Diacylglycerol and Its Formation by Phospholipase C Regulate Rab- and SNARE-dependent Yeast Vacuole Fusion*

Although diacylglycerol (DAG) can trigger liposome fusion, biological membrane fusion requires Rab and SNARE proteins. We have investigated whether DAG and phosphoinositide-specific phospholipase C (PLC) have a role in the Rab- and SNARE-dependent homo-typic vacuole fusion in Saccharomyces cerevisiae. Vacuole fusion was blocked when DAG was sequestered by a recombinant C1b domain. DAG underwent ATP-dependent turnover during vacuole fusion, but was replenished by the hydrolysis of phosphatidylinositol 4,5-bisphosphate to DAG by PLC. The PLC inhibitors 3-nitrocoumarin and U73122 blocked vacuole fusion in vitro, whereas their inactive homologues did not. Plc1p is the only known PLC in yeast. Yeast cells lacking the PLC1 gene have many small vacuoles, indicating defects in protein trafficking to the vacuole or vacuole fusion, and purified Plc1p stimulates vacuole fusion. Docking-dependent Ca2+ efflux is absent in plc1Δ vacuoles and was restored only upon the addition of both Plc1p and the Vam7p SNARE. However, vacuoles purified from plc1Δ strains still retain PLC activity and significant 3-nitrocoumarin- and U73122-sensitive fusion, suggesting that there is another PLC in S. cerevisiae with an important role in vacuole fusion.

nisms. The homotypic fusion of yeast vacuoles, reconstituted in vitro, occurs in three stages: priming, docking, and bilayer fusion and contents mixing (1,2). During priming, Sec18p (N-ethylmaleimide-sensitive factor) disassembles cis-SNARE complexes containing the three Q-SNAREs (Vti1p, Vam7p, and Vam3p) and an R-SNARE (Nyv1p or Ykt6p). Priming releases Sec17p (␣-SNAP) and a soluble SNARE, Vam7p, from vacuoles. The docking stage begins with tethering, which is regulated by the small Rab GTPase Ypt7p. Tethering is followed by the accumulation of fusion factors into a vacuole membrane microdomain termed the "vertex ring," which surrounds the closely apposed membranes of the tethered vacuoles (3). The final step of docking is the formation of SNARE complexes in trans between apposed vacuoles. Docking requires the HOPS/ VpsC complex, composed of the subunits Vps11p, Vps16p, Vps18p, Vps33p, Vps39p, and Vps41p. HOPS fulfills several functions. It serves as a nucleotide exchange factor for Ypt7p (4). HOPS also binds to Ypt7p-GTP and is thus a Ypt7p effector (5). The Vps33p subunit of HOPS is a Sec1/Munc18 family protein that allows HOPS to interact with the SNARE complexes. trans-SNARE interactions trigger a Ca 2ϩ efflux from the vacuole lumen (6). The late stages of fusion may require additional factors such as the V 0 sector of the V-ATPase (7,8), the Vtc complex (9), the armadillo repeat protein Vac8p (10,11), and actin and actin assembly regulators (12).
Specific "regulatory lipids" are also essential for vacuole fusion. Ergosterol, a yeast sterol, is required for priming (13). At least two phosphoinositides are involved in vacuole docking. Phosphatidylinositol 3-phosphate (PI(3)P) is recognized by the Phox homology domain of Vam7p to facilitate the rebinding of this SNARE to the vacuole (14,15). Phosphatidylinositol 4,5bisphosphate (PI(4,5)P 2 ) regulates priming and docking (16), although its functions have been unknown. Theses lipids are required for the enrichment of fusion factors in the vertex ring during docking. 2 Enzymes that generate or modify these lipids are therefore likely to play essential roles in vacuole fusion. Cells deleted for any of the genes for ergosterol biosynthesis have fragmented vacuoles (13), which often indicate a fusion defect. Vacuoles are also fragmented in vps34⌬ cells that lack phosphatidylinositol (PI) 3-kinase activity (17), and vacuoles isolated from spheroplasts pretreated with the PI 3-kinase inhibitor wortmannin are incompetent for fusion in vitro (15). PI(4,5)P 2 participates in the formation of the vertex ring microdomain 2 and may serve as a membrane anchor for fusion catalysts such as those that remodel actin (18) or as a biosynthetic precursor of other lipids that are needed for fusion. PI(4,5)P 2 is synthesized in yeast through the sequential phosphorylation of PI by the PI 4-kinase Stt4p and the PI(4)P 5-kinase Mss4p (19). PI(4,5)P 2 is hydrolyzed to diacylglycerol * This work was supported in part by an NIGMS grant GM23377 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: SNAREs, soluble N-ethylmaleimidesensitive factor attachment protein receptors; ␣-SNAP, ␣-soluble Nethylmaleimide-sensitive factor attachment factor; HOPS, homotypic fusion and vacuole protein sorting complex; PI, phosphatidylinositol; PI(3)P, phosphatidylinositol 3-phosphate; PI(4,5)P 2 , phosphatidylinositol 4,5-bisphosphate; DAG, diacylglycerol; PLC, phosphoinositide-specific phospholipase C; GFP, green fluorescent protein; HA, hemagglutinin; PIPES, 1,4-piperazinediethanesulfonic acid; rVam7p, recombinant Vam7p; NBD, 12-(N-methyl-N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)); IP 3 , inositol 1,4,5-triphosphate.
(DAG) and IP 3 by phosphoinositide-specific phospholipase C (PLC). We refer here to all phospholipase C activities as PLC, whereas PLC1 encodes the only known yeast PLC. DAG is a hydrophobic lipid of intrinsic negative curvature that promotes the fusion of bilayer membranes (20). Biophysical studies have shown that DAG promotes the fusion of synthetic vesicles by altering the physical properties of a membrane lipid bilayer, modifying its fluidity, or inducing lateral phase separations and transitions (21). DAG may fulfill these roles in biological membrane fusion events as well. DAG has been implicated in several membrane trafficking events. It promotes Ca 2ϩ -induced fusion of chromaffin granules with other membranes (22) and facilitates exocytosis of amylase from parotid gland secretory granules (23). DAG also has an essential role in protein transport from the yeast Golgi complex (24).
We now report that DAG and PLC, which hydrolyzes PI(4,5)P 2 to DAG, are required for vacuole fusion, suggesting that the requirement of PI(4,5)P 2 for vacuole fusion may reflect an essential role of DAG. DAG is produced by PLC on yeast vacuoles and is consumed in an ATP-dependent manner. There is only one known PLC, encoded by PLC1, in Saccharomyces cerevisiae (25)(26)(27). Yeast cells lacking the PLC1 gene have a fragmented vacuole phenotype, which often indicates a vacuole fusion defect. However, vacuoles isolated from plc1⌬ cells still retain PLC activity and exhibit reduced but meaningful fusion, indicating that there is an additional gene, structurally unrelated to PLC1, encoding PLC activity. These studies establish that DAG regulation through PLC is required for Rab-and SNARE-dependent biological membrane fusion.
Calcium Efflux Assay-Ca 2ϩ efflux assays were performed as described for fusion, but with 10 g rather than 6 g of vacuoles from BJ3505 or YJY1 (BJ3505 plc1⌬::URA3) in 30-l reactions. Aequorin luminescence assays were performed as described previously (9,43). Samples were analyzed in 96-well low protein binding conical bottom plates (Nunc, Rochester, NY) in a luminometer (Molecular Devices). IGOR Pro 4 (WaveMetrics) and JMP5 (SAS Institute) were used for data analysis. Calibration was done using buffered Ca 2ϩ -EGTA standard solutions (Molecular Probes, Inc.) as described previously (6).

RESULTS
DAG enhances the fusion of liposomes by changing their surface properties (21), suggesting that DAG may also fulfill this function for biological membrane fusion. DAG becomes enriched in vertex rings during yeast vacuole fusion and regulates the vertex enrichment of other lipids and of proteins such as Ypt7p, SNAREs, and HOPS. 2 We have tested the role of DAG in membrane fusion by examining the effect of DAG ligands and the need for DAG production from PI(4,5)P 2 . We first examined whether DAG is required for vacuole fusion, using an in vitro fusion assay and a specific DAG ligand, the C1b domain from mammalian protein kinase C␤II (33). We assayed the homotypic fusion of vacuoles by mixing vacuoles from two strains. One strain has the vacuolar luminal protease Pep4p, but is deleted for the gene encoding the Pho8p phosphatase. The other strain lacks the protease Pep4p and therefore accumulates catalytically inactive pro-Pho8p. During incubation with ATP, fusion allows the Pep4p protease to gain access to pro-Pho8p, converting it to the active form, which is assayed colorimetrically. The recombinant C1b domain inhibited the fusion reaction ( Fig. 1), indicating that DAG is involved in vacuole fusion. PI(4,5)P 2 , which is hydrolyzed by PLC to DAG and inositol 1,4,5-triphosphate (IP 3 ), is required for early stages of vacuole fusion (16). PI(4,5)P 2 may be needed as a substrate for PLC to produce DAG. To examine whether PLC activity is required for vacuole fusion, we have employed two specific PLC inhibitors, 3-nitrocoumarin and U73122. 3-Nitrocoumarin inhibits the PLC activity of Plc1p in vitro with a K i of 57.6 M (44) and also inhibits mammalian PLC␥ (45). Vacuole fusion was blocked by 3-nitrocoumarin at a comparable concentration (K i Ͻ 40 M), whereas its inactive analogue 7-hydroxy-3-nitrocoumarin had little effect on vacuole fusion ( Fig.  2A). Although not previously tested on yeast PLC, U73122 has been used as an inhibitor for mammalian PLCs (46) and plant PLCs (47). U73122 prevented vacuole fusion with a K i of 40 M, whereas its structural analogue U73343 did not inhibit (Fig. 2B). U73122 also inhibited the activity of purified His 6 -Plc1p (Fig. 2C) and the PLC activity that is present on isolated yeast vacuoles (see Fig. 8E below).
We developed an assay of vacuole surface DAG to determine whether it is dynamically altered during in vitro fusion reactions. C1b labeled with the Alexa 488 fluorophore was incubated with vacuoles, and the vacuole-bound Alexa 488-labeled C1b was assayed by sedimenting the organelle. During a standard fusion reaction, there was an ATP-dependent loss of vacuole surface DAG (Fig. 3A, circles, ϩATP; and squares, ϪATP). A greater loss of vacuole surface DAG was seen in the presence of the PLC inhibitor U73122 (Fig. 3B, bar 3 versus bar 2), suggesting that the DAG pool was being replenished by the action of one or more PLCs. This is supported by the finding that the addition of purified His 6 -Plc1p increased the level of accessible DAG (Fig. 3B, bars 4 and 5). DAG is thus synthesized by PLC on isolated vacuoles and undergoes further ATPdependent modification or internalization.
Vacuole fusion occurs in three stages. Priming, initiated by ATP-dependent Sec18p action, is a prerequisite for docking and fusion (48). To determine the stage at which PLC functions during vacuole fusion and whether PLC function is also regulated by the priming factor Sec18p, the reaction was arrested by blocking priming using either anti-Sec18p (Fig. 4A) or anti-Sec17p (Fig. 4B) antibody. Although anti-Sec18p antibody blocked the reaction (Fig. 4A, bar 3), fusion could be restored after a 30-min incubation by the addition of excess Sec18p (bar 10) as reported (49). Without anti-Sec18p antibody, substantial docking and even fusion occurred during a 30-min incubation (bar 16), but the presence of anti-Sec18p antibody prevented priming, and thereby docking and fusion, during this period (bar 9). Therefore, upon reversal of an anti-Sec18p antibody block with Sec18p, fusion reactions were still fully sensitive to docking inhibitors such as anti-Ypt7p (bar 11), anti-Vam3p (bar 12), or anti-Vps33p (bar 13) antibody. However, fusion reactions had acquired substantial resistance to 3-nitrocoumarin or U73122 during a 30-min incubation with anti-Sec18p antibody (bars 14 and 15), showing that PLC function is not regulated by Sec18p action. To further determine when PLC functions during vacuole fusion, we exploited the recent observation that rVam7p can bypass the requirement for Sec17/18pmediated priming (6,50). Fusion reactions were arrested by anti-Sec17p antibody (Fig. 4B, bar 4) and at least partially restored after a 30-min incubation by the addition of rVam7p (bar 9). This fusion reaction was largely resistant to U73122 (bar 11), although it was fully sensitive to anti-Vam3p antibody, a docking inhibitor (bar 10), or to the C1b domain (bar 12), suggesting that PLC function is largely completed before rVam7p-driven docking, but that DAG, a product of PLC function, is still needed during or after docking. These experiments show that the production of DAG by PLC is an early event that is not regulated by Sec18p and that DAG generated at earlier stages is required for the later stages of vacuole fusion.
S. cerevisiae has only one known PLC, encoded by PLC1 (25)(26)(27). Wild-type yeast cells have one to three large vacuoles (Fig. 5A), but deletion of the PLC1 gene caused striking vacuole fragmentation (Fig. 5B), which may reflect a defect either in vacuole fusion or in biosynthetic trafficking to the vacuole (17). Introduction of the PLC1 gene into plc1⌬ cells restored the normal vacuole morphology (Fig. 5C). A lipase-inactive mutant of PLC1 failed to restore the normal vacuole phenotype (Fig.  5D), suggesting that the enzymatic activity of Plc1p is required for normal vacuole structure in vivo.
Since Plc1p might support vacuole structure either directly or indirectly, e.g. through biosynthetic trafficking to the vacuole, we characterized its presence and functions on purified vacuoles. Vacuoles were isolated from strains in which the PLC1 gene was replaced on the chromosome with a gene encoding GFP-Plc1p or HA-Plc1p under the control of the native PLC1 promoter. Vacuoles were analyzed by SDS-PAGE and immunoblotted with anti-GFP or anti-HA antibody. GFP-Plc1p (Fig. 6, upper panel) or HA-Plc1p (data not shown) was detected in purified vacuole extracts, although neither was enriched in vacuoles as was the vacuolar protease Pep4p (Fig. 6, lower  panel). Although our vacuoles were purified ϳ30-fold from the cell lysate, Plc1p was present on vacuoles at the same relative abundance as in the total cell extract, making it unlikely that it is simply present because of cytosolic contamination.
Low concentrations of purified His 6 -Plc1p stimulated in vitro vacuole fusion (Fig. 7), although it inhibited at higher concentrations. The inhibition may be due to depletion of PI(4,5)P 2 , which may have more than one role in vacuole fusion (16), 2 or through production of excess DAG, which has an inhibitory effect on the membrane fusion of synthetic vesicles (51). Heattreated His 6 -Plc1p had neither stimulatory nor inhibitory activity (data not shown).
To examine whether Plc1p is essential for vacuole fusion in vitro, we generated plc1⌬ pep4⌬ and plc1⌬ pho8⌬ strains, isolated vacuoles from these double deletion strains, and assayed their fusion. Surprisingly, vacuoles from plc1⌬ strains were competent for in vitro fusion, although the fusion was not as efficient as for wild-type vacuoles (2-50% as efficient; 17% average from nine independent experiments with successive vacuole preparations) (Fig. 8A, bars 3 and 4) . This reduced fusion of plc1⌬ vacuoles was not restored by the addition of purified Plc1p (Fig. 8A, bar 6), suggesting that PLC1 deletion may affect the biosynthetic delivery or stability of other essential fusion catalysts. Indeed, plc1⌬ vacuoles showed modest reductions in their levels of several SNAREs (Fig. 8B). The fusion of plc1⌬ vacuoles could be restored by  5) and assayed for exposed DAG by binding Alexa 488-labeled C1b as described for A. the addition of excess rVam7p (Fig. 8A, bar 8), consistent with the SNARE deficiency and with the recent report that rVam7p-mediated "bypass" fusion is largely resistant to PLC inhibitors (50). Since DAG is required for vertex enrichment of three vacuolar Q-SNAREs (Vam3p, Vti1p, and Vam7p) and for vacuolar association of soluble Vam7p, 2 one of the roles of DAG in vacuole fusion may be to promote SNARE complex formation at the vertex ring. By driving SNARE assembly by mass action, high levels of rVam7p may reduce the requirement for DAG. This idea was further corroborated by studying the ability of rVam7p to restore fusion to C1b-treated vacuoles (Fig. 8C). Vacuole fusion (Fig. 8C, bar 8) was almost completely blocked by 30 M C1b (bar 2). The addition of rVam7p after a 25-min incubation with C1b partially restored fusion (bars 3-6), whereas rVam7p did not stimulate fusion in reactions that had not received C1b (bars 8 -12). The fusion of plc1⌬ vacuoles was still sensitive to anti-Sec17p, anti-Sec18p, or anti-Vam7p antibody or to the PLC inhibitor U73122 (Fig. 8D). Although Plc1p may contribute to vacuole fusion, it is apparently not essential. However, the sensitivity of the fusion of plc1⌬ vacuoles to U73122 suggests that some PLC activity is still required. There may be another PLC whose activity is required during vacuole fusion in vitro in the absence of Plc1p. The existence of another PLC is strengthened by the observation that vacuoles purified from a plc1⌬ strain retain substantial PLC activity (Fig. 8E, lane

8).
Furthermore, the PLC activity present in plc1⌬ vacuoles was inhibited by either U73122 or 3-nitrocoumarin (lanes 9 and 10). The reduced but measurable fusion shown in plc1⌬ vacuoles is consistent with the vacuole phenotype of plc1⌬ cells as Class B (17), reflecting a modest defect in vacuole fusion (52), instead of Class C, which is more severe. The unknown gene encoding the novel PLC activity found in plc1⌬ cells must be structurally unrelated to PLC1, as there are no other genes that show significant sequence homology to PLC1 in S. cerevisiae.
In a complex reaction pathway, a step that is not ratelimiting may be severely diminished, whereas the overall reaction is far less compromised. In this regard, vacuoles exhibited a broad peak of net Ca 2ϩ efflux after ϳ40 min of the fusion reaction (Fig. 9A, closed circles). This Ca 2ϩ efflux has been shown to depend on trans-associations of SNAREs (6) and thus was stimulated by the addition of rVam7p (closed squares) or inhibited by anti-Vam3p antibody (open circles). This peak of Ca 2ϩ efflux was also stimulated by His 6 -Plc1p (closed triangles), and there was synergistic stimulation of Ca 2ϩ efflux when both Plc1p and rVam7p were added (closed diamonds). In all cases, the efflux of Ca 2ϩ was blocked by anti-Vam3p antibody (open symbols), showing that it depends on the authentic reaction pathway. Strikingly, this Ca 2ϩ efflux was almost absent in vacuoles from plc1⌬ cells (Fig. 9B, closed circles). There was detectable Ca 2ϩ efflux when either rVam7p (closed squares) or Plc1p (closed triangles) was added, but full restoration required both Plc1p and rVam7p (closed diamonds). Ca 2ϩ efflux was still completely sensitive to Gdi1p (open diamonds), a characterized docking inhibitor (48), as well as anti-Vam3p antibody (data not shown). DISCUSSION Fusion between two lipid bilayer membranes is not a spontaneous event. It requires energy to overcome hydration repulsion between apposed membranes and to disrupt the nor-FIG. 5. PLC1 and its PLC activity are required for normal vacuole morphology in vivo. Wild-type strain BY4742 (A) and BY4742 plc1⌬::neo (B) were grown in yeast extract/peptone/dextrose medium at 30°C. BY4742 plc1⌬::neo bearing pRS316-PLC1 (C) or pRS316-PLC1(lipase-inactive mutant) (D) was grown in synthetic dextrose medium lacking uracil at 30°C. After 12 h, 4 M FM4-64 was added; incubation was continued for 4 h at 30°C, and vacuole morphology was examined as described (3). After 30 min, rVam7p was added at the indicated concentrations. Reactions were incubated at 27°C for an additional 70 min before assaying for alkaline mal bilayer structure. The energy needed for membrane fusion, which can be provided by proteins such as SNAREs (53), may be reduced by modifying lipid bilayer properties, e.g. by generating lipids such as DAG that introduce negative curvature in the membrane (21). Although a number of biophysical studies have shown that some lipid species can promote the fusion of lipid bilayer membranes, the role of such lipids in Rab-and SNARE-dependent biological membrane fusion is still largely unknown.
Homotypic vacuole fusion in S. cerevisiae is Ypt/Rab-and SNARE-dependent, but has also been shown to require specific regulatory lipids such as PI(4,5)P 2 (16), PI(3)P (15), and ergosterol (13). We have shown that DAG is also required for vacuole fusion. DAG and the other regulatory lipids become enriched in the vertex ring microdomain of docked vacuoles and are required for the assembly of selected fusion proteins into these vertex rings and thus for subsequent vacuole fusion. 2 DAG present on vacuole membranes was both consumed and produced during normal vacuole fusion reactions (Fig. 3). Either the recombinant C1b domain (a ligand for DAG) or the PLC inhibitors 3-nitrocoumarin and U73122 (which prevent DAG production) blocked vacuole fusion. ET-18-OCH 3 , a third PLC inhibitor (54), has been shown previously to block the vacuole fusion reaction (17). Although ET-18-OCH 3 inhibits PI 3-kinase (55) as well as PLC, well studied inhibitors of PI 3-kinase such as wortmannin and LY-294002 (16) do not affect the fusion reaction, and the block of vacuole fusion by ET-18-OCH3 is therefore likely due to inactivation of PLC. Moderate levels of purified His 6 -Plc1p stimulated fusion, whereas high levels of added Plc1p inhibited vacuole fusion (Fig. 7). Such stimulation at low PLC levels and inhibition at high levels may be reminiscent of the DAG effects on fusion seen in liposome studies (51), where fusion is promoted by low DAG levels and inhibited by high DAG levels. These findings have been explained by the need for specific DAG concentrations to be produced at the correct time and place in the reaction pathway, i.e. during docking and (for vacuoles) at vertices, to promote non-bilayer lipid phases. Plc1p is found on vacuoles, and its absence through gene deletion results in a vacuole fragmentation phenotype in vivo and in substantial reduction, although not abolition, of the fusion of purified vacuoles in vitro. However, the ability of Vam7p, but not Plc1p, to restore fusion, combined with immunoblot analysis showing a general diminution of vacuolar SNARE levels, suggests that PLC1 deletion may inhibit fusion through lowered SNARE levels. We have also found that excess rVam7p can bypass the PLC requirement, as shown by the ability of rVam7p to restore fusion to 3-nitrocoumarin-or U73122-treated vacuoles (50). The ability of rVam7p to restore fusion to C1b-treated vacuoles suggests that a major function of DAG may be to help gather Ypt7p, PI(3)P, and the other SNAREs into vertices, where Ypt7p and PI(3)P can catalyze the assembly of the low concentrations of endogenous Vam7p into SNARE complexes. DAG can still be produced by plc1⌬ vacuoles, although further studies will be needed to discover whether this represents a novel PLC activity or the sequential activity of enzyme pairs such as a phosphoinositidespecific phospholipase D (56) and a phosphatidic-acid phosphatase. The continued sensitivity to 3-nitrocoumarin and U73122 of plc1⌬ vacuole fusion and of plc1⌬ vacuole conversion of PI(4,5)P 2 to DAG suggests the existence of a second PLC. Although there is no homologue of PLC1 in yeast, the continued  6 and 10). A portion (3 l) of each reaction was spotted on a TLC plate and assayed as described under "Experimental Procedures." As controls, NBD-PI(4,5)P 2 and Plc1p-treated NBD-PI(4,5)P 2 were also analyzed (lanes 1 and 2, respectively). sensitivity of the fusion of vacuoles from plc1⌬ strains to 3-nitrocoumarin and U73122 suggests that crucial active-site residues must be conserved in the second PLC. It is unclear why plc1⌬ vacuoles show a strikingly synergistic requirement for the addition of both Plc1p and rVam7p to restore the normal dynamics of trans-SNARE pairing, assayed by Ca 2ϩ efflux. In this regard, we have recently shown that, by driving SNARE complex assembly, rVam7p can bypass or substantially reduce the requirements for several fusion factors such as Ypt7p, ATP, Sec17p, and Sec18p (50). DAG may promote SNARE complex formation through promoting SNARE concentration in vertex rings; by driving SNARE assembly by mass action, high levels of rVam7p may reduce the requirement for DAG (Fig. 8C). This is further supported by the observations that bypass fusion that is driven by the addition of excess rVam7p is largely resistant to 3-nitrocoumarin and U73122 (50) and that C1b prevented the enrichment of SNAREs in vertex rings. 2 The vacuole lumen is an intracellular calcium store in yeast (57), as is the endoplasmic reticulum in mammalian cells. It is also possible that Plc1p induces or enhances Ca 2ϩ efflux by generating IP 3 , which triggers Ca 2ϩ efflux in mammalian cells, as well as DAG. Indeed, Plc1p increased docking-dependent Ca 2ϩ efflux (Fig. 9A, compare closed circles and closed triangles), and U73122 completely abolished Ca 2ϩ efflux. 3 However, this is unlikely, as yeast has no obvious homologue of the mammalian IP 3 receptor calcium channel proteins (58), and exogenous IP 3 did not affect docking-dependent Ca 2ϩ efflux. 3 We have not seen stimulation of vacuole fusion upon the addition of exogenous IP 3 and DAG. 3 Although a minor component of biological membranes, DAG is a key intermediate in lipid metabolism, cell signaling, and membrane fusion. DAG was identified as a cell fusogen in classical studies (59). Since then, it has been implicated in membrane fusion in various systems. sn-1,2-DAG, but not sn-1,3-DAG, stimulates Ca 2ϩ -induced fusion of phosphatidylserine-containing vesicles (60). Treatment of phospholipid vesicles with phospholipase C, thereby forming DAG within the membrane, also causes vesicle fusion (61). sn-1,2-dioleoylglycerol promotes Ca 2ϩ -induced fusion of chromaffin granules with other membranes (22). Several DAG species facilitate the release of amylase from parotid gland secretory granules (23). The propensity of DAG to form non-bilayer phases and to destabilize the bilayer structure of fusing membranes and its ability to cause negative curvature can promote membrane fusion.
We previously showed that PI(4,5)P 2 , a precursor of DAG, is required for two early steps of vacuole fusion, priming and docking (16). Although all of the functions of PI(4,5)P 2 remain to be determined, it may promote the actin remodeling (18) that is required for vacuole fusion (12). The observation that excess Plc1p fully inhibited fusion (Fig. 7) may either reflect a role for PI(4,5)P 2 in processes such as actin reassembly during a late stage of vacuole fusion or indicate that excessive DAG can inhibit fusion, as reported in liposome fusion studies (51). PI(4,5)P 2 is also required as a source of DAG since PLC, which hydrolyzes PI(4,5)P 2 to generate DAG, is required for vacuole fusion. This idea is further supported by the observations that a PLC inhibitor could block DAG generation during fusion reactions and that His 6 -Plc1p increased DAG levels on the vacuolar membrane during the fusion reaction.
Plc1p is the only known PLC of S. cerevisiae. Vacuoles from plc1⌬ are capable of in vitro fusion. However, this fusion is still sensitive to U73122, and these vacuoles still retain PLC activity (Fig. 8E, lane 8), implying that there is another PLC in S. cerevisiae. The unknown gene encoding the second PLC must be unrelated to PLC1 because there is no gene with significant sequence homology. In a previous report, PLC activity was not detected in PLC1 temperature-sensitive mutants at non-permissive temperature (62). We have detected the second PLC activity in a vacuole preparation from plc1⌬ cells, whereas Yoko-o et al. used a high speed supernatant fraction from a cell lysate to detect PLC activity. If the second PLC is an integral or peripheral membrane protein specifically enriched on membranes, the soluble fraction may have little activity.
Our studies show that vacuolar DAG is generated from PI(4,5)P 2 by PLC activity, presumably including Plc1p, and is consumed by an ATP-dependent activity. DAG is needed for vacuole fusion. Each of the regulatory lipids (DAG, PI(3)P, PI(4,5)P 2 , and ergosterol) is interdependent with each other and with proteins (Ypt/Rab Ypt7p, the HOPS complex, and SNAREs) for the assembly of the vertex ring microdomain around the apposed membranes of docked vacuoles. 2 DAG may contribute to the enrichment of Ypt7p, PI(3)P, and SNAREs such as Vam3p, Vti1p, and Nyv1p in vertex rings, allowing the vacuole re-association of soluble Vam7p and its incorporation into SNARE complexes. trans-SNARE pairing may trigger hemifusion, and the hemifusion intermediate may be resolved by the opening and expansion of fusion pore(s) around the vertex ring. As a lipid of negative curvature, DAG may also contribute to the formation of stalk and hemifusion intermediates. There is extensive documentation of the capacity of even a few mole percent of DAG to promote fusion in model membranes and of the promotion of fusion by PLC-mediated generation of DAG (21). Our present studies establish DAG and PLC as essential for the fusion of vacuoles, along with other lipids and with Rab GTPases, their effectors, and SNAREs.