Vtc5, a Novel Subunit of the Vacuolar Transporter Chaperone Complex, Regulates Polyphosphate Synthesis and Phosphate Homeostasis in Yeast*

SPX domains control phosphate homeostasis in eukaryotes. Ten genes in yeast encode SPX-containing proteins, among which YDR089W is the only one of unknown function. Here, we show that YDR089W encodes a novel subunit of the vacuole transporter chaperone (VTC) complex that produces inorganic polyphosphate (polyP). The polyP synthesis transfers inorganic phosphate (Pi) from the cytosol into the acidocalcisome- and lysosome-related vacuoles of yeast, where it can be released again. It was therefore proposed for buffer changes in cytosolic Pi concentration (Thomas, M. R., and O'Shea, E. K. (2005) Proc. Natl. Acad. Sci. U.S.A. 102, 9565–9570). Vtc5 physically interacts with the VTC complex and accelerates the accumulation of polyP synthesized by it. Deletion of VTC5 reduces polyP accumulation in vivo and in vitro. Its overexpression hyperactivates polyP production and triggers the phosphate starvation response via the PHO pathway. Because this Vtc5-induced starvation response can be reverted by shutting down polyP synthesis genetically or pharmacologically, we propose that polyP synthesis rather than Vtc5 itself is a regulator of the PHO pathway. Our observations suggest that polyP synthesis not only serves to establish a buffer for transient drops in cytosolic Pi levels but that it can actively decrease or increase the steady state of cytosolic Pi.

SPX domains control phosphate homeostasis in eukaryotes. Ten genes in yeast encode SPX-containing proteins, among which YDR089W is the only one of unknown function. Here, we show that YDR089W encodes a novel subunit of the vacuole transporter chaperone (VTC) complex that produces inorganic polyphosphate (polyP). The polyP synthesis transfers inorganic phosphate (P i ) from the cytosol into the acidocalcisome-and lysosome-related vacuoles of yeast, where it can be released again. It was therefore proposed for buffer changes in cytosolic P i concentration ( Its overexpression hyperactivates polyP production and triggers the phosphate starvation response via the PHO pathway. Because this Vtc5-induced starvation response can be reverted by shutting down polyP synthesis genetically or pharmacologically, we propose that polyP synthesis rather than Vtc5 itself is a regulator of the PHO pathway. Our observations suggest that polyP synthesis not only serves to establish a buffer for transient drops in cytosolic P i levels but that it can actively decrease or increase the steady state of cytosolic P i . Phosphate is a limiting factor for the growth of living organisms. It is mainly taken up by the cells as P i and incorporated in biological molecules, including ATP, nucleic acids, and phospholipids. P i plays a major role in the regulation of biochemical pathways through phosphorylation, pyrophosphorylation, and polyphosphorylation. Its concentration affects the free energy liberated by the hydrolysis of ATP and other nucleoside di-and triphosphates. Therefore, P i homeostasis (import, usage, storage, and export) is regulated (1,2), and P i can be accumulated and stored as a polymer of P i units called polyphosphate (polyP). 3 One mode of regulation operates on the transcrip-tional level. The PHO pathway, which is a P i -dependent transcriptional program in yeast, has been extensively characterized (2). Transcription of the PHO genes is regulated by the transcription factors Pho4 and Pho2. Phosphorylated Pho2 interacts with Pho4 to induce expression of the PHO genes. Pho2 is a target of Cdc28 kinase in vitro suggesting a coordination between cell cycle progression and nutrient availability (3). Pho4 is itself regulated by the cyclin-dependent kinase Pho85 and its cyclin Pho80. Under P i -rich conditions, the Pho80-Pho85 complex phosphorylates Pho4, thereby restricting the localization of the transcription factor to the cytosol (4,5). Under P i limitation, Pho80/Pho85 is inhibited, and the nonphosphorylated Pho4 is imported into the nucleus, allowing transcription of PHO genes. Pho80/85 is regulated by the cyclin-dependent kinase inhibitor Pho81. Inhibition of Pho80/85 by Pho81 is facilitated by the inositol pyrophosphate 1-IP 7 , which is produced by Vip1 (6). However, a recent report found that induction of the PHO pathway shows only a minor delay in vip1⌬ mutants (7). In plants and yeast, many proteins involved in P i homeostasis contain SPX domains, which have been proposed to regulate P i metabolism by serving as receptors to inositol pyrophosphates (8).
The genome of Saccharomyces cerevisiae contains 10 open reading frames encoding SPX domain proteins (9), named after Syg1, Pho81, XPR1. Eight of these SPX proteins have functions related to P i metabolism as follows: Pho87 and Pho90 are plasma membrane P i importers (10,11); Pho91 is a vacuolar P i exporter (12); Pho81 is a cyclin-dependent kinase inhibitor (13); Gde1 is a glycerophosphocholine phosphodiesterase that hydrolyzes glycerophosphocholine, thereby releasing choline and glycerol 3-phosphate that can be used as a P i source (14). Three other SPX domain containing proteins (Vtc2, Vtc3, and Vtc4) are part of a large complex that produces polyP, the Vacuole Transporter Chaperone (VTC) complex (15)(16)(17)(18). AlthoughVtc4 holds the catalytic activity, Vtc2 and Vtc3 are assumed to be accessory subunits with a regulatory function (19). In line with this assumption, VTC exists in two isoforms, composed of Vtc4/Vtc3/Vtc1 or Vtc4/Vtc2/Vtc1. Vtc1 is a small protein that contributes only three transmembrane domains to the complex, which resemble the transmembrane domains of the larger VTC proteins but lack their large hydro-philic N-terminal extensions. VTC is located in the membrane of the acidocalcisome-and lysosome-like vacuole and in a peripheral location presumed to be the endoplasmic reticulum (ER), with Vtc4/Vtc3/Vtc1 being more enriched in the vacuole and Vtc4/Vtc2/Vtc1 being more prominent in the ER (19,20). VTC couples synthesis of polyP to its translocation across the membrane, thereby sequestering polyP inside the lumen of the vacuole and avoiding its accumulation in the cytosol, where polyP can be toxic (21). PolyP is degraded when cells are shifted from P i -rich media to P i -starvation media (22). Therefore, polyP can influence the kinetics of changes in cytosolic P i (1,22). Finally, Syg1 is an SPX domain protein of unknown function. It is a homolog of the mammalian phosphate exporter Xpr1. Because both proteins localize to the plasma membrane and interact with ␤ subunits of G-proteins (23)(24)(25), they may have similar functions in exporting P i .
In this work, we focused on a yeast gene coding an SPX domain-containing protein of unknown function, YDR089W. We identify its product as a novel subunit of the VTC complex.
Genetic Manipulations-Yeast transformations were carried out using the lithium acetate method (25). Gene deletions and tagging were performed following established procedures (28 -30). Strains and plasmids used in this study are listed in Tables 1 and 2, respectively. Oligonucleotides used are listed in Table 3.
The plasmid used to overexpress VTC5, pRS416 GPD1-Vtc5, was obtained by cloning the VTC5 open reading frame between the XhoI and SpeI restriction sites of the pRS416 GPD1 plasmid (Euroscarf). To obtain pRS416 P PHO4 -PHO4-GFP, GFP from the plasmid pKT127 was first amplified by PCR using oligo-24 and oligo-25 and inserted into pRS416 between the EcoRI and HindIII restriction sites, yielding pRS416-GFP. Then, P PHO4 - PHO4 was amplified from genomic DNA using oligo-26 and oligo-27 and ligated between the SacI and EcoRI restriction sites of pRS416-GFP, yielding pRS416 P PHO4 -PHO4-GFP.
Vacuole Isolation-Yeast cells were grown overnight in 2-liter flasks containing 1 liter of YPD to an OD 600 nm of 0.5-1.5. Cells from l liter of culture were collected by centrifugation (1 min, 4000 ϫ g, room temperature, JLA-10.500 rotor) and resuspended in 50 ml of 30 mM Tris-HCl buffer, pH 8.9, containing 10 mM DTT. Suspensions were incubated for 5 min at 30°C and centrifuged as before. Cells were incubated in 15 ml of spheroplasting buffer (50 mM potassium phosphate, pH 7.5, 600 mM sorbitol in YPD with 0.2% D-glucose, and 3600 units ml Ϫ1 lyticase) for 25 min at 30°C. Spheroplasts were collected by centrifugation (2 min, 1500 ϫ g, 4°C), resuspended in PS buffer (10 mM PIPES-KOH, pH 6.8, 200 mM sorbitol) containing 15% Ficoll 400, and 80 g of DEAE-dextran was added to each tube. Samples were incubated (30°C, 2 min) and fractionated on a discontinuous Ficoll 400 gradient (90 min, 150,000 ϫ g, 2°C). Vacuoles were collected from the 4-0% Ficoll 400 interface. Protein concentration in the samples was determined using the Bradford reagent, using BSA as a standard.
Immunoprecipitations-200 g of isolated vacuoles were diluted into 1 ml of PS buffer and sedimented by centrifugation (7000 ϫ g, 7 min, 4°C). Pellets were resuspended in 500 l of solubilization buffer (10 mM PIPES, pH 6.8, 200 mM sorbitol, 5% glycerol, 0.5% Triton X-100). Lysis was allowed to proceed for 20 min at 4°C with gentle shaking. Insoluble material was removed by centrifugation (20,000 ϫ g, 20 min, 4°C), and supernatants were transferred to a fresh tube. Small aliquots of the supernatants were withdrawn to serve as loading controls. Anti-HA monoclonal antibodies (3 l of serum from mouse ascites, Covance) and 30 l of protein G-agarose beads, preequilibrated in solubilization buffer, were added to the remaining fraction of the solubilized samples. Tubes were placed on a rotating shaker (1 h, 4°C). Beads were washed with solubilization buffer, and immunoprecipitated proteins were eluted by incubating the beads in 2ϫ SDS-PAGE sample loading buffer for 10 min at 65°C.
Polyphosphate Synthesis Assay in Vitro-PolyP synthesis by isolated yeast vacuoles was measured by DAPI-polyP fluorescence as described previously (21). Briefly, reaction master mix was prepared in the reaction buffer (10 mM PIPES/KOH, pH 6.8, 150 mM KCl, 0.5 mM MnCl 2 , 200 mM sorbitol) containing an ATP-regenerating system (1 mM ATP-MgCl 2 , 40 mM creatine phosphate and 0.25 mg/ml creatine kinase). The reactions were started by adding isolated vacuoles to a final protein concentration of 0.005 mg/ml, and the samples were incubated at 27°C. At indicated time points, an aliquot was withdrawn and mixed with 2 volumes of stop solution (10 mM PIPES/KOH, pH 6.8,  Vtc5 Is a Novel Subunit of the Polyphosphate Polymerase X-100, and 15 M DAPI). Fluorescence of the DAPI-polyP complex was measured in a black 96-well plate using ex ϭ 415 nm and em ϭ 550 nm (cutoff ϭ 530 nm) at 27°C. A calibration curve was produced using synthetic polyP-60 as a standard. Determination of Polyphosphate Levels in Living Cells-Cells were grown overnight in SC medium until an OD 600 nm ϳ1.0 was reached. One equivalent OD of cells was collected by centrifugation (11,000 ϫ g, 1 min, room temperature). PolyP was extracted by adding 50 l of 2 M sulfuric acid and vortexing for 10 s. The samples were neutralized by adding 50 l of 2 M NaOH and vortexing for 10 s. The extracts were supplemented with 100 l of 1 M Tris malate, pH 7.4. PolyP was purified through PCR DNA purification columns (Qiagen PCR purification kit). 600 l of binding buffer was added to the extracts, and the samples were loaded onto the columns. After two washes with washing buffer (10 mM Tris-HCl, pH 7.0, 50% ethanol, 10 mM NaCl, 0.3 mM EDTA), polyP was eluted with 110 l of the manufacturer's elution buffer (10 mM Tris-HCl, pH 8.5). The amount of orthophosphate released after incubation with recombinant Ppx1 (0.01 g of Ppx1 in 5 mM Tris, pH 7.0, 0.15 mM MgCl 2 ) was determined by the Malachite Green assay.
Determination of PolyP Chain Lengths-PolyP was purified from logarithmically growing cells by glass bead lysis and phenol extraction as described (31) and dissolved in 10 mM Tris-HCl, pH 8.0, 0.5 mM EDTA. RNA concentration was determined by measuring absorbance at 260 nm. The aliquots containing 2 g of RNA were treated with RNase A (0.04 mg/ml, 30 min at 37°C), followed by proteinase K (0.1 mg/ml, 30 min at 50°C). The samples were fractionated on 25% polyacrylamide gel, and polyP was visualized by negative DAPI staining (31,32).
Quantification of Pho5 Activity-Cells were grown overnight in YPD to logarithmic phase (OD 600 nm ϭ 1-2). 50 l of the cell suspension was mixed with 200 l of 20 mM p-nitrophenol in 0.1 mM sodium acetate, pH 4.2. Samples were incubated at 30°C for 15 min. The reactions were stopped by addition of 200 l of a 10% trichloroacetic acid solution. 400 l of 2 M sodium carbonate was added, and insoluble material was removed by centrifugation(1min,11,000ϫg,roomtemperature).Theabsorbance of the supernatant was measured at 420 nm. Pho5 activity was expressed as OD 420 nm /OD 600 nm ϫ 1000.
Fluorescence-activated Cell Sorting-Cells were grown in YPD medium to logarithmic phase. Just before measurement, 50 l of the cell suspension was added to 950 l of TBS. Samples were vortexed and immediately analyzed using an LSRII flow cytometer (BD Biosciences). 20,000 events were recorded for each sample. The mean fluorescence value was used to estimate protein expression levels.
Microscopy-Where indicated, vacuoles were stained with the vital dye FM4-64. Cells were grown overnight to an OD 600 nm of 0.6 -1.0 and diluted to OD 600 nm 0.2 in fresh medium. Vacuoles were stained with 10 M FM4-64 during 1 h at 30°C. Cells were washed twice with dye-free medium and chased in fresh dye-free medium for 1.5 h at 30°C. Microscopy was done on a LEICA DMI6000B inverted microscope equipped with a Hamamatsu digital CCD camera (ORCA-R2 C10600-10B), an X-Cite series 120Q UV lamp, and a Leica 100 ϫ 1.4 NA lens.
Protein Extraction-Two OD 600 nm units of cells were harvested by centrifugation (11,000 ϫ g, 1 min, room temperature). Pellets were resuspended in 200 l of 0.1 M NaOH, and the suspensions were incubated for 5 min at 25°C. Cells were collected as before and resuspended in 50 l of 2ϫ loading sample buffer. Samples were heated at 95°C for 5 min, and insoluble material was removed by centrifugation (11,000 ϫ g, 30 s, room temperature). Supernatants were loaded on SDS-polyacrylamide gels.
Statistics-Tests for significance of observed differences were performed by an unpaired Student's t test. Error bars represent the standard deviation. Differences with p values higher than 0.1 were considered as not significant.

Vtc5 Is a Transmembrane Protein Localized in the Vacuolar
Membrane-No significant sequence homology of YDR089W to other yeast proteins was found outside the SPX domain. Using secondary structure predictions (TMHMM) software, we noticed that YDR089W topologically resembled Vtc2, Vtc3, and Vtc4. The YDR089W ORF is of similar length as VTC2, VTC3, and VTC4, and like these genes, it encodes an SPX domain at its N terminus and three C-terminal transmembrane regions (Fig. 1A). Therefore, we explored its possible relationship to polyP metabolism and refer to YDR089W as VTC5. To determine the localization of Vtc5, the protein was genomically tagged with yeGFP at its C terminus in protease-deficient strains (BJ3505) and protease-proficient strains (DKY6281). Vtc5-yeGFP yielded a relatively weak signal around the vacuole rim in BJ3505 cells. No fluorescence signal was detected in other parts of the cell. In a protease-proficient strain (DKY6281), GFP localized mainly to the vacuole lumen (Fig.  1B). This suggests that the C-terminal end of Vtc5-yeGFP is sensitive to cleavage by vacuolar proteases and faces the vacuole lumen. When Vtc5 was tagged at the N terminus, GFP was detected on the vacuolar membrane in a protease-deficient strain as well as in a protease-proficient strain (Fig. 1B). This suggests that the N-terminal domain of Vtc5 is inaccessible to vacuolar proteases. Thus, Vtc5 is likely to be a trans-membrane protein with its C terminus in the vacuole lumen and its N terminus facing the cytosol, resembling the topology of Vtc2-4 (17).
Vtc5 Physically Interacts with the VTC Complex-A possible interaction of Vtc5 with the VTC complex was tested by immunoprecipitation. Vacuoles were isolated from a strain expressing HA-tagged Vtc5 under control of the ADH1 promoter, solubilized in Triton X-100, and immunoprecipitated with antibodies to the HA tag ( Fig. 2A). SDS-PAGE and Western blotting revealed that Vtc3, Vtc4, and Vtc1 co-precipitated with Vtc5-HA but not with untagged Vtc5. Other vacuolar proteins, such as Vam3 and Ypt7 and the peripherally associated Sec17, were not detected in the Vtc5-HA precipitate with comparable efficiency. To determine whether Vtc5 associates preferentially with the Vtc2-or Vtc3-containing isoform of VTC (19), we immunoprecipitated Vtc2-HA and Vtc3-HA from a strain expressing Vtc5-GFP. All three tagged proteins were expressed Vtc5 Is a Novel Subunit of the Polyphosphate Polymerase OCTOBER 14, 2016 • VOLUME 291 • NUMBER 42 under the control of their endogenous promoters. Vtc5 was found associated with both proteins (Fig. 2B). For comparison, we performed the same experiment in strains carrying Vtc1-HA or Vtc4-HA, which are found in both VTC isoforms (Fig. 2, C and D). In all cases, the immunoprecipitates enriched Vtc5-GFP with the bait protein. This suggests that Vtc5 can associate with both Vtc2-and Vtc3-containing VTC complexes.
Next, we tested whether the abundance of Vtc5 could alter the interaction between other VTC subunits. We deleted or overexpressed Vtc5 in a Vtc4-HA strain, isolated vacuoles, and immunoprecipitated Vtc4-HA. The fractions of Vtc3 associated with Vtc4 were similar in both cases (Fig. 2E). We also measured the amount of Vtc5 interacting with Vtc4 in a strain expressing genomic levels of Vtc5 and in a strain overexpressing Vtc5 under the control of the strong GPD1 promoter. The amount of Vtc5 associated with Vtc4-HA increased as the total amount of Vtc5 in the vacuole membrane increased (Fig. 2F). Because the vacuolar amounts of Vtc4-HA were not affected by VTC5 overexpression, this suggests that not all VTC complexes on the vacuoles are saturated with Vtc5 and that Vtc5 association with the complex might regulate VTC activity. This notion is supported by mass spectrometry data, which found Vtc5 to be 20ϫ less abundant than Vtc1-4 (33).
Vtc5 Is Required for Normal Expression Levels of the VTC Complex-We asked whether Vtc5 is required for the stability of the VTC complex. Vacuoles were isolated from a strain deleted for VTC5, and the levels of the other VTC subunits were determined by Western blotting. Deletion of VTC5 decreased the levels of Vtc2, Vtc3, and Vtc4 and of the vacuolar alkaline phosphatase Pho8 by around 30% (Fig. 3, A-C) when compared with the levels of the vacuolar marker proteins Vam3 and Nyv1. No significant reduction of Vtc4 was detected in whole-cell extracts, suggesting that absence of Vtc5 interferes with sorting of the VTC complex to vacuoles. Overexpression of Vtc5 from the GPD1 promoter did not change the levels of other VTC subunits or of Pho8. We also analyzed the levels of VTC5 on vacuoles from vtc2⌬, vtc3⌬, and vtc4⌬ cells but observed no significant changes (Fig. 3D). Thus, expression and sorting of Vtc5 to vacuoles do not depend on the other constituents of the VTC complex, but the presence of Vtc5 increases the abundance of the other VTC subunits on vacuoles.
Vtc5 Stimulates the Accumulation of PolyP Synthesized by the VTC Complex-To test the role of Vtc5 in polyP accumulation in vivo, yeast cells were grown in SC medium to exponential phase, and polyP was extracted, purified, and quantified (34). vtc5⌬ cells showed only 20% of the polyP content of wild-type cells (Fig. 4A). This phenotype was reversed by introducing a centromeric plasmid encoding VTC5 under control of the GPD1 promoter. Overexpressing VTC5 or GFP-VTC5 from the GPD1 promoter increased the polyP content of wild-type cells 3.5-fold. This increase in polyP content depended on the catalytic activity of Vtc4 because it was not observed upon overexpressing VTC5 in a vtc4 R264A mutant, which leaves the VTC complex physically intact and correctly sorted but eliminates the catalytic activity of Vtc4 (Fig. 4B) (19). We determined the length of the polyP chains produced by the vtc5 mutants by PAGE and negative staining with DAPI (Fig. 4C). Yeast cells synthesize polyP chains of up to several hundred residues. Chains that differ by a single unit can be resolved up to a chain length of around 40 units, forming a ladder of individual bands. Longer chains run as a smear. Synthetic polyP of an average chain length of 60 units was used as a marker. Deletion or overexpression of Vtc5 had no impact on the length distribution of cellular polyP. This suggests that Vtc5 does not affect the processivity of the VTC complex nor does it seem to influence turnover of these chains by vacuolar endo-or exo-polyphosphatases (35)(36)(37).
PolyP synthesis can be assayed in vitro using isolated organelles (21). The system recapitulates the stimulation of polyP accumulation by inositol pyrophosphates, which is observed in vivo (31,38). The inositol pyrophosphate 5-IP 7 is a potent agonist and accelerates polyP accumulation by WT vacuoles over 50-fold (8). We used this in vitro system to test the influence of Vtc5 on polyP synthesis. Neither deletion nor overexpression of Vtc5 abolished the stimulation of polyP synthesis by 5-IP 7 (Fig.  5, A and C). However, Vtc5 levels influenced the rates of polyP FIGURE 1. Localization and topology of Vtc5. A, schematic representation of the predicted structural features of Vtc5 and comparison with previously known VTC subunits. Three transmembrane helices were predicted using the TMHMM server. B, Vtc5-GFP and GFP-Vtc5 localization was determined by confocal fluorescence microscopy in a pep4⌬ strain (BJ3505) and in a strain expressing PEP4 (DKY6281). Cells were cultivated in YPD to logarithmic phase, and vacuoles were stained with the vital dye FM4-64. The N-terminally tagged version of Vtc5 (GFP-Vtc5) was expressed from the genomically integrated construct under control of the ADH1 promoter. The ADH1 promoter results in moderate overexpression. This explains the fusion of the vacuoles into a single larger organelle, which occurs as a consequence of enhanced polyP production (52). DIC, differential interference contrast. Scale bar, 5 m.
accumulation. vtc5⌬ vacuoles showed 50% reduced activity in the presence of 5-IP 7 . In the absence of 5-IP 7 , wild-type vacuoles showed only very low activity (5 pmol/g/min), which dropped to undetectable levels in vtc5⌬ (Ͻ1 pmol/g/min) (Fig. 5, B and C). Note that the activities shown have been normalized to the levels of Vtc4, which means that the observed effects cannot be explained by the reduced abundance of this catalytic subunit on vtc5⌬ vacuoles. Overexpression of Vtc5 increased the rate of polyP accumulation in the absence of 5-IP 7 over 10-fold, whereas the 5-IP 7 -stimulated activity increased only 1.2-fold. The effect of Vtc5 on polyP synthesis did not result from a change in the stability of the complex. Isolated vacuoles from strains deleted for VTC5 or overexpressing Vtc5 were incubated under the conditions used for the in vitro polyP assay, and proteins were TCA-precipitated at different time points before SDS-PAGE analysis. The levels of Vtc4 and Vtc3 were then quantified by Western blotting (Fig. 5D). Degradation of the two subunits could not be detected in either of the strains tested. These results suggest that Vtc5 can augment the catalytic activity of the VTC complex by a mechanism independent of 5-IP 7 .

Vtc5-mediated Stimulation of PolyP Synthesis
Activates the PHO Pathway-Because overexpression of VTC5 allowed us to uncouple VTC activity from its inositol pyrophosphate-mediated control by intracellular P i levels, we exploited this feature to study the impact of polyP synthesis on cytosolic P i homeostasis. The transcriptional program of the PHO pathway is induced by a decline in intracellular P i levels (39). This induction can be read out via the production of Pho5, which is the major PHO pathway-controlled acid phosphatase that is secreted into the medium (40,41). Deletion of VTC5 led to a low but reproducible reduction of secreted phosphatase activity by 20% (Fig. 6A). By contrast, overexpression of VTC5 increased it 2-fold, indicating activation of the PHO pathway. As controls, we used a pho80⌬ strain, in which the PHO pathway is constantly active, and the pho4⌬ and pho81⌬ strains, in which the PHO pathway is constitutively inactive. We also measured the expression level of Pho84, a high affinity P i importer that is strongly induced under low P i conditions (1,42,43). vtc5⌬ mutants carrying Pho84-GFP were grown to exponential phase in YPD, i.e. under P i -rich conditions, and Pho84-GFP fluorescence intensity was measured by fluorescence-activated T a g g e d T a g g e d U n t a g g e d U n t a g g e d Ypt7 B . E, Vtc5 does not influence the interaction between other VTC subunits. VTC5 was deleted or overexpressed from the GPD1 promoter in BJ3505 cells carrying HA-tagged or untagged Vtc4. Vacuoles were isolated and subjected to immunoprecipitation as in A, using antibodies directed against GFP or against the HA tag. F, influence of Vtc5 expression level on the Vtc4-Vtc5 interaction. VTC5 was expressed in BJ3505 VTC4-HA under the control of its endogenous promoter (Vtc5-GFP) or the strong GPD1 promoter (GPD1-GFP-Vtc5). Vtc4-HA was pulled down from isolated vacuoles as in C, and adsorbed Vtc5-GFP was detected by Western blotting for its GFP tag. The migration of relevant molecular mass markers (size in kDa) is indicated on the right-hand side of the panels.

U n t a g g e d V t c 2 -H A V t c 3 -H A U n t a g g e d V t c 2 -H A V t c 3 -H
cell sorting (FACS). Deletion of VTC5 reduced Pho84-GFP by 70%, whereas its overexpression increased fluorescence intensity 10-fold (Fig. 6B). To verify that these observations resulted from increased transcription of the PHO genes and not from defective degradation of Pho84, we measured activation of the Pho84 promoter via a GFP reporter (PHO84pr-GFP) (1), using FACS (Fig. 6C) and Western blotting (Fig. 6D). In wild-type cells grown in P i -rich medium, GFP expression was low. Deletion of VTC4 or VTC5 reduced expression of the reporter to 30 and 9% of the wild type, respectively. This is equivalent to the levels observed in a pho81⌬ strain, in which the PHO pathway is constitutively repressed. Overexpression of VTC5 increased GFP levels similarly as deletion of PHO80, which renders the PHO pathway constitutively active. Thus, up-or down-regulation of Vtc5 levels activates or inactivates the PHO pathway, respectively. Given that the PHO genes, including the P i importer Pho84, are up-regulated in cells overexpressing Vtc5, the increased polyP levels in this mutant might result from higher abundance of PHO proteins and higher import of P i into the cytosol. We tested this hypothesis in pho81⌬ and pho4⌬ cells, which cannot up-regulate the PHO genes. Overexpression of Vtc5 in these mutants did not induce Pho5, but it did induce polyP overaccumulation (Fig. 6E). This suggests that Vtc5 regulates polyP synthesis independently of the PHO pathway.
Vtc5-induced Activation of the PHO Pathway Depends on PolyP Accumulation-A GFP-tagged version of the P i -sensitive transcription factor Pho4 (Pho4-GFP) shifts to the cytosol in P i -rich conditions and accumulates in the nucleus under P i limitation, providing a rapidly responding bona fide readout for cytosolic P i concentration (1). In line with this, pho81⌬ cells, in which the phosphate starvation program is constitutively inactive, showed a strictly cytosolic staining in P i -rich as well as P i -limiting SC medium (Fig. 7). Conversely, pho80⌬ cells, in which the phosphate starvation program is constitutively active, carried Pho4-GFP in the nucleus even on P i -rich media. Wild-type cells showed an increase in nuclear Pho4-GFP localization from 20 to 100% when shifted from P i -rich to P i -limiting medium for 15 min, indicating that cytosolic P i concentration

FIGURE 3. Levels of VTC subunits in vtc mutants.
A, influence of Vtc5 on proteins on isolated vacuoles. 30 g of vacuolar protein from BY4741, vtc5⌬, and BY4741 P GPD1 -Vtc5 were analyzed by SDS-PAGE and Western blotting with antibodies against the indicated proteins or the HA tag. B, influence of Vtc5 on proteins in whole-cell extracts. Whole-cell extracts were prepared from cells expressing Vtc4 from its endogenous promoter with an HA tag, grown in YPD to logarithmic phase. Proteins were analyzed as in A. Note that we used Vtc4-HA in this case because Vtc4 is difficult to detect in whole-cell extracts with our antibodies. C, quantification of the Vtc4 levels from A (white bars) and B (black bars) from three independent preparations each. n.s., not significant. D, influence of Vtc2, Vtc3, and Vtc4 on the level of Vtc5. Vacuoles were isolated from the indicated BY4741 strains expressing Vtc5 with or without a GFP tag. Vacuolar proteins were analyzed as in A. Vam3 was used as a loading control. The migration of relevant molecular mass markers (size in kDa) is indicated on the right-hand side of the panels. . Cells were grown in SC medium, and polyP was quantified as in A; n ϭ 3. ***, p Ͻ 0.001. C, polyP chain length distribution. PolyP was extracted from the indicated BY4741 strains, which had been grown logarithmically in YPD. PolyP was fractionated on a 25% polyacrylamide gel and visualized by negative staining with DAPI. Equal amounts of extracted polyP were loaded for each strain. An additional sample consisting of a three times higher amount of polyP was loaded for the vtc5⌬ strain (vtc5⌬ 3x). Synthetic polyP of an average chain length of 60 phosphate units was loaded as a length marker.

Vtc5 Is a Novel Subunit of the Polyphosphate Polymerase
had dropped, and the starvation response had been initiated. vtc5⌬ and vtc4⌬ cells behaved similarly but showed nuclear Pho4-GFP in less than 10% of the cells on P i -rich medium. By contrast, among cells overexpressing Vtc5, more than 90% showed nuclear Pho4-GFP under P i -rich as well as under P ilimited conditions. This nuclear accumulation could not be reverted by addition of 10 mM P i , although the same treatment led to rapid redistribution of Pho4-GFP to the cytosol in wildtype, vtc4⌬, and vtc5⌬ cells. Thus, similar to a pho80⌬ mutation, VTC5-overexpression constitutively activates the P i starvation response, suggesting that it leads to a constitutive depletion of cytosolic P i . Next, we asked whether this change in Pho4-GFP localization could be ascribed to altered polyP synthesis or whether it might be due to other effects of Vtc5, such as a direct interaction with cytosolic P i signaling components. To this end, we manipulated polyP synthesis pharmacologically and also through mutations that leave the VTC complex intact but increase or decrease its catalytic activity (8,19). VTC5 was overexpressed in vtc4⌬ strains that were reconstituted with plasmids expressing the wild-type allele Vtc4 WT or a catalytically inactive vtc4 R264A allele. In addition, we used a hyperactive VTC variant that carries a point mutation in the SPX domains of Vtc3 and Vtc4 (vtc3 K126A /vtc4 K129A ) (8). As observed before, VTC5 overexpression increased Pho5 activity (Fig. 8A). This effect was not observed upon VTC5 overexpression in vtc4⌬ or vtc4 R264A cells. Conversely, if polyP synthesis was raised independently of Vtc5 by using the vtc3 K126A /vtc4 K129A alleles, Pho5 activity rose to a similar degree as in VTC5-overexpressing wild-type cells.
The rapidly responding Pho4-GFP allowed us to test the effect of an acute pharmacological shutdown of polyP synthesis. Because polyP synthesis and translocation are driven by the V-ATPase-dependent electrochemical gradient across the vacuolar membrane (21), disruption of this gradient immediately arrests polyP synthesis. Within less than 15 min, addition of the V-ATPase pump inhibitor concanamycin A (ConA) to wildtype cells growing on SC-P i led to a re-localization of Pho4-GFP from the nucleus to the cytosol, suggesting that a sudden arrest in polyP synthesis rapidly increases cytosolic P i concentration. Pho4-GFP did not re-localize in P i -limited vtc4⌬ cells. Furthermore, ConA did not change the predominantly cytosolic localization of Pho4-GFP in wild-type cells growing in P i -rich SC (Fig. 8B), and it had no effect on the localization of Pho4-GFP in pho80⌬ and pho81⌬ mutants, indicating that the drug itself does not promote changes in Pho4-GFP localization. However, ConA abolished the nuclear accumulation of Pho4-GFP in VTC5-overexpressing cells.
To better understand the link between polyP accumulation and activation of the PHO pathway, we targeted an exopolyphosphatase (vt-Ppx1) to the vacuole lumen of a strain overexpressing VTC5. This is expected to result in the degradation of polyP (21) and export of released P i to the cytosol. Expression of vt-Ppx1 in a wild-type strain reduced polyP content by 80% but decreased rAPase activity by less than 20% (Fig. 8C). In VTC5 overexpressing cells, vt-PPX1 reduced polyP by Ͼ90%. In these cells, rAPase activity was also reduced by Ͼ75%, bringing the absolute rAPase activity to a level only half as high as in wild-

Vtc5 Is a Novel Subunit of the Polyphosphate Polymerase
type cells. This suggests that polyP production, when followed by uncontrolled degradation in the vacuole, increases cytosolic P i above the normal level and suppresses the PHO pathway.

Discussion
Although the response to phosphate starvation has been well characterized in yeast, the mechanisms of sensing P i and managing P i stores remain poorly understood. PolyP is the main P i reservoir in yeast. The VTC complex, which is mainly localized on the vacuole membrane (15)(16)(17)44), synthesizes polyP from ATP and translocates the nascent polyP chain into the lumen (19,21). Here, we show that the ORF YDR089W encodes a previously unknown subunit of the VTC complex, Vtc5. Vtc5 is not essential for catalytic activity of VTC because a vtc5⌬ strain retains ϳ20% of polyP, and vacuoles isolated from this strain still accumulate polyP, although at a lower rate. Conversely, overexpression of VTC5 increases polyP accumulation in vivo but is not sufficient for it. VTC complexes containing Vtc3 are mainly located on vacuoles, but Vtc2-containing complexes are abundant in the cell periphery, probably in the peripheral ER (20). Even when overexpressed, we found Vtc5 only in the vacuole membrane, and its sorting to the vacuole was independent of the presence of other VTC subunits. This suggests that Vtc5 stimulates polyP synthesis specifically on vacuoles, which is required to accumulate the major polyP stores of the cell. In vitro, VTC5 overexpression accelerates polyP accumulation drastically and partially substitutes for the stimulation of VTC by 5-IP 7 . The abundance of Vtc5 is low compared with the other subunits (40). In line with this, we found that the amount of Vtc5 interacting with the other subunits increases with the increased expression of VTC5 on the vacuole membrane. This suggests that VTC complexes are not saturated with Vtc5 and that Vtc5 may regulate their activity by interacting with them transiently. This invites the consideration that inositol pyrophosphates might regulate polyP synthesis in the context of cytosolic P i homeostasis, whereas Vtc5 might regulate polyP synthesis with respect to other cellular parameters. PolyP synthesis is activated in correlation with P i availability by the binding of inositol pyrophosphates to SPX domains of the VTC complex (8). BecauseVtc2, Vtc3, Vtc4, and Vtc5 carry SPX domains, we tested whether their interactions might be affected by 5-IP 7 (data not shown). This was not the case, although polyP synthesis on the isolated vacuoles used for this assay strongly responded to overexpression of VTC5 and to 5-IP 7 . Neither overexpression nor deletion of VTC5 abolished the stimulation of VTC by 5-IP 7 . Therefore, we presume that association with Vtc5 might represent an independent mechanism for regulat-ing polyP synthesis. We speculate that this might constitute a means to integrate multiple regulatory inputs on the VTC complex. This is relevant because polyP synthesis is not the only determinant of cytosolic P i homeostasis, where the SPX domains and inositol pyrophosphates play a role. In amino acid and metal homeostasis, polyP also constitutes an important polymeric counter-ion to basic amino acids and divalent metal ions (45,46). Basic amino acids can be stored in vacuoles and acidocalcisomes in concentrations of several hundred millimolars. This accumulation is only possible if sufficient polyP is synthesized, probably because polyP is needed to adsorb and osmotically inactivate the amino acids, which otherwise could build up strong osmotic pressure inside the organelle (47). Regulated association of Vtc5 with the other VTC subunits might

Vtc5 Is a Novel Subunit of the Polyphosphate Polymerase
modify polyP synthesis in response to such requirements that are independent of P i levels. In line with this, VTC5 has not been found among genes undergoing strong transcriptional activation upon phosphate starvation, whereas the other four VTC genes are all strongly induced (15). We confirmed the lack of induction of Vtc5 by quantifying the levels of Vtc5-GFP expressed from the native VTC5 promoter and found no change upon transferring the cells to low-P i medium (data not shown). Furthermore, Vtc2, Vtc3, and Vtc4 share a similar central domain, whereas the central domain of Vtc5 is of similar size, and yet its primary sequence and the secondary structure prediction reveal no conservation between this domain and the central domains of Vtc2, Vtc3, and Vtc4. PolyP also serves as a P i source to facilitate rapid DNA synthesis in S-phase, which transiently drains more P i than the cells can take up via their transporters (48). During S-phase, cytosolic P i concentrations are held constant, although the polyP reserves are consumed, which suggests that net polyP synthesis should be down-regulated. This might implicate Vtc5 because the hydrophilic domain that is found between its SPX and transmembrane domains is extensively phosphorylated and, for at least one of these phospho-sites, the cell cycle regulator Cdc28 has been identified as the kinase (49). Furthermore, both polyP synthesis and the cell cycle-regulating proteins Cdc28 and Cdc42 influence vacuolar structure through the vacuolar fusion machinery (50 -52). This, together with the cyclic consumption of polyP (48), can explain the transient decay of vacuoles into multiple smaller fragments during S-phase (53), because vacuole structure is determined by an equilibrium of fusion and fission activities (54 -56). Our working hypothesis, which we will explore in future studies, is that Cdc28 influences vacuolar morphology by regulating VTC and the polyP store produced by it.
Several studies have implicated enzymes of polyP metabolism in the onset and stabilization of the yeast phosphate starvation program, the PHO pathway (1,7,57). Consistent with this, we found that de-regulation of VTC5 expression alters activation of the PHO pathway. vtc5⌬ cells show the transcription factor Pho4 in the cytosol, like cells on P i -rich medium, whereas cells overexpressing VTC5 show Pho4 in the nucleus and hence mimic cells starving for P i , even in P i -rich conditions. The fact that this effect depends on PHO81 provides a genetic argument suggesting that VTC5 acts upstream of PHO81. Straightforward hypotheses to be derived from this observation would be that either cytosolic P i is strongly affected in vtc5 mutants or that the vacuolar polyP that is produced from it serves as an indirect readout of cytosolic P i . A signal about vacuolar polyP levels would then have to be communicated to the PHO pathway. An argument against the latter possibility is that a vtc4⌬ strain, which does not make any polyP, does not show Pho4-GFP constitutively in the nucleus and hence does not induce the PHO pathway. Also the vacuole-targeted exopolyphosphatase vt-Ppx1, which degrades most of the vacuolar polyP store, does not induce the PHO pathway. To the contrary, vt-Ppx1 even represses it when used in a strain overexpressing VTC5. This suggests that the increased production of polyP in VTC5-overexpressing cells, in conjunction with non-controlled degradation of this pool by vt-Ppx1, increases cytosolic P i as compared with a wild-type strain. This argues for the notion that cytosolic P i determines the regulation of the PHO pathway.
Several mechanisms to detect changes in P i have been identified in yeast. The high affinity P i transporter Pho84 was proposed to be a P i transceptor meaning that it would be able to sense ambient P i and transport it to the cytosol. Pho84 can indeed transduce information about extracellular P i availability to the protein kinase A pathway (58). Subsequent studies attributed regulation of the PHO pathway to the levels of cytosolic P i , independently of Pho84, and postulated the existence of an unidentified cytosolic P i receptor (39,59). How cells measure their cytosolic P i concentration and ensure its homeostasis remains poorly understood. This work reinforces the idea that cytosolic P i is sensed by the cells. We believe that our finding that Vtc5 may modify the PHO pathway by altering the intracellular P i concentration provides a good tool to tackle this problem.