 2ϩ uptake system may sequester cytosolic Ca 2ϩ more efficiently than the Golgi apparatus under all but the most severe conditions. This could occur, for example, if the vacuolar transporters were activated at a lower cytosolic Ca 2ϩ concentration than Pmr1p. By either mechanism, a high level of Golgi Ca 2ϩ storage would not be observed under most growth conditions that did not subject the cells to high Ca 2ϩ stress. Such an overlapping hierarchy of transporter activation to control Ca 2ϩ homeostasis (either at the level of synthesis or function) would be consistent with the observations obtained in the current study. Such a mechanism would also explain why Pmr1p was not attributed a role in the maintenance of cellular Ca 2ϩ homeostasis in previous studies.
Other results obtained in this study are also consistent with a hierarchical control of Ca 2ϩ homeostasis. First, we found that strains carrying the vps33⌬ mutation exhibit a 2-3-fold higher basal level of cytosolic Ca 2ϩ when incubated in a medium containing only 10 M Ca 2ϩ (Fig. 2). This finding provides evidence that Pmr1p function may be activated at a higher cytosolic Ca 2ϩ concentration than the vacuolar Ca 2ϩ transporters. Because our results suggest that the secondary system utilizing Pmr1p plays a larger role in Ca 2ϩ homeostasis in strains carrying the vps33⌬ mutation, it would be expected that the basal cytosolic Ca 2ϩ would be maintained near the concentration that activates this transporter. We also found that strains carrying the vps33⌬ mutation exhibited a severe defect in the maintenance of cytosolic Ca 2ϩ homeostasis when exposed to 50 mM extracellular CaCl 2 . Under these conditions, we found that the cytosolic Ca 2ϩ concentration of the vps33⌬ strain quickly rose to 1.75 M, a level that was 6-fold higher than the wild type strain. Furthermore, the rate of recovery was slower, and the new steady-state level that was reached was also much higher than the control strain. Again, these results suggest that this secondary system of Ca 2ϩ sequestration cannot remove excess Ca 2ϩ from the cytosol as quickly as the vacuolar system. Despite these limitations, this system remains capable of maintaining intracellular Ca 2ϩ homeostasis (at least to the extent required to maintain growth) in strains carrying the vps33⌬ mutation when challenged by environmental concentrations as high as 50 mM Ca 2ϩ (see Fig. 9).
Previous studies have shown that strains carrying vps33 mutations mislocalize the vacuolar membrane protein alkaline phosphatase to the cell surface (24 -26). Unfortunately, neither the extent of the mislocalization of other vacuolar membrane proteins nor the composition of vesicles that accumulate in the vps33⌬ strain has been characterized further. Nevertheless, it is possible that vacuolar Ca 2ϩ transporters may also be mislocalized to the cell surface and thus could potentially contribute to the increased cytosolic Ca 2ϩ levels observed upon exposure to high extracellular Ca 2ϩ . However, we found that the peak cytosolic Ca 2ϩ level was still 5-fold higher than the wild type strain in the vps33⌬/pmc1⌬/vcx1⌬ strain. This indicates that the mislocalization of the Pmc1p and Vcx1p transporters to the plasma membrane is not responsible for most of the elevated cytosolic Ca 2ϩ observed in strains carrying the vps33⌬ mutation. Two additional lines of evidence suggest that the vps33⌬ mutation does not significantly alter the general permeability of the plasma membrane. First, the steady-state cellular concentrations of two other cations, K ϩ and Na ϩ , were unaffected by the vps33⌬ mutation. In addition, the rate of 45 Ca 2ϩ uptake measured in strains carrying the vps33⌬ mutation was identical to that of the wild type strain. Taken together, these results suggest that the vps33⌬ mutation does not significantly alter the permeability of the plasma membrane in the vps33⌬ strain. As discussed above, it is more likely that the higher peak in cytosolic Ca 2ϩ is caused by a reduced capacity to sequester the Ca 2ϩ into other cellular compartments rapidly.
Several studies have reported that the loss of calcineurin function increases the total cellular Ca 2ϩ level (7,14,(37)(38)(39). Our finding that the vps33⌬ strain shows an increased sensitivity to CsA on standard YPD medium is also consistent with the model that this secondary system of Ca 2ϩ sequestration is not capable of transporting Ca 2ϩ from the cytosol into intracellular compartments as efficiently as the wild type strain. In addition, it has been shown that the induction of PMR1 expression is prevented by the immunosuppressive drug FK506 (which functions to inhibit calcineurin activation in a manner analogous to CsA) (14). Thus, the combined effects of increased cellular Ca 2ϩ uptake and lack of PMR1 induction could account for the increased CsA sensitivity that was observed in the vps33⌬ strains.
The results of this study provide evidence that the Golgi apparatus plays a significant role in the maintenance of cellular Ca 2ϩ homeostasis under conditions where the accumulation of cytosolic Ca 2ϩ exceeds the capacity of the vacuole. This suggests that the vacuolar storage system that normally mediates the bulk of Ca 2ϩ homeostasis in yeast may have been superimposed upon another system that is functionally related to the Ca 2ϩ storage and signaling system found within the secretory pathway of mammalian cells. Further studies are required to determine whether other intracellular organelles of yeast (such as mitochondria) also participate in the maintenance of cellular Ca 2ϩ homeostasis under conditions of extreme Ca 2ϩ stress.