Pore models of membrane fusion postulate that cylinders of integral membrane proteins can initiate a fusion pore after conformational rearrangement of pore subunits. In the fusion of yeast vacuoles, V-ATPase V0 sectors, which contain a central cylinder of membrane integral proteolipid subunits, associate to form a transcomplex that might resemble an intermediate postulated in some pore models. We tested the role of V0 sectors in vacuole fusion. V0 functions in fusion and proton translocation could be experimentally separated via the differential effects of mutations and inhibitory antibodies. Inactivation of the V0 subunit Vph1p blocked fusion in the terminal reaction stage that is independent of a proton gradient. Δvph1 mutants were capable of docking and trans-SNARE pairing and of subsequent release of lumenal Ca2+, but they did not fuse. The Ca2+-releasing channel appears to be tightly coupled to V0 because inactivation of Vph1p by antibodies blocked Ca2+ release. Vph1 deletion on only one fusion partner sufficed to severely reduce fusion activity. The functional requirement for Vph1p correlates to V0 transcomplex formation in that both occur after docking and Ca2+ release. These observations establish V0 as a crucial factor in vacuole fusion acting downstream of trans-SNARE pairing.
The fusion of intracellular membranes depends on a proteinaceous machinery which has been partially conserved in evolution. Homologous components act on different stations of membrane traffic in the secretory and endocytic pathways (Gotte et al., 2000; Chen and Scheller, 2001; Pelham, 2001). This led to the suggestion that all intracellular fusion reactions of eukaryotic cells might follow a universal reaction mechanism implemented by variations of the same core machinery, the SNARE proteins (Rothman, 1994). After activation by the ATPase Sec18p/NSF, cognate SNAREs from opposing membranes can form transcomplexes which are believed to force the membranes into close apposition and cause fusion. In line with this proposal, a basal fusion activity of SNAREs could be reconstituted from purified components inserted into liposomes (Weber et al., 1998).
The fusion of vacuoles from the yeast Saccharomyces cerevisiae shares many key features with other fusion reactions (Mayer, 2001). Thus, it can serve to test hypotheses about the fusion mechanism and about the role of specific conserved components. Vacuole fusion depends on the activation of t- and v-SNAREs by the ATPase Sec18p/NSF and its cofactor Sec17p/α-SNAP and on a Rab-GTPase, Ypt7p (Haas et al., 1995; Haas and Wickner, 1996; Mayer et al., 1996; Ungermann et al., 1999a). Ypt7p cooperates with the HOPS complex, an oligomeric assembly of tethering factors containing the class C Vps proteins (Price et al., 2000a,b; Sato et al., 2000; Seals et al., 2000; Wurmser et al., 2000). During priming, ATP hydrolysis by Sec18p/NSF disrupts cis-SNARE complexes (Nichols et al., 1997; Ungermann et al., 1998a) and yields SNAREs in a labile, activated state which is stabilized by the LMA1 complex (Xu and Wickner, 1996; Slusarewicz et al., 1997; Xu et al., 1997, 1998). Priming also releases the armadillo repeat protein Vac8p from SNAREs and triggers its palmitoylation (Veit et al., 2001; Rohde et al., 2003), a modification that might be relevant to the function of Vac8p in later stages of fusion (Wang et al., 2000). Priming facilitates tethering, the initial and less stable attachment of the fusion partners that depends on Ypt7p and the HOPS complex (Mayer and Wickner, 1997; Ungermann et al., 1998b; Price et al., 2000a). Specific interactions between HOPS and SNAREs involve the NH2-terminal domain of the SNARE Vam3p (Laage and Ungermann, 2001; Wang et al., 2001a). Tethering is a prerequisite for subsequent docking, a tighter binding of vacuoles that requires SNAREs and might involve the formation of trans-SNARE complexes, i.e., complexes of cognate t- and v-SNAREs on the opposing membranes (Ungermann et al., 1998b; Laage and Ungermann, 2001). Tethering and docking are accompanied by a concentration of many fusion-relevant components around the contact zones between vacuoles (Wang et al., 2002). Trans-SNARE complexes accumulate to low abundance during the fusion reaction (Ungermann et al., 1998b; Rohde et al., 2003). An enormous advantage of the vacuole fusion system is that trans-SNARE pairing can be directly assayed as an intermediate which is well integrated into the reaction pathway, a property that distinguishes it from the other major systems used to study membrane fusion. Notably, trans-SNARE pairs between vacuoles can be disassembled after docking without blocking further progression of fusion (Ungermann et al., 1998b). This indicates that SNAREs are required at least up to the docking stage but that trans-SNARE pairing may be dispensable for completion of the reaction.
Priming and docking also show specific lipid requirements, in particular for phosphatidylinositol 4,5-bisphosphate (Mayer et al., 2000), ergosterol (Kato and Wickner, 2001), and phosphatidylinositol 3-phosphate (Cheever et al., 2001; Boeddinghaus et al., 2002). Like exocytosis (Adamo et al., 1999, 2001; Guo et al., 2001; Zhang et al., 2001), vacuole fusion requires more than one small GTPase. In addition to the Rab-GTPase Ypt7p, the Rho-GTPases Cdc42p and Rho1p are involved (Eitzen et al., 2001; Muller et al., 2001), probably by regulating the remodeling of vacuolar actin. Dynamic changes of vacuolar actin occur during fusion (Eitzen et al., 2002; Seeley et al., 2002).
Vacuole docking triggers an efflux of calcium from the lumen of the organelle which fosters the binding of calmodulin to the membranes (Peters and Mayer, 1998). Calmodulin binds to a high molecular weight complex which contains the protein phosphatase 1 Glc7p (Peters et al., 1999) and V0 sectors, the membrane integral part of the vacuolar H+-ATPase (V-ATPase). Calmodulin was also found in association with the membrane integral VTC complex (Peters et al., 2001). The VTC complex binds to the V-ATPase, is required for the priming activity of Sec18p/NSF, and influences the binding of LMA1 to the membrane (Muller et al., 2002). VTC proteins may, therefore, couple Sec18p/NSF to V0, perhaps to activate it for a potential role in fusion. This speculation appears particularly attractive because a subset of V0 sectors, which is enriched in calmodulin and in the vacuolar t-SNARE Vam3p, form transcomplexes. V0 sectors from the opposing membranes bind to each other before fusion has been completed (Peters et al., 2001). A multimeric (probably hexameric) cylinder of proteolipids is a central part of the V0 sector (Hirata et al., 1997; Powell et al., 2000). Thus, V0 transcomplexes resemble the situation postulated in some pore models of fusion (Lindau and Almers, 1995; Mayer, 2001; Zimmerberg, 2001) which suggest that continuous proteinaceous pores composed of multiple subunits could mediate membrane fusion. This could occur, for example, by radial expansion of the cylinder and partial lateral separation of their subunits (Lindau and Almers, 1995) or by conformational changes of the pore subunits, resulting in rotation or the transient shifting of hydrophobic and hydrophilic surfaces (Zimmerberg, 2001). If vacuole fusion followed this or a similar pathway, proteolipids or entire V0 sectors should act in the postdocking phase of this reaction. We have begun to test this hypothesis by manipulating different V-ATPase subunits and assaying the effects on vacuolar membrane fusion.
Dissecting the role of V0 in vacuolar membrane fusion poses the problem to separate the requirement for a proton motive force (pmf)*
Abbreviations used in this paper: BCECF, 2′7′-bis-(2-carboxyethyl)-5-(and 6)-carboxyfluorescein; FCCP, carbonylcyanide-4-trifluormethoxyphenylhydrazon; pmf, proton motive force.
We pursued in vivo and in vitro approaches to address this question. V-ATPase–deficient mutants die on neutral media and survive with slow growth on acidic media (Stevens and Forgac, 1997; Kane, 1999). Due to altered physiological properties of the cells, this complicates an in vivo analysis of vacuole structure. All deletions of core V-ATPase subunits result in this “vma phenotype,” with one important exception: The a-subunit of the V-ATPase exists in two isoforms, the vacuolar isoform Vph1p and the Golgi or endosomal isoform Stv1p (Manolson et al., 1994). Both isoforms can substitute for one another in proton translocation, although they differ with respect to regulated V1/V0 dissociation that occurs upon glucose depletion (Parra and Kane, 1998; Kawasaki-Nishi et al., 2001a,b). A single deletion of either Vph1 or Stv1 does not result in the vma phenotype and permits healthy growth of the cells, also on neutral media (Manolson et al., 1994).
Deletion of the V0 subunit Vph1p results in vacuolar fusion defects
Absence of the vma phenotype makes vph1 mutants suitable for probing a role of V0 in vacuolar fusion in vivo. Mutation of fusion-relevant components frequently results in a fragmentation of vacuoles into multiple small vesicles in vivo (Raymond et al., 1992; Wada et al., 1992b). In some mutants, fragmentation does not occur, for example, after mutation of the v-SNARE Nyv1, Vtc1, or Vtc4 (Nichols et al., 1997; Muller et al., 2002). We assayed the in vivo effects of V-ATPase mutations by light microscopy after staining the vacuolar membranes with the red fluorescent vital dye FM4–64. Mutants lacking the vacuolar V0 subunit Vph1p showed numerous small vacuolar fragments which formed clusters (Fig. 1)
. This phenotype was observed with high frequency and resembles that of mutants in other genes with a function in the postdocking phase of vacuole fusion, such as Vtc3, Glc7 (protein phosphatase 1), calmodulin, or Vac8 (Peters and Mayer, 1998; Peters et al., 1999, 2001; Wang et al., 2001b; Muller et al., 2002). In contrast, mutation of Stv1, the Golgi/endosomal isoform of Vph1, did not result in vacuolar fragmentation (Fig. 1), nor did deletion of the V1 subunit Vma1p (unpublished data; with the strong caveat that this last mutant has the severe vma phenotype). Our results differ from those published in another study (Perzov et al., 2002) that reported vacuolar fragmentation for Δstv1 cells but not for Δvph1 cells. Due to this discrepancy, we generated Δvph1 mutants in three independent strain backgrounds and consistently observed vacuolar fragmentation. Currently, we have no explanation for this difference. However, our results are strongly corroborated by an unbiased microscopic screen for deletion mutants showing vacuolar fragmentation (Seeley et al., 2002). This genome wide screen identified mutants in four out of five Vo subunits as defective in vacuolar morphology, reporting a particularly strong phenotype for Δvph1 but none for Δstv1 or Δvma1.
In vivo experiments provide only indirect information about the proton uptake and fusion activities of the vacuoles. To overcome this limitation, we analyzed this aspect in an in vitro system reconstituting vacuolar fusion (Conradt et al., 1992). Vacuoles can be extracted from yeast spheroplasts after enzymatic digestion of the cell wall and gentle lysis by DEAE-dextrane, a treatment that leaves vacuoles intact and fusion competent. The isolated organelles fuse upon incubation with an ATP-regenerating system and cytosolic extracts. Since both V1 and V0 sectors are essential for proton translocation activity of the V-ATPase (Stevens and Forgac, 1997; Kane, 1999), we compared the fusion competence of vacuoles from V1 and V0 mutants, testing whether this V-ATPase activity was required for vacuole fusion. Vacuoles purified from the V0 mutant Δvph1 were deficient for in vitro fusion. The fusion defect was even observed if only one of the fusion partners was a Δvph1 vacuole and the other fusion partner was a wild-type vacuole (Fig. 2
A). This suggests that Vph1p is required for fusion on both fusion partners. Vacuoles from the V1 mutant Δvma1 fused with ∼50% of wild-type activity (Fig. 2 A). Addition of a cytosolic extract increased the fusion activity of Δvma1 vacuoles to 70%. This activation was not due to readdition of intact V1 subunits via the cytosolic extract because immunodepletion of V1 from these extracts (Fig. 2 B) did not diminish their activities (Fig. 2 A). Δvph1 vacuoles remained fusion incompetent under all conditions. Since both deletions of the V0 subunit Vph1 and deletion of the V1 subunit Vma1 completely abolish the pump activity of the V-ATPase (Stevens and Forgac, 1997; Kane, 1999), the difference in fusion activity between these strains is best explained by a function of Vph1p for vacuolar fusion that is independent of proton translocation.
Vacuolar proton uptake is not limiting for fusion of Δvph1 vacuoles
Since vacuole fusion depends on a pmf across the vacuolar membrane (Conradt et al., 1994; Ungermann et al., 1999b) and the V-ATPase is the major vacuolar proton pump (Stevens and Forgac, 1997), we asked whether the fusion activity of Δvph1 vacuoles could be limited by lack of a pmf. We tested whether Δvph1 vacuoles are able to generate a basal pmf by a V-ATPase–independent pathway and whether this pmf would be above the threshold required for fusion. V-ATPase–independent antiport mechanisms can reconstitute a basal proton uptake activity by antiport of, for example, amino acids and protons (Ohsumi and Anraku, 1981, 1983; Sato et al., 1984; Klionsky et al., 1990; Wada et al., 1992a; Wada and Anraku, 1994) and/or by combination of an active vacuolar Ca2+ pump and a Ca2+/H+ antiporter (Ohsumi and Anraku, 1983; Cunningham and Fink, 1994; Forster and Kane, 2000). The pmf as such is important to drive fusion, but for the purpose of our experiments it is irrelevant whether proton uptake occurs via the V-ATPase or alternative mechanisms. We measured the apparent proton uptake activity of whole vacuoles under the conditions of our fusion reaction by following the change in absorbance of acridine orange at 491 and 540 nm, a common assay for the vacuolar pmf (Gluck et al., 1982). Since acridine orange is also fluorescent, we could monitor its distribution in the membrane preparation by fluorescence microscopy. At least 92% of the dye localized to structures that, by size and morphology, could be identified as vacuoles (unpublished data). Therefore, acridine orange will mostly report changes of the pmf on vacuoles. At standard dye to membrane ratios wild-type vacuoles showed a clear ATP-dependent decrease in the absorbance signal, indicating that a pmf had formed (Fig. 3
A, bottom panel). Addition of the protonophor carbonylcyanide-4-trifluormethoxyphenylhydrazon (FCCP) at the end of the assay period collapsed the pmf and restored the signal to the initial values. No signal was observed in the absence of ATP or after addition of even low concentrations (0.1 μM) of the V-ATPase inhibitor concanamycin A. Concanamycin A blocks V-ATPase–dependent H+ translocation at nanomolar concentrations (Drose and Altendorf, 1997). Our observation is consistent with earlier results from various other groups using subvacuolar vesicles (Drose and Altendorf, 1997). These studies demonstrated that the V-ATPase was the main source of the vacuolar pmf and could be completely inhibited by bafilomycin A or concanamycin A. At lower dye to vacuole ratios (Fig. 3 A, top two panels), however, the assay was more sensitive and could still detect a pmf in the presence of 0.1 μM concanamycin A. 1 and 5 μM of the inhibitor reduced the signal further and yielded overlapping curves, suggesting that these concentrations had saturated the system and that V-ATPase had been completely inhibited. Nevertheless, significant ATP-dependent proton uptake activity remained even at these high concentrations. This concanamycin-insensitive activity was comparable to the levels of ATP-dependent proton uptake observable with vacuoles prepared from V1 (Δvma2) or V0 (Δvph1) mutant cells (Fig. 3 B). The signal in V-ATPase mutant vacuoles could not be further reduced by concanamycin A. Therefore, this signal appears to reflect the V-ATPase–independent part of vacuolar proton uptake.
Next, we asked whether the V-ATPase–independent proton uptake activity could suffice to drive fusion. We compared the fusion and proton translocation activities of Δvph1 vacuoles to those of wild-type vacuoles in which V-ATPase pump activity had been pharmacologically suppressed with concanamycin A. In this case, we measured apparent proton uptake activity by following alkalinization of the medium with the pH-sensitive dye 2′7′-bis-[2-carboxyethyl]-5-[and 6]-carboxyfluorescein (BCECF; Peters et al., 2001). Unlike acridine orange, which inserts into the vacuolar membrane (unpublished data), this dye is coupled to high molecular weight dextrane, does not interfere with vacuole fusion, and permits assay of fusion and proton translocation in the same samples. In this assay, the apparent proton translocation activity of Δvph1 vacuoles was equal to that of wild-type vacuoles treated with 1 μM concanamycin A (Fig. 4
B). As in the acridine orange assay, this signal was ATP dependent and FCCP sensitive, suggesting that it reflected the same V-ATPase–independent component of vacuolar proton uptake. Comparison of the fusion activities revealed that Δvph1 vacuoles did not fuse, but wild-type vacuoles with 1 μM concanamycin A retained full fusion activity (Fig. 4 A). FCCP dissipates the proton gradient across the vacuolar membrane, eliminating both its V-ATPase–dependent and V-ATPase–independent components. In agreement with earlier studies (Conradt et al., 1994; Mayer et al., 1996; Ungermann et al., 1999b), FCCP inhibited fusion. Together, vacuole fusion seems to depend on a pmf across the vacuolar membrane but not on V-ATPase proton pump activity. Hence, the fusion defect of Δvph1 vacuoles cannot be explained by a reduction in the apparent proton uptake activity, supporting a direct function of V0 in vacuole fusion.
We sought to specifically interfere with the fusion relevant aspect of V0. Available low molecular weight V-ATPase inhibitors affect only the proton translocation function of the V-ATPase, (e.g., concanamycin A) and/or are slowly reacting and of unsatisfactory specificity (e.g., dicyclohexylcarbodiimide). We raised antibodies to V0 subunits in order to generate highly specific agents that might interfere with V0 function and be suitable for analysis of the fusion reaction. We immunized goats with the recombinantly expressed and purified NH2-terminal cytosolic part of Vph1p and prepared affinity-purified polyclonal antibodies from the sera. If titrated into in vitro fusion reactions, these antibodies inhibited fusion in a concentration-dependent manner (IC50= 6 μM; Fig. 5
A). Nonimmune antibodies had no effect at comparable concentrations (Fig. 5 B). To exclude potential interference due to the divalent nature of whole IgG, we also prepared monovalent Fab fragments from these antibodies. They also inhibited fusion at concentrations comparable to those observed for whole IgGs (Fig. 5 C), matching our general experience for many other fusion relevant components: the inhibitory potency of whole IgGs and Fab fragments is not significantly different (Haas et al., 1995; Haas and Wickner, 1996; Mayer and Wickner, 1997; Nichols et al., 1997).
As described for the Δvph1 mutants above, we tested whether antibodies to Vph1p would interfere with the apparent vacuolar proton uptake. This was not the case, although the antibodies blocked fusion (Fig. 6)
. The lack of effect of anti-Vph1p on proton uptake is probably due to the fact that the antibodies recognize only solitary V0 sectors but not the assembled V0/V1 holoenzyme (unpublished data). The antibodies were also tested on vacuoles from which the V1 sector had been removed by cold extraction with KSCN and ATP, a treatment that leaves the V0 sector in the membranes (Arai et al., 1989; Moriyama and Nelson, 1989; Adachi et al., 1990; Peters et al., 2001) but inactivates V-ATPase–dependent proton translocation (Fig. 7
B). Fusion of such vacuoles was inhibited by anti-Vph1p as efficiently as by GTPγS or antibodies to Sec18p/NSF (Fig. 7 A), two established antagonists of vacuole fusion (Haas et al., 1994; Mayer et al., 1996). Thus, antibodies to the cytosolic part of Vph1p provide an additional tool to separate the fusion function of V0 from its role in proton pumping.
Vph1p acts in the postdocking phase of vacuole fusion
The antibodies to Vph1p were used for kinetic analyses in order to explore which stage of vacuole fusion required Vph1p. First, we tested when the reaction became resistant to anti-Vph1p (Fig. 8
A). The majority of vacuoles completes distinct reaction steps within defined intervals and becomes resistant to inhibitors of these steps (Conradt et al., 1994; Mayer et al., 1996), revealing the sequence of events. Inhibitors were added to an aliquot of an ongoing fusion reaction at different times after the start point. Then, the incubation was continued until the end of a standard fusion period (70 min). Another aliquot was chilled to stop fusion at the time of inhibitor addition. All inhibitors abolished fusion when added at the start of the reaction (Fig. 8 A). After 35 min, the reaction was resistant to antibodies to the t-SNARE Vam3p, indicating the completion of docking (Mayer and Wickner, 1997; Ungermann et al., 1998b). The inhibition curve for anti-Vph1p lagged significantly behind that for anti-Vam3p, suggesting that Vph1p was required past the docking stage. This behavior resembled the inhibition curves obtained previously with calmodulin antagonists or with the Ca2+ chelator BAPTA, which inhibit completion of fusion after the docking step (Peters and Mayer, 1998; Ungermann et al., 1999b).
Trans-SNARE pairing is believed to be a key event in docking membranes and committing them to fusion (Sollner et al., 1993; Weber et al., 1998). It can be assayed as a molecular indicator of docking (Ungermann et al., 1998b). To test whether deletion of Vph1 affected a step before or concomitant with docking, or rather an event required for subsequent fusion per se, we tested whether Δvph1 vacuoles were able to form trans-SNARE complexes. Yeast vacuoles offer the possibility to measure trans-SNARE pairing by use of a combination of vacuoles lacking either the vacuolar v-SNARE Nyv1p or the vacuolar t-SNARE Vam3p (Ungermann et al., 1998b). These vacuoles are prepared from separate strains and mixed in vitro. v/t-SNARE complexes can only be formed if a v-SNARE–containing vacuole docks to a t-SNARE–containing vacuole. If this experiment is performed in the presence of a late acting fusion inhibitor, such as GTPγS, accumulating SNARE complexes represent transcomplexes. We generated a strain carrying deletions of Vam3 (t-SNARE) and Vph1. Vacuoles from this strain were fused with vacuoles from a Δnyv1 (v-SNARE) strain. In agreement with previous results (Ungermann et al., 1998b), SNARE complexes accumulated at 27°C but not on ice (Fig. 8 B). There was no significant difference in the yield of trans-SNARE pairing between VPH1–containing vacuoles and Δvph1 vacuoles. The docking inhibitor Gdi1p, which extracts the Rab-GTPase Ypt7p from the membranes, suppressed formation of trans-SNARE complexes, demonstrating that the signal from the coimmunoprecipitation assay was specific and a consequence of docking. GTPγS, which permits docking and the formation of V0 transcomplexes (Peters et al., 2001) but not fusion, did not interfere with trans-SNARE pairing. Since trans-SNARE pairing depends on a pmf (Ungermann et al., 1999b), this experiment also indicates that the apparent proton uptake activity of Δvph1 vacuoles sufficed to drive trans-SNARE pairing. More importantly, however, it demonstrates that Δvph1 vacuoles do not fuse (Fig. 2), although they form trans-SNARE complexes (Fig. 8 B), indicating that a step subsequent to or independent of trans-SNARE pairing is blocked by deletion of Vph1.
Coupling of Vph1p to the Ca2+-releasing channel
Docking triggers a Ca2+ release from the vacuolar lumen that strengthens the binding of calmodulin to the vacuolar membrane and triggers fusion (Peters and Mayer, 1998). We tested whether Vph1p function would even be required after the Ca2+/calmodulin-dependent step by accumulating reaction intermediates in the presence of the Ca2+ chelator BAPTA. This reversible block can be relieved by removing BAPTA and replenishing Ca2+ (Ungermann et al., 1999b). Completion of the reaction from this stage still showed significant sensitivity to anti-Vph1p (Fig. 8 C), although the effect was not as pronounced as for the low molecular weight inhibitor GTPγS. This suggests that the antibody-sensitive stage might be tightly coupled to the Ca2+-requiring step and/or be overcome rapidly after removal of BAPTA. The requirement for Vph1p can obviously not be completely satisfied before completion of the Ca2+/calmodulin-dependent step. Together, the results from the kinetic analysis (Fig. 8 A), the direct assay of trans-SNARE pairing (Fig. 8 B), and the staging experiments (Fig. 8 C) all suggest that Vph1p functions after the completion of docking and trans-SNARE pairing, concomitant with or after the release of lumenal Ca2+ and close to the completion of fusion.
The relationship of Ca2+ release and Vph1p function in fusion was further characterized by monitoring Ca2+ release in the course of fusion. Fusion reactions were run in the presence of aequorin and coelenterazine (Muller et al., 2001). Aequorin is a protein that oxidizes its substrate coelenterazine in a Ca2+-dependent fashion, emitting light (Shimomura et al., 1993). This permits continuous recording of Ca2+ levels in the buffer of ongoing fusion reactions. In agreement with previous results obtained with a different approach (Peters and Mayer, 1998), treatments preventing priming, such as antibodies to Sec18p/NSF or Sec17p/α-SNAP, suppressed the release of Ca2+ from the vacuolar lumen (Fig. 9
A). Docking inhibitors, such as antibodies to the t-SNAREs Vam3p and Vam7p had the same effect (not depicted). Surprisingly, antibodies to Vph1p also inhibited Ca2+ release, suggesting that either Vph1p or V0 itself might be involved in Ca2+ release or that Vph1p might be coupled to the Ca2+ channel. In both cases, adding an antibody to Vph1p might block a conformational change required for activation of the channel. Only in the first case, however, should deletion of Vph1p, which eliminates the vacuolar V0 sector as a whole (Stevens and Forgac, 1997; Kane and Parra, 2000), abolish Ca2+ release. This was not the case. Δvph1 vacuoles showed Ca2+efflux just as their wild-type counterparts (Fig. 9 B). Efflux was sensitive to the Rab-GTPase inhibitor Gdi1p, demonstrating that it depended on docking and was triggered by the authentic fusion pathway. This permits several conclusions. First, it suggests that the fusion defect of Δvph1 vacuoles is not due to the inability to release calcium. Second, Vph1p itself is not part of the Ca2+-releasing channel. Third, Vph1p may be coupled to the Ca2+ channel, leaving calcium release intact if Vph1p is absent but blocking it if Vph1p is inactivated by an antibody.
A previous study demonstrated that V0 sectors from opposing vacuolar membranes can associate into transcomplexes before fusion has occurred (Peters et al., 2001). Vacuolar V0 sectors consist of a cylinder composed of (probably six) proteolipids, the large multispanning integral membrane protein Vph1p, and the peripheral subunit Vma6p (Kane et al., 1989; Hirata et al., 1997; Powell et al., 2000). V0 transcomplexes might thus connect proteolipid cylinders from opposing membranes, creating a condition that might resemble the situation postulated in pore models of fusion (Lindau and Almers, 1995; Zimmerberg, 2001). Formation of V0 transcomplexes requires ATP, the Rab-GTPase Ypt7p, and trans-SNARE pairing. Therefore, it depends on the early reaction steps of priming and docking (Peters et al., 2001). Maintenance of V0 transcomplexes is independent of trans-SNARE pairing. These observations link V0 transcomplex formation into the reaction sequence. It is important, however, to investigate whether V0 is required in the final stage of fusion. The results presented here address this aspect.
First, we demonstrate that V0 does not only form transcomplexes but is critically involved in vacuole fusion in vitro and in vivo. The V0 requirement kinetically mapped to the reaction phase that follows completion of docking and trans-SNARE pairing (Ungermann et al., 1998b) and is independent of a proton gradient (Mayer et al., 1996; Mayer and Wickner, 1997; Ungermann et al., 1999b). However, this terminal reaction phase depends on Ca2+ release from the vacuolar lumen which triggers bilayer fusion (Peters and Mayer, 1998). This is consistent with the observation that V0 transcomplexes form after docking and depend on Ca2+ release (Peters et al., 2001). It establishes a correlation between the timing of V0 function in fusion and V0 transcomplex formation, strengthening the hypothesis that V0 transcomplex formation is tied to V0 function in vacuole fusion. Second, we demonstrate that Vph1p is required on both fusion partners. This symmetric requirement for Vph1p is consistent with a feature of some published pore models, namely that two membrane integral multisubunit cylinders connect the two membranes before fusion (Lindau and Almers, 1995; Zimmerberg, 2001).
Several lines of evidence suggest that trans-SNARE pairing is a necessary intermediate in vacuole fusion but that further events are required to drive the transition from docking to full fusion. First, quenching of the Ca2+ efflux by BAPTA reduces fusion activity to 10% of the control samples (Peters and Mayer, 1998), although BAPTA permits trans-SNARE pairing (Ungermann et al., 1999b). However, BAPTA blocks V0 transcomplex formation completely (Peters et al., 2001). Second, deletion of Vtc3 permits docking of the vacuoles, Ca2+ release, and V0 transcomplex formation but not fusion (Muller et al., 2002). Third, inactivation of V0 by deletion of Vph1p reduced fusion activity to 10–15% of the control value, although trans-SNARE pairing was unaffected. This demonstrates that V0 is crucial for most of the observable fusion activity of vacuoles. Fourth, trans-SNARE complexes appear to be largely dispensable after the BAPTA-sensitive stage (Ungermann et al., 1998a, 1999b) but V0 is not. This correlates to the behavior of V0 transcomplexes because docking and trans-SNARE pairing are required to establish V0 transcomplexes. Once formed, however, V0 transcomplexes persist in the absence of SNARE complexes (Peters et al., 2001).
The 10–15% of fusion activity observed after V0 inactivation is probably caused by trans-SNARE complexes, reflecting a basal level of fusion that may result from forcing the membranes into close proximity. It is possible that this basal activity is what could be reconstituted with purified SNAREs in liposomes. Although these liposomes contained high concentrations of SNAREs, they yielded significantly lower fusion rates than the equivalent complete fusion system can show in situ (Weber et al., 1998; Parlati et al., 1999). Lowering the v-SNARE concentration in liposomes to roughly physiological levels abolished this fusion activity (Mahal et al., 2002). It made the system completely dependent on an additional factor, the exocytic Ca2+ sensor synaptotagmin. Synaptotagmin strongly stimulated fusion, though in this case in a Ca2+-independent fashion.
It appears that both in triggered fusion events, such as regulated exocytosis, and in “constitutive” fusion events, such as vacuole fusion, the membranes require additional factors to efficiently progress from docking and trans-SNARE pairing to fusion. Such a function may be fulfilled by synaptotagmin in regulated exocytosis (Fernandez-Chacon et al., 2001; Chapman, 2002) or by V0 and Vtc3p in vacuole fusion. It is possible that proteins with related properties participate in other membrane fusion reactions. Candidates exist, as exemplified by Got1p and Sft2p, tetraspanning membrane proteins with a genetic interaction with the t-SNARE Sed5p and some similarities to the V0 proteolipids. Got1p is required in a postdocking stage of ER-Golgi trafficking and forms multimeric assemblies in the membrane (Conchon et al., 1999). SCAMP2 was identified as a factor required for endocytosis but also for a very late step in mast cell exocytosis. SCAMP2 is also a tetraspanning membrane protein that belongs to a protein family with members on various compartments, exists as a multimeric complex in the membrane, and binds to the t-SNARE subunit SNAP-23 (Singleton et al., 1997; Fernandez-Chacon et al., 2000; Fernandez-Chacon and Sudhof, 2000; Guo et al., 2002).
Ca2+ is released from the vacuolar lumen in a way that depends on priming and docking (Peters and Mayer, 1998). Now we have discovered that inactivation of the V0 subunit Vph1p by antibodies prevents this release, but deletion of Vph1p does not. Since inactivation of Vph1p by deletion or antibodies interferes neither with priming nor with docking (Fig. 8 B; unpublished data), we propose that the effect of Vph1p antibodies on Ca2+ release may reflect a coupling of Vph1p to the putative Ca2+ channel. Whether this coupling would be direct or indirect via shared binding partners cannot be judged at the moment. But it is striking that V0 is associated with the t-SNARE Vam3p (Peters et al., 2001) and that t-SNAREs in regulated exocytosis were shown to specifically bind to calcium channels (Leveque et al., 1994; Sheng et al., 1994). Our hypothesis is further supported by the differential effects of Ca2+ chelators on fusion (Peters and Mayer, 1998). Only BAPTA, a Ca2+ chelator with a rapid association rate, can block fusion. EGTA, which has a comparable Kd for Ca2+ but a much slower association rate, is not effective. This difference is best explained by assuming that the site of Ca2+ release and the Ca2+ effector proteins (in case of the vacuoles the effectors are most likely V0 and calmodulin [Peters et al., 2001]) are tightly associated. Then, chelators must act rapidly to be able to quench Ca2+ before it diffuses to and binds its receptor site (Neher, 1998). An analogous situation exists in regulated exocytosis. The triggering stage of this reaction is only sensitive to fast Ca2+ chelators (Adler et al., 1991), and the Ca2+ channels are associated with syntaxin (t-SNARE) in the plasma membrane (Leveque et al., 1994; Sheng et al., 1994). They should hence be close to the fusion site. This provides a physical correlate for the differential sensitivity of exocytosis to BAPTA and EGTA.
The reconstituted proteolipid cylinder of the V-ATPase can form large pores in response to binding of Ca2+ and calmodulin, demonstrating a potential for significant conformational changes in the membrane (Peters et al., 2001). On the other hand, a fraction of V0 in vacuolar membranes is associated with the VTC complex, a membrane integral complex that is required for the priming activity of Sec18p/NSF (Muller et al., 2002) and for membrane binding of LMA1, i.e., it is coupled to two components that are directly involved in generating (Sec18p; Mayer et al., 1996; Ungermann et al., 1998a) or maintaining (LMA1; Mayer and Wickner, 1997; Xu et al., 1997) an activated state of the fusion machinery. Therefore, we speculate that Vph1p, or the entire V0 sector, might undergo significant conformational changes during fusion, perhaps driven by Sec18p/NSF. Antibodies to Vph1p may prevent such conformational changes and thereby also block Ca2+ release through a channel associated with V0. Such coupling should be functionally relevant because Ca2+ release from the lumen of an organelle can only be transient. If the fusion system is not competent to respond to Ca2+ at the time of release, the reaction might be nonproductive. One might expect higher efficiency of the overall process if Ca2+ release and the Ca2+ controlled fusion system were activated in a coordinated fashion. This question can be answered with certainty only once the Ca2+-releasing channel will have been identified and assays for conformational changes of V0 during fusion will be available.
Materials And Methods
BJ3505 and DKY6281 are standard fusion strains (Mayer et al., 1996). Strain no. 120 is BJ3505 (Δnyv1) (Nichols et al., 1997). Strain no. 418 is DKY 6281 Δvam3 provided by Christian Ungerman (Biochemiezentrum, Heidelberg, Germany). SBY521 (Δvam3, Δvph1) was derived from no. 418 by replacing VPH1 with a kanamycin resistance cassette as described below. BY4742 (Mat α his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) and its Δvph1::kanMX4 or Δstv1::kanMX4 derivates were purchased from EUROSCARF.
Deletion strains of VPH1 were made by replacing the gene with a loxP-kanMX-loxP cassette from plasmid pUG6 (Guldener et al., 1996) using the following primers: Vph1 forward K.O., GCCTCTCAATATATCTATACTTGCTTAGAGGGCTACCTGTGTCAGCTGAAGCTTCGTACGC and Vph1 reverse K.O., CTGGTGGATTGGATTGCAAGTCTAACGTTTTCATGAGATAAGGCATAGGCCACTAGTGGATCTG. Vma1 and Vma2 were deleted by replacement with the HIS3 cassette from plasmid pRS 303 using the following primers: VMA1 forward, ATGGCTGGTGCAATTGAAAAGCTCGTAAGGAAATAAAAAGAATCAGATTGTACTGAGAGTGCAC, VMA1 reverse, CAGACAAGGCCGATTTACAACTAATTGAACAACTTGTTCCTG-TGCGGTATTTCACACCG, VMA2 forward, GGTGTGAACGGTCCATT-AGTCATTTGGAAAAGGTCAAGTTCCCAGATTGTACTGAGAGTGCAC, and VMA2 reverse, GAATCTTTGGGGAGATTCTATTCAACATCTCCTTAGGGTAGCTGTGCGGTATTTCACACCG. All deletions were verified by diagnostic colony PCR and by Western blotting of purified vacuoles from the mutants.
Vacuole isolation, cytosol preparation, fusion, purification of antibodies or Gdi1p, inhibitors, and FM4–64 staining were as described (Peters and Mayer, 1998; Peters et al., 1999). PS buffer is 10 mM Pipes/KOH, pH 6.8, and 200 mM sorbitol. 1× PIC is a protease inhibitor cocktail: 100 μM pefabloc SC, 100 ng/ml leupeptin, 50 μM o-phenanthroline, and 500 ng/ml pepstatin A. Vacuoles without V1 sector were prepared as described (Peters et al., 2001). In experiments with BAPTA, the alkaline phosphatase assay mixture contained 10 mM CaCl2. All samples were centrifuged (1 min, 14, 000 g) before spectrophotometry to remove precipitated Ca2+ salts.
Proton uptake activity of vacuoles
For the determination of pump activity, 100-μl fusion reactions were incubated at 25°C with 0.5 μM BCECF-dextran (70 K; Molecular Probes). The change of fluorescence was measured at an excitation of 490 nm and an emission of 520 nm in a Perkin Elmer LS50B fluorometer or in a plate reader fluorometer (Spectra Max Gemini; Molecular Devices). Initial pump activity was determined as described (Peters et al., 2001). In brief, the pH shift in the medium was followed via the pH-dependent shift in fluorescence of BCECF. This shift could be calibrated by buffering the solution to different pH values using PS buffer with 150 mM KCl.
Acridine orange assay.
Proton uptake of vacuoles was measured essentially as described. The absorbance changes of acridine orange at 491 and 540 nm were followed in a Beckman Coulter DU-600 spectrophotometer (Gluck et al., 1982). The difference between these values at any time point serves as a measure of the vacuolar pmf. The reaction mixture in a final volume of 100 μl contained 20 μg of vacuoles under standard fusion conditions with 15 μM acridine orange. The reaction was started by the addition of 5 μl of ATP regenerating system. At the end, 30 μM FCCP were added. Proton uptake activity was defined as the absorbance change in the first 20 s of the reaction.
120-μl fusion reactions (containing 24-μg vacuoles) with cytosol (0.5 μg/μl) were incubated for 50 min. The reactions were stopped by chilling the samples on ice. 380 μl of precipitation buffer were added (150 mM KCl, 1.3% [wt/vol] Triton X-100, 1.3 mM PMSF, 1.2× PIC, 13 mM EDTA, 200 mM Sorbitol, 10 mM Pipes/KOH, pH 6.8). Samples were shaken (10 min, 4°C) and centrifuged (21,000 g, 10 min). 450 μl of the solubilisate were transferred to a new reaction tube. 3% of the volume was kept as load control. 14 μg of affinity-purified anti-Vam3p antibody and 100 μl of a 1:4 slurry of protein A–Sepharose CL-4B (Amersham Biosciences) preequilibrated in washing buffer (150 mM KCl, 1% [wt/vol] Triton X-100, 1 mM PMSF, 1× PIC, 10 mM EDTA, 200 mM Sorbitol, 10 mM Pipes/KOH, pH 6.8) were added to each sample. After 1 h of shaking at 4°C, the protein A–Sepharose was washed three times with washing buffer. Residual liquid was removed with a Hamilton pipette. 80 μl SDS sample buffer was added, and aliquots were analyzed by gel electrophoresis and Western blotting.
Generation of antibodies to Vph1p
Vph1 cytosolic domain (Vph1-Cyt) was cloned with an NH2-terminal His6 tag for expression in Escherichia coli. The coding region of this domain was amplified by PCR using primers cyt Vph1 forward, GCAGAGAAGGAGGAAGCGATTTTTC, cyt Vph1 reverse, TCACTGGACTTCAATTTGATCGGT, and genomic DNA as a template. The PCR product was cloned into pCR T7/NT-TOPO (Invitrogen). The resulting plasmid was transformed into E. coli BL21. Cells were grown in LB/Ampicillin medium (37°C, 225 rpm) to an OD600 of ∼0.6. Expression was induced with 1 mM IPTG at 37°C for 4 h. Cells were harvested by centrifugation (10,000 g, 7 min), washed with twofold-concentrated PBS, and lysed by two freeze–thaw cycles and sonication in twofold-concentrated PBS, 8 M urea. The lysate (50 ml out of 8 liters culture) was centrifuged (20,000 g, 15 min, 4°C) and batch incubated with Ni-NTA resin (1 ml bed volume, 2 h, 4°C). The resin was washed with 20 ml of twofold-concentrated PBS, 0.05% SDS and with 20 ml twofold-concentrated PBS, 0.05% SDS, and 10 mM imidazole. Urea was omitted during washing to avoid interference in the following coupling reaction. 0.05% SDS was added to avoid precipitation during elution from the column. The protein was eluted with 150 mM imidazole. The eluate (containing 3 mg of protein) was directly coupled (1 h, RT) to 1.5 ml CH-activated Sepharose 4 B (Amersham Biosciences).
Antibodies to Vph1-Cyt were raised by injecting purified recombinant protein eluate (300–500 μg per boost) into a goat. Serum (10 ml) was affinity purified using the recombinant protein immobilized on activated CH–Sepharose 4B. The serum was diluted with 1 vol PBS and passed over the column at RT. The column was washed with 20–25 column volumes of PBS. Antibodies were eluted with 0.1 M glycine pH 2.75/150 mM NaCl, and the pH was immediately neutralized by adding 1 M Tris/HCl, pH 8.8. The antibodies were transferred into PS buffer with 150 mM KCl by repeated dilution and ultrafiltration in centricon-30 (Amicon) and stored at −20°C. Control antibodies were affinity purified with protein G from preimmune serum and treated as above.
Preparation of Fab fragments
Fab fragments were prepared from Vph1p antibodies essentially according to the manufacturer's instructions (ImmunoPure Fab Preparation kit; Pierce Chemical Co.). 20 mg of purified antibody were diluted with 1 vol of digestion buffer (as supplied). The solution (2 ml) was incubated overnight at 30°C with 0.5 ml immobilized papain. The sample was centrifuged, and the supernatant was passed over a protein G column to remove uncleaved antibodies and Fc Fragments. The flowthrough containing the Fab fragments was repeatedly diluted and concentrated to exchange the buffer for PS with 150 mM KCl.
Ca2+ release from vacuoles in the course of the fusion reaction was assayed similarly as described (Peters and Mayer, 1998) with the following modifications. Ca2+ was directly measured with aequorin in ongoing fusion reactions running in parallel in a microtiter plate luminometer (EG&G Berthold). Reactions (30 μl) contained 10 μg of vacuoles from strain BJ3505 in PS buffer with 150 mM KCl and 0.5 μl cytosol (30 mg/ml). Respective inhibitors were included. After 2 min of preincubation on ice, 2 μl of ATP regenerating system (same composition as in the fusion assay) and 3 μl of a freshly prepared 2:2:1 mix of the following stock solutions were added: 2.5 μM aequorin (Aqualite; Molecular Probes), 1 mM EGTA/KOH, pH 7.5, and 500 μM native coelenterazine (Molecular Probes) in EtOH. Samples were pipetted into a microtiter plate, and light emission of aequorin was measured with the luminometer for 60 min at 27°C. The assay was calibrated with solutions of different calcium concentrations as described (Peters and Mayer, 1998).
Cells were grown logarithmically in liquid YPD overnight (OD600 < 2) at 30°C. Cells were harvested (2 min, 3,000 g), resuspended in YPD with 10 μM FM4–64 at OD600 = 1, and shaken for 3 h at 30°C. Cells were harvested as above, resuspended in YPD at OD600 = 0.2, and shaken for 2–3 h at 30°C. For microscopy, individual samples were removed from the shaker. Cells were harvested (1 min, 3,000 g), resuspended in YPD at OD600 = 10, and immediately analyzed by fluorescence microscopy. For quantitation of vacuole morphology, photos of at least 20 random fields were taken. Vacuoles were counted in at least 200 cells per experiment and condition.
We thank Michael Harrison (School of Biochemistry and Molecular Biology, University of Leeds, UK) for plasmids and Christa Baradoy for technical assistance.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB184) and from the Human Frontier Science Program.
M.J. Bayer, C. Reese, and S. Bühler contributed equally to this paper.