INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In addition to amino acids, nucleotides, and sugars, fatty acids are one of four families of basic cellular components. As such, they serve different biological functions; e.g., they form the hydrophobic core of lipid membranes, serve as an efficient storage form of metabolic energy, and may direct membrane association of otherwise soluble proteins.

Synthesis of fatty acids is catalyzed by a cytosolic multifunctional fatty acid synthetase complex (FAS) which in yeast is encoded by two genes, FAS1 ( subunit [6]) and FAS2 ( subunit [30]), whose products form the active hexameric ₆₆ complex. Following the condensation of acetyl coenzyme A (acetyl-CoA) with the sulfhydryl group of the FAS-bound phosphopantetheine prosthetic group, sequential two-carbon unit elongation of the growing acyl chain occurs by condensation with malonyl-CoA, under the concomitant release of CO₂. Seven elongation cycles are thus required to produce palmitoyl-CoA (C_16:0). FAS elongates acetyl-CoA-primed substrates only (43, 51). Acyl chains of intermediate length are elongated by a different, membrane-bound system that also uses malonyl-CoA for elongation (8, 40, 47). Null mutations in either FAS1 or FAS2 are lethal unless the cells are supplemented with exogenous fatty acids 12 to 18 carbon atoms in length (42).

Malonyl-CoA, required for chain elongation, is supplied by acetyl-CoA carboxylase (Acc1p in yeast; ACC in other organisms), the rate-limiting enzyme of fatty acid synthesis. In prokaryotes, ACC activity requires the functional association of three different proteins, which provide a biotin-carboxylase, biotin-binding site, and transcarboxylase function to the active heteromeric complex. In eukaryotes, on the other hand, ACC activity is encoded by a single, trifunctional protein. Yeast Acc1p has a subunit molecular mass of 250 kDa, is active as a tetramer, and is subject to short-term regulation by phosphorylation (4, 29, 46, 56). ACC1 transcription is regulated positively by Ino2p/Ino4p and negatively by Opi1p and is thus under the general transcriptional control of phospholipid biosynthetic genes (13). In mammalian cells, active ACC forms long filamentous structures (19). Similar structures have been observed in yeast cells that overexpress the protein (40). In wild-type yeast, Acc1p is a cytosolic enzyme that is in association with the endoplasmic reticulum membrane (17).

The first acc1 mutant strains have been isolated in screens for fatty acid auxotrophic cells (28, 35). These early acc1 alleles represent reduced-function rather than loss-of-function mutations, as indicated by the observation that an acc1 allele is lethal even when fatty acids are supplemented (4, 13, 40). More recently, a novel temperature-sensitive (ts) allele of ACC1, mtr7 (hereafter referred to as acc1^ts), has been isolated in a screen for mutants affected in nuclear mRNA transport (mtr mutants) (18, 40). Analysis of this conditional mutant revealed that the second essential function of Acc1p is to provide malonyl-CoA for elongation of long-chain (C₁₆ to C₁₈) fatty acids to very long chain (C₂₆) fatty acids (40, 41). The synthesis of C₂₆ is essential, as it forms part of the yeast ceramide (7, 38). Acyl chain elongation of C₁₆/C₁₈ to C₂₆ is catalyzed by two microsomal membrane proteins with overlapping substrate specificity, Elo2p and Elo3p (31). The function of Acc1p in C₂₆ synthesis cannot be restored by exogenous supplementation with long-chain and very long chain fatty acids, probably due to limited uptake and/or activation of the very hydrophobic C₂₆ compound (40). The observations that acc1^ts, but not the fatty acid auxotrophic acc1 alleles, affects nuclear export of mRNA and that acc1^ts mutant cells have an altered morphology of the nuclear membrane have been taken to suggest a requirement of C₂₆ synthesis in nuclear membrane-nuclear pore complex function (40, 41).

The biochemical, morphological, and genetic characterization of a novel cold-sensitive (cs) allele of acetyl-CoA carboxylase, acc1-200^cs (hereafter referred to as acc1^cs), is reported. This cs allele has been isolated in a screen for mutations that exhibit synthetic lethal interaction with hpr1, a hyperrecombination mutation of Saccharomyces cerevisiae (1-3, 12). The synthetic lethal interaction between hpr1 and acc1^cs is not allele specific, and we have previously shown that hpr1 and acc1^cs mutant cells affect nuclear export of polyadenylated RNA (39).

We now report that similar to the ts allele, the cs allele affects the synthesis of the C₂₆ fatty acid and that acc1^cs mutant cells have greatly reduced steady-state levels of C₂₆ under permissive conditions. Electron microscopic analysis of the morphology of acc1^cs mutant cells revealed a fragmented vacuolar phenotype. This altered vacuolar morphology was rescued by the exogenous addition of fatty acids. We provide biochemical evidence that this multilobed vacuolar phenotype is due to defective acylation of Vac8p, a myristoylated and palmitoylated vacuolar protein whose acylation has previously been shown to be required for vacuolar morphology and inheritance (32, 52).

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains, plasmids, and genetic techniques. Yeast strains and plasmids used in this study are listed in Tables 1 and 2. Media were prepared as described elsewhere (36). Media supplemented with fatty acids contained 1% Tween 40, 0.015% palmitic acid, and 0.015% stearic acid (35). Soraphen A, a kind gift from A. Hinnen, was added to media from a 10-mg/ml stock solution in methanol (48). Yeast cells were grown in liquid YEPD medium at 30°C. Optical density at 600 nm (OD₆₀₀) was monitored every hour for growth rate determinations. Yeast cells were transformed as described elsewhere (15).

Tn10-LUK mutagenesis. Plasmid pCG002 was mutagenized by Tn10-LUK transposon insertion (16). The transposon insertion site in the target plasmid was determined by restriction enzyme analysis and in some cases by sequencing the region upstream of the IS10-lacZ region of the transposon. DNA sequencing was performed with a Sequenase version 2.0 kit (U.S. Biochemical Corp. Cleveland, Ohio) and the oligonucleotide 5'-TGTAAAACGACGGGATC-3' as primer. Plasmids carrying different transpositions were used to transform an acc1^cs strain (479-2A). Transformants were scored for the cold-sensitive phenotype on YEPD medium at 20°C and for -galactosidase activity on 5-bromo-4-chloro-3-indolyl--D-galactopyranoside (X-Gal) indicator medium at 30°C. -Galactosidase assays were performed as described elsewhere (36).

Fatty acid analyses. Glycerophospholipids and sphingolipids were extracted as previously described (14). After alkaline hydrolysis of lipids, fatty acids were converted to methyl esters by BF₃-catalyzed methanolysis and separated by gas-liquid chromatography on a Hewlett-Packard Ultra 2 capillary column (5% phenyl dimethyl silicone) with a temperature gradient (20 min at 200°C, 10°C/min to 280°C, 15 min at 280°C). Fatty acids were identified by comparison to commercially available methyl ester standards (NuCheck, Inc., Elysian, Minn.).

Subcellular fractionation and Western blot analysis. To determine the membrane association of Vac8p, exponentially growing cells were lysed by vigorous agitation with 0.5-mm-diameter glass beads in lysis buffer (0.3 M sorbitol, 10 mM Tris [pH 7.5], 0.1 M NaCl, 1 mM MgCl₂, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride) containing protease cocktail (57). Extracts were precleared by centrifugation at 500 × g for 10 min and separated into pellet (P13) and supernatant (S13) fractions by centrifugation at 13,000 × g for 10 min. The S13 fraction was then further separated into pellet (P100) and supernatant (S100) fractions by centrifugation at 100,000 × g for 30 min as previously described (52).

Enzyme activity measurements. Cytosol used for ACC and FAS enzyme activity assays was prepared as follows. Cells were harvested at 1,200 × g for 10 min, washed with 0.1 M K₂HPO₄-KH₂PO₄ buffer (pH 6.5), mixed with breaking buffer (50 mM Tris-HCl, 100 mM NaF, 1 mM EDTA, 10 mM -mercaptoethanol, 0.25 M sucrose, 1 mM phenylmethylsulfonyl fluoride [pH 7.5]) and glass beads (0.30-mm diameter) in a ratio of 1:1:1 (wt/vol/wt), and disrupted by four 1-min bursts in a Braun-Melsungen homogenizer under CO₂ cooling. Glass beads were collected by centrifugation at 5,000 rpm for 5 min; the supernatant was centrifuged at 20,000 × g for 20 min and then centrifuged again at 195,000 × g for 80 min. Saturated (NH₄)₂SO₄ was added in three portions within 20 min to 50% saturation, and samples were stirred for an additional 30 min. The precipitate was collected by centrifugation at 15,000 × g, dissolved in a HEPES buffer (50 mM HEPES, 1 mM EDTA, 0.02% sodium azide, 50% glycerol [pH 7.0]), and stored frozen at 20°C. Enzymatic activities of Acc1p and Fas1/2p were stable over a period of 3 weeks after freezing of samples. All steps were carried out at 4°C.

Fluorescence microscopy and vital staining. Vital staining of vacuoles with the lipophilic styryl dye N-(3-triethylammoniumpropyl)-4-(p-diethylaminophenyl-hexatrienyl) pyridinium dibromide (FM4-64; Molecular Probes, Eugene, Oreg.) was performed as described elsewhere (49). In vivo localization of the Vac8p-GFP fusion and microscopic analysis of FM4-64-stained samples was performed using a Leica TCS 4d confocal microscope with a PL APO 100×/1.40 objective. Figures were composed using Adobe Photoshop 5.0 (Adobe Systems, San Jose, Calif.) on a Macintosh PowerPC (Apple Computer Co., Cupertino, Calif.) and printed on an HP8500 color laser printer (Hewlett-Packard, Palo Alto, Calif.). Corresponding pictures were recorded using identical pin hole openings and amplification settings.

Electron microscopy. For ultrastructural investigations, cells were fixed in 4% paraformaldehyde-5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.0)-1 mM CaCl₂ for 90 min at room temperature, then washed in buffer with 1 mM CaCl₂ for 1 h, and incubated for 1 h with a 2% (aqueous) solution of KMnO₄. Fixed cells were washed in distilled water for 30 min and incubated in 1% sodium metaperiodate for 20 min. Then samples were rinsed in distilled water for 15 min and postfixed for 2 h in 2% OsO₄ buffered with 0.1 M cacodylate at pH 7.0. After another wash with buffer for 30 min, the samples were dehydrated in a graded series of ethanol (50 to 100%, with en bloc staining in 2% uranyl acetate in 70% ethanol overnight) and embedded in Spurr's resin. Ultrathin sections were stained with lead citrate and viewed with a Philips CM10 transmission electron microscope.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Growth of the acc1^cs mutant is not fully rescued by fatty acid supplementation. The acc1^cs allele, acc1-200^cs, was recovered in a screen for mutants that displayed synthetic lethal interaction with a null allele of HPR1 (12). The acc1^cs strain showed a slight decrease in mitotic growth rate at the permissive temperature but grew poorly at the nonpermissive temperature of 20°C. Addition of fatty acids to the YEPD medium improved growth at 20°C but did not restore wild-type growth rates (Fig. 1A).

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Effects of fatty acids and of soraphen A on growth of the acc1cs mutant. (A) Haploid yeast colonies of the indicated genotype were resuspended in water, and aliquots of dilutions were plated on medium with and without fatty acid supplementation; 10 µl containing ca. 104, 103, 102, or 101 cells was spotted onto plates, which were incubated for 3 days at 30°C or for 7 days at 20°C. (B) The soraphen A-sensitive phenotype of acc1cs is semidominant. Diploid strains of indicated genotypes were streaked onto YEPD medium containing 0.25 µg of soraphen A per ml, and the plate was incubated for 2 days at 30°C.

acc1^cs is defective in malonyl-CoA-dependent acyl chain elongation and the synthesis of C₂₆. Supplementation with long-chain fatty acids is not sufficient to fully complement the loss of Acc1p function but is sufficient to rescue fas null alleles (13, 40, 42).

Fate of Acc1p mutant proteins. Previous analysis of the fate of ts mutant cells revealed that a shift to a nonpermissive temperature results in a rapid and irreversible loss of cell viability (40). Shifting the cs mutant strain to nonpermissive conditions, on the other hand, resulted in a reversible growth inhibition but did not affect cell viability (data not shown). To understand the difference in behavior of the two conditional alleles, the fate of the two mutant proteins was investigated in more detail. First, steady-state levels of Acc1p upon shifting cells to nonpermissive conditions were investigated by Western blot analysis. Levels of Acc1p in the cs strain did not visibly change upon a 4-h shift to nonpermissive conditions. Those in the ts mutant, however, declined to nondetectable levels within the same period of time at 37°C. This block in synthesis and/or increased turnover of Acc1p appeared to be specific, as steady-state levels of Fas1/2p displayed an apparently compensatory increase rather than decrease in acc1^ts (Fig. 2).

View larger version (32K): [in this window] [in a new window]   FIG. 2.   Steady-state levels of Acc1p in conditional acc1 mutants, determined by Western blotting to detect Acc1p and Fas1/2p expression in wild-type (W303; lane 1 and 2) acc1cs (479-2A; lane 3 and 4), and acc1ts (YRXS12; lanes 5 to 9) strains at the indicated times at permissive and nonpermissive conditions.

Interallelic complementation of different acc1 alleles. To map the mutation responsible for the Cs phenotype, transposon insertion mutagenesis was performed. During the course of this work, an insertion within the coding region of ACC1 that complemented the Cs phenotype of acc1^cs was recovered. DNA sequence analysis of the insertion junction indicated that this insertion would create a carboxy-terminally truncated ACC1 allele (acc1^C-term) that lacked the 574 amino acids after position 1772, including the CoA-binding site of the intact protein. Since the CoA-binding site is essential for the transcarboxylase activity of Acc1p, the truncated subunit would be expected to be enzymatically inactive (4, 23). Accordingly, a plasmid carrying this truncated acc1 allele failed to complement the lethality of an acc1 allele or the Ts phenotype of acc1^ts.

View larger version (38K): [in this window] [in a new window]   FIG. 3.   A biotinylation-deficient allele of ACC1, acc1K735R, partially rescues the acc1ts allele. Wild-type (wt; W303), the acc1ts mutant strain (ts; YRXS12), and the acc1ts mutant harboring a plasmid encoding the biotinylation deficient allele of ACC1 [ts(K735R); pRXS89] were serially diluted 10-fold and spotted onto YEPD and YEPD plates supplemented with fatty acids. Plates were incubated for 3 days at 30 or 37°C.

View larger version (19K): [in this window] [in a new window]   FIG. 4.   Map (drawn to scale) summarizing the positions of mutations in different acc1 alleles. The biotin prosthetic group is represented by a diamond-shaped symbol, the biotin-carboxylase domain is represented by a triangle (positions 100-550), and the transcarboxylase domain is denoted by a circle (positions 1450 to 2050). Lesions associated with different alleles are indicated.

Interallelic complementation by stabilization of the heteroallelic enzymatic complex. The fact that the cs mutation fell into the transcarboxylase domain but was nevertheless complemented by the carboxy-terminally truncated version of the enzyme suggested that in this case, interallelic complementation was not due solely to the functional substitution of one mutant protein domain by the corresponding wild-type domain of the complementing partner. Instead, the complementation map was consistent with the idea that interallelic complementation between different alleles could also be due to a stabilization of the enzymatically active heteroallelic complex. To test this hypothesis, we followed the fate of the thermolabile Acc1^tsp mutant in the presence or absence of the carboxy-terminally truncated version of the protein. Western blot analysis of acc1^ts mutant cells that coexpressed Acc1^C-termp revealed a marked stabilization of the full-length thermolabile enzyme upon a shift to nonpermissive conditions (Fig. 5).

View larger version (63K): [in this window] [in a new window]   FIG. 5.   Acc1C-termp stabilizes the thermolabile enzymatic complex formed by Acc1tsp. For Western blot analysis of acc1ts (YRXS12) transformed with an empty vector or with pCG002::Tn10-LUK#21 encoding the C-terminally truncated version of Acc1p, cells were grown at permissive conditions in synthetic medium lacking leucine and shifted to nonpermissive conditions for 8 h. Protein extracts were prepared and separated by SDS-PAGE, and blots were probed with peroxidase-conjugated ExtrAvidin to detect biotinylated Acc1p and the two isoforms of pyruvate carboxylase (Pyc1/2p).

Conditional acc1 alleles affect vacuolar morphology. Our previous analysis of the morphological alterations observed in acc1^ts at nonpermissive conditions revealed a striking alteration of the nuclear membrane. The two nuclear membranes were frequently separated from each other, and the newly formed intermembrane space contained vesicle-like structures (40, 41). To determine whether the cs allele exhibited a similar phenotype, cells were fixed and prepared for thin-section electron microscopy. As shown in Fig. 6B, acc1^cs mutant cells shifted to nonpermissive conditions for 4 h displayed a fragmented vacuole, but no alteration of the nuclear envelope was observed. The vacuolar phenotype was exhibited by the majority of cells and was not observed in similarly treated wild-type cells (Fig. 6A). Vacuolar fragmentation appeared to be dependent on long-chain fatty acid synthesis, as it was rescued in cells supplemented with long-chain fatty acids (Fig. 6C). In addition, some of the acc1^cs mutant cells contained electron-dense, granular cytosolic material, reminiscent of the structures that we previously observed in cells overexpressing Acc1p (40), suggesting that the cs mutant protein forms cytosolic aggregates at nonpermissive conditions (Fig. 6D).

View larger version (141K): [in this window] [in a new window]   FIG. 6.   Morphological analysis of acc1cs mutant cells. Transmission electron micrographs show wild-type (wt; A) and acc1cs mutant cells (B to D) shifted to nonpermissive conditions for 4 h in the absence (B and D) or presence (C) of supplemented fatty acids (fa). Fragmented vacuoles are indicated by arrows in panel B. Electron-dense granular structures reminiscent of Acc1p filaments are indicated by white stars in panel D. Bars: A to C, 1 µm; D, 0.1 µm.

View larger version (106K): [in this window] [in a new window]   FIG. 7.   Vacuolar morphology in wild-type, acc1cs, and acc1ts cells stained with FM4-64 Wild-type, acc1cs, and acc1ts cells were cultivated in YEPD or YEPD supplemented with fatty acids to early logarithmic growth phase at 30°C. Cells were then incubated with 30 µM FM4-64 for 30 min, followed by a chase for 1 h in YEPD or YEPD supplemented with fatty acids at 30°C. They were then incubated at the permissive temperature (30°C) or shifted to nonpermissive conditions (17°C for 3 h or 37°C for 30 min) and examined by confocal microscopy. Multilobed vacuoles are indicated by arrowheads in panels E, M, and Q. Dead cells in panels V and X are indicated by arrows. DIC (differential interference contrast) pictures of the visual fields are shown to the right of the fluorescence images. Bar, 10 µm.

Acylation and vacuolar membrane association of Vac8p is impaired in acc1^cs mutants. Several vacuolar inheritance mutants, isolated through selection with a fluorescence-activated cell sorter, have previously been divided into three classes based on vacuolar morphology (53). Among these, the class I vac8 mutant appears to arrest early in vacuolar inheritance and displays multilobed vacuoles, similar to what we observed in the conditional acc1 alleles (53). Interestingly, Vac8p/Yeb3p/Yel013p is an armadillo repeat-containing protein that requires both myristoylation and palmitoylation for localization to the vacuolar membrane (10, 32, 52). Moreover, a palmitoylation-deficient allele of VAC8, vac8-3, displays a multilobed vacuolar phenotype similar to that of a vac8 allele, indicating that palmitoylation is essential for a wild-type vacuolar morphology (52).

View larger version (26K): [in this window] [in a new window]   FIG. 8.   Membrane association and acylation of Vac8p, but not processing of API, is affected in acc1cs mutant cells. (A) Membranes isolated from exponentially growing wild-type, vac8-3 (expressing a nonpalmitoylatable allele of Vac8p) (52), and acc1cs cells were subfractionated into P100 and S100 fractions, and membrane association of Vac8p was determined by immunoblot analysis with an anti-Vac8p serum. (B) Whole cell extracts of wild-type (wt) and cs mutant cells shifted to nonpermissive conditions for 4 h in the presence (+fa) or absence of exogenously added fatty acids were separated by SDS-PAGE, and the relative mobility of Vac8p was assessed by immunoblot analysis with an anti-Vac8p serum. (C) API processing in wild-type (wt), vac8, vac8-3, and acc1cs mutant cells (cs) incubated at either permissive (30°C) or nonpermissive (17°C) conditions for 4 h was determined by immunoblot analysis with an anti-API serum. The positions of the precursor (p) and mature (m) forms of API are indicated.

acc1^cs affects the subcellular localization of a Vac8p-GFP fusion. To determine the in vivo localization of Vac8p in acc1^cs mutant cells, GFP was fused to the carboxy terminus of Vac8p by genomic integration of a GFP-KanMX6 module just prior to the stop codon of VAC8 (50). The resulting Vac8p-GFP fusion was functional in both API maturation and vacuolar morphology and inheritance (data not shown). Remarkably, subcellular localization of this fusion protein by confocal microscopy revealed a somewhat polarized distribution of Vac8p-GFP on the vacuolar membrane of wild-type cells, compared to the more uniform signal obtained from FM4-64 staining of the membrane. On relaxed vacuolar membranes of wild-type cells, the more intense Vac8p-GFP signal was always located on the site of the vacuole that faced the nucleus and the bud site (Fig. 9A, D, and G). In acc1^cs mutant cells at permissive conditions, Vac8p-GFP colocalized with the multilobed vacuole. Under nonpermissive conditions, however, a significant fraction of the cells displayed aberrant, uniform labeling of Vac8p-GFP throughout the cell (Fig. 9J). This mislocalization of Vac8p-GFP was not observed in acc1^cs mutant cells supplemented with fatty acids (data not shown).

View larger version (104K): [in this window] [in a new window]   FIG. 9.   Subcellular distribution of Vac8p fused to GFP in wild-type and acc1cs mutant cells. Wild-type and acc1cs mutant cells expressing a functional Vac8p-GFP fusion protein were grown to early logarithmic growth phase at 30°C. Cells were then incubated with 30 µM FM4-64 for 30 min, followed by a chase for 1 h in YEPD. The cultures were split and incubated at either permissive (30°C) or nonpermissive (17°C) conditions for 4 h. Vacuolar morphology and the subcellular distribution of Vac8p-GFP was then analyzed by confocal microscopy. The polarized distribution of Vac8p-GFP on the relaxed vacuolar membrane of wild-type cells is indicated by arrows in panels A, D, and G. Small arrowheads in panels D and J point to a polarized distribution of Vac8p-GFP on multilobed vacuolar structures. Aberrant distribution of Vac8p-GFP in acc1cs mutant cells at nonpermissive conditions is indicated by larger arrowheads in panel J. DIC (differential interference contrast) pictures of the visual fields are shown to the right of the fluorescence images. Bar, 5 µm.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Acc1p is a cytoplasmic enzyme that catalyzes the rate-limiting step of fatty acid synthesis. The gene is essential, and we have shown in this report that synthesis of the gene product is rate limiting. The inability to fully complement the cold-sensitive growth of the acc1^cs mutant by exogenous long-chain fatty acids reflects deficiencies in other essential malonyl-CoA-dependent processes, i.e., the microsomal fatty acid elongation systems (8, 31, 40). Consistent with such a defect in acyl chain elongation, the chain length profile of fatty acids in the acc1^cs mutant was shifted toward shorter chains. Most importantly, steady-state levels of C₂₆ were reduced approximately twofold in the mutant, suggesting that the malonyl-CoA pool in acc1^cs is limiting at permissive conditions and becomes depleted at nonpermissive conditions.

In an attempt to map the cold-sensitive lesion, a transposon insertion in the coding sequence of the ACC1 gene was recovered. This carboxy-terminally truncated allele complemented the cold-sensitive growth of the acc1^cs mutant but did not complement the temperature-sensitive growth of the acc1^ts mutant, suggesting that the cs allele affected a different domain of the enzyme than the ts allele. Consistent with this hypothesis, an acc1^cs/acc1^ts diploid strain was wild type for growth at all temperatures tested, indicating that the two alleles fully complemented each other's defect by interallelic complementation (58). The fact that the enzymatically inactive carboxy-terminally truncated version complemented the cs allele, despite bearing a defect in the same domain of the protein, suggested that in this case, complementation was due to a stabilizing effect of the truncated protein on the heteroallelic enzyme complex. In cs mutant cells, cytosolic structures resembling the filamentous form of Acc1p were apparent by electron microscopy, suggesting that the mutant enzyme aggregated at the nonpermissive temperature. In vitro sedimentation experiments, however, did not allow us to recapitulate this possible aggregation of the cs mutant protein, thus precluding a direct assay of the stabilizing effect of the truncated allele on the cs mutant protein (data not shown). A stabilizing effect of Acc1^C-termp on a heteroallelic enzymatic complex, however, could be demonstrated with the heat-labile enzyme, indicating that stabilization of an otherwise labile enzymatic complex is one possible mechanism to account for the complementation between conditional alleles and the nonfunctional alleles of ACC1.

Electron microscopic analysis of the morphology of acc1^cs mutant cells revealed a fragmented vacuole. This vacuolar phenotype was also observed by fluorescence microscopy of cells stained with FM4-64. The phenotype was not specific for the cs allele but was also observed in ts mutant cells. Addition of fatty acids fully rescued the vacuolar fragmentation of acc1^cs mutant cells as determined by electron microscopy and FM4-64 staining of viable cells. Acidification of the vacuolar compartment, however, was not affected in the conditional acc1 mutants, as revealed by quinacrine staining (data not shown) (54).

There were remarkable differences between the two conditional acc1 alleles with respect to the way in which they affected vacuolar morphology and inheritance: (i) in contrast to the ts mutant allele, the cs mutant cells displayed no defect in vacuolar inheritance; (ii) the multilobed vacuolar morphology of the ts mutant was not fully rescued by fatty acid addition; and (iii) shifting acc1^ts to nonpermissive conditions rapidly resulted in the relaxation of the vacuolar membrane. The signals that affect vacuolar structure may be manifold, and acylation of Vac8p is probably only one of these. The observation that the presence of fatty acids is not sufficient to fully relax the vacuolar membrane in the ts mutant suggests that additional signals required for vacuolar relaxation are also impaired in this mutant.

This apparent discrepancy between the two conditional acc1 alleles may be explained by recalling that the ts allele is the stronger of the two alleles, as indicated by the decline of steady-state levels of Acc1^tsp at 37°C and the fact that ts mutant cells shifted to nonpermissive conditions rapidly die. The acyl-CoA pool in this mutant thus appears to become depleted more efficiently than in the cs mutant cells, which require a shift to low temperature, a condition that by itself reduces metabolic turnover. The same argument may also explain why the nuclear envelope alteration is observed only in ts mutant cells (40, 41). Other phenotypes, such as the synthetic lethal interaction with hpr1 (39), however, are shared between the two alleles, suggesting that different levels of depletion of malonyl-CoA may result in distinguishable phenotypic consequences.

The multilobed vacuolar phenotype observed in the conditional acc1 mutants is a hallmark of the class I vacuolar inheritance (vac) mutation, vac8 (53). Vac8p is an armadillo repeat-containing protein whose function is required both for the Cvt pathway and for vacuolar morphology and inheritance (52, 53). Interestingly, Vac8p is both myristoylated and palmitoylated, and acylation of the protein is required for its localization to the vacuolar membrane but not for its function in protein targeting (32, 52). Characterization of the membrane association of Vac8p and its mobility in SDS-gels revealed that Vac8p was mostly soluble and not fully acylated in the acc1^cs mutant, suggesting that a reduced level of acylation of Vac8p may result in the observed multilobed vacuolar phenotype of acc1^cs mutant cells. Consistent with this proposed role of ACC1 in vacuolar morphology, API processing was not affected in acc1^cs mutant cells.

In contrast to a myristoylation-deficient mutant, a palmitoylation-deficient point mutant allele of VAC8, vac8-3, renders the mutant completely defective in vacuolar inheritance (52). Given that vacuolar inheritance is affected in the ts but not the cs mutant strain, one might suggest that the ts allele affects palmitoylation of Vac8p more strongly than does the cs mutation at 30°C.

Analysis of the subcellular distribution of a GFP-tagged functional version of Vac8p revealed a polarized distribution of Vac8p on the relaxed vacuolar membrane of wild-type cells, where it appeared to be concentrated on those parts of the membrane that faced the nucleus and the growing bud site. The subcellular localization of a carboxy-terminal fusion of Vac8p with GFP has previously been analyzed (32). In this study, the Vac8p fusion was observed to be concentrated ~5 to 7-fold in bands located between clustered vacuoles. While we see a similar clustering of Vac8p-GFP on multilobed membranes (Fig. 9D and J), our analysis extends this observation of the polarized distribution of Vac8 on relaxed vacuolar membranes. Consistently, previous localization of Vac8p by immunogold electron microscopy revealed that Vac8p appeared to be enriched on the vacuolar membrane at sites of membrane-membrane contact (10). The functional significance of this remarkably polarized localization of Vac8p remains to be determined. However, the fact that Vac8p is required for polarization of the vacuole to the presumptive bud site and specifically associates with actin filaments in vitro has been taken to suggest that Vac8p may mediate vacuole membrane-actin interactions (52).

In the cs mutant, mislocalization of Vac8p-GFP throughout the cell was observed at nonpermissive conditions. This mislocalization was not observed when the cells were supplemented with fatty acids (data not shown). The microscopic assay for membrane association of Vac8p, however, appears less sensitive than the biochemical fractionation, since minor amounts of soluble Vac8p will escape microscopic detection due to dilution of the protein in the cytosol, but these minor amounts will be revealed quantitatively by the immunoblot analysis. The mutant cells that displayed mislocalized Vac8p-GFP appeared to be more generally impaired, as indicated by the aberrant FM4-64 staining of the vacuolar membrane.

The relationship between fatty acid synthesis and Vac8p function was further investigated by epistatic analysis. The vacuolar morphology of an acc1^cs vac8-3 double mutant was strongly multilobed already at 30°C, indicating that vac8-3 is downstream of acc1^cs. Similarly, transformation of acc1^cs with a VAC8-harboring high-copy-number plasmid did not rescue the vacuolar phenotype of the cs mutant (data not shown). These results are consistent with the proposed function of ACC1 in acylation of Vac8p. Moreover, analysis of the vacuolar morphology of a conditional N-myristoylation-deficient (nmt1-181) strain (9) also revealed a multilobed vacuolar phenotype. Consistent with the myristic acid auxotrophy of this mutant strain, the vacuolar phenotype of nmt1-181 was fully rescued by supplementation with C_14:0 but not C_16:0 fatty acids (our unpublished observations). Thus, like the ts mutant acc1 allele, the vacuolar membrane of nmt1-181 did not relax upon fatty acid (C_16:0) supplementation. Finally, analysis of an acc1^cs nmt1-181 double mutant revealed a vacuolar morphology characteristic of nmt1-181, indicating that myristoylation is downstream of acc1^cs. Taken together, these genetic data are consistent with the proposed function of ACC1 in affecting vacuolar morphology through acylation of Vac8p.

Unlike myristoylation, palmitoylation appears to be reversible and thus may be regulated in a dynamic manner (27, 45). Protein acylation is generally thought of as a mechanism to shuttle a protein on and off a membrane. In case of Vac8p, however, acylation may be important for functions in addition to membrane targeting. The observed polarized localization of Vac8p is intriguing and might possibly stand in relation to the acyl anchor of Vac8p and the proposed properties of saturated acyl chains to form structurally and functionally defined membrane domains (5).