Organelle size control – increasing vacuole content activates SNAREs to augment organelle volume through homotypic fusion

Maintaining membrane tension of an organelle while avoiding its lysis requires correspondence between the amount of content and the volume provided by the boundary membrane. Maintaining this correspondence is likely to be functionally important because the activity of many membrane proteins, such as transporters and channels, can depend on the tension of the membrane they reside in (Hamill and Martinac, 2001; Sukharev and Sachs, 2012), and the complexes and activities of soluble proteins in the organelle lumen are greatly influenced by macromolecular crowding (Zhou et al., 2008). Furthermore, specific interactions of macromolecules inside the organelle are affected by their concentration. Adjustment of the surface-to-volume ratio of an organelle can assure optimal membrane tension and appropriate concentrations for the soluble content of the organelle. Fusion and fission of the multiple copies of a compartment provide a rapid and efficient way of achieving this because they are independent of new membrane synthesis and readily reversible. These mechanisms are also expected to serve in the adjustment of organelle volume to correlate with cell size, a ratio that is kept surprisingly constant under steady-state growth conditions but which can change considerably as the growth conditions of a cell change (Marshall, 2012). However, we can expect that other processes also contribute to the adjustment, such as the net influx and efflux of membrane, or modifications of organelle shape. Changes in organellar size and/or organellar surface-to-volume ratio have been observed – e.g. for the Golgi, vacuoles, mitochondria and endosomes (Banta et al., 1988; Bevis et al., 2002; Bhave et al., 2014; Chan and Marshall, 2014; Gaynor et al., 1998; Seeley et al., 2002).

The size of several organelles or protein assemblies has been reported to be actively regulated (Katsura, 1987; Levy and Heald, 2010; Ludington et al., 2012), but it remains poorly understood how size control is achieved at the molecular level. Under constant growth conditions, the size of membrane-enclosed organelles scales with cell size (Chan and Marshall, 2012). Upon changes in growth conditions, the macromolecular content and the volume occupied by an organelle type can be altered both in absolute terms and relative to the cell surface. Because organelles remain intact during such changes and do not burst, cells might be capable of controlling the volume of the organellar boundary membrane relative to organelle content. One way in which this can be achieved are shape changes, because all deviations from a spherical shape will decrease the volume-to-surface ratio. Non-spherical shapes of membranes can be dictated by proteins or by changes in lipid distribution (Derganc et al., 2013; Frolov et al., 2011; Heald and Cohen-Fix, 2014; Rafelski and Marshall, 2008). Manipulation of the balance of membrane influx and efflux pathways can also increase or decrease the volume of the organellar envelope. For example, a mutation reducing vesicular traffic from vacuoles to endosomes or the Golgi (atg18Δ) increases the volume of yeast vacuoles, whereas reductions in membrane influx (apl5Δ) decrease it (Chan and Marshall, 2014; Dove et al., 2004; Efe et al., 2007; Tamura et al., 2013; Zieger and Mayer, 2012). A third possibility is the fusion or fission of the organelle. Fusion of several organelle copies will increase the volume-to-surface ratio without a need for new membrane synthesis, fission will do the reverse. So far, however, these mechanisms have not been implicated in adjusting the volume of the organellar envelope to the amount of content to be enclosed. We have begun to study this question with the lysosome-like vacuoles of yeast cells.

Lysosomes and vacuoles are dynamic organelles (Settembre et al., 2012; Weisman, 2003). During starvation and autophagy, lysosomal hydrolases are induced to a large extent. The resulting augmentation of macromolecular content is accompanied by a volume increase of the lysosomal boundary membrane (Yu et al., 2010). During this volume increase, as well as during its reversal, the boundary membrane neither explodes nor does it collapse into a ruffled morphology. It appears to retain a well-inflated appearance, which suggests that the volume-to-surface ratio of the organelle envelope might be controlled. This can help to maintain the membrane tension and the concentration of lumenal macromolecules that are crucial for organelle function (Ellis, 2001; García-Pérez et al., 1999; Heald and Cohen-Fix, 2014; Upadhyaya and Sheetz, 2004).

We explore the hypothesis that the volume of the boundary membrane of vacuoles adapts to the amount of content to be enclosed through use of membrane fusion or fission. Vacuoles are good objects to study this question. Their size and copy number are influenced by an equilibrium of membrane fusion and fission (Baars et al., 2007; Efe et al., 2007; Zieger and Mayer, 2012) and by environmental changes (Peters et al., 2004; Weisman, 2003). A crucial advantage is that vacuole content is dominated by only few compounds, polyphosphate (polyP) and the basic amino acids that associate with it (Saito et al., 2005). PolyP is a polymer of up to hundreds of inorganic phosphate units linked by phospho-anhydride bonds. It is a storage form of phosphate, a compound that is often growth-limiting for plants, fungi and bacteria. PolyP occurs also in mammalian lysosomes, but here its function is unknown (Pisoni and Lindley, 1992). Yeast synthesizes polyP through the vacuolar transporter chaperone (VTC) complex, which is located in the vacuolar membrane and comprises three subunits (Vtc1, Vtc3 and Vtc4) (Hothorn et al., 2009). Vacuolar polyP can constitute up to 15% of the dry weight of a yeast cell. Because the vacuole occupies less than a third of the volume of a yeast cell but contains virtually all cellular polyP, polyP can constitute >50% of the vacuolar dry weight. Thus, manipulating the synthesis of polyP grossly changes vacuole content.

Vacuoles respond to increases in their osmotic pressure through fusion, and in vitro experiments have identified the stimulation of Rab-GTPase-mediated vacuole docking as one of the effects responsible for this change (Brett and Merz, 2008). Our earlier work has identified the VTC complex as a factor that is necessary for vacuole fusion and for priming of SNARE proteins (Müller et al., 2002,, 2003). Given that the VTC complex interacts with parts of the vacuolar fusion machinery, such as the R-SNARE Nyv1 and the V₀ sector of the V-ATPase, we had interpreted the effects of VTC deletion on fusion as a consequence of the loss of these interactions. Because VTC has since been identified as the major polyP polymerase of yeast (Hothorn et al., 2009), we have reconsidered the role of VTC in vacuole fusion and have now tested whether it could stimulate vacuole fusion by changing the abundance of vacuole content. We exploited the availability of structural information on the catalytic center of VTC and of methods to reliably measure synthesis and translocation of polyP by yeast vacuoles (Gerasimaite et al., 2014) in order to explore a possible link between the vacuolar fusion machinery and polyP accumulation in the vacuolar lumen.

We asked whether the synthesis of polyP might trigger a corresponding increase in vacuolar storage space for this compound. Fusing the multiple vacuoles of a cell into a single organelle of larger volume could achieve this (Fig. 1A). We used the vital dye FM4-64 (Vida and Emr, 1995) to stain vacuoles in yeast cells that did not accumulate polyP because they contain inactive VTC complexes (Fig. 1B,C). When grown on synthetic Hartwell's complete medium, numerous small vacuoles were found in cells carrying the catalytically inactive vtc4^R264A as well as in the deletion mutants vtc4Δ and vtc1Δ. Wild-type cells showed mostly 1–2 large vacuoles. This is in contrast with our earlier observations that, similar to nyv1Δ cells, vtc4Δ cells cultured in yeast extract peptone dextrose (YPD) medium carry a single large vacuole (Müller et al., 2001) (Fig. S1). This difference could be explained by an osmotic imbalance, resulting from the composition of YPD medium, which contains four-times more of the amino acids Arg, Lys and His than Hartwell's complete medium. These three amino acids accumulate in vacuoles to very high concentrations, and this effect is much more pronounced on YPD than on synthetic medium (Kitamoto et al., 1988). In absence of polyP, which normally sequesters and osmotically inactivates these amino acids (Dürr et al., 1979), the excess of vacuolar amino acids might create an osmotic pressure that hinders the organelle from fragmenting in YPD-grown vtc mutants.

The tendency of vtc1Δ and vtc4Δ cells to show fragmented vacuoles is an indicator for reduced vacuolar fusion activity. We tested whether vacuolar polyP stimulates vacuole fusion by comparing the fusion activity of isolated vacuoles to their rate of polyP accumulation. Purified vacuoles fuse when incubated with ATP (Wickner, 2010). Under the same conditions, they also synthesise polyP efficiently, reaching up to 0.4 µg of polyP per µg of vacuolar protein within 20 min (Fig. 2A). Structure–function analyses of the VTC complex identified missense mutations and/or single residue substitutions that impair or stimulate polyP synthesis without affecting the structural integrity of the complex (Hothorn et al., 2009). These substitutions lie either in the catalytic domain of Vtc4, which is oriented towards the cytosol and converts ATP into polyP and ADP; in the cytosolic SPX domains, which are assumed to have regulatory function; or in the transmembrane domains of Vtc1, which are believed to translocate the polyP chain across the membrane (Gerasimaite et al., 2014; Hothorn et al., 2009; Wild et al., 2016). This set of mutations includes the catalytic domain substitution vtc4^K200A, which permits Vtc4 to hydrolyze ATP, but without producing polyP (Hothorn et al., 2009). Fusion activity of vacuoles from the respective mutants increased with the rate of polyP accumulation (Fig. 2B). Because the single amino acid substitutions in the Vtc proteins allow assembly of the VTC complex and its sorting to the vacuole (Hothorn et al., 2009; Wild et al., 2016), this result suggests that the vacuolar fusion machinery might be controlled by the accumulation of polyP inside the vacuoles independently of the physical presence of the VTC complex.

Next, we compared the effects of the lumenal fraction of polyP and of an artificially generated cytosolic pool of polyP on vacuole fusion. VTC has been proposed to act as an obligatorily coupled synthase and translocase (Gerasimaite et al., 2014). Its cytosolic domains synthesize polyP and feed it directly into the vacuolar lumen through the VTC transmembrane domains. Therefore, VTC produces lumenal polyP rather than freely diffusible cytosolic polyP (Saito et al., 2005). The luminal pool can be reduced by engineering the delivery of a polyphosphatase into the vacuolar lumen. In this way, the VTC complex remains unmodified and continues to synthesize polyP, but polyP accumulation in the lumen is nevertheless reduced. We used a fusion of the polyphosphatase Ppx1 with the pre-pro-sequence of the vacuolar carboxypeptidase Y (Prc1), which targets the fusion protein through the ER, Golgi and endosomes into the vacuolar lumen (Gerasimaite et al., 2014). This construct can only be used in strains deficient in vacuolar proteases, such as BJ3505, because it is degraded in protease-proficient vacuoles. Vacuolar Ppx1 converts polyP into inorganic phosphate and oligophosphates (Wurst and Kornberg, 1994). Inorganic phosphate can then be exported into the cytosol by Pho91 (Hürlimann et al., 2007). A GFP-tagged version of this construct yielded strong GFP-staining of the vacuolar lumen, no staining in the cytosol and only very weak signals in the cell periphery and around nuclei, which we interpreted as being staining of the ER (Fig. 2C) (Gerasimaite et al., 2014). Whole-cell extracts showed no significant degradation of the fusion protein (Fig. S2), suggesting that the construct is correctly targeted to vacuoles. Vacuole-targeted Ppx1 reduced the in vivo polyP content of BJ3505 cells by 90%. It reduced the in vitro polyP accumulation of their vacuoles by 80% (Fig. 2D) and their fusion activity by 40% (Fig. 2E), suggesting that the lumenal pool of polyP influenced fusion activity. It should be noted that in this experiment, polyP accumulation could be reduced only in one of the fusion partners – i.e. in the protease-deficient BJ3505 vacuoles – whereas the construct could not be used in the other protease-competent fusion partner from DKY6281 cells, where it would be degraded. This can explain why fusion activity is less affected than polyP accumulation. Furthermore, the reduction in polyP might be overestimated because Ppx1 shows >100 times higher K_m values for short chains (<10 residues) than for long chains (>50 residues), and it cannot degrade short end-products, such as pyrophosphate (Wurst and Kornberg, 1994). Such short species might remain in the vacuoles of cells expressing vacuole-targeted Ppx1, but they do not induce the shift in DAPI fluorescence that is the basis for detection by our polyP assay (Omelon et al., 2016).

If the presence of polyP at the cytosolic side constitutes a signal for vacuolar fusion, the addition of polyP to the buffer should substitute for polyP synthesis by VTC. We isolated vacuoles from vtc4Δ and vtc4^R264A cells and incubated them in fusion reactions containing synthetic polyP of an average chain length of 60 and 300 phosphate residues (polyP-60 and polyP-300, respectively), which covers the range of chain lengths produced by the VTC complex (Hothorn et al., 2009; Lonetti et al., 2011) but which is not imported by the vacuoles (Gerasimaite et al., 2014). In tests over a wide range of different concentrations, polyP-60 and polyP-300 that had been added to the cytosolic side could not stimulate fusion (Fig. 2F). Taken together, these observations suggest that only lumenal polyP promotes fusion of vacuoles.

Next, we asked which step of the fusion reaction was affected by polyP. Vacuole fusion can be resolved into distinct sub-reactions, such as SNARE activation, trans-SNARE pairing, hemifusion and fusion-pore opening (Ostrowicz et al., 2008; Wickner, 2010). SNARE activation depends on ATP hydrolysis by Sec18/NSF. This chaperone dissociates and activates SNAREs that exist in cis-complexes on vacuoles (Mayer et al., 1996; Ungermann et al., 1998). SNARE activation can be bypassed by adding the purified recombinant Q_c-SNARE subunit Vam7, which recruits an existing pool of non-paired, free SNAREs into trans-SNARE complexes and drives fusion in the absence of Sec18/NSF activity (Thorngren et al., 2004). Vam7 addition restored the fusion activity of vtc4Δ, vtc4^R264A and vtc4^K200A vacuoles to wild-type levels (Fig. 3A). This suggested that trans-SNARE pairing, hemifusion and fusion-pore opening proceed in the absence of polyP and that SNARE activation might be the only step that is affected. We tested SNARE activation directly, using Sec17/α-SNAP, a co-factor of Sec18/NSF. The release of Sec17 from the membrane concurs with the dissociation of cis-SNARE complexes (Mayer et al., 1996; Ungermann et al., 1998). Vacuoles were incubated under fusion conditions and sedimented, and Sec17 was assayed in the membrane pellets and supernatants (Fig. 3B). Release of Sec17 was reduced by 80–95% in vacuoles from vtc1Δ, vtc4Δ and vtc4^R264A cells. Vacuoles from vtc1^R98E cells, which showed an intermediate reduction of polyP synthesis and fusion, showed a reduction in Sec17 release by 30% (Figs 3B and 2B). Also, expression of the vacuole-targeted polyphosphatase Ppx1 reduced Sec17 release by around 30%. Next, we assayed the dissociation of cis-SNARE complexes by performing co-immunoprecipitation of the R-SNARE Nyv1 and the Q_a-SNARE Vam3 (Ungermann et al., 1998). After 15 min of incubation under fusion conditions, vacuoles were solubilized in detergent, and Nyv1 was immunoprecipitated (Fig. 3C). Vam3 co-precipitated with Nyv1 if the preceding incubation of the vacuoles had been performed in the absence of ATP, which leaves Sec18/NSF inactive. Upon incubation in the presence of ATP, the efficiency of co-immunoprecipitation dropped significantly in wild-type vacuoles, but not in vtc4^R264A or vtc4Δ vacuoles (Fig. 3C). This suggests that the ATP-dependent disruption of SNARE complexes depends on the synthesis of polyP and its translocation into the vacuole lumen.

Vacuoles from mutants deficient in polyP synthesis carry normal amounts of Sec18/NSF (Fig. 4A). Therefore, we tested whether their deficiency in SNARE activation might be due to altered interaction of Sec18/NSF with the SNARE complex. Vacuoles were isolated from cells that expressed the SNARE Vam3 with a C-terminal HA tag (Pieren et al., 2010). The membranes were treated with the cleavable crosslinker DTSSP, solubilized in detergent, and immunoprecipitated with anti-HA antibodies. In wild-type vacuoles, Vam3–HA was associated with Sec18/NSF, as expected (Ostrowicz et al., 2008). This association was only detected upon crosslinking. On vtc4^R264A and vtc4Δ vacuoles, Vam3–HA did not interact with Sec18/NSF, whereas it did associate with Nyv1, Vam7 and Sec17, as on wild-type vacuoles (Fig. 4A). This suggests that polyphosphate accumulation facilitates SNARE activation by promoting the interaction of Sec18/NSF with the vacuolar SNARE complex. Therefore, we tested whether fusion between polyP-deficient vacuoles could be activated by excess Sec18/NSF. Overexpressing SEC18 in a vtc1Δ mutant by a factor of >10 over the wild-type level (Fig. 4B–D) rescued the fragmented vacuole phenotype of the vtc1Δ mutant and restored 1–2 large vacuoles per cell. Also, the in vitro fusion of vacuoles that had been isolated from vtc4^R264A, vtc4^K200A and vtc4Δ cells could be restored by incubating them with purified recombinant Sec18/NSF at concentrations that exceeded the endogenous levels by 50 times (Fig. 4E,F).

The VTC complex physically interacts with Nyv1 (Müller et al., 2003) and genetically interacts with Pho80 (Huang et al., 2002), the cyclin associating with the kinase Pho85. We tested whether these proteins are implicated in the effect of polyphosphates on priming. In absence of Nyv1, the SNAREs remaining on nyv1Δ vacuoles – Vam3, Vam7 and Vti1 – must still be activated by Sec18/NSF and Sec17 (Mayer and Wickner, 1997; Nichols et al., 1997; Ungermann et al., 1998). Therefore, SNARE activation can be assayed in nyv1Δ vacuoles in the same manner as in wild-type vacuoles. Vacuoles from a nyv1Δ vtc1Δ double mutant (ΔΔ) showed a significant rescue of Sec17 release to ∼70% of that of the wild-type level (Fig. 5A,B). In testing whether deletion of NYV1 might rescue fusion activity of a vtc1Δ vacuoles, we had to consider that vacuole fusion requires Nyv1 on at least one of the fusion partners (Nichols et al., 1997), and we therefore compared the fusion of vtc1Δ and ΔΔ vacuoles with wild-type organelles. Although a vtc1Δ×wild type combination showed only 40–50% of the fusion signal of a wild type×wild type combination, fusion was rescued to 80–90% in a ΔΔ×wild type combination (Fig. 5C). No fusion was observed between two ΔΔ or two vtc1Δ vacuoles. This suggests that Nyv1 is necessary for impairing the interaction of Sec18 with the SNAREs in the absence of polyP.

We tested available mutants that were null for Pho85 cyclin genes in a lipid mixing assay to test for in vitro vacuole fusion (Reese et al., 2005). Among eight cyclin mutants tested, only vacuoles from pho80Δ strains were deficient in lipid mixing (Fig. S3). In content mixing assays, pho80Δ vacuoles showed 20% of the activity of wild-type vacuoles (Fig. 6A). pho80Δ vacuoles showed normal Sec17 release (Fig. 6B) and SNARE complex disruption (not shown), suggesting that Pho80 was not required for SNARE activation. It might, however, negatively regulate SNARE activation. In this case, a pho80Δ vtc1Δ double knockout might remedy the defects of vtc1Δ vacuoles. Vacuoles from pho80Δ vtc1Δ cells indeed showed a partial rescue of Sec17 release and of fusion to 30–40% of the wild-type level (Fig. 6C,D). In vivo, a pho80Δ vtc1Δ mutant showed less fragmented vacuoles than vtc1Δ cells (Fig. 6E). These data support the hypothesis that polyP accumulation inside vacuoles leads to Pho80–Pho85-dependent changes in the cytosol that influence SNARE activation.

We investigated whether the block of priming was due to a stable modification of the SNARE complex by immuno-adsorbing Vam3–HA-tagged SNARE complexes to protein G-agarose beads. The adsorbed complexes could not be dissociated by simple addition of an ATP regenerating system, suggesting that Sec18/NSF activity had been lost during purification (Fig. 7A). We added back pure Sec18/NSF in increasing concentrations in order to test whether SNARE complexes from vtc1Δ and wild-type cells were equally sensitive to activation by this protein. SNARE activation, measured by release of Sec17 and Vam7 from the beads, showed the same concentration dependence for wild-type and vtc1Δ complexes, suggesting that vtc1Δ complexes had not retained a stable modification that hindered Sec18/NSF from acting on them. SNARE complex disruption could also be induced by adding wild-type cytosol and ATP, irrespective of whether the vacuoles came from vtc1Δ or wild-type cells (Fig. 7B,D). Cytosol from vtc1Δ cells was inactive, although it contained the same amount of Sec18 as wild-type cytosol (Fig. 7B,C). By contrast, cytosol from pho80Δ vtc1Δ cells supported SNARE complex disruption (Fig. 7E), which corroborates the partial rescue of vacuole fusion on pho80Δ vtc1Δ vacuoles described above. Heating (5 min at 95°C) abolished the activity of cytosol (Fig. 7D), suggesting that the activating effect was not due to the presence of heat-stable low-molecular-mass compounds. Taken together, these data support the hypothesis that cytosolic factors regulate SNARE activation as a function of polyP accumulation in vacuoles, and they implicate Pho80–Pho85 kinase and Nyv1 in this regulation.

Vacuole fusion increases the volume of the vacuolar system in response to starvation, accumulation of storage compounds such as polyP, or upon hypotonic shock. The ways in which this volume increase is triggered differ, however. Upon starvation, vacuolar hydrolases are massively upregulated, vacuoles fuse and the volume of the organelle expands, probably in order to accommodate the need for increased degradation through autophagy. In this case, the adjustment occurs over hours rather than minutes and operates through TORC1 kinase. TORC1 is necessary for vacuolar fission but not for vacuole fusion (Michaillat et al., 2012). Because TORC1 activity is downregulated by starvation, starvation tilts the fusion–fission equilibrium towards fusion, leading to a slow net fusion of vacuoles. Interestingly, mammalian lysosomes fuse and undergo a similar increase in volume upon induction of autophagy (Sardiello et al., 2009; Yu et al., 2010). In contrast to this slow adaptation to starvation, hypotonic shock increases vacuolar membrane tension very rapidly. Vacuoles fuse in the range of seconds to minutes, probably in order to avoid their lysis. This rapid fusion is induced by a stimulation of the machinery for Rab-GTPase-dependent vacuole docking (Brett and Merz, 2008). The activation of SNAREs is unaffected by membrane tension that is induced by hypotonicity (Brett and Merz, 2008). We have now found that the accumulation of a major vacuolar storage compound stimulates fusion through SNARE activation, controlling the interaction of Sec18/NSF with cis-SNARE complexes. Thus, the vacuolar surface-to-volume ratio can be adjusted through the fusion–fission equilibrium using at least three different molecular pathways. A priori, these regulatory inputs are not mutually exclusive. However, polyP-induced fusion of vacuoles is unlikely to depend on osmotic water influx into the lumen of the organelle because polyP chains synthesized by vacuoles are dozens to hundreds of phosphate residues long. The increase of vacuolar osmolarity by such macromolecules should be negligible. PolyP, which is a highly charged polyanion, could also increase the osmotic value of vacuoles by driving an influx of compensatory cations, such as lysine or arginine. Although these amino acids accumulate inside vacuoles up to molar concentrations, they do not lead to a corresponding increase in osmotic pressure, probably because their sequestration by polyP renders them osmotically inert (Dürr et al., 1979). A further argument against the activation of fusion through an osmotic effect of polyP accumulation is that polyP specifically affects SNARE activation and NSF/Sec18 binding, whereas an osmotic increase of membrane tension does not affect this reaction (Brett and Merz, 2008). It is also unlikely that polyP promotes vacuole fusion through TORC1 because vacuole fusion does not require TORC1 activity (Michaillat et al., 2012). Thus, the stimulation of NSF-dependent SNARE activation through the synthesis and lumenal accumulation of polyP represents a previously unidentified regulatory pathway for vacuole size and number that is distinct from the inputs of starvation and osmotic stress.

The fusion defect of vtcΔ vacuoles had initially been ascribed to a physical interaction of VTC with the vacuolar H⁺-ATPase (Müller et al., 2002,, 2003), which is required for vacuole fusion (Bayer et al., 2003; Peters et al., 2001; Sreelatha et al., 2014; Strasser et al., 2011). A function of VTC as the vacuolar polyP polymerase had been dismissed owing to the fact that the complex is entirely oriented towards the cytosol, whereas polyP is located inside the vacuoles. Progress in our understanding of VTC structure and polyP metabolism (Gerasimaite et al., 2014; Hothorn et al., 2009) has now led to the revelation that the crucial factor is indeed vacuolar polyP. This is supported by the effects of point mutants in the catalytic center as well as by artificial expression of a polyphosphatase in the vacuole lumen. Our observations support a model in which polyP accumulation controls fusion through the interaction of Sec18/NSF with the SNARE complex (Fig. 8). We must then postulate that information about the accumulation of polyP inside the organelle is transmitted to the cytosolic face of the membrane where Sec18/NSF and SNAREs interact. An argument in favor of this notion stems from the observation that a cytosolic extract from a wild-type strain promotes SNARE dissociation, an extract from a vtc1Δ strain does not and that an extract from a vtc1Δ pho80Δ mutant restores SNARE activation. This suggests that a crucial factor related to the accumulation of polyP in vacuoles is present in the soluble fraction. How then is the information transmitted? The VTC complex interacts physically with Nyv1 and genetically with Pho80–Pho85 kinase (Huang et al., 2002; Müller et al., 2003). Our data implicate both proteins in the control of SNARE activation because deletion of either of the genes encoding Nyv1 or Pho80 partially rescued the priming defect of vtcΔ vacuoles.

The role of Pho80 could be rationalized in the context of its function in the Pho80–Pho85 kinase, which is activated by cytosolic phosphate (Schneider et al., 1994). Because polyP synthesis transfers large amounts of phosphate into vacuoles, thereby lowering cytosolic phosphate concentration (Thomas and O'Shea, 2005), polyP accumulation in vacuoles could be indirectly reported through the cytosolic phosphate levels (Fig. 8). The lowered cytosolic concentration of phosphate that results from polyP synthesis would reduce Pho80–Pho85 activity, whereas both parameters are expected to increase when polyP synthesis is shut down. Deletion of Pho80 eliminates Pho80–Pho85 kinase activity and would thus mimic a state where polyP is synthesized, promoting SNARE activation. Our attempts to identify relevant Pho80–Pho85-dependent phosphorylation sites have been frustrated by the complexity of the phosphorylation patterns. On the VTC complex alone, we found >20 phosphorylation sites that are strongly up- or downregulated in pho80Δ vacuoles relative to wild-type organelles (A.M., unpublished data). Furthermore, we have detected that phosphorylation of residue Ser197 of Nyv1 is upregulated in vtc4Δ vacuoles. However, replacing this residue with Ala or Asp did not change vacuolar morphology, fusion activity or the block of priming in vtcΔ mutants. A challenge for future analyses will be to take a comprehensive approach that addresses the regulation of the vacuolar fusion machinery by phosphorylation.

The following chemicals were bought from the indicated sources: creatine phosphate, creatine kinase, adenosine-5′-triphosphate and protein G-agarose (Roche); dithiothreitol, amino acids, uracil and adenine (AppliChem). 3,3′-dithiobis[sulfosuccinimidylpropionate] (DTSSP) (Pierce). Rhodamine-phosphatidylethanolamine (Rh–PE) and N-(3-triethylammoniumpropyl)-4-(p-diethylaminophenylhexatrienyl)-pyridinium dibromide (FM4-64) (Invitrogen); 4′,6′- diamidino-2-phenylindole (DAPI) (Sigma-Aldrich). Monoclonal anti-hemagglutinin (anti-HA) antibodies (mouse ascites, Covance); poly-peptone and yeast extract (Pronadisa); dextrose and yeast nitrogen base without amino acids and ammonium sulfate (Difco). Polyclonal antibodies against Nyv1, Vam3, Sec17, Sec18, Vtc1 and Vtc4 were raised in rabbits or goats and were prepared as described previously (Müller et al., 2002; Pieren et al., 2010). The lytic enzyme (lyticase), recombinant Sec18, recombinant Ppx1 and recombinant Vam7 were expressed and purified as described previously (Gerasimaite et al., 2014; Müller et al., 2002; Stroupe et al., 2006).

Strains, primers and plasmids used in this study are listed in Tables S1, S2 and S3, respectively. Their generation is described in the appropriate supplementary tables.

Unless indicated otherwise, yeast cells were grown in YPD medium [2% poly-peptone, 1% yeast extract and 2% dextrose (Difco)]. Hartwell's complete medium contained 0.67% yeast nitrogen base without amino acids and ammonium sulfate, 2% dextrose, 0.002% methionine, 0.006% tyrosine, 0.008% isoleucine, 0.005% phenylalanine, 0.01% glutamic acid, 0.02% threonine, 0.01% aspartic acid, 0.015% valine, 0.04% serine, 0.002% arginine, 0.0035% uracil, 0.002% adenine, 0.012% lysine, 0.008% tryptophan, 0.008% leucine and 0.002% histidine.

Vacuoles were isolated as described previously (Pieren et al., 2010). Briefly, yeast were grown overnight in YPD at 30°C. Cells were harvested (1 min, 4000 g, room temperature, JLA-10.500 rotor) at OD_600 nm=1.0–1.5, resuspended in 50 ml of 30 mM Tris-HCl pH 8.9 containing 10 mM DTT. After a 5-min incubation at 30°C, cells were sedimented and resuspended 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 U ml⁻¹ of lyticase). Spheroplasts were obtained after incubation for 26 min at 30°C and sedimented by centrifugation (1 min, 3500 g, 4°C, JA-25.50 rotor). The pellet was gently resuspended in 2.5 ml of 15% Ficoll 400 in PS buffer (10 mM PIPES-KOH pH 6.8, 200 mM sorbitol). Tubes were kept on ice for 2 min before addition of 80 μg or 100 μg DEAE dextran for BJ3505 and DKY6281 or BY4742 strains, respectively. Tubes were transferred to a water bath at 30°C for 90 s and chilled on ice. Spheroplasts were transferred to a SW41 tube and overlaid with 3 ml of 8% Ficoll, 3 ml of 4% Ficoll, and 2 ml 0% Ficoll. After centrifugation (90 min, 150,000 g, 2°C) organelles were collected from the 4%-Ficoll–0%-Ficoll interface. Protein concentrations were determined using a Bradford assay using bovine serum albumin as a standard.

To assay content mixing in vitro, 3 μg of vacuoles from BJ3505 and DKY6281 strains were mixed in 30 μl of fusion buffer (PS buffer containing 150 mM KCl and 0.5 mM MnCl₂). After addition of an ATP regeneration system (0.5 mM ATP, 0.5 mM MgCl₂, 20 mM creatine phosphate, 0.125 mg ml⁻¹ creatine kinase, final concentrations), fusion reactions were incubated at 27°C for 60 min. Then, 470 μl of the phosphatase assay mix (10 mM p-nitrophenyl-phosphate, 10 mM MgCl₂, 0.4% Triton X-100 in 250 mM Tris-HCl pH 8.9) was added to each tube, and the incubation was prolonged for 5 min at the same temperature. The reaction was stopped by the addition of 500 μl of 1 M glycine-KOH pH 11.5. Absorbance at 400 nm was read with a Spectramax photometer. Values were corrected for a background value obtained with a sample that had been kept on ice during the fusion reaction.
 Reporter protein expression of unprocessed Pho8 (pro-Pho8) was assayed by stimulating its activity with Zn²⁺. 10 mM ZnCl₂ was added to 3 µg of vacuoles in 30 µl PS buffer, supplemented with phosphatase assay mix, and alkaline phosphatase activity was developed as described above (Strasser et al., 2011). One unit is equal to the formation of 1 µmole of p-nitrophenol per min and per µg of BJ3505 vacuoles.

Lipid mixing was measured as described previously (Reese et al., 2005). Vacuoles from BY4742 or vtc mutants were isolated in the presence of 1× protease inhibitor cocktail (PIC) (7.5 μM pefablock SC, 17.6 nM leupeptin, 3.75 μM o-phenanthroline, 54.7 nM pepstatin A) and 1 mM PMSF. An aliquot of the suspension (186 μg of protein) was diluted to a final volume of 800 μl of PS buffer and equilibrated at 32°C for 40 s. Sixty microliters of Rhodamine–phosphatidylethanolamine (from a 3 mM stock in DMSO) was added drop-wise under continuous slow vortexing. The sample was incubated at 27°C for 2 min. Labeled vacuoles were re-isolated by centrifugation on a discontinuous Ficoll gradient. Mixtures of vacuoles were prepared at a ratio of 1:5 (labeled:unlabeled) in fusion buffer. Fusion reactions were performed in non-coated black 96-well plates that had been pre-incubated with 5% milk in water. Reactions were started by adding an ATP-regeneration system to the samples. Fluorescence changes (λ_exc=540 nm, λ_em=590 nm) were recorded with a SpectraMax Gemini XS plate reader at 27°C. Measurements were taken every 2 min for 32 min. At the end, 0.5% Triton X-100 (final concentration) was added to the sample, and fluorescence intensities were recorded for 10 min.

PolyP synthesis was followed using DAPI (Gerasimaite et al., 2014). 5 μg of vacuoles were suspended in a total volume of 100 μl at a concentration of 0.05 mg ml⁻¹ in fusion buffer containing an ATP-regeneration system (1 mM ATP, 1 mM MgCl₂, 40 mM creatine phosphate and 0.25 mg ml⁻¹ creatine kinase). Samples were incubated at 27°C for the indicated time. The reaction was stopped by the addition of Stop Solution (8 mM EDTA, 0.1% Triton X-100 and 10 μM DAPI, final concentrations). 250 μl was transferred to a black 96-well plate, and fluorescence intensity (λ_exc=415 nm, λ_em=550 nm) was measured with a SpectraMax GEMINI XS fluorescence plate reader.

Polyphosphate accumulation in whole cells was measured as described previously (Hothorn et al., 2009), except that the cells were grown in Hartwell's complete medium or in Hartwell's complete medium lacking uracil until an OD_600nm of around 1.5 was reached. Then, 1 OD unit of cells was harvested by centrifugation (14,000 g, 1 min, 20°C), resuspended in 50 μl of 1 M sulfuric acid, and vortexed for 10 s. An equal volume of 2 M sodium hydroxide was added to restore a neutral pH, and samples were vortexed again for 10 s. Two volumes of 1 M Tris-malate buffer pH 7.4 were added before the samples were briefly vortexed. Polyphosphate was purified on DNA-binding columns (Qiagen PCR purification Kit). The washing buffer was 10 mM Tris-HCl pH 7.0 containing 50% ethanol, 10 mM NaCl and 0.3 mM EDTA. The final elution was performed in 110 μl of the commercial Elution Buffer. 50 μl of the extracts was mixed in a white 96-well plate with an equal volume of 2× polyphosphatase buffer (10 mM Tris-HCl pH 7.0, 3 mM MgCl₂) containing 0.01 μg μl⁻¹ of the recombinant exopolyphosphatase Ppx1, or buffer only. Samples were incubated at 37°C for 1.5 h before addition of 150 μl of the malachite green reagent (86 μl of 28 mM ammonium heptamolybdate in 2.1 M sulfuric acid and 64 μl of 0.76 mM malachite green in 0.35% polyvinyl alcohol). Absorption at 595 nm was measured using a microplate photometer.

Vacuoles were suspended in 60 μl of fusion buffer at a concentration of 0.24 mg ml⁻¹. ATP-regeneration system or buffer was added to the reaction mixtures before the tubes were placed at 27°C for 15 min. Samples were then put back on ice and diluted to 400 μl with fusion buffer and immediately centrifuged (20,000 g, 20 min, 4°C). Pellets were washed once with fusion buffer and transferred to a new tube. Proteins in the supernatant were trichloroacetic-acid-precipitated (12.5% final concentration). Samples were then resuspended in loading buffer and analyzed by SDS-PAGE and western blotting. Signals were detected by using a LICOR Odyssey infrared fluorescence scanner and quantified with its software.

For cis-SNARE disruption assays, 90 μg of vacuoles were diluted in fusion buffer to a concentration of 0.2 mg ml⁻¹. Then, ATP-regeneration system or buffer was added as described for the content mixing assay, and samples were incubated for 20 min at 27°C. Vacuoles were harvested by centrifugation (7 min, 7000 g, 4°C) and resuspended in lysis buffer (20 mM HEPES buffer pH 7.4, 150 mM KCl, 0.5% Triton X-100, 10% glycerol, 1 mM PMSF, 1× PIC). Samples were solubilized by gentle agitation for 20 min at 4°C, and insoluble material was removed by centrifugation (15 min, 20,000 g, 4°C). Proteins in the supernatant were immunoprecipitated using immunoprecipitated using 1.5 μl of mouse ascites (Covance) containing anti-HA antibodies or 2 μg of affinity-purified anti-Nyv1 antibodies and 20 μl of protein-G agarose beads. After a 45-min incubation period at 4°C, beads were washed three times with a washing buffer (20 mM HEPES pH 7.4, 150 mM KCl, 0.2% Triton X- 100, 10% glycerol, 1 mM PMSF, 1× PIC). Proteins were eluted by heating the beads at 95°C for 5 min in 2× loading buffer for SDS-PAGE.

For cis-SNARE disruption assays on beads, vacuoles were solubilized in the lysis buffer, and insoluble materials were removed by centrifugation (15 min, 20,000 g, 4°C). Protein-G agarose beads and anti-HA antibodies were added to the supernatant, and immunoprecipitation was performed as described above. Beads were then washed three times with lysis buffer. MnCl₂ was then added to a final concentration of 0.5 mM. Samples were then incubated at 27°C for 20 min with additives where indicated. Beads were recovered by centrifugation and washed three times with the washing buffer. Samples were then processed as described above. When needed, cytosols were boiled (5 min at 95°C), and insoluble material was removed by centrifugation (15 min, 20,000 g, 4°C). The supernatant was used as ‘boiled cytosol’.

Proteins on isolated vacuoles were first crosslinked with the cleavable amine-reactive crosslinker DTSSP [3,3′-dithiobis (sulfosuccinimidylpropionate)]. 200 μg of isolated vacuoles was pelleted by centrifugation (7000 g, 10 min, 4°C) and resuspended in 700 μl of 20 mM HEPES buffer pH 7.4 containing 200 mM sorbitol 150 mM KCl and 2 mM DTSSP. Samples were incubated at 4°C with end-over-end agitation for 60 min. The reaction was quenched by the addition of 20 mM Tris-HCl pH 7.5, 150 mM KCl, 200 mM sorbitol (final concentrations), and the incubation was continued at 30°C for 30 min. Vacuoles were isolated by centrifugation (10,000 g, 10 min, 4°C), resuspended in lysis buffer, and immunoprecipitation was performed as described above. Samples were reduced in SDS- PAGE loading buffer containing 100 mM DTT for 30 min at 50°C.

Vacuole morphology was determined by FM4-64 staining of the vacuole membrane. Cells were grown overnight in Hartwell's complete medium to an OD_600nm of 0.4–1.0. Cultures were diluted to an OD_600nm of 0.3, and FM4-64 was added at a final concentration of 10 μM. Cells were incubated for 1.5 h at 25°C, washed in medium without dye and finally resuspended in fresh medium. The cells were further incubated at 25°C for 2 h, concentrated in their medium by short and mild centrifugation, resuspended in one-tenth of the supernatant and then analyzed with microscopy. For the localization of GFP-tagged proteins, cells were grown in Hartwell's complete medium lacking uracil (HC-URA) to logarithmic phase. Vacuoles were labeled as previously described, and the cells were analyzed with fluorescence microscopy. 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 100×1.4 NA lens. Pictures were collected with Perkin Elmer Volocity software and analyzed with the open-source software ImageJ.

Unless stated otherwise, data points with error bars represent the mean of three independent experiments. Error bars represent the standard deviation.

We thank Andrea Schmidt and Véronique Comte for assistance and the members of our laboratory for discussion.