Introduction

Cells actively manage the concentration of P_i in their cytosol because they must strike a balance between conflicting goals (Austin & Mayer, 2020). On the one hand, P_i is a product of nucleotide-hydrolysing reactions. Its concentration has a significant impact on the free energy that these reactions can provide to drive metabolism. Since, for his reason, excessive P_i concentrations might stall metabolism, we can expect that cytosolic P_i must remain limited. On the other hand, P_i is an essential macronutrient. As a major constituent of nucleic acids and phospholipids, and as an important modifier of proteins, carbohydrates and many metabolites, it is consumed in large quantities for anabolic reactions. This can lead to situations where progression of the S-phase of the cell cycle is limited by the P_i uptake capacity of the cells (Bru et al, 2016, 2017; Gillies et al, 1981). Rapidly dividing cells, such as yeast cells, remedy this problem by maintaining phosphate stores in the form of inorganic polyP. PolyPs are chains of P_i linked through phosphoric anhydride bonds, which can be stored in acidocalcisome-like vacuoles (Urech et al, 1978; Okorokov et al, 1980). These polyP stores are accessed by cells when P_i becomes limiting in the environment, or when they face a sudden enhanced P_i consumption, e.g. during metabolic transitions from respiratory growth to fermentation, which requires much more phosphorylated metabolites (Shirahama et al, 1996; Thomas & O’Shea, 2005; Gillies et al, 1981; Nicolay et al, 1982, 1983). In such cases, cellular polyphosphate content decreases. It has therefore been proposed that polyPs are re-converted into P_i to buffer shortages in cytosolic P_i (Nicolay et al, 1983, 1982).

While this hypothesis is straightforward, we are lacking a coherent concept of how a P_i buffer based on an acidocalcisome-like organelle might work. Pioneering work on these organelles has revealed several characteristic and conserved features, such as their acidity, and their high content of basic amino acids and divalent cations (Docampo & Huang, 2016; Docampo, 2024). Of direct relevance to P_i homeostasis is the high capacity of acidocalcisome-like organelles for storing polyP. PolyP can vary greatly in length, from two to hundreds of phosphate units. The membrane of acidocalcisome-like organelles can carry polyP polymerases, such as VTC, and P_i transporters, such as Pho91 (Jimenez & Docampo, 2015; Hürlimann et al, 2007; Wang et al, 2015; Huang & Docampo, 2015; Gerasimaite & Mayer, 2016; Müller et al, 2003, 2002). VTC is a coupled polyP polymerase and translocase (Gerasimaite et al, 2014), which synthesizes polyP by transferring the ψ-phosphate of cytosolic ATP onto an elongating polyP chain (Hothorn et al, 2009). PolyP is synthesized by a catalytic domain in the centre of the Vtc4 subunit of the complex (Hothorn et al, 2009). The activity of this subunit is controlled through the SPX domains of VTC (Wild et al, 2016), which may associate into a dimerized inactive state (Pipercevic et al, 2023). They can be released from this state through the action of inositol pyrophosphates, signalling molecules that accumulate when P_i is abundant in the cytosol and signal that P_i storage in the form of polyP should be activated. The P_i state of the cytosol is communicated to VTC through a specific inositol pyrophosphate, 1,5-IP₈ (Gerasimaite et al, 2017; Chabert et al, 2023).

The postulated direct channeling of polyP from the site of its synthesis through the vacuolar membrane (Gerasimaite et al, 2014) is facilitated through the structure of the VTC complex. The catalytic domain of VTC has its exit for polyP right at the entry of a polyP-conducting channel, which is formed through the transmembrane domains of the VTC complex itself (Liu et al, 2023; Guan et al, 2023; Müller et al, 2002, 2003; Gerasimaite et al, 2014). This channel was proposed to exist in an open and closed conformation and be gated through polyP (Liu et al, 2023). PolyP translocation may be driven by the electrochemical potential across the membrane, which could move the highly negatively charged polyP chain through electrophoresis. This could explain why efficient polyP synthesis depends on proton pumps such as the V-ATPase or H⁺-pumping pyrophosphatases (Gerasimaite & Mayer, 2016; Gerasimaite et al, 2014; Freimoser et al, 2006; Lemercier et al, 2004).

The transporter Pho91 was proposed to export vacuolar P_i (Hürlimann et al, 2007). Several properties of yeast Pho91 and its homologs from other organisms are consistent with this view. Its ablation increases vacuolar P_i and polyP content, in yeast as well as in trypanosomes (Hürlimann et al, 2007; Jimenez & Docampo, 2015; Farofonova et al, 2023), and it weakly induces the phosphate starvation program in yeast (Pinson et al, 2004), suggesting that it may induce cytosolic P_i scarcity. Furthermore, patch-clamp analyses of Pho91-like channels from yeast, trypanosomes and plants have shown their P_i permeability and the dependence of their directionality on a pH gradient across the membrane (Potapenko et al, 2018, 2019; Wang et al, 2015). In the presence of a pH gradient, the Pho91 homolog from rice, OsSpx-MFS3, mediates P_i flux along the proton gradient, consistent with a function in exporting P_i from the acidic lumen of vacuoles (Wang et al, 2015) towards the cytosol.

Acidocalcisome-like organelles also contain polyphosphatases in their lumen (Lander et al, 2016; Gerasimaite & Mayer, 2016; Kulakovskaya et al, 2021; McCarthy & Downey, 2023), which can convert polyP back into P_i. In baker’s yeast, two vacuolar polyphosphatases are known, Ppn1 and Ppn2 (Gerasimaite & Mayer, 2017; Sethuraman et al, 2001; Andreeva et al, 2019). This means that a chain of polyP, when being synthesized by VTC and arriving in the vacuolar lumen, is immediately exposed to hydrolytic enzymes that will degrade it. While this seems at first sight paradoxical, we explored the hypothesis that the co-existence of polyP-synthesizing and polyP-hydrolysing activities might be a key feature conveying to acidocalcisome-like organelles the capacity to buffer cytosolic P_i. That these organelles have a critical role to play in this process is illustrated by observations in yeast, where artificial up- or down-regulation of vacuolar polyP synthesis suffices to drive the cytosol into a state of P_i starvation or P_i excess, respectively (Desfougères et al, 2016a). Furthermore, the presence of polyP reserves delays the activation of the transcriptional phosphate starvation response, the PHO pathway (Thomas & O’Shea, 2005). We hence explored the capacity of isolated yeast vacuoles to interconvert polyP and P_i, and we characterized the roles played by the vacuolar polyphosphatases Ppn1 and Ppn2 and the vacuolar P_i transporter Pho91. Our observations can be combined with previous findings to yield a coherent model of how an acidocalcisome-like organelle can operate as a P_i buffer for the cytosol.

Results

We explored the interplay of VTC, polyphosphatases and Pho91 in the accumulation of polyP and P_i inside vacuoles using an in vitro system with purified organelles. Vacuoles can be isolated in intact form when the cells are gently opened by enzymatic digestion of the cell wall and disruption of the cell membrane by low concentrations of DEAE-dextran (Dürr et al, 1975). Organelles isolated in this way can perform many of their normal cellular functions, such as membrane fusion, membrane fission and autophagy (Michaillat et al, 2012; Sattler & Mayer, 2000; D’Agostino & Mayer, 2019; Kunz et al, 2004). They also contain active polyP polymerase (VTC) and active polyphosphatases (Ppn1 and Ppn2) and they can synthesize and import polyP (Gerasimaite et al, 2014; Gerasimaite & Mayer, 2017; Gerasimaite et al, 2017).

Vacuolar P_i accumulation depends on polyP synthesis

Purified vacuoles were incubated with an ATP-regenerating system under conditions that allow these organelles to synthesize polyP in vitro (Gerasimaite et al, 2014). After different periods of incubation, the organelles were sedimented and solubilized in detergent. Vacuolar accumulation of polyP was assayed through DAPI fluorescence and P_i was measured through malachite green. VTC is stimulated by a variety of inositol pyrophosphates, most efficiently by 1,5-InsP₈ (Gerasimaite et al, 2014, 2017). Since 1,5-InsP₈ is not commercially available, we used saturating concentrations of 5-InsP₇ for our experiments (Pavlovic et al, 2015). This compound stimulates VTC with a higher EC₅₀ but to the same maximal activity as 1,5-InsP₈, and it was more readily available to us for routine experiments. Under the conditions used here, the organelles rapidly and efficiently produced polyP (Fig. 1), in line with previous results (Gerasimaite et al, 2014, 2017). This production was stimulated by 5-InsP₇, reaching 0.7 nmol of phosphate units per µg of vacuolar protein and min. This means that already within 10 min, wildtype vacuoles produced a mass of polyP equivalent to their total protein content, indicating how efficiently these organelles synthesize polyP. This signal was entirely dependent on VTC, because it was not observed in a mutant lacking the catalytic subunit of VTC (vtc41).

Fig. 1:

Isolated wildtype vacuoles also accumulated P_i with significant efficiency, at an initial rate of at least 0.3 nmol µg⁻¹ min⁻¹ (Fig. 2A). Vacuoles from vtc41 cells showed only 10% of this P_i signal, which was not time-dependent and hence may represent a background signal from the organelle preparation. In sum, vacuolar P_i accumulation depended on polyP synthesis through VTC.

Figure 2.

The kinetics of polyP and P_i accumulation were remarkably different. Whereas polyP continued to accumulate throughout the entire incubation period, P_i accumulated rapidly in the initial phase but reached a plateau within 30 min (Fig. 2A, B). Thus, although polyP production was essential for P_i accumulation, it could not become the limiting factor for it in the second phase. This is further illustrated by P_i accumulation in strains showing enhanced polyP synthesis, such as overexpressors of VTC5 (Fig. 2A, B) or strains carrying hyperactivating substitutions in the SPX domains of VTC (vtc3^K126A vtc4^K129A; Fig. 2 C, D) (Desfougères et al, 2016b; Wild et al, 2016).

Vacuoles from these strains accumulated polyP at a two to seven times higher initial rate than the wildtype and to higher concentrations. Nevertheless, their accumulation of P_i arrested at the same level as that of wildtype vacuoles, at 5 nmol µg⁻¹ vacuolar protein. To estimate the corresponding luminal concentration of P_i, we measured the diameters of 100 isolated wildtype vacuoles and counted the number of vacuoles per µg of vacuolar protein. With an average diameter of 0.8 µm and 5*10⁷ vacuoles/µg of protein we can estimate that P_i accumulation in vacuoles incubated with 5-InsP₇ reached a plateau at a luminal concentration of around 30 mM. This is a substantial concentration, which is in the range of the in vivo P_i concentration of 25 mM that was measured by ³¹P-NMR spectroscopy of yeast cells under conditions where vacuolar P_i dominates the signal (Okorokov et al, 1980). Since, upon P_i scarcity, cytosolic P_i drops to 1 mM (Okorokov et al, 1980), a substantial P_i concentration gradient across the vacuolar membrane could drive rapid replenishment of the cytosol from a readily accessible vacuolar pool of P_i.

Vacuolar P_i accumulation is limited by feedback inhibition of Ppn1 and Ppn2

We asked why vacuolar P_i accumulation quickly forms a plateau whereas vacuolar polyP continues to accumulate. To this end, we explored the hypothesis that vacuolar polyphosphatases become product-inhibited once 30 mM of their product, P_i, has accumulated. To test polyphosphatase activity we liberated Ppn1 and Ppn2 from isolated vacuoles by detergent lysis, followed by incubation with synthetic polyP as a substrate. When incubated in the presence of Zn²⁺, which permits activity of Ppn1 and Ppn2 (Gerasimaite & Mayer, 2017), the substrate was consumed in less than 3 min (Fig. 3). Degradation was delayed in vacuoles from ppn11 or ppn21 mutants, and it was suppressed in vacuoles from a ppn11 ppn21 mutant, in which both polyphosphatases were ablated. In incubations with only Mg²⁺ instead of Zn²⁺, which stimulates the activity of Ppn1 but much more than that of Ppn2, polyP degradation was slower and genetic ablation of Ppn1 sufficed to stabilize polyP. The addition of 30 mM potassium phosphate attenuated degradation of polyP 5- to 10-fold, both through Ppn1 (assayed in ppn21 and in samples without Zn²⁺) and through Ppn2 (assayed in ppn11). That both polyphosphatases are inhibited by P_i at this concentration is consistent with the notion that product inhibition of Ppn1 and Ppn2 might limit the conversion of polyP into P_i in the vacuolar lumen and thus define the concentration of the vacuolar P_i reservoir.

Figure 3.

Pho91 limits vacuolar P_i accumulation in an inositol pyrophosphate-dependent manner

To allow vacuoles to function as a P_i buffer for the cytosol, the P_i in the vacuolar lumen should become accessible in a regulated manner. The P_i transporter Pho91 is a prime candidate for mediating regulatable efflux because an overexpressed GFP fusion was reported to localize to vacuoles and Pho91 is inactivated by InsPPs through its SPX domain (Potapenko et al, 2018, 2019; Wang et al, 2015; Hürlimann et al, 2007). Since overexpression of membrane proteins in yeast easily leads to their erroneous accumulation in vacuoles, we re-investigated Pho91 localization in the absence of a strong, overexpressing promotor. When we expressed C- or N-terminal Pho91-GFP fusions from the authentic PHO91 promotor, we observed different and complex localization patterns (see Supplementary Figure 1 for examples).

Supplementary Figure 1:

These localization patterns depended on the nature of the fluorescent tag and the linker peptides used to attach it to Pho91, and they often showed significant accumulation in the ER. This suggests that the fluorescent protein tags interfere with intracellular trafficking of Pho91. We hence analysed purified vacuoles to test whether the non-tagged, endogenous Pho91 is a vacuolar protein. Mass spectrometry was used to determine the enrichment of proteins in the vacuolar fraction relative to a whole cell extract (Suppl. Table 1). Peptides from Pho91 were enriched in the purified vacuoles to a similar degree (39-fold) as peptides from vacuolar marker proteins, such as the vacuolar polyP polymerase subunit Vtc3 (35-fold), the vacuolar amino acid transporter Avt3 (48-fold), the v-ATPase subunit Vma9 (46-fold), or the alkaline phosphatase Pho8 (34-fold). By contrast, typical plasma membrane proteins were barely enriched, such as the iron permease Ftr1 (2.5-fold), the P_i importer Pho87 (2.4-fold) or the polyamine importer Tpo5 (2.2-fold) (Supplementary data file 1). This co-enrichment suggests that the major fraction of non-tagged Pho91 indeed resides in the vacuole. In line with such an intracellular localization, Pho91 is the only one of the five known yeast P_i transporters that cannot feed the cells with P_i. This was tested in a quintuple knockout strain lacking these five P_i transporters (Pho84, Pho87, Pho89, Pho90, Pho91), which was kept alive through expression of Pho89 from a URA-based plasmid. Plasmid shuffling allowed to exchange this plasmid against others expressing an individual P_i transporter. Whereas individual expression of the plasma-membrane P_i transporters Pho84, Pho87, Pho89, and Pho90 supports normal colony formation (Wykoff et al, 2007), Pho91 did not support growth, at all (Fig. S2). This suggests that Pho91 is indeed a vacuolar P_i transporter but that C- or N-terminal protein fusions interfere with correct sorting and cannot serve as reliable reporters for localization of this protein.

Supplementary Figure 2:

We tested the impact of Pho91 on P_i accumulation by vacuoles in vitro, using the same approach as above (Fig. 4). Vacuoles lacking Pho91 (pho911) accumulated P_i two times faster than the wildtype and the maximal accumulated concentration was two times higher. This is consistent with a function of Pho91 as vacuolar P_i exporter (Hürlimann et al, 2007; Potapenko et al, 2018; Wang et al, 2015).

Fig. 4:

The difference between pho911 and wildtype vacuoles vanished when the vacuoles were incubated in the presence of the inositol pyrophosphate 5-InsP₇. Since, as we showed above, vacuolar P_i accumulation depends on polyP synthesis through VTC, the relative enhancement of P_i accumulation in wildtype vacuoles through 5-InsP₇ could be caused by downregulation of Pho91, which would make the wildtype vacuoles behave similarly as pho911 vacuoles. Alternatively, 5-InsP₇ might stimulate polyP synthesis in wildtype more than in pho911 vacuoles. We could rule out the latter explanation based on two observations: An assay of polyP synthesis during the incubation (Fig. 5) revealed that 5-InsP₇ stimulated polyP synthesis in wildtype and pho911 vacuoles to similar degrees. Furthermore, as shown above (Fig. 2), polyP synthesis activity in wildtype cells is not rate-limiting for vacuolar P_i accumulation. Therefore, we attribute the enhancement of P_i accumulation through 5-InsP₇ to Pho91. Upon P_i scarcity, the declining inositol pyrophosphate levels should then activate Pho91 to replenish cytosolic P_i.

Fig. 5:

Interference with vacuolar polyP turnover provokes cytosolic P_i scarcity

We tested the effects of this postulated cycle of polyP synthesis, polyphosphatase activity and Pho91-mediated P_i export on P_i homeostasis in vivo. Pho4-GFP was used as a reporter, because this transcription factor shuttles between nucleus and cytosol. Under cellular P_i scarcity and correspondingly low inositol pyrophosphate levels it is predominantly nuclear, but it shifts to the cytosol under P_i-replete conditions (O’Neill et al, 1996; Chabert et al, 2023; Desfougères et al, 2016a; Auesukaree et al, 2004). Pho4-GFP localisation can hence serve as a readout for cytosolic P_i signalling. In wildtype cultures growing logarithmically on P_i-rich media, a small but significant fraction of the cells show nuclear accumulation of Pho4-GFP. Such activation of the PHO pathway in P_i-replete media can reflect a transient shortage of P_i during S-phase, when P_i utilization for biosynthesis may exceed the uptake capacity of the cell (Bru et al, 2016; Neef & Kladde, 2003; Pondugula et al, 2009). In addition, feedback signalling can maintain a fraction of these cells in a stable state of PHO pathway activation that had been triggered by a preceding period of transient P_i scarcity (Wykoff et al, 2007; Vardi et al, 2013, 2014).

We scored the fraction of cells showing this sign of P_i scarcity even in P_i-replete medium. This fraction was small (7%) for wildtype cells (Fig. 6). It increased fourfold upon hyperactivation of polyP synthesis, which we achieved by overexpression of the regulatory VTC subunit Vtc5 (Desfougères et al, 2016a). Ablation of polyP synthesis by deletion of the catalytic subunit Vtc4 had the opposite effect and reduced the frequency of cells with nuclear Pho4-GFP by half.

Fig. 6:

Deletion of the vacuolar polyphosphatases Ppn1 and Ppn2 prevents polyP turnover and leads to the accumulation of extremely long polyP chains (Gerasimaite & Mayer, 2017; Sethuraman et al, 2001). We may thus expect phosphate to remain fixed in the form of polyP instead of being made available for P_i reflux into the cytosol.

In line with this, the fraction of ppn11 ppn21 double mutants that showed nuclear Pho4-GFP was 3-fold higher than in wildtype. pho911 cells, in which we expect P_i export from the vacuoles to be impaired, showed two times higher frequency of nuclear Pho4-GFP than wildtype. pho911 vtc41 double mutants showed an even lower frequency of nuclear Pho4-GFP than wildtype, suggesting that the state of P_i starvation that pho911 favours is dependent on vacuolar polyP accumulation. Cells expressing pho91^K237A as the sole source of Pho91 also showed a 50% lower frequency of nuclear Pho4-GFP. The pho91^K237A allele generates an amino acid substitution in the InsPP-binding patch of the Pho91 SPX domain. It mimics the InsPP-free state (Wild et al, 2016) and hence low-P_i conditions (Chabert et al, 2023). Collectively, our results are consistent with the notion that loss of InsPP binding activates Pho91 to export P_i from vacuoles to the cytosol, enhancing repression of the PHO pathway.

Since InsPP levels in yeast decline in response to cellular phosphate availability (Lonetti et al, 2011; Chabert et al, 2023), we tested the impact of Pho91, Ppn1 and Ppn2 on the cellular levels of these metabolites using CE-MS, capillary electrophoresis coupled to mass spectrometry (CE-MS) (Qiu et al, 2020, 2023, 2021). The cells were harvested from the same P_i-replete growth conditions as for the microscopic assays above (Fig. 7).

Fig. 7:

The ppn11 ppn21 double mutants showed a decrease by around one third in the inositol pyrophosphates 1,5-InsP₈, 5-InsP₇ and 1-InsP₇, whereas InsP₆ remained at a similar level as in wildtype. This is consistent with the 3-fold increase of cells with nuclear Pho4-GFP that these cells showed (Fig. 6). pho911 cells did not show significant changes for all four metabolites although they showed a partial shift of Pho4-GFP into the nucleus (Fig. 6). We attribute this discrepancy to the different nature of the assays. The microscopic assay for Pho4-GFP localization can pick up effects in a fraction of the cells because it offers single cell resolution. Inositol pyrophosphate analysis is an ensemble assay, in which changes in a smaller fraction of the population become diluted through the major pool that does not show the effect. For this reason, a change in inositol pyrophosphates affecting only 10% of the pho911 cells that show nuclear Pho4-GFP remains undetectable. The microscopic assay suggests that cytosolic P_i scarcity in ppn11 ppn21 affects more cells, perhaps because it is more profound, and it hence becomes detectable even in a whole-population-analysis.

Discussion

Our observations can be integrated with existing data on the properties of VTC (Wild et al, 2016; Hothorn et al, 2009; Müller et al, 2002; Gerasimaite et al, 2014, 2017; Liu et al, 2023; Guan et al, 2023; Pipercevic et al, 2023) and Pho91 (Hürlimann et al, 2007; Wang et al, 2015; Potapenko et al, 2018) to generate a working model explaining how an acidocalcisome-like organelle such as the yeast vacuole is set up to function as a P_i buffer for the cytosol. Under P_i-replete conditions, high InsPP levels activate VTC to polymerize P_i into polyP and translocate it into the vacuolar lumen. Here, the vacuolar polyphosphatases degrade polyP into P_i, filling the lumen with P_i (Sethuraman et al, 2001; Shi & Kornberg, 2005; Lichko et al, 2010; Gerasimaite & Mayer, 2017). Since the P_i exporter Pho91 is downregulated through InsPP binding to its SPX domain (Potapenko et al, 2018; Hürlimann et al, 2007; Wang et al, 2015), the P_i liberated through polyP hydrolysis accumulates in the vacuoles. Product inhibition of the polyphosphatases attenuates polyP hydrolysis once the vacuolar lumen has reached a P_i concentration above 30 mM. When the cells experience P_i scarcity, InsPP levels decline (Chabert et al, 2023). This activates Pho91 to release P_i from the vacuolar pool into the cytosol and stabilizes cytosolic P_i.

Fig. 8:

Yeast cells do not only accumulate P_i as a rapidly accessible buffer for the cytosol. Under P_i-replete conditions they accumulate hundreds of millimolar of phosphate in the form of polyP (Urech et al, 1978). In contrast to the vacuolar reserve of P_i, which is presumed to be accessible immediately, mobilizing the polyP store takes minutes to hours (Nicolay et al, 1982; Pondugula et al, 2009; Bru et al, 2016). But the polyP store offers advantages in the form of high capacity - hundreds of millimolar of phosphate units can be stored in the form of polyP - and low osmotic activity of polyP (Dürr et al, 1979). Keeping such a large stock of a critical resource, which is often growth-limiting in nature, is relevant for the cells. In case of phosphate shortage, the vacuolar polyP store can be mobilized to enable the cells to complete the cell cycle and transition into G₀ phase (Spain et al, 2015; Westenberg et al, 1989; Müller et al, 1992). This can consume substantial amounts of phosphate, because we can estimate that replicating the entire DNA (1.2*10⁷ base pairs) immobilizes roughly 1 mM P_i in the cells; and cellular RNA is even 50 times more abundant than DNA (Warner, 1999), accounting for 50 mM phosphate. Phospholipids, which must also be synthesized to complete a cell cycle, fix phosphate in similar amounts (Lange & Heijnen, 2001). Thus, a large polyP store is necessary to guarantee that the cells can finish S-phase upon a shortage of phosphate sources. In accord with this notion, absence of the polyP store impairs cell cycle progression, nucleotide synthesis and induces genome instability (Bru et al, 2016, 2017). Also, a shift from non-fermentable carbon sources to fermentation of glucose leads to a strong requirement for P_i because the activation of glucose uptake and glycolysis depends on large amounts of phosphate-containing sugars and glycolytic intermediates (Nicolay et al, 1982, 1983; Gillies et al, 1981). Shortage of P_i restrains the abundance of these metabolites (Kim et al, 2023).

The properties of the regulatory circuit described above imply an inbuilt switch from vacuolar P_i accumulation to large-scale stocking of vacuolar polyP. P_i-replete conditions generate high cellular InsPP levels. These will not only reduce P_i efflux from the vacuoles through Pho91 and inactivate the vacuolar polyphosphatases, but at the same time stimulate continued polyP synthesis by VTC. Coincidence of these effects will favour storage and high accumulation of phosphate in the form of polyP. Conversely, depletion of the vacuolar P_i reservoir upon P_i scarcity in the medium will activate the vacuolar polyphosphatases. In combination with the downregulation of the polyP polymerase VTC through the decline of InsPPs this will mobilize the large vacuolar polyP reserve once the immediately available vacuolar P_i pool is gradually depleted.

The concentration of P_i inside vacuoles as a rapidly accessible P_i reserve, and the synthesis of a large polyP stock, come at an energetic cost because the transformation of P_i into polyP requires the formation of phosphoric anhydride bonds (Hothorn et al, 2009; Gerasimaite et al, 2014) and vacuolar P_i reaches 30 mM. This exceeds the cytosolic P_i concentration, which was measured through ³¹P-NMR in a variety of yeasts, yielding values of 5-17 mM (Nicolay et al, 1982, 1983). Cytosolic P_i can also be estimated based on data from several other studies (Auesukaree et al, 2004; Hürlimann et al, 2009; Theobald et al, 1996; Zhang et al, 2015; Pinson et al, 2004). Assuming that the cytosolic volume of a BY4741 yeast cell is 40 fL (Chabert et al, 2023), and 1 g of dry weight contains 40 * 10⁹ yeast cells, these studies point to cytosolic values of 10-15 mM in P_i-replete media. Upon P_i starvation, this value rapidly drops up to fivefold, resulting in a strong P_i gradient across the vacuolar membrane (Okorokov et al, 1980; Shirahama et al, 1996). To replenish the cytosolic pool under P_i scarcity, Pho91 can exploit not only this P_i concentration gradient, but also the vacuolar electrochemical potential, which was shown to stimulate P_i export through the Pho91 homolog OsSPX-MFS3 from plant vacuoles (Wang et al, 2015).

Vacuolar P_i accumulation is driven indirectly through ATP in two ways. VTC uses ATP as a substrate and transfers the phosphoric anhydride bond of the ψ-phosphate onto a polyP chain (Hothorn et al, 2009). The growing polyP chain exits from the catalytic site directly towards the transmembrane part of VTC (Liu et al, 2023; Guan et al, 2023). This transmembrane part likely forms a controlled channel that can guide polyP through the membrane (Liu et al, 2023). Coupled synthesis and translocation require the V-ATPase (Gerasimaite et al, 2014), probably because polyP is highly negatively charged and therefore follows the electrochemical potential across the vacuolar membrane of 180 mV (inside positive) and 1.7 pH units (Kakinuma et al, 1981), which is generated through the proton pumping V-ATPase. Thus, the combination of the VTC complex and vacuolar polyphosphatases can be considered as a P_i pump that is driven by ATP through polyP synthesis and through the electrochemical potential for polyP translocation and P_i export.

It is likely that acidocalcisome-like organelles of other organisms act as buffers for cytosolic P_i similarly as described in our model for yeasts. This notion is supported by the conserved molecular setup of acidocalcisome-like organelles as well as by phenotypic similarities. The acidocalcisomes of Trypanosomes contain VTC, a Pho91 homolog and proton pumps in their membranes, and polyphosphatases in their lumen (Huang & Docampo, 2015; Billington et al, 2023; Fang et al, 2007; Lander et al, 2013; Ulrich et al, 2013; Scott et al, 1997). Also the acidocalcisome-like organelles of the alga Chlamydomonas contain such proteins and they accumulate polyP through VTC as a function of the availability of P_i, a proton gradient and metal ions (Zúñiga-Burgos et al, 2024; Blaby-Haas & Merchant, 2014; Long et al, 2023; Hong-Hermesdorf et al, 2014; Aksoy et al, 2014; Ruiz et al, 2001; Goodenough et al, 2019). Like in yeast, the polyP stores are mobilized upon P_i limitation (Sanz-Luque et al, 2020; Plouviez et al, 2021). We hence propose that acidocalcisome-like vacuoles may have a general role as feedback-controlled, rapidly accessible phosphate buffers for the cytosol, addressing a critical parameter for metabolism. However, given that acidocalcisome-like organelles accumulate not only phosphate but also multiple other metabolites and ions (Docampo, 2024), they are probably interlinked with cellular metabolism in multiple ways and might form an important hub for its homeostasis.

Materials and methods

Yeast strains and growth conditions

Saccharomyces cerevisiae cells were grown on yeast extract-peptone-dextrose (YPD: 1% yeast extract, 2% peptone and 2% dextrose) or in synthetic complete (SC) medium from Formedium, supplemented with sodium phosphate as needed). Yeast backgrounds used in this study were BY4741 and BY4742. Genetic manipulations of yeast were performed by homologous recombination according to published procedures and/or transformation with the indicated plasmids (Gietz & Schiestl, 2007; Güldener et al, 1996). Strains used are listed in Table 1.

Table 1

PHO4 localization

Cells transformed with pRS415-P_PHO4-PHO4-GFP, a plasmid expressing a PHO4-GFP fusion from the PHO4 promotor (Chabert et al, 2023), were grown exponentially overnight in SC medium without leucine (SC-Leu). Care was taken that the culture did not grow beyond a density of OD₆₀₀=0.7. At this point, the cells were collected by brief centrifugation (15 sec, 3000 x g) in a tabletop centrifuge and resuspended in 1/10th to 1/20th of their own supernatant. Pho4-GFP localization was immediately checked by microscopy. An aliquot of the cells was washed twice with SC-Leu without P_i and then diluted in SC-Leu with 200 μM phosphate to OD₆₀₀=0.7. After 30 minutes of incubation in this medium, Pho4-GFP localization was analysed by fluorescence microscopy on a LEICA DMI6000B inverted microscope equipped with a Hamamatsu ORCA-R2 (C10600-10B) camera, an XCite ® series 120Q UV lamp and a Leica 100x 1.4 NA lens.

Vacuole preparation

Vacuoles were purified from yeast cells essentially as described (D’Agostino & Mayer, 2019). Briefly, yeast cells were grown in 1L of YPD to an OD₆₀₀ of 1.5. 330 ml of cells were collected by centrifugation and resuspended in 50 ml of 30 mM Tris-HCl pH 8.9, 10 mM DTT buffer. Suspensions were incubated in a 30° C water bath for 5 min and collected by centrifugation. The pellet was resuspended in 15 ml of spheroplasting solution (50 mM K-phosphate pH 7.5, 600 mM sorbitol in YPD with 0.2% D-glucose and 3600 U/ml lyticase) and incubated for 25 min at 30°C. Spheroplasts were collected by centrifugation (2500 x g, 3 min) and resuspended in 2.5 ml of 15% Ficoll 400 in PS buffer (10 mM PIPES-KOH pH 6.8, 200 mM sorbitol). 80 μg of DEAE-dextran were added under gentle mixing. After incubation on ice for 2 min and then at 30 °C for 80 sec, spheroplasts were transferred into Beckman SW41.1 tubes, overlaid with cushions of 8%, 4% and 0% Ficoll 400 in PS buffer, and centrifuged (150’000 x g, 90 min, 2 °C). Vacuoles were collected from the 0-4 % Ficoll interface. Their protein concentration was determined through Bradford assay using BSA as a standard.

Polyphosphatase activity of vacuolar lysates

Polyphosphatase activity was assayed as described previously (Gerasimaite & Mayer, 2017), with the following modifications. Isolated vacuoles were diluted to a final protein concentration of 0.002 mg/ml in 1 ml of reaction buffer (20 mM PIPES/KOH pH 6.8, 150 mM KCl, 1 mM ZnCl₂ or MgCl₂, 0.1% Triton X-100, 1xPIC, 1 mM PMSF and 300 µM polyP₃₀₀) in the presence or absence of 30 mM of KH₂PO₄ and incubated at 27°C. At the indicated times, 80 µl aliquots were collected and the reaction was stopped by dilution with 160 µl of stop solution (10 mM PIPES/KOH pH 6.8, 150 mM KCl, 12 mM EDTA pH 8.0, 0.1% Triton X-100, 15 µM DAPI). Remaining polyP was quantified by measuring polyP-DAPI fluorescence (λ_exc. 415 nm, λ_em. 550 nm) in a black 96 well plate in a Spectramax Gemini microplate fluorometer (Molecular Devices). A reaction containing boiled vacuoles was used as negative control.

Polyphosphate synthesis by isolated vacuoles

Polyphosphate synthesis was assayed as described (Gerasimaite et al, 2014). Isolated vacuoles were diluted to final protein concentration of 0.02 mg/ml on 1 ml of reaction buffer (10 mM PIPES/KOH pH 6.8, 150 mM KCl, 0.5 mM MnCl2, 200 mM sorbitol) and the reaction was started by adding an ATP regenerating system (1 mM ATP-MgCl₂, 40 mM creatine phosphate and 0.25 mg/ml creatine kinase). The mix was incubated at 27 °C. At different timepoints, 80 µl aliquots were mixed with 160 µl of stop solution (10 mM PIPES/KOH pH 6.8, 150 mM KCl, 200 mM sorbitol, 12 mM EDTA, 0.15% Triton X-100 and 15 μM DAPI). PolyP synthesis was quantified through polyP-DAPI fluorescence (λ_exc. 415 nm, λ_em. 550 nm) in a black 96 well plate. A calibration curve was prepared using commercial polyP₆₀ as a standard.

Phosphate quantification in isolated vacuoles

Isolated vacuoles were incubated as described for the polyP synthesis assay above. At different time points, 80 µl aliquots were centrifuged (3 min, 2000 x g, 2°C), the pellets were washed with 500 µl of washing solution (10 mM PIPES/KOH pH 6.8, 200 mM Sorbitol, 150 mM KCl) and centrifuged as before. The final pellet was resuspended with 100 µl of lysis buffer (10 mM PIPES/KOH pH 6.8, 200 mM sorbitol, 150 mM KCl, 12 mM EDTA, 0.1 % Triton). Free phosphate was quantified by adding 150 µl of molybdate-malachite green solution (1 mM malachite green, 10 mM ammonium molybdate,1M HCl) and reading the absorbance at 595 nm in a microplate photometer.

Inositol pyrophosphate synthesis, extraction and quantification

5-InsP7 was synthesized as described (Capolicchio et al, 2013; Wang et al, 2014). For quantification from cells, InsPPs extraction was performed as described previously (Kim et al, 2023), with the following modifications. Briefly, 3 ml of yeast cells at OD_600nm=1 were collected using rapid vacuum-filtration on a polytetrafluoroethylene membrane filter (1.2 μm; Piper Filter GmbH, Germany). After snap freezing in liquid nitrogen, yeast cells on the membrane were resuspended in 400 μL of 1 M perchloric acid and lysed by bead beating (glass beads; 0.25-0.5 mm) for 10 min at 4 °C. After centrifugation at 13,000 rpm for 3 min at 4 °C, the supernatant was transferred into a new tube containing 3 mg of titanium dioxide (TiO₂) beads (GL Sciences, Japan) which had been pre-washed twice with H₂O and 1 M perchloric acid. The sample was gently rotated for 15 min at 4 °C. The TiO₂ beads were collected by centrifugation at 13,000 rpm for 1 min at 4 °C and washed twice using 1 M perchloric acid. After the second washing step, the TiO₂ beads were resuspended in 300 μL of 3 % (v/v) NH₄OH and rotated gently at room temperature. After centrifugation at 13,000 rpm for 1 min, the eluants were transferred into a new tube and dried in SpeedVac (Labogene, Denmark) at 42 °C. InsPPs were measured through capillary electrophoresis coupled to mass spectrometry (CE-MS) as described (Qiu et al, 2021).

Quantification of the vacuolar proteome

Vacuoles were prepared as described above. During the purification procedure, a sample of the sedimented spheroplasts was withdrawn before DEAE dextran was added. These withdrawn spheroplasts constitute the “whole cell” extract. After the flotation step, the vacuoles withdrawn from the 4%-0% Ficoll interface were used as the vacuole fraction. For each fraction samples with 100 µg protein were precipitated by adding a final concentration 12.5% TCA for 10 min on ice. The proteins were sedimented (12’000 x g, 5 min, room temperature), the supernatant discarded, and the pellets were washed with ice-cold acetone and centrifuged as above twice. The final pellet was dried and dissolved in reducing SDS sample buffer (10% glycerol, 50 mM Tris-HCl pH 6.8, 2 mM DTT, 2% SDS, 0.002% bromophenol blue).

Samples dissolved in Laemmli buffer (50 mM Tris pH 6.8, 10 mM DTT, 2 % SDS, 0.1 % bromophenol blue, 10 % glycerol) were equalized in concentration by dilution in SP3 buffer (2% SDS, 10 mM DTT, 50 mM Tris, pH 7.5). 150 µg of protein per sample were heated at 95 °C for 5 min and cooled down. Reduced cysteine residues were alkylated by adding iodoacetamide (30 mM final) and incubating for 45 min at room temperature in the dark. Digestion was done by the SP3 method (Hughes et al, 2019) using magnetic Sera-Mag Speedbeads (Cytiva 45152105050250, 50 mg/ml). Beads were added at a ratio 10:1 (w:w) to samples, and proteins were precipitated on beads with ethanol (final concentration: 60 %). After 3 washes with 80% ethanol, beads were digested in 50 μl of 100 mM ammonium bicarbonate with 3.0 μg of trypsin (Promega #V5073). After 1 h of incubation at 37 °C, the same amount of trypsin was added to the samples for an additional 1 h of incubation. Supernatants were then recovered, transferred to new tubes, acidified with formic acid (0.5% final concentration), and dried by centrifugal evaporation. To remove traces of SDS, two sample volumes of isopropanol containing 1% TFA were added to the digests, and the samples were desalted on a strong cation exchange (SCX) plate (Oasis MCX; Waters Corp., Milford, MA) by centrifugation. After washing with isopropanol/1%TFA and 2% acetonitrile/0.1% FA, peptides were eluted in 200µl of 80% MeCN, 19% water, 1% (v/v) ammonia, and dried by centrifugal evaporation. Data-dependent LC-MS/MS analyses of samples were carried out on a Fusion Tribrid Orbitrap mass spectrometer (Thermo Fisher Scientific) interfaced through a nano-electrospray ion source to an Ultimate 3000 RSLCnano HPLC system (Dionex). Peptides were separated on a reversed-phase custom packed 45 cm C18 column (75 μm ID, 100Å, Reprosil Pur 1.9 μm particles, Dr. Maisch, Germany) with a 4-90% acetonitrile gradient in 0.1% formic acid at a flow rate of 250 nl/min (total time 140 min). Full MS survey scans were performed at 120’000 resolution. A data-dependent acquisition method controlled by Xcalibur software (Thermo Fisher Scientific) was used that optimized the number of precursors selected (“top speed”) of charge 2+ to 5+ while maintaining a fixed scan cycle of 0.6 s. Peptides were fragmented by higher energy collision dissociation (HCD) with a normalized energy of 32%. The precursor isolation window used was 1.6 Th, and the MS2 scans were done in the ion trap. The m/z of fragmented precursors was then dynamically excluded from selection during 60 s.

Data files were analysed with MaxQuant 1.6.14.0 (Cox & Mann, 2008) incorporating the Andromeda search engine (Cox et al, 2011). Cysteine carbamidomethylation was selected as fixed modification while methionine oxidation and protein N-terminal acetylation were specified as variable modifications. The sequence databases used for searching were the S. cerevisiae reference proteome based on the UniProt database (www.uniprot.org, version of June 6th, 2021, containing 6050 sequences), and a “contaminant” database containing the most usual environmental contaminants and enzymes used for digestion (keratins, trypsin, etc). Mass tolerance was 4.5 ppm on precursors (after recalibration) and 20 ppm on MS/MS fragments. Both peptide and protein identifications were filtered at 1% FDR relative to hits against a decoy database built by reversing protein sequences. The match between runs feature was enabled.

Data from the proteomic analysis have been deposited at the PRIDE database under the identifier PXD060102.

Acknowledgements

Mass spectrometry-based proteomics work was performed by the Protein Analysis Facility of the Faculty of Biology and Medicine, University of Lausanne. This study was supported by grants from the SNSF (320030-228119, 31003A_179306 and 310030_204713) and ERC (788442) to AM, by the HFSP (LT000588/2019) to GDK, by the Deutsche Forschungsgemeinschaft (CIBSS, EXC-2189, Project ID 390939984, to HJJ) and the Volkswagen Foundation (VW Momentum Grant 98604 to HJJ).

Additional information

Author contributions

AM conceived the study, analysed data and wrote the paper. SB performed in vitro and in vivo experiments with yeast. GDK, LM, GL, and HJJ analysed inositol pyrophosphates. All authors assembled and corrected the manuscript.

Funding

Swiss National Science Foundation (320030-228119)

Swiss National Science Foundation (31003A_179306)

Swiss National Science Foundation (310030_204713)

European Research Council

https://doi.org/10.3030/788442

Deutsche Forschungsgemeinschaft (EXC-2189 Project 390939984)

Volkswagen Foundation (98604)

International Human Frontier Science Program Organization (LT000588/2019)

Additional files

Supplemental datafile 1