Palmitoylation of either Cys 4 or Cys 5 is required for Vac8p association with the vacuole membrane
Our previous studies demonstrated that Vac8p is myristoylated and palmitoylated and strongly suggested that Cys 4, 5 and 7 are each sites of palmitoylation (11). In order to determine whether each of these cysteines was a site of palmitoylation and to determine the role of acylation in Vac8p localization, all possible single and double mutants of Cys 4, 5 and 7 were constructed. Analysis of these mutants showed that acylation is not required for Vac8p expression or its stability. As measured by immunoblot analysis, Vac8p mutants that are not myristoylated or not palmitoylated are present in cells at similar steady state levels as wild-type Vac8p (Figure 1B).
Acylation of Vac8p is critical to its proper localization. Wild-type Vac8p was readily observed on the vacuole membrane by conventional immunofluorescence microscopy (Figure 2). However, using the same technique, neither the vac8-G2A nor the vac8-C4G/C5T/C7S mutants could be detected (data not shown). Therefore, we used a more sensitive ‘sandwich’ immunofluorescence microscopy technique (see Materials and Methods). Use of this technique revealed that a fraction of the Vac8p mutant that cannot be myristoylated, vac8-G2A, was still associated with the vacuole membrane. This mutant is not palmitoylated to the same extent as wild type (11). Therefore, this finding suggests that myristoylation and/or normal levels of palmitoylation are required for full association of Vac8p with the vacuole membrane. In contrast, in the absence of any myristoylation and palmitoylation, Vac8p cannot associate with the vacuole membrane. The vac8-G2A/C4G/C5T/C7S mutant, which cannot be acylated, was detected only as small, dispersed spots in the cytoplasm. Loss of only one putative palmitoylation site had little effect on Vac8p localization: all three of the single mutants (vac8-C4G, C5T and C7S) had levels of vacuole membrane-associated Vac8p that were detectable by conventional immunofluorescence microscopy. Loss of two potential palmitoylation sites resulted in decreased localization of Vac8p to the vacuole membrane, such that detection of the protein required the use of a sandwich technique. Two of the double mutants (vac8-C5T/C7S and vac8-C4G/C7S) were found to be localized to both the vacuole membrane and the cytoplasm. The remaining double mutant (vac8-C4G/C5T) had a more severe phenotype. Some Vac8p was detected on the vacuole membrane, but most of the protein was present as small spots dispersed in the cytoplasm. This indicates that either Cys 4 or Cys 5 is crucial for the proper localization of Vac8p to the vacuole membrane. These immunofluorescence localization studies parallel those observed using subcellular fractionation (26).
Figure 2. Palmitoylation at either Cys 4 or Cys 5 is required for vacuole membrane association of Vac8p. The vacuole membrane (red) was visualized with mouse anti-Vma2p and rhodamine-red-conjugated donkey anti-mouse IgG. Vac8p (green) was labeled with rabbit anti-Vac8p and Alexa-488-conjugated goat anti-rabbit IgG. * indicates that a ‘sandwich’ technique was used. For the sandwich technique, Vac8p was labeled with rabbit anti-Vac8p, followed by goat anti-rabbit IgG and Alexa-488-conjugated donkey anti-goat IgG. Scale bar is 5 μm.
Palmitoylation of either Cys 4 or Cys 5 is required for vacuole inheritance, vacuole fusion and caffeine resistance
Palmitoylation is required for Vac8p function (11). Simultaneous mutation of both the myristoylation and the palmitoylation sites abolishes Vac8p function in vacuole inheritance (11), homotypic vacuole fusion (13) (also see Figure 3A) and caffeine resistance (Figure 3B). Because Vac8p-G2A is not palmitoylated to the same extent as wild-type Vac8p (11), it is difficult to test the role of myristoylation in Vac8p function. Consistent with a partial defect in palmitoylation, the vac8-G2A mutant, while capable of supporting vacuole inheritance, had a modest defect in homotypic vacuole fusion as measured by vacuole fragmentation (11,13) (also see Figure 3A). Moreover, myristoylation alone is not sufficient for Vac8p function: the palmitoylation-minus mutant vac8-C4G/C5T/C7S was defective in vacuole inheritance, homotypic vacuole fusion and caffeine resistance (11,13) (Figure 3A–B).
Figure 3. Functional analysis of Vac8p palmitoylation mutants. A) Vacuole inheritance and vacuole homotypic fusion in wild-type and mutant vac8 cells. The indicated plasmids were transformed into a vac8Δ strain. Transformants were labeled with FM4-64 for 1.5 h and chased for one cell doubling. For each strain, more than 100 cells were scored for vacuole inheritance and for vacuole fusion. The vacuole of a wild-type (WT) cell usually contains three to six lobes. Cells that contained more than six lobes were scored defective in homotypic vacuole fusion. The average and standard deviations from three independent experiments are shown. ΔG: G2A; ΔC: C4G/C5T/C7S and ΔGΔC: G2A/C4G/C5T/C7S. B) Caffeine resistance of WT and mutant vac8 cells. vac8Δ cells carrying the indicated plasmids were spotted onto SC-URA (Synthetic Complete-Uracil) and SC-URA + 0.2% caffeine plates and incubated at 24°C for 3 days (SC-URA plates) or 4 days (SC-URA + 0.2% caffeine plates). C) Localization of Nvj1p–GFP in wild-type and mutant vac8 cells. pNVJ1–GFP was cotransformed with the indicated vac8 mutants. Transformants were labeled with FM4-64 for 1.5 h and chased for one cell doubling time. Scale bar is 5 μm.
Each of the single Cys point mutants functioned similarly to wild-type VAC8 in vacuole inheritance (Figure 3A). These mutants were virtually normal for homotypic vacuole fusion (Figure 3A) and caffeine resistance (Figure 3B). Likewise, two of the double mutants vac8-C5T/C7S and vac8-C4G/C7S behaved the same as the single mutants (Figure 3A–B). Importantly, however, the remaining double mutant vac8-C4G/C5T was severely defective in these functions: less than 20% of cells displayed normal vacuole inheritance and less than 40% showed normal vacuole morphology (Figure 3A). This double mutant also was less resistant to caffeine and had a clear growth defect on 0.2% caffeine plates (Figure 3B). These results indicate that palmitoylation at either Cys 4 or Cys 5 is required for vacuole inheritance, homotypic vacuole fusion and caffeine resistance. Palmitoylation at Cys 7 alone results in a partial defect in these Vac8p functions.
The levels of Vac8p on the vacuole membrane are crucial for formation of the nucleus–vacuole junction
Vac8p–green fluorescent protein (GFP) is enriched in at least two regions of the vacuole membrane: the nucleus–vacuole junction (10) and vacuole–vacuole junctions (vertices) (10,14). The nucleus–vacuole junction is the site of piecemeal microautophagy of the nucleus, whereby a portion of the nuclear membrane is engulfed by the vacuole and degraded by vacuolar hydrolases (27). Nvj1p, an outer nuclear membrane protein, is a component of the nucleus–vacuole junction and interacts with Vac8p in a yeast two-hybrid test (16). The proper localization of Nvj1p at the nucleus–vacuole junction requires enrichment of Vac8p at the same region (16).
Consistent with the above studies, in the presence of wild-type VAC8, Nvj1p–GFP concentrated at the nucleus–vacuole junction (Figure 3C). In the vac8-G2A mutant or in cells containing any of the single Cys mutants, Nvj1p–GFP was predominantly localized to the nucleus–vacuole junctions, with a small amount of Nvj1p–GFP dispersed around the nuclear envelope (Figure 3C). Two of the double mutants (vac8-C5T/C7S and vac8-C4G/C7S) were also nearly normal with most of the Nvj1p–GFP at the nucleus–vacuole junction. In contrast, in the double mutant vac8-C4G/C5T and the mutants in which all three Cys residues are replaced, Nvj1–GFP was dispersed throughout the outer nuclear membrane (Figure 3C). These findings suggest that palmitoylation of Vac8p at either Cys 4 or Cys 5 is required for localization of Vac8p and Nvj1p to the nucleus–vacuole junction.
Defects observed in nucleus–vacuole junction formation in the Vac8p mutant that cannot be palmitoylated at both Cys 4 and Cys 5 could be due directly to a loss of palmitoylation or due to the fact that levels of the mutant on the vacuole membrane are significantly lower than those of wild-type Vac8p. To distinguish between these possibilities, we compared the vac8-C4G/C5T mutant with a second vac8 mutant PVAC17-VAC8, which is composed of the wild-type VAC8 open-reading frame driven by the VAC17 promoter. The levels of VAC8 expressed from the VAC17 promoter are ∼10-fold lower than those of wild type (17). As measured by immunoblot analysis of isolated vacuoles, PVAC17-VAC8 and vac8-C4G/C5T mutants have similar levels of vacuole membrane-associated Vac8p (Figure 4A). Notably, this low level of Vac8p is not sufficient for formation of the nucleus–vacuole junction; Nvj1p–GFP was mislocalized in both mutants (Figure 3C). The above results show that the levels of Vac8p on the vacuole membrane are critical for formation of the nucleus–vacuole junction. Thus, we could not test whether palmitoylation plays a direct role.
Figure 4. Palmitoylation of Vac8p is required for vacuole inheritance and homotypic vacuole fusion. A) Comparison of vacuole membrane association of Vac8p-C4G/C5T and PVAC17-Vac8p. Vacuoles from a vac8Δ strain carrying the indicated plasmids were isolated and analyzed by immunoblotting using anti-Vac8p and anti-Pho8p antibodies. B) Functional differences of Vac8p-C4G/C5T and PVAC17-Vac8p. Vacuole inheritance and vacuole morphology were measured as described in Figure 3. The average and standard deviations from three independent experiments are shown. C) Localization of vacuoles and Myo2p–GFP in wild-type and vac8 mutant cells. The indicated plasmids were introduced into a vac8Δ strain that contains genomically integrated MYO2–GFP (LWY7814). Cells were labeled with FM4-64 for 1.5 h and chased for one cell doubling time. More than 15 cells were scored in each sample. The representative examples are shown. Scale bar is 5 μm.
Palmitoylation plays a role in targeting Vac8p to microdomains on the vacuole membrane
To further test the roles of Vac8p palmitoylation, we compared additional functions of the PVAC17-VAC8 and vac8-C4G/C5T mutants. These two mutants demonstrated functional differences in vacuole fusion and vacuole inheritance. PVAC17-VAC8 was significantly better at these two functions than vac8-C4G/C5T (Figure 4B). To analyze the partial defect of vac8-C4G/C5T in vacuole inheritance, we compared the localization of Myo2p relative to the vacuole in wild-type, vac8-C4G/C5T/C7S, vac8-C4G/C5T, PVAC17-VAC8 and vac8Δ cells. In wild-type cells, a portion of the vacuole extends to the bud tip, where it colocalizes with Myo2p–GFP. Likewise in the PVAC17-VAC8 mutant, which contains all three potential palmitoylation sites, a portion of the vacuole colocalizes with Myo2p–GFP. In contrast, in the absence of Vac8p, the vacuole remains in the mother cell, while Myo2p–GFP is concentrated at the bud tip. The vac8-C4G/C5T/C7S mutant shows the same phenotype. Notably, the vac8-C4G/C5T mutant has an intermediate phenotype. In approximately 50% of the cells, a small portion of the vacuole moves into the bud and colocalizes with Myo2p–GFP (Figure 4C). These results strongly support the hypothesis that Cys 7 can serve as a palmitoylation site. It is tempting to speculate that a single palmitoyl moiety at Cys 7 is not sufficient to fully anchor Vac8p on the vacuole membrane.
The fact that Vac8p is palmitoylated raises the possibility that vacuole microdomains may exist and, furthermore, may share common properties with the better characterized plasma membrane microdomains, specifically lipid rafts. Palmitoylated plasma membrane proteins, such as G proteins and Src kinases, reside in lipid rafts that are rich in sphingolipids and cholesterol (18). Sphingolipids and ergosterol (the yeast equivalent of cholesterol) are found in very low levels in the vacuole membrane (28). Intriguingly, a block in the biosynthesis of ergosterol (erg6Δ) or very long chain fatty acids (sur4Δ), sphingolipid precursors, results in fragmented vacuoles (29), a phenotype similar to that observed in the vac8Δ mutant. Moreover, like Vac8p, ergosterol accumulates at vacuole–vacuole vertices (30). These results suggest that ergosterol, sphingolipids and additional lipids form specific microdomains on the vacuole membrane and that Vac8p is enriched in these microdomains. Thus, we conducted the following experiments.
Plasma membrane lipid rafts are insoluble in Triton X-100. We tested the behavior of Vac8p in Triton X-100 and in other more mild detergents, Brij 98, Lubrol WX and Tween-20. These detergents have been successfully used to separate distinct types of plasma membrane rafts in mammalian and yeast cells (31–33). Each of these plasma membrane rafts has a unique protein and lipid composition (33). Moreover, ‘Lubrol rafts’ and ‘Brij 98 rafts’ have distinct physiological roles compared to ‘Triton rafts’ in vivo (34–37).
The solubility of Vac8p was compared with that of Pma1p, a plasma membrane protein that resides in Triton-insoluble rafts (28), and that of Pgk1p, a cytosolic protein. We found that regardless of the detergent, Pma1p remained associated with lipid and floated in the low-density (membrane-containing) fraction, while Pgk1p appeared with the soluble components (Figure 5A). The behavior of Vac8p in detergent was more complex. In the absence of detergent, most of Vac8p floated and cofractionated with Pma1p (Figure 5A). In the presence of Triton X-100 (Figure 5A) or Brij 98 (data not shown), Vac8p was completely soluble. In contrast, in the presence of the milder detergents, Lubrol WX or Tween-20, a significant amount of Vac8p cofractionated with Pma1p (Figure 5A), suggesting that a fraction of the Vac8p resides in ‘Lubrol’ rafts or ‘Tween 20 rafts’.
Figure 5. Palmitoylation plays a role in targeting Vac8p to specific microdomains. A) Solubility of Vac8p in the absence or presence of Triton X-100, Lubrol WX and Tween 20. Wild-type cells were lysed and fractionated in a 0–40% (w/v) Optiprep gradient in the absence or presence of 1% (w/v) Triton X-100, 0.5% (w/v) Lubrol WX or 1% (w/v) Tween 20. Samples were collected from the top (fraction 1) to the bottom (fraction 9) of the gradient, and proteins were subjected to immunoblot analysis for Pma1p, Vac8p and Pgk1p. B) Solubility of Vac8p-C4G/C5T and PVAC17-Vac8p in the absence or presence of 0.5% Lubrol WX. C) The localization of Vac8p-C4G/C5T and PVAC17-Vac8p by immunofluorescence microscopy. Vac8p (green) was detected by a ‘sandwich’ technique. Vac8p spots are much less frequent in the vac8-C4G/C5T mutant. Vacuole-localized Vac8p ‘spots’ were not detected in 36 randomly chosen vac8-C4G/C5T cells, while vacuole-localized Vac8p spots were present in 31% of PVAC17-VAC8 cells (n = 88). Scale bar is 5 μm.
Next, we tested whether the Lubrol WX- or Tween 20-insoluble portion of Vac8p accounted for the functional difference between the vac8-C4G/C5T and the PVAC17-VAC8 mutants. We applied the same gradient centrifugation to the mutants in the presence of Lubrol WX (Figure 5B) or Tween-20 (data not shown). A modest but reproducible difference was observed. We consistently detected more PVAC17-Vac8p in the same fractions as Pma1p.
In order to further test whether fully palmitoylated Vac8p resides in subdomains on the membrane, we compared the localization of PVAC17-Vac8p with Vac8p-C4G/C5T. By immunoblot analysis, both proteins were found to be at similar levels on the vacuole membrane (Figure 4A). However, when detected by immunofluorescence microscopy, PVAC17-Vac8p was observed as many spots on the vacuole membrane, whereas the Vac8p-C4G/C5T mutant had far fewer spots, presumably because it is more diffusively distributed (Figure 5C). Interestingly, this distribution of PVAC17-Vac8p to small spots on the vacuole membrane is similar to what has been observed for its binding partner Vac17p (6). We postulate that these small spots are Vac8p microdomains that facilitate the interaction between Vac8p and Vac17p.