Phosphoinositide-signaling lipids function in diverse cellular pathways. Dynamic changes in the levels of these signaling lipids regulate multiple processes. In particular, when Saccharomyces cerevisiae cells are exposed to hyperosmotic shock, PI3,5P2 (phosphatidylinositol [PI] 3,5-bisphosphate) levels transiently increase 20-fold. This causes the vacuole to undergo multiple acute changes. Control of PI3,5P2 levels occurs through regulation of both its synthesis and turnover. Synthesis is catalyzed by the PI3P 5-kinase Fab1p, and turnover is catalyzed by the PI3,5P2 5-phosphatase Fig4p. In this study, we show that two putative Fab1p activators, Vac7p and Vac14p, independently regulate Fab1p activity. Although Vac7p only regulates Fab1p, surprisingly, we find that Vac14 regulates both Fab1p and Fig4p. Moreover, Fig4p itself functions in both PI3,5P2 synthesis and turnover. In both the absence and presence of Vac7p, the Vac14p–Fig4p complex controls the hyperosmotic shock–induced increase in PI3,5P2 levels. These findings suggest that the dynamic changes in PI3,5P2 are controlled through a tight coupling of synthesis and turnover.
Phosphoinositides are required in cellular events such as signal transduction, cytoskeletal rearrangements, and membrane trafficking. Phosphoinositides function by recruiting protein effectors to specific membrane domains. Most phosphoinositides bind to multiple effectors. This enables cells to control multiple functions simultaneously by regulating the levels of a single phosphoinositide isomer. Therefore, tight control of phosphoinositide levels is critical for the ability of a cell to orchestrate diverse signaling pathways.
The most recently discovered phosphoinositide, PI3,5P2 (phosphatidylinositol [PI] 3,5-bisphosphate), is required for signaling pathways that invoke a response within the endomembrane system. For example, PI3,5P2 is important for insulin-induced trafficking of the glucose transporter GLUT4 to the plasma membrane (Ikonomov et al., 2002; Berwick et al., 2004). In addition, increases in PI3,5P2 levels accompany a cellular response to UV irradiation (Jones et al., 1999) and EGF stimulation (Tsujita et al., 2004). A similar increase in PI3,5P2 is observed in yeast and plant cells in response to acute increases in environmental osmolarity (Dove et al., 1997; Meijer et al., 1999; Zonia and Munnik, 2004).
PI3,5P2 levels are also important for constitutive membrane traffic to the lysosome. Mammalian EGF receptor trafficking to the lysosome is impaired in cells that overexpress the PI3,5P2 3-phosphatase mytotubularin (Tsujita et al., 2004). In addition, Salmonella enterica secrete a phosphoinositide phosphatase (SopB) that decreases host PI3,5P2 levels (Hernandez et al., 2004) and diverts maturing phagosomes away from the lysosomal degradation pathway (Scott et al., 2002; Vergne et al., 2003).
In Saccharomyces cerevisiae, the levels of PI3,5P2 under normal conditions are low (Dove et al., 1997; Whiteford et al., 1997)—20-fold lower than the other detectable phosphoinositides PI3P, PI4P, and PI4,5P2 (Bonangelino et al., 2002). Yeast mutants that cannot produce PI3,5P2 have grossly enlarged vacuoles. Moreover, the vacuoles are not acidified properly despite the proper localization of the vacuolar ATPase (Bonangelino et al., 1997; Gary et al., 1998). Furthermore, some proteins that are normally transported to the lumen of the vacuole through multivesicular bodies are mislocalized to the limiting membrane of the vacuole (Odorizzi et al., 1998; Shaw et al., 2003; Dove et al., 2004). Vesicular traffic out of the vacuole is also compromised in the absence of PI3,5P2 (Bryant et al., 1998; Dove et al., 2004). These observations demonstrate that PI3,5P2 controls multiple vacuolar functions.
The low levels of PI3,5P2 in S. cerevisiae increase dramatically when cells are acutely exposed to hyperosmotic conditions (Dove et al., 1997). Within 5 min, there is a 20-fold increase in PI3,5P2 levels (Duex et al., 2006). These elevated levels are briefly maintained, but by 30 min, PI3,5P2 levels return to their normal, low levels.
It is likely that the transient increase in PI3,5P2 provides protection against osmotic changes in the environment. For example, vacuole volume decreases dramatically as PI3,5P2 levels increase (Bonangelino et al., 2002; Duex et al., 2006). PI3,5P2 may also regulate channels that release ions, other osmolytes, and water from the vacuole into the cytoplasm (Thumm, 2000).
In S. cerevisiae, Fab1p is the sole kinase that generates PI3,5P2 from a pool of PI3P (Yamamoto et al., 1995). Overexpression of FAB1 does not lead to an increase in PI3,5P2 levels (Gary et al., 1998). This suggests that Fab1p activity is tightly regulated. In fact, two putative activators of Fab1p have been identified: Vac7p and Vac14p. Like fab1Δ cells, vac7Δ and vac14Δ cells have no detectable PI3,5P2 under basal conditions. Moreover, in response to hyperosmotic shock, vac7Δ and vac14Δ cells are defective in producing the 20-fold increase in PI3,5P2 levels observed in wild-type cells (Gary et al., 1998; Bonangelino et al., 2002; Duex et al., 2006). Furthermore, human Vac14 was recently shown to physically associate with mammalian Fab1 (PIKfyve), and overexpression of human Vac14 increased the PI3P 5-kinase activity of PIKfyve (Sbrissa et al., 2004). These observations support the hypothesis that Vac7p and Vac14p regulate PI3,5P2 levels through the activation of Fab1p. Neither Vac7p nor Vac14p have identifiable conserved domains, and it is not yet known how they regulate PI3,5P2 levels.
Fig4p is a known PI3,5P2 5-phosphatase (Rudge et al., 2004). Thus, our recent observation that fig4Δ cells are defective in the hyperosmotic shock–induced increase in PI3,5P2 levels was surprising (Duex et al., 2006). It raised the possibility that Fig4p also functions as an activator of Fab1p. Moreover, Fig4p forms a complex with Vac14p (Dove et al., 2002; Gary et al., 2002; Rudge et al., 2004), suggesting that Vac14p may regulate Fig4p or that Fig4p may regulate Vac14p. Therefore, we sought to determine the roles of Vac7p, Vac14p, and Fig4p in regulation of the lipid kinase activity of Fab1p and determine the potential roles of Vac14p and Vac7p in regulation of the lipid phosphatase activity of Fig4p.
In this study, we identify specific mutations in two distinct regions of Fab1p that enhance its activity. The resulting hyperactive Fab1p mutants produce elevated levels of PI3,5P2 under basal conditions and respond normally to hyperosmotic shock when Vac7p, Vac14p, and Fig4p are present. In their absence, the Fab1p mutants are still able to produce detectable levels of PI3,5P2 but have lost their ability to respond to hyperosmotic shock. These hyperactive Fab1p mutants allowed us to individually test the roles of the putative Fab1p activators in the regulation of PI3,5P2 levels. We found that Vac7p and Vac14p independently regulate PI3,5P2 levels. We also show that Vac14p is absolutely required for Fig4p-mediated turnover of PI3,5P2 after hyperosmotic shock. Lastly, we show that Fig4p point mutants defective in PI3,5P2 turnover are simultaneously defective in hyperosmotic shock–induced elevation of PI3,5P2 levels, suggesting a direct role for Fig4p in both PI3,5P2 synthesis and turnover. The ability of Fig4p and Vac14p to act in both the synthesis and turnover of PI3,5P2 may be key to the ability of cells to acutely regulate transient changes in PI3,5P2 levels.
Fab1p mutants that suppress vac14Δ phenotypes
In vivo, Fab1p cannot produce PI3,5P2 in the absence of either Vac7p or Vac14p. Furthermore, in the absence of Fig4p, there is a dramatic defect in PI3,5P2 elevation after hyperosmotic shock. Therefore, to analyze the roles of Vac7p, Vac14p, and Fig4p as potential activators of Fab1p, Fab1p mutants that function in the absence of these proteins were sought. These mutants would enable us to study the individual roles of Vac7p, Vac14p, and Fig4p in controlling PI3,5P2 levels and may also reveal regulatory regions within the Fab1p protein. A Fab1p mutant that is partially active in the absence of Vac7p had been previously identified (Gary et al., 2002) but had multiple mutations that were distributed throughout two thirds of the protein. Thus, we isolated new Fab1p mutants with single amino acid changes.
Random mutagenesis was performed on the region of FAB1 that encodes the kinase domain (<30% of the total gene; Fig. 1 A
). FAB1 constructs subjected to mutagenesis were transformed into fab1Δvac14Δ cells (Fig. 1 B). Mutant alleles that exhibited wild-type growth rates at 37°C and suppressed the large vacuole volume of vac14Δ cells were sequenced (Fig. 1 C). Two of these mutant alleles, fab1-6 (T2250A) and fab1-7 (G2238W), were each the result of a single amino acid substitution of highly conserved residues in the COOH-terminal end of the kinase domain. Notably, all previously reported kinase domain mutations had a deleterious effect on Fab1p function (Gary et al., 1998; Friant et al., 2003).
A second type of informative mutant (e.g., fab1-8) had mutations exclusively in the region proximal to the kinase domain. Both sets of mutants also displayed reduced vacuole volume in fig4Δ, wild-type, and, to a lesser extent, vac7Δ cells (Fig. 1 C). The identification of these mutants supports a hypothesis that specific residues in the kinase domain and the region proximal to the kinase domain are involved in the up-regulation of Fab1p kinase activity.
Fab1p mutants that suppress vac7Δ phenotypes
The Fab1p mutants identified above restored normal vacuole volume in vac14Δ cells but had much less effect in vac7Δ cells (Fig. 1 C). There are two possibilities as to why the mutants suppressed vac14Δ better than vac7Δ mutants. First, vac7Δ cells have a greater defect in the production of PI3,5P2. Therefore, the basal PI3P 5-kinase activity of these Fab1p mutants may be increased enough to restore normal PI3,5P2 levels and vacuole volume in vac14Δ cells but not in vac7Δ cells. Alternatively, the mutants may specifically bypass Vac14p function but not Vac7p function. Therefore, we performed an additional screen for Fab1p mutants that could restore normal vacuole volume in vac7Δ cells. We randomly mutagenized fab1-6 to ensure that new mutations would be obtained. New mutations might specifically bypass vac7Δ defects or may be more general and bypass vac7Δ defects while further bypassing vac14Δ defects.
The screen yielded mutants in this latter category. The mutants contained the original fab1-6 kinase domain mutation plus mutations in the region proximal to the kinase domain (Fig. 1 D). These Fab1p mutants led to a decrease in vacuole volume in both vac7Δ and vac14Δ cells as well as wild-type and fig4Δ cells. This suggests that specific mutations in either domain result in Fab1p molecules with elevated kinase activity.
Mutations in each domain were additive. For example, the most active mutant (fab1-14) contained mutations in both the kinase domain and in the region proximal to the kinase domain. The kinase domain mutation (T2250A) was identical to that of fab1-6. The proximal region mutated residues (E1822V/F1833L) were nearly identical to those of a proximal region mutant, fab1-8 (E1822K/N1832Y). In fact, a fab1E1822V/F1833L mutant and a fab1E1822K/N1832Y mutant (fab1-8) had identical suppression of vac7Δ and vac14Δ vacuole volume defects (not depicted). In addition, suppression of vacuole volume defects by fab1-14 was equal to that of fab1-6 and fab1-8 combined (Fig. 1, C and D).
Under basal conditions, fab1-6 and fab1-14 mutants produce elevated levels of PI3,5P2
The ability of isolated Fab1p mutants to suppress vacuole volume defects of vac7Δ and vac14Δ cells suggests that these Fab1p mutants produce elevated levels of PI3,5P2. Indeed, the fab1-6 (0.24 units) and fab1-14 (0.68 units) mutants had significantly higher PI3,5P2 levels than wild-type FAB1 (0.04 units; Fig. 2 A
). The observation that fab1-14 results in three times more basal PI3,5P2 than fab1-6 likely explains why fab1-14 suppresses both vac7Δ and vac14Δ vacuole volume defects, whereas fab1-6 suppresses only vac14Δ.
In the presence of FAB1, basal levels of PI3,5P2 in fig4Δ cells are similar to wild type (Rudge et al., 2004; Duex et al., 2006), whereas basal levels of PI3,5P2 in vac7Δ or vac14Δ cells are significantly lower (Bonangelino et al., 2002). These relative differences between the strains were also present when either fab1-6 or fab-14 was expressed as the sole copy of FAB1 in fig4Δ, vac14Δ, or vac7Δ cells (Fig. 2 A). In each case, substitution of FAB1 with fab1-6 resulted in a modest increase in steady-state levels of PI3,5P2, whereas substitution of FAB1 with fab1-14 resulted in a larger increase in PI3,5P2 levels. These findings suggest that neither mutant specifically bypasses the requirement for Vac7p, Vac14p, or Fig4p under basal conditions.
Although the fab1-6 and fab1-14 alleles result in increased Fab1p PI3P 5-kinase activity, our data do not indicate whether the increased activity is a result of changes in the catalytic activity of Fab1p itself or whether the mutations affect modulators of Fab1p activity. For example, the Fab1p mutants presented in this study may have increased PI3P 5-kinase activity because of a reduction in negative regulation by Atg18p. The observation that atg18Δ cells have extremely high levels of PI3,5P2 led to the proposal that Atg18p negatively regulates Fab1p (Dove et al., 2004; Efe et al., 2005). Furthermore, Atg18p has also been reported to bind Fab1p in the region proximal to the kinase domain, which is the site of mutations in fab1-14 (Georgakopoulos et al., 2001).
Hyperosmotic shock–induced elevation in the levels of PI3,5P2 requires Vac7p, Vac14p, and Fig4p
We tested whether the fab1-6 or fab1-14 mutants produced even higher levels of PI3,5P2 after hyperosmotic shock. Indeed, in a wild-type background, both mutants responded to hyperosmotic shock (Fig. 2 B). Unexpectedly, the mutants produced the same level of PI3,5P2 as wild-type FAB1. This was surprising because under basal conditions, the fab1-6 and fab1-14 alleles had elevated levels of PI3,5P2 compared with wild-type FAB1. Perhaps these Fab1p mutants are already partially activated under basal conditions and cannot be further activated beyond a specific maximum. Alternatively, the regulation of Fab1p under basal conditions may be different than the activation of Fab1p after hyperosmotic shock. Another alternative is that there may be regulated inhibition of Fab1p activity to maintain specific levels of PI3,5P2 at specific times. Along similar lines, perhaps regulated turnover of PI3,5P2 prevents the levels from exceeding a maximum. It should also be noted that the levels of PI4P and PI4,5P2 did not change significantly with the expression of fab1-6 or fab1-14 when compared with FAB1 (not depicted).
In the absence of Vac7p, Vac14p, or Fig4p, neither fab1-6 nor fab1-14 achieved the normal levels of PI3,5P2 induced by hyperosmotic shock (Fig. 2 B). Furthermore, the relative differences in hyperosmotic shock–induced levels of PI3,5P2 observed when fab1-6 or fab1-14 were expressed in vac7Δ, vac14Δ, or fig4Δ mutants parallel those observed under basal conditions. Specifically, when these mutant alleles are expressed in vac7Δ cells, they are most defective in the elevation of PI3,5P2 levels; when expressed in vac14Δ cells, they are moderately defective; and when expressed in fig4Δ cells, they are least defective in the elevation of PI3,5P2 levels.
A previous study showed that mouse Fab1 (PIKfyve) PI3P 5-kinase activity is increased upon its dephosphorylation (Sbrissa et al., 2000). Based on this information, we tested the hypothesis that the fab1-6 allele (fab1T2250A) is more active because it prevents a possible inhibitory phosphorylation at the conserved Thr2250 residue. We generated a fab1T2250D mutant to mimic inhibitory phosphorylation and observed that the fab1T2250D mutant had elevated levels of PI3,5P2 that were almost identical to that of the fab1T2250A mutant before and after hyperosmotic shock (unpublished data). Therefore, it is unlikely that the threonine at position 2,250 regulates the activity of Fab1p via its phosphorylation/dephosphorylation.
In the absence of Vac7p, Vac14p, and Fig4p, the fab1-14 mutant is not activated by hyperosmotic shock
A vac7Δ strain containing the hyperactive fab1-14 mutant exhibited a small increase in PI3,5P2 levels after hyperosmotic shock (Fig. 2, A and B). Likewise, when the fab1-14 mutant was integrated into the FAB1 chromosomal locus of a vac7Δ strain, hyperosmotic shock induced a modest increase in PI3,5P2 levels (0.08–0.12 units) and a corresponding decrease in vacuole volume (Fig. 3 A
). These results suggest that a second mechanism of elevating PI3,5P2 exists that is independent of Vac7p. This other mechanism may involve the Vac14p–Fig4p complex. To test this possibility, we simultaneously deleted Vac7p, Vac14p, and Fig4p and measured the levels of PI3,5P2. The vac7Δvac14Δfig4Δfab1-14 strain produced low but detectable levels of PI3,5P2 under basal conditions (Fig. 3 B). This is in contrast with wild-type FAB1, where PI3,5P2 cannot be detected in the absence of Vac7p, Vac14p, and Fig4p. Notably, hyperosmotic shock of the vac7Δvac14Δfig4Δfab1-14 strain did not induce an increase in the levels of PI3,5P2 or a decrease in vacuole volume (Fig. 3 B). These observations strongly suggest that a hyperosmotic shock–induced increase in the levels of PI3,5P2 requires Vac7p, Vac14p, and/or Fig4p.
Vac7p and Vac14p independently increase the levels of PI3,5P2
To determine whether Vac7p, Vac14p, or Fig4p alone elicit an increase in the levels of PI3,5P2, we individually expressed each gene in the vac7Δvac14Δfig4Δfab1-14 strain. Because the localization of Vac14p and Fig4p are partially dependent on each other (Rudge et al., 2004; Duex et al., 2006), each gene was overexpressed to mitigate potential problems in their localization. Overexpression of FIG4 in the vac7Δvac14Δfig4Δfab1-14 strain had no effect on the levels of PI3,5P2 before or after hyperosmotic shock (Fig. 4 A
). This indicates that Fig4p alone cannot elevate the levels of PI3,5P2. Overexpression of VAC14 resulted in a 4.5-fold increase in the basal levels of PI3,5P2. This suggests that Vac14p can elevate levels of PI3,5P2 in the absence of Vac7p and Fig4p. After hyperosmotic shock, there was no further increase in PI3,5P2 levels. Overexpression of VAC7 resulted in a 4.5-fold increase in the basal levels of PI3,5P2. Moreover, after hyperosmotic shock, there was a further increase in PI3,5P2 levels. This suggests that Vac7p can elevate levels of PI3,5P2 in the absence of Vac14p and Fig4p. Moreover, it also suggests that the majority of the hyperosmotic shock–induced elevation of PI3,5P2 levels occurs via Vac7p.
A maximal increase in the levels of PI3,5P2 requires both Vac7p and the Vac14p–Fig4p complex
To test whether the potential regulators act in concert with each other, we overexpressed pairwise combinations of VAC7, VAC14, and FIG4 in the vac7Δvac14Δfig4Δfab1-14 strain. When FIG4 and VAC14 were overexpressed simultaneously, the basal levels of PI3,5P2 were lower than when VAC14 was expressed alone (Fig. 4 B). This is likely caused by the ability of the Vac14p–Fig4p complex to turn over PI3,5P2 under basal conditions. Upon hyperosmotic shock, the levels of PI3,5P2 increased. This is in contrast to the overexpression of Vac14p or Fig4p alone where hyperosmotic shock had no effect on the levels of PI3,5P2 (Fig. 4 A). This finding suggests that the Vac14p–Fig4p complex plays a role in the hyperosmotic shock–induced increase in PI3,5P2 levels. This postulate is also supported by the observation of a Vac7p-independent hyperosmotic shock–induced increase in PI3,5P2 levels in vac7Δfab1-14 cells (Fig. 3, A and B). Analysis of a Fig4p point mutant is also consistent with a role of the Vac14p–Fig4p complex in the hyperosmotic shock–induced increase in PI3,5P2 levels (see the last section of Results).
When VAC7 and VAC14 were overexpressed simultaneously, the levels of PI3,5P2 were the sum of the levels observed with VAC7 or VAC14 alone. This is consistent with the hypothesis that Vac7p and Vac14p independently elevate the levels of PI3,5P2. The highest levels of PI3,5P2 observed under both basal and hyperosmotic shock conditions occurs when all three regulators were overexpressed simultaneously. This further supports the hypothesis that all three proteins play important roles in the elevation of PI3,5P2 levels.
All of the aforementioned results support a model in which Vac7p and Vac14p independently increase PI3,5P2 levels. Under basal conditions, Vac7p and Vac14p play an equal role. After hyperosmotic shock, Vac7p is the main protein required to elevate PI3,5P2 levels. In addition, after hyperosmotic shock, Fig4p is important for the elevation of PI3,5P2 levels and requires Vac14p for this function.
The Vac14p–Fig4p complex is required for turnover of PI3,5P2
Analysis of vac14Δ cells revealed that after hyperosmotic shock, there is a delay in the turnover of PI3,5P2 (Fig. 5 A
). This delay is similar to that observed in fig4Δ cells where the levels of PI3,5P2 at 30 min after hyperosmotic shock are the same or even higher than the levels achieved at 10 min. This delay in the turnover of PI3,5P2 can also be observed when the fab1-14 allele was expressed in vac14Δ or fig4Δ cells (Fig. 5 B). These findings clearly demonstrate that in either a vac14Δ or fig4Δ mutant, there is a delay in the turnover of PI3,5P2. Therefore, both Vac14p and Fig4p are required for the hyperosmotic shock–induced turnover of PI3,5P2.
The Fig4p sequence is similar to known lipid phosphatases. Moreover, Fig4p functions in vitro as a PI3,5P2 5-phosphatase (Rudge et al., 2004). These findings, combined with the aforementioned analysis, strongly suggests that the rapid turnover of PI3,5P2 after hyperosmotic shock is the result of the Fig4p-dependent conversion of PI3,5P2 to PI3P.
The requirement for Vac14p in hyperosmotic shock–induced turnover of PI3,5P2 was surprising because Vac14p was first proposed to be an activator of Fab1p. More recently, Vac14p had also been shown to interact with Fig4p (Dove et al., 2002; Rudge et al., 2004). Here, we find that the levels of Fig4p are greatly diminished in a vac14Δ mutant (Fig. 5 C). This suggests that Vac14p may play a role in the turnover of PI3,5P2 either by stabilization and/or regulation of Fig4p.
The observation that hyperosmotic shock–induced PI3,5P2 levels eventually decrease even in the absence of Fig4p suggests that there are other pathways that degrade PI3,5P2. Five other proteins (Sac1p, Sjl1p, Sjl2p, Sjl3p, and Inp54p) are predicted to have phosphoinositide 5-phosphatase activity. However, defects in PI3,5P2 turnover have not been observed in the corresponding knockout strains. Only when multiple genes are knocked out are elevated PI3,5P2 levels observed (Gary et al., 2002; Stefan et al., 2002; Parrish et al., 2004). Furthermore, none of these phosphoinositide phosphatases localize to the vacuole membrane, which is the site of PI3,5P2 synthesis and function. Although it is likely that at least some of these proteins can degrade PI3,5P2, none of them are likely to have a primary role in the rapid turnover of PI3,5P2 after hyperosmotic shock.
The vac7Δfab1-14 strain produced very little PI3,5P2 in response to hyperosmotic shock, but these levels decreased at 30 min (Fig. 5 B). This suggests that Vac7p is not required for the rapid turnover of PI3,5P2.
Fig4p requires Vac14p to function in the hyperosmotic shock–induced turnover of PI3,5P2
We used the vac7Δvac14Δfig4Δfab1-14 strain to further test the function of Vac14p and Fig4p in the hyperosmotic shock–induced turnover of PI3,5P2. Simultaneous overexpression of FIG4, VAC7, and VAC14 in the vac7Δvac14Δfig4Δfab1-14 strain resulted in the hyperosmotic shock–induced turnover of PI3,5P2 (Fig. 6 A
). Overexpression of FIG4 alone resulted in no turnover of PI3,5P2 (Fig. 6 B). Similarly, overexpression of VAC14 alone resulted in no turnover of PI3,5P2. However, simultaneous overexpression of FIG4 and VAC14 resulted in the turnover of PI3,5P2. This supports the hypothesis that the Vac14p–Fig4p complex functions in the turnover of PI3,5P2 after hyperosmotic shock.
When VAC7 was overexpressed alone in the vac7Δvac14Δfig4Δfab1-14 strain, a modest decrease in PI3,5P2 levels was observed (Fig. 6 C). However, the amount of turnover (0.19 units) was less than half the amount (0.41 units) observed when FIG4, VAC7, and VAC14 were overexpressed (Fig. 6 A). The observed turnover is likely the result of a Fig4p-independent mechanism (discussed in the previous section).
Simultaneous overexpression of VAC7 and FIG4 had a rate of PI3,5P2 turnover that was similar to VAC7 alone. Simultaneous overexpression of VAC7 and VAC14 also had a rate of PI3,5P2 turnover that was similar to VAC7 alone. These data suggest that Fig4p is the major enzyme responsible for the rapid turnover of hyperosmotic shock–induced PI3,5P2 levels and that Fig4p stability and function require Vac14p.
The fig4G519R phosphatase mutant is partially defective in both hyperosmotic shock–induced PI3,5P2 elevation and turnover
Perturbation in the localization of Vac14p in a fig4Δ strain (Rudge et al., 2004; Duex et al., 2006) makes it difficult to directly assess a role for Fig4p in the hyperosmotic shock–induced elevation and turnover of PI3,5P2. Therefore, we used a Fig4p point mutant in which the localization of Vac14p is normal (Fig. 7 A
). The fig4G519R point mutant (Gary et al., 2002) is defective in PI3,5P2 5-phosphatase activity in vitro (Rudge et al., 2004). Therefore, we tested the in vivo levels of PI3,5P2 in a fig4G519R mutant. When compared with wild-type FIG4 (Fig. 7 B), basal levels of PI3,5P2 in a fig4G519R mutant are elevated (Fig. 7 C). This observation is consistent with a defect in PI3,5P2 5-phosphatase activity. However, fig4G519R appears to retain some lipid phosphatase activity. At 30 min after hyperosmotic shock, fig4G519R cells have degraded 0.5 units of PI3,5P2 compared with 0.0 units in fig4Δ cells (Fig. 7 D) and 1.3 units in FIG4 cells.
In an attempt to produce a Fig4p mutant with no phosphatase activity, we generated a second Fig4p mutant, fig4D469N. Based on a similar mutation in Sac1p (Stock et al., 1999), the fig4D469N mutant was predicted to have no lipid phosphatase activity. Moreover, the D469 residue is conserved in all Sac1 domain–containing proteins, and the residue lies within the conserved CX5RT catalytic site of lipid (Maehama and Dixon, 1998) and protein (Stone and Dixon, 1994) phosphatases. The fig4D469N mutant behaved nearly identically to the fig4G519R mutant (unpublished data). Both mutants had elevated levels of PI3,5P2 under basal conditions, a decreased rate of hyperosmotic shock–induced turnover of PI3,5P2, and a defect in the elevation of PI3,5P2 in response to hyperosmotic shock (∼65% of wild type). Notably, the extent of the partial defect observed in PI3,5P2 turnover parallels the extent of the partial defect in PI3,5P2 elevation. These observations suggest that the Sac1 phosphatase domain of Fig4p is important for both an increase in PI3,5P2 levels and their turnover.
The dramatic and transient change in PI3,5P2 levels after hyperosmotic shock raises the question of how acute changes in phosphoinositides are regulated. In this study, we investigated the putative PI3,5P2 regulators Vac7p, Vac14p, and Fig4p to determine the role each plays in PI3,5P2 synthesis and turnover.
Transient elevation in PI3,5P2 levels are primarily caused by its increased synthesis
Steady-state levels of PI3,5P2 are dependent on both its rate of synthesis and turnover. In the studies reported here, the large increases in the levels of PI3,5P2 observed after hyperosmotic shock or in the presence of the activated Fab1p mutants were accompanied by a decrease in the levels of PI3P (Fig. 7 B). Decreases in PI3P could be caused by the activation of Fab1p and a corresponding depletion of its substrate or caused by the inhibition of a PI3,5P2 5-phosphatase and a corresponding loss in the production of its product. Strains that are defective in normal hyperosmotic shock–induced PI3,5P2 elevation, including fab1Δ in which no PI3,5P2 is produced, have up to 35% higher levels of PI3P than wild-type cells (Fig. 7 D). This observation strongly suggests that the steady-state levels of PI3P are caused by its synthesis from PI rather than from the turnover of PI3,5P2. Also, in wild-type cells under basal conditions, the levels of PI3,5P2 are 20-fold lower than the levels of PI3P, making it unlikely that the turnover of PI3,5P2 is a significant source of PI3P. These results are most consistent with the hypothesis that the increases in PI3,5P2 levels and concomitant decreases in PI3P levels that occur during hyperosmotic shock are caused by increased Fab1p activity rather than the inhibition of a PI3,5P2 5-phosphatase.
Vac7p likely elevates PI3,5P2 via the activation of Fab1p
Our data support the hypothesis that Vac7p increases PI3,5P2 levels via the activation of Fab1p. In the vac7Δvac14Δfig4Δfab1-14 strain, PI3P but not PI3,5P2 levels are elevated by hyperosmotic shock. However, when VAC7 is overexpressed in this strain, the levels of PI3,5P2 are elevated after hyperosmotic shock, and there is a corresponding decrease in PI3P levels (unpublished data). These changes are most consistent with the hypothesis that Vac7p activates Fab1p.
Vac7p is the primary activator of Fab1p. Vac7p activates the hyperactive fab1-14 mutant in the absence of Vac14p and Fig4p. Furthermore, Vac7p activates Fab1p to a greater degree than Vac14p, Fig4p, or the Vac14p–Fig4p complex (Fig. 8
). Notably, Vac7p homologues have only been observed in fungi. Perhaps there is a functional homologue of Vac7p in other eukaryotes that cannot be found through sequence comparison. Alternatively, in organisms that lack Vac7p, the conserved Vac14p–Fig4p complex may serve as the main regulator of Fab1p.
Vac14p-dependent elevation of PI3,5P2 levels is independent of Vac7p
Overexpression of Vac14p alone in the vac7Δvac14Δfig4Δfab1-14 strain results in an increase in PI3,5P2 levels. Overexpression of Vac14p and Vac7p together results in an increase in PI3,5P2 levels that is equal to the sum of overexpression of either protein alone. Therefore, Vac14p and Vac7p function independently. Analysis of the Vac14p–Fig4p complex strongly suggests that Vac14p increases PI3,5P2 levels via the activation of Fab1p (see the last section of Discussion). In addition, Vac14 physically associates with and increases the activity of mammalian Fab1p (PIKfyve; Sbrissa et al., 2004).
In contrast to Vac7p, the overexpression of Vac14p alone does not result in a further increase in PI3,5P2 levels in response to hyperosmotic shock. Perhaps the fab1-14 allele partially mimics the Vac14p-induced activation of Fab1p. Alternatively, Vac14p may require Fig4p for its function in hyperosmotic stress. Note that PI3,5P2 levels increase in response to hyperosmotic shock when both Vac14p and Fig4p are present (Fig. 4).
The observation that Vac7p and the Vac14p–Fig4p complex independently increase PI3,5P2 levels suggests that there may be multiple pathways for Fab1p activation. The Vac7p and Vac14p–Fig4p pathways may generate distinct pools of PI3,5P2 or may be downstream of different signaling cascades.
It is tempting to speculate that Vac7p and Vac14p–Fig4p each bind to unique regions of Fab1p. Fab1p is a large protein (2,278 aa) and may contain multiple binding sites for multiple partners. Although the hyperactive mutants identified do not specifically bypass either Vac7p or Vac14p, only 30% of the protein was mutagenized. Perhaps Vac7p and/or Vac14p activate Fab1p through interaction with specific regions in the other 70% of the protein.
Fig4p requires Vac14p for its PI3,5P2 5-phosphatase activity
Turnover of PI3,5P2 after hyperosmotic shock is predominantly achieved through the activity of Fig4p, a PI3,5P2 5-phosphatase. Surprisingly, Vac14p is required for the Fig4p-mediated turnover of PI3,5P2. In vac14Δ cells, the rate of PI3,5P2 turnover after hyperosmotic shock is decreased to the same degree as it is in fig4Δ cells. The defect in the turnover of PI3,5P2 is unlikely caused by the reduced levels of Fig4p observed in the absence of Vac14p. When FIG4 is overexpressed in a vac7Δvac14Δfig4Δfab1-14 strain, membrane-bound Fig4p levels are higher than membrane-bound Fig4p levels in a wild-type strain (unpublished data). Yet, the levels of PI3,5P2 are identical in a vac7Δvac14Δfig4Δfab1-14 strain with or without the overexpression of FIG4. Thus, it appears that the defect in hyperosmotic shock–induced turnover of PI3,5P2 observed in vac14Δ cells is likely caused by the absence of Vac14p. Further supporting the idea that Vac14p is a regulator of Fig4p is the observation that Vac14p has weak homology to the regulatory subunit of protein phosphatase 2A (PR65/A).
The Vac14p–Fig4p complex likely activates Fab1p
The most surprising outcome of our studies is the observation that Fig4p is required for the hyperosmotic shock–induced elevation of PI3,5P2 levels. This elevation is likely the result of a Fig4p-dependent direct or indirect activation of Fab1p and not of the role of Fig4p in PI3,5P2 turnover. First, if hyperosmotic shock–induced increases in the levels of PI3,5P2 are caused by the inhibition of PI3,5P2 turnover, then in the fig4Δ and fig4G519R mutants, where turnover of PI3,5P2 is already inhibited, one would expect normal elevation of PI3,5P2. However, this is not observed. Second, the fig4G519R mutant is partially defective in both the hyperosmotic shock–induced elevation of PI3,5P2 and the corresponding decrease in PI3P that follows hyperosmotic shock (Fig. 7 C). Also note that in fig4Δ cells, basal levels of PI3P are 30% higher than wild type and remain high after hyperosmotic shock. This strongly suggests that production of PI3P is not perturbed in fig4Δ cells. The fact that the fig4Δ mutant has high levels of PI3P provides additional support for the model that turnover of PI3,5P2 is not a significant source of PI3P production and that the defect in the decrease in PI3P levels is caused by a defect in the synthesis of PI3,5P2.
We propose two models for how the Vac14p–Fig4p complex functions in the elevation of PI3,5P2 levels after hyperosmotic shock. In all of these models, Vac14p functions as an activator of Fab1p under both basal and hyperosmotic conditions. In one model, Fig4p inhibits Vac14p activity under basal conditions and releases this inhibition upon hyperosmotic shock. This mechanism would require Fig4p to function under basal conditions as both a PI3,5P2 phosphatase and an inhibitor of Vac14p. In a second model, Fig4p either enhances Vac14p activity or directly activates Fab1p after hyperosmotic shock. It is tempting to speculate that Fig4p performs these functions as a protein phosphatase.
Phosphoinositide phosphatases may function as protein phosphatases. The mammalian proteins PTEN and myotubularin have protein phosphatase activity in vitro (Cui et al., 1998; Laporte et al., 1998; Ramaswamy et al., 1999) and phosphoinositide phosphatase activity both in vitro and in vivo (Maehama and Dixon, 1998; Myers et al., 1998; Blondeau et al., 2000; Taylor et al., 2000). Furthermore, the critical active site residues of tyrosine phosphatases, serine/threonine/tyrosine phosphatases, and phosphoinositide phosphatases share an active site CX5RT/S motif. The Fig4p phosphatase point mutations presented in this study are located in or near this active site. Therefore, these mutations could affect both lipid and protein phosphatase activity.
Perhaps Fig4p functions as both a lipid phosphatase and protein phosphatase, dephosphorylating Vac14p or Fab1p to increase Fab1p activity. Notably, dephosphorylation of mammalian Fab1p increases its in vitro PI3P 5-kinase activity three to fourfold (Sbrissa et al., 2000). According to this model, cells lacking Fig4p phosphatase activity would have diminished Fab1p activation after hyperosmotic shock and defects in PI3,5P2 turnover. This is precisely what is observed in the Fig4p phosphatase point mutants. Moreover, the defect in Fab1p activation (∼35%) parallels the defect in PI3,5P2 turnover (∼30%).
In both models, the Vac14p–Fig4p complex regulates both PI3,5P2 turnover and synthesis. Such dual regulation may provide a mechanism to control acute and transient changes in PI3,5P2 levels. A similar model has been proposed for control of the onset of cytokinesis (Maddox and Oegema, 2003). In that case, the small GTPase RhoA is positively regulated by a complex containing the guanine nucleotide exchange factor Ect2 and negatively regulated by the GTPase-activating protein MgcRacGAP. Changes in the local concentration of RhoA, coupled with changes in the activities of MgcRacGAP and Ect2, establish the formation of a RhoA activity zone, thereby controlling the onset of cytokinesis (Minoshima et al., 2003; Bement et al., 2005). Tight coupling of the activation and inactivation of regulatory elements such as RhoGTPases, phosphoinositide kinases, and phosphatases is likely a common mechanism used to provide spatial and temporal control of signaling pathways.
Materials And Methods
Strains, media, and vectors
Strains are listed in Table I
. Strains were grown at 24°C in either YPD (yeast extract/peptone/glucose) or synthetic complete (SC) minimal media. All yeast plasmids were derived from the pRS400 series of vectors (Christianson et al., 1992).
PCR reactions used PfuUltra HF (Stratagene). Restriction enzymes and buffers were obtained from New England Biolabs, Inc. For pRS415-fig4G519R, a 660-bp BsaBI–NcoI fragment from pET-15b-fig4-1 (Rudge et al., 2004) was cloned into BsaBI–NcoI-gapped pRS415-FIG4. The fig4D469N mutant was generated by site-directed mutagenesis. A 542-bp ClaI fragment from pRS415-FIG4 was cloned into pUC18. The GAC codon (aspartic acid) was changed to AAC (asparagine) using the QuikChange Site-Directed Mutagenesis Kit (Stratagene). The ClaI fragment was cloned into a pUC18 plasmid containing the SacI fragment of FIG4. This mutated SacI fragment was then cloned into pRS415 to generate pRS415-fig4D469N. The mutation was confirmed with DNA sequencing.
Labeling of yeast vacuoles
Log phase cells were labeled with FM4-64 as described previously (Duex et al., 2006). Salt treatment of labeled cells was performed by resuspending cells in media containing 0.45 M NaCl, which elicits the same changes in vacuole volume as 0.9 M NaCl. A large change in refractive index when using 0.9 M NaCl makes it difficult to achieve focused images on the microscope. Cells were viewed at room temperature on a microscope (Axioplan II; Carl Zeiss MicroImaging, Inc.) with a 100× NA 1.40 ∞/0.17 plan Apochromat objective using a TRITC filter. Images were captured with a camera (RT-Spot; Diagnostic Instruments) and analyzed with MetaMorph software (Universal Imaging).
Subcellular fractionation and Western blot analysis
Experiments were performed as described previously (Duex et al., 2006). Log phase cells were lysed in cytosol buffer, and the extracts were spun at 300 g to sediment unbroken cells. The supernatant was spun at 13,000 g for 10 min at 4°C. The supernatant (S13) was moved to a fresh tube, and the pellet (P13) was resuspended in cytosol buffer in an equal volume to the supernatant. Fractions were subjected to SDS-PAGE and transferred to nitrocellulose for Western blot analysis. Blots were probed with rabbit anti-Fig4p sera or goat anti-Vac14p sera (Bonangelino et al., 2002). HRP-conjugated secondary antibodies (Invitrogen) followed by SuperSignal Chemiluminescence (Pierce Chemical Co.) was used to visualize immunoreactive proteins.
Random mutagenesis of FAB1
FAB1 was PCR amplified from pRS416-FAB1, forward primer CTCAAACCGAAGAACGTCCG, reverse primer GGCACCCCAGGCTTTACAC (in pRS416), and standard Taq polymerase conditions (Boehringer). This produced a 3.5-kbp product that included 2,053 bp FAB1 (encoding aa 1,595–2,278). The PCR product was cotransformed into a fab1Δvac14Δ strain with NheI–NotI-gapped pRS416-FAB1 and transformants selected on SC-uracil plates. Transformants were replica plated onto SC-uracil plates containing 100 μM of the iron chelator bathophenanthroline disulfonic acid (Fluka). Wild-type colonies have normal growth rates and are white, whereas fab1Δ, vac7Δ, or vac14Δ cells are red. After a 36-h incubation at 37°C, large white colonies were selected as vac14Δ bypass candidates. Their ability to bypass the large vacuole morphology of vac14Δ cells was verified with FM4-64 labeling. Candidate plasmids were isolated and retransformed into fab1Δvac14Δ cells to confirm that a plasmid encoded a hyperactive Fab1p mutant. The mutagenized regions were sequenced. The same approach was used to screen for fab1 mutants that bypass vac7Δ, except pRS416-fab1-6 was used as the starting PCR template.
Chromosomal integration of the fab1-6 and fab1-14 hyperactive mutants
The kanamycin cassette integrated along with the FAB1 mutations was PCR amplified from the yeast haploid knockout collection (Open Biosystems) using primers that had 45 bp of flanking sequence that would anneal to either the 3′ end of FAB1 or the 3′ untranslated region of FAB1. This PCR product was transformed into wild-type yeast along with BsrGI-digested (partial) pRS416-fab1-6 or pRS416-fab1-14 and transformants selected on YPD supplemented with gentamycin. Plasmid preps from candidate transformants were performed, and the pRS416-fab1:KAN cassettes were verified by restriction digest and DNA sequence. Positive plasmids were digested with SwaI–NotI, and the resulting 3.5-kbp fragment was transformed into vac7Δ (LWY2054) and transformants selected on YPD supplemented with gentamycin. Proper integration of the fab1:KAN mutants was verified with PCR. Generation of the fig4Δvac7Δvac14Δfab1-14 mutant (LWY7197) was performed by mating fig4Δvac14Δ (LWY6538) with vac7Δfab1-14 (LWY7179). Diploids were sporulated, and genotypes of the resultant tetrads were determined by the presence or absence of expected nutritional markers and PCR.
PI was extracted and analyzed as described previously (Duex et al., 2006). In brief, total PI was extracted from [3H]inositol-labeled cells and deacylated with methylamine. The inositol head groups were then separated with HPLC, and radioactive counts were measured. Values for PI3,5P2 are reported as the percentage of total PI extracted from each strain to normalize for [3H]inositol incorporation, the number of cells, and the amount of extracted inositol injected into HPLC.
We would like to thank Drs. Rob Piper, Mark Stamnes, Dave Kusner, and John Engelhardt for insightful discussions and advice. We thank Yanling Zhang, Omar Itani, Nicole Bruse, and Abby Stork for help in isolating Fab1p mutants and Marissa Schoenfish for help in constructing the Fig4p mutants. We also thank Jem Efe and Dr. Scott Emr (University of California, San Diego, La Jolla, CA) for Fig4p antibody, the fig4-1 mutant, and helpful discussions.
This work was supported by the National Institutes of Health grant GM50403 to L.S. Weisman. J.E. Duex was supported by the American Heart Predoctoral Fellowships 0215265Z and 0315239Z.
Abbreviations used in this paper: SC, synthetic complete; PI, phosphatidylinositol; PI3,5P2, PI 3,5-bisphosphate.