|Home | About | Journals | Submit | Contact Us | Français|
The pandemic of lipid-related disease necessitates a determination of how cholesterol and other lipids are transported and stored within cells. The first step in this determination is the identification of the genes involved in these transport and storage processes. Using genome-wide screens, we identified 56 yeast (Saccharomyces cerevisiae) genes involved in sterol-lipid biosynthesis, intracellular trafficking, and/or neutral-lipid storage. Direct biochemical and cytological examination of mutant cells revealed an unanticipated link between secretory protein glycosylation and triacylglycerol (TAG)/steryl ester (SE) synthesis for the storage of lipids. Together with the analysis of other deletion mutants, these results suggested at least two distinct events for the biogenesis of lipid storage particles: a step affecting neutral-lipid synthesis, generating the lipid core of storage particles, and another step for particle assembly. In addition to the lipid storage mutants, we identified mutations that affect the localization of unesterified sterols, which are normally concentrated in the plasma membrane. These findings implicated phospholipase C and the protein phosphatase Ptc1p in the regulation of sterol distribution within cells. This study identified novel sterol-related genes that define several distinct processes maintaining sterol homeostasis.
Both cholesterol biosynthesis and storage are controlled in response to levels and localization of regulatory pools of sterols (33, 37, 54, 65). In response to high cholesterol levels in the endoplasmic reticulum (ER) membrane, the enzyme acyl coenzyme A (CoA):sterol O-acyltransferase (ASAT) initiates sterol esterification and storage by covalently coupling fatty acids to cholesterol. Through an active process, the esterified cholesterol is amalgamated with other neutral lipids into lipid storage droplets that are released from the ER membrane (42, 88). The trafficking of unesterified sterols also affects the sterol distribution in regulatory pools. Although cholesterol is synthesized in the ER, the highest level of unesterified cholesterol is found in the plasma membrane (33) and maintenance of normal sterol levels requires the efficient transport of cholesterol from the ER membrane to the plasma membrane. The maintenance of cholesterol levels in the plasma membrane is affected by sorting from endosomal compartments and recycling back to the cell surface (33, 54), and feedback regulation of cholesterol on its own biosynthesis and storage also controls levels of cellular sterols (16, 68). These findings suggest that the maintenance of cellular cholesterol homeostasis requires the regulatory integration of cholesterol synthesis, storage, and transport pathways.
As in mammalian cells, the budding yeast Saccharomyces cerevisiae synthesizes its own cholesterol-like lipids but, under normal aerobic conditions, yeast does not internalize exogenous sterol lipids. Apart from this difference, other elements of sterol homeostasis, including lipid storage and transport pathways, appear to be conserved (70). In yeast, ASAT is encoded by two homologous genes, ARE1 and ARE2, which together generate steryl esters for lipid storage droplets (84). Lipid droplets are also comprised of triacylglycerols, which in yeast are produced by the acyl-CoA:diacylglycerol acyltransferase 2 (DGAT2) homologue encoded by DGA1 and by the phospholipid:diacylglycerol acyltransferase (PDAT) homologue encoded by LRO1 (46). The genes encoding sterol and diacylglycerol acyltransferases are not essential, and a viable strain has been constructed that lacks all genes required for neutral-lipid biosynthesis (61, 66). These findings indicate that lipid storage is itself not required for yeast growth under normal culture conditions. A likely explanation for why neutral-lipid/sterol storage is dispensable for yeast viability is that it represents only one of several independent mechanisms that contribute to the maintenance of lipid and sterol homeostasis. This leads to the prediction that sterol regulatory pathways are functionally redundant and that growth defects occur only when several of these pathways are disrupted in concert.
In the case of sterol storage and other sterol regulatory pathways, functional redundancy has been successfully exploited to identify novel sterol-associated genes in yeast. ARV1, which affects the distribution of unesterified sterols, was originally identified as a deletion mutation that is lethal in combination with deletions of both ARE1 and ARE2 (72). This finding suggests that both sterol storage and trafficking make overlapping contributions to sterol homeostasis. ECM22 and UPC2 encode transcription factors that control another aspect of sterol homeostasis through the coordinate regulation of several sterol biosynthesis genes (79). Although the combined deletion of ECM22 and UPC2 is not lethal, upc2Δ ecm22Δ cells are inviable with the additional perturbation of sterols caused by the deletion of ERG2, which encodes the otherwise nonessential enzyme C-8 sterol isomerase (79). Together these results affirm that the disruption of just one sterol regulatory pathway is not detrimental unless there are additional defects in sterol homeostasis.
In this study, we carried out a functional genomics screen to identify yeast deletion mutants that cannot tolerate drug-induced disruptions in sterol homeostasis. This screen successfully identified 56 known and novel genes that are required for maintenance of sterol homeostasis. The identified deletion mutants were analyzed by cellular and biochemical approaches to establish their specific roles in sterol-lipid biosynthesis, trafficking, and/or storage. In this study, we defined distinct steps required for lipid storage droplet biogenesis and established a link between ASAT/DGAT lipid esterification and secretory protein glycosylation. Our findings provide insights into mechanisms affecting sterol transport, synthesis, and neutral-lipid storage, which together maintain sterol homeostasis and are potentially linked to human lipid disorders.
Culture media and genetic manipulations were as described previously (1). To select for the kan-MX4 gene, yeast were grown on yeast rich medium (YPD) containing 200 μg/ml Geneticin sulfate (G418) (Gibco BRL Life Technologies, Inc., Rockville, MD). YPD solid medium containing nystatin (Sigma Chemicals, Inc., St. Louis, MO) or lovastatin (a gift of Merck & Co., Inc., NJ) was prepared as previously described (8).
Functional genomics screens were conducted using the nonessential kan-MX4-marked homozygous diploid deletion collection (isogenic derivates of BY4743), and subsequent analysis involved the MATa nonessential haploid deletion strain collection (isogenic derivates of BY4741) (81). The genotypes of other yeast mutant strains not obtained from the deletion mutant collections are listed in Table Table1.1. Strains bearing multiple gene disruptions were generated through standard genetic crosses.
DNA cloning techniques and bacterial transformations were performed by standard procedures (60) (Table (Table2).2). Restriction enzymes were obtained from New England Biolabs (Beverly, MA). Oligonucleotide primers for PCR were purchased from Operon Biotechnologies, Inc. (Huntsville, AL), and the yeast genomic DNA template for amplification was isolated from BY4741. All oligonucleotide primers used for PCR amplifications are listed in Table Table33.
To construct a yeast plasmid that would rescue CNB1 mutant defects, the primer combination of CBP263 and CBP264 was used to amplify the CNB1 gene by PCR. The amplified 1.1-kb fragment included all promoter and terminator sequences for wild-type expression and was cloned into the EcoRI-XhoI sites of pRS416 (64) to generate the plasmid pCB456. One set of primers used to amplify the CAX4 gene were CBP267 and CBP268, and the amplified 1.9-kb fragment was cloned into the BamHI site of pRS416 to generate pCB419. The YHP1 and YHP2 primers were used to generate the CAX4 gene, after which the amplified fragment was cloned into the HindIII and BamHI restriction sites of YCplac111 to generate the plasmid YCplac111-CWH8. The primer combination used to amplify VMA21 was CBP287 and CBP288, which produced a 0.6-kb fragment that was cloned into the EcoRI site of pRS416 to generate pCB523. Using primers CBP276 and CBP277, a 3.1-kb PLC1 fragment was amplified and cloned into the EcoRI site of pRS416, producing pCB526.
To screen the homozygous deletion collection for sterol-sensitive mutants, a pin replicator was used to transfer equivalent inocula from strains arrayed and grown on solid medium into 200 μl of sterile water. The resuspended cells were further diluted 100-fold in sterile water into individual wells of microtiter plates. Using a pin replicator, strains were spotted and arrayed onto YPD solid rich medium and onto YPD containing 5 U/ml nystatin or 20 U/ml nystatin. Strains cultured on YPD solid medium or YPD containing 5 U/ml nystatin were incubated for 1, 1 to 2, and 3 days at 37, 30, and 23°C, respectively. Strains cultured on YPD solid medium containing 20 U/ml nystatin were incubated at 37, 30, or 23°C for 2, 3, and 4 days, respectively. Resistance to nystatin was recorded only if the mutant grew in the presence of 20 U/ml nystatin, whereas sensitivity was recorded only for strains that grew poorly on medium containing 5 U/ml nystatin. Growth defects were assessed relative to the wild-type control (BY4743) and in comparison to growth of each respective deletion strain on YPD without nystatin. If the growth of a specific deletion strain was affected by nystatin when cultured at two or more of the temperatures tested, the strain was picked and retested on nystatin-containing medium to confirm the results. Once confirmed, these deletion strains were then arrayed on solid medium and equivalent inocula were transferred to wells of microtiter plates and diluted 100-fold in sterile water. Using a pin replicator, these strains were spotted onto YPD solid rich medium and YPD medium containing 150 μg/ml lovastatin and incubated at 37, 30, and 23°C for 2, 3, and 4 days, respectively. Relative to the wild-type control (BY4743), deletion mutants that were susceptible to lovastatin for at least two of the three culture temperatures tested were retested (no lovastatin-resistant deletion mutants were identified). The confirmed list of nystatin/lovastatin-affected homozygous deletion mutants includes all of the deletions shown in Table Table44.
To examine sterol-lipid distribution, yeast cells were fixed and treated with filipin complex as previously described (6). For filipin and FM4-64 colocalization, 5.0 units of log-phase cells at an optical density at 600 nm grown in synthetic complete medium at 30°C were pelleted and cultured at 30°C with 32 μM FXM4-64 (Molecular Probes/Invitrogen, Carlsbad, CA) (78) for either 5 or 25 min. After the timed FXM4-64 uptake, cells were washed once with water, pelleted, and diluted to an optical density at 600 nm of 0.7 units/ml with fresh medium. Cells were then fixed for 10 min following the addition of formaldehyde to a final concentration of 3.75%. These cells were treated with filipin complex as described previously (6).
Lipid storage droplets were visualized by fluorescence microscopy after treatment with the lipophilic dye Nile red (Sigma Chemicals, St. Louis, MO). Cells from mid-logarithmic-phase-grown cultures were centrifuged, and the cell pellet was resuspended with water before the addition of Nile red to a final concentration of 2 μg/ml. Nile red-stained cells were washed once with water before visualization. Osmotically susceptible mutant strains were fixed with 3.75% formaldehyde for 15 min and washed in an equal volume of water before and after Nile red addition.
For all fluorescence microscopy, samples were mounted on poly-lysine-coated slides, sealed under coverslips with nail polish, and imaged on a Leica DMRA2 microscope microscope (Leica Microsystems, Wetzlar, Germany) equipped with a Orca-ER charge-coupled device digital camera (Hamamatsu Photonics, Hamamatsu City, Japan). Filipin and FM4-64 fluorescence was observed with a UV and fluorescein isothiocyanate (FITC) filter set using neutral-density filters to preserve fluorescence. For each experimental trial shown, equal exposure times were used to compare cellular fluorescence. Image analysis was performed using Improvision (Lexington, MA) Open Lab image analysis software.
The incorporation of [3H]oleate into steryl ester and triacylglycerol was used as a measurement of sterol and diacylglycerol esterification rates as described previously (46). Cells (5 ml) were grown in YPD liquid medium to mid-logarithmic phase and then incubated at 30°C for 30 min with 5 μCi of [3H]oleate. To remove residual [3H]oleate, cells were washed twice with 0.5% Tergitol, washed once with water, and then lyophilized. Dried cell pellets were resuspended in 50 μl of lyticase solution (1,700 U/ml in 10% glycerol, 0.02% sodium azide), chilled for 1 h at −70°C, and then incubated at 30°C for 15 min. Lipids were extracted by hexane and analyzed by thin-layer chromatography (TLC). The plates were developed in hexane-diethyl ether-acetic acid (70:30:1) and stained with iodine vapor. Incorporation of label into lipids was determined after scintillation counting and normalization to a [14C]cholesterol internal standard and cell dry weight. For each assay, at least three independent strains of each genotype were used. Statistical analysis was performed using the paired t test.
Lipid extractions were performed as described by Zhang et al. (86), and the quantification of neutral lipids and unesterified sterols were assayed by the methods of Zweytick et al. (88) with modifications. Log-phase cells were grown in rich medium and pelleted and then resuspended and pelleted twice in 0.5% Nonidet P-40 and once in distilled water before lyophilization. The dried cell pellets were resuspended in 50 μl of lyticase (1,700 U/ml in 10% glycerol; Sigma Chemicals), incubated at 37°C for 15 min, and then freeze/thaw lysed at −70°C for 1 h and then at 37°C for 15 min. Lipids were extracted with hexane, blown dry with N2, and dissolved in 100 μl of chloroform-methanol (2:1 [vol/vol]). Samples were applied to Silica gel 60 F254 plates (Merck), and chromatograms were developed in hexane-diethyl ether-acetic acid (85:15:1) with cholesterol, triolein, and cholesteryl ester (Sigma Chemicals) as the standard. Quantitative analysis of unesterified sterol was carried out by densitometric scanning at 275 nm with a CAMAG TLC scanner. For quantitation of steryl ester and TAG, plates were dipped into methanolic MnCl2 solution (0.63 g MnCl2 · 4H2O, 60 ml water, 60 ml methanol, and 4 ml concentrated sulfuric acid), dried, and heated at 120°C for 15 min. Densitometric scanning was performed at 500 nm.
Yeast extracts for Western blots were prepared as described by Ohashi et al. (47). Prior to loading for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), samples were incubated in sample buffer at 60°C for 10 min before loading. Transfer and immunoblot wash conditions were as previously described (7). Polyclonal antibodies were raised in rabbits against glutathione S-transferase (GST)-fused Are1p (amino acids 12 to 191) and against GST-fused Lro1p (amino acids 440 to 661). Rabbit anti-Are1p polyserum was used at a 1:500 dilution; rabbit anti-Lro1p polyserum was used at 1:1,000. Rabbit anti-Vti1p was a gift from Wanjin Hong (Institute of Molecular and Cell Biology, Singapore) and was used at a dilution of 1:3,000. Bands were visualized with a 1:3,000 dilution of horseradish peroxidase-conjugated antirabbit secondary antibody, followed by chemiluminescent detection (Pierce Chemical Co., Rockford, IL). Endoglycosidase H (endo H) removal of N-linked glycosylation was performed on protein extracts as described by the manufacturer (Sigma Chemicals, Inc.). Following deglycosylation, equivalent amounts of protein (20 μg) were loaded per lane for SDS-PAGE and, after transfer, immunoblots were probed with anti-Are1p and anti-Lro1p antibodies and detected by chemiluminescence as described above.
Samples were prepared for electron microscopy as described previously (55). In brief, cells were fixed and embedded after treatment with osmium-thiocarbohydrazide and dehydration. Thin sections were stained with uranyl acetate and lead citrate prior to viewing on a Philips CM12 transmission electron microscope.
Sterol lipids are essential for viability in almost all eukaryotic cells. The overall regulation of sterols, however, involves the control of sterol synthesis, as well as sterol transport and storage pathways. The individual pathways appear to be dispensable for normal cell growth only because each pathway compensates for the others. We predicted that yeast mutations that disrupt any of these pathways might result in cells that are sensitized to further sterol perturbations and would be unable to compensate for imbalances in sterol homeostasis. With this in mind, a yeast functional/chemical genomics approach was applied to identify nonessential genes representing each of the general pathways that conspire to maintain sterol homeostasis.
Ergosterol is a cholesterol-like sterol that is the bulk product of sterol biosynthesis in yeast. To identify potential candidate sterol-regulatory genes, we screened the ~4,700 diploid homozygous deletion strains (81) for sterol-related defects. This mutant collection represents deletions corresponding to almost all individual nonessential genes in the yeast genome (81). We screened the homozygous diploid collection, as opposed to haploid deletions, to reduce false mutant identifications due to nonspecific spontaneous recessive mutations. For the initial screening of the deletion collection, we analyzed mutant growth in the presence of the ergosterol-binding antibiotic, nystatin (80, 82). Although the mechanism of nystatin toxicity is complex, it exerts its effects by direct binding to plasma membrane ergosterol and nystatin has been successfully used to select viable mutants defective in ergosterol biosynthesis (38). In this broad-based screen (see Materials and Methods), we identified 262 nystatin-susceptible mutants and 95 nystatin-resistant mutants. Some of the same deletion mutants (e.g., ARV1, ERG3, CDC50, CNB1, DRS2, NUT1, PHO88, RIM9, SIF2, UME6, ZAP1, etc.) were independently identified in previous genomic studies that examined the effects of various sterol-affecting drugs on yeast growth (21, 34, 49, 50). However, to eliminate from consideration those deletion mutants that did not specifically affect sterols, we performed a secondary screen using lovastatin. In yeast (and mammals), lovastatin reduces total amounts of sterols by inhibiting the rate-limiting enzyme in sterol synthesis, 3-hydroxy-3-methyl-glutaryl-CoA reductase (5, 74). In this regard, nystatin and lovastatin have different inhibitory mechanisms but, because both drugs disrupt normal sterol regulation, both would affect deletion strains defective in sterol homeostasis. Thus, the 357 deletion strains affected by nystatin were tested for lovastatin sensitivity (none were lovastatin resistant). Fifty-seven of the nystatin-susceptible deletion strains were lovastatin sensitive, whereas only 5 nystatin-resistant strains were also lovastatin sensitive. Six of the mutants had been previously reported to exhibit nonspecific multidrug sensitivities (49), and they were not analyzed further unless independent evidence suggested otherwise (see below). Therefore, the 56 deletion mutations that affected sterol regulation or homeostasis are listed in Table Table4;4; 8 of these mutants had mutations that corresponded to genes with established links to lipid synthesis, regulation, or transport (ARV1, BTS1, CDC50, DRS2, ERG3, ERG6, PLC1, and RAM1), whereas 3 corresponded to novel genes (YEL045C, YDL133W, and YJL175W). The remainder of mutants represented known genes that have not been previously reported to have a role in sterol or lipid function. Our study is complementary to a previous genomic approach that surveyed all nonessential deletion mutants for those that affected sterol uptake during anaerobic growth conditions (56). Except for the tkl1Δ and rlr1Δ mutants, the subset of mutants identified by these different approaches had no overlap. To determine how the deletion mutations we identified affect sterols, each of the 56 deletion mutants was analyzed for specific cellular defects in sterol storage, sterol synthesis, or in the intracellular membrane distribution of sterols.
Esterified sterols within lipid droplets represent a major pool of cellular sterols in most eukaryotes, including yeast (42, 88). To determine if lipid storage was affected by any of the deletion mutations we identified, mutant cells were examined after incubation with Nile red, a fluorescent dye that stains neutral-lipid storage droplets. Nile red has been successfully used both in yeast (84) and in Caenorhabditis elegans genomic screens to examine lipid storage defects (3). Haploid deletion mutants, representing each of the 56 sterol-related homozygous deletion strains identified in the genomic screens, were individually cultured in rich medium to the logarithmic phase and stained with Nile red. The number, size, and intensity of Nile red-stained lipid droplets were determined by fluorescence microscopy and image analysis (Table (Table4).4). In wild-type cells, an average of 2.7 Nile red droplets was observed by fluorescence microscopy in a single optical section. Although many of the deletion strains had relatively modest but reproducible deviations from the wild-type control in droplet number or Nile red staining intensity, seven deletion mutants had severely reduced numbers of lipid droplets (fewer than half that of wild type) and 2 strains (ume6Δ and cdc50Δ) had a significant increase in both intensity and droplet number (Table (Table4).4). These findings suggested that at least some of the deletion mutants originally identified were susceptible to sterol-specific inhibitors because of lipid storage defects. As a pragmatic approach, we conducted detailed analyses on just those mutants having the greatest effects on the number and fluorescence intensity of Nile red-stained lipid droplets.
As shown in Fig. Fig.1,1, the deletion of CAX4 drastically reduced lipid storage droplets. CAX4 encodes dolichyl pyrophosphate phosphatase (19), and the corresponding deletion mutant exhibited the greatest reduction in lipid droplet numbers of those analyzed. Compared to wild-type cells, there were 9.1-fold-fewer lipid droplets in cax4Δ cells (Table (Table44 and Fig. Fig.1).1). Of all the deletion mutants with reduced numbers of lipid droplets, the deletion of CAX4 had the greatest impact. To confirm that the observed Nile red staining defects were a direct result of the specified deletion and were not due to another unlinked random mutation, we transformed the cax4Δ strain with a low-copy plasmid containing its respective wild-type gene. The cax4Δ strain transformed with the CAX4 gene, but not the vector alone control, fully rescued the lipid droplet defect phenotype (Fig. (Fig.1).1). These results implicated CAX4 as being required for lipid droplet biogenesis.
In contrast to cax4Δ cells, deletion of either UME6 or CDC50 caused enhanced Nile red fluorescence and a proliferation of lipid droplets relative to wild-type cells (Table (Table4).4). In this regard, the ume6Δ and cdc50Δ mutants were unique in that, in addition to the many lipid droplets, the intensity of Nile red fluorescence was significantly greater than those observed in the other mutants. For these reasons, the sterol defects in ume6Δ and cdc50Δ cells were analyzed in detail. CDC50 encodes a protein that is involved in cell polarization and also regulates the cellular localization of Drs2p, a lipid translocase (45, 58). The finding that the number of lipid droplets increased in cdc50Δ cells was particularly noteworthy because it was recently shown that CDC50 genetically interacts with sterol biosynthetic genes (31). In cdc50Δ cells, wild-type CDC50 expressed from a plasmid rescued the lipid droplet proliferation and the observed increase in Nile red staining (Fig. (Fig.1).1). These findings indicated that the lipid droplet defect was linked to the CDC50 locus. Significant increases in both lipid droplet number and intensity were also observed in ume6Δ cells (Table (Table4).4). UME6 encodes a transcriptional regulator that induces early meiotic genes (77), but it also has a role as a regulator of specific mitotic genes (18, 69, 71). All told, our results established that several yeast genes, in addition to those directly involved in neutral-lipid biosynthesis, are required for lipid storage particle biogenesis.
A core component of lipid storage particles is esterified ergosterol and the enzyme ASAT catalyzes the coupling of sterols with fatty acids (83, 84). If cax4Δ mutations inhibit lipid droplet formation by blocking sterol transesterification, then ASAT activity might be reduced in mutant cells. To measure ASAT activity, the rate of [3H]oleate incorporation into steryl esters was determined for wild-type and mutant strains. Compared to the wild-type strain and the other deletion mutants tested, the cax4Δ mutant had significantly reduced ASAT activity (Fig. (Fig.2A).2A). After a pulse-labeling for 30 min at 30°C, the amount of [3H]oleate incorporated into steryl esters in the cax4Δ mutant was only 32% of that measured for the wild-type strain. Consistent with the reduction in ASAT activity, steady-state levels of steryl esters in log-phase cax4Δ cells were markedly reduced when separated and measured by TLC after lipid extraction (Fig. (Fig.3).3). These findings indicated that the lipid droplet defect observed in cax4Δ cells was caused at least in part by a reduction in ASAT activity.
Another major lipid component of yeast lipid droplets is triacylglycerol, which is synthesized by DGAT and PDAT (67). To determine whether defects in DGAT/PDAT activity contributed to the observed reduction of lipid storage particles in the cax4Δ strain, cells were pulse-labeled with [3H]oleate to measure the rate of its incorporation into triacylglycerol. In the cax4Δ strain, DGAT/PDAT activity was reduced (29% of wild type) (Fig. (Fig.2B)2B) and steady-state levels of triacylglycerol were barely detectable by TLC (Fig. (Fig.3).3). These results indicated that the CAX4 deletion affects all aspects of neutral-lipid synthesis, including ASAT and DGAT/PDAT activities.
Cax4p is a dolichyl pyrophosphate phosphatase that regenerates free dolichol, a lipid required for N-linked core glycosylation of secretory proteins in the ER (19, 76). To determine whether the generation of lipid storage particles is linked to secretory protein glycosylation, or to another aspect of dolichol metabolism, we examined temperature-sensitive (ts) sec53-6ts mutant cells (RSY12) for lipid droplet defects. SEC53 encodes an essential phosphomannomutase required for core glycosylation of secretory proteins, but SEC53 is not directly associated with dolichol metabolism (30). If protein glycosylation and lipid particle generation are indeed linked, then sec53-6ts cells might be defective for both. After 1 h at 37°C, sec53-6ts cells stained with Nile red exhibited a clear reduction in lipid droplets of 0.54 per cell (standard deviation [SD], 1.2; n = 100) compared to the wild-type control (RSY255) (2.5 droplets per cell [SD, 1.6; n = 100]). Even under permissive growth conditions, sec53-6ts cells had a reduction in lipid droplets comparable to the cax4Δ mutant. To test whether a general block in secretion would also block lipid droplet biogenesis, lipid droplets were counted after Nile red staining in sec18ts cells (JRY4130). After 1 h at 37°C, sec18ts cells are defective in multiple secretory transport events (23), but under these conditions there were no detectable defects in lipid droplet number or intensity of Nile red staining (2.7 lipid droplets per cell [SD, 1.3] compared to the wild-type, RSY255, value above). Thus, there is a specific link between core glycosylation and lipid droplet biogenesis that is independent of general secretory transport.
The CAX4 dependence of neutral-lipid synthesis might reflect a requirement for Cax4p in the enzymatic activation of ASAT and DGAT/PDAT or for the expression of these proteins. To determine if CAX4 is required for ASAT or PDAT protein expression, we examined cax4Δ cells for the expression levels of Are1p, which is a representative ASAT protein (84), and Lro1p, a phospholipid diacylglycerol (DAG) acyltransferase (46). By immunoblot analysis, Are1p and Lro1p were expressed at comparable levels in protein extracts derived from either wild-type cells or a cax4Δ strain transformed with a plasmid containing the wild-type CAX4 gene (Fig. (Fig.4A).4A). In contrast, levels of both Are1p and Lro1p were significantly reduced in protein extracts from the cax4Δ strain, whereas levels of the internal control protein (Vti1p) remained unchanged (Fig. (Fig.4A).4A). These results suggested that in the cax4Δ mutant, inhibition of neutral-lipid synthesis was a result of a global reduction in ASAT and DGAT/PDAT protein expression. Thus, CAX4 defines a mechanism that links secretory protein glycosylation with neutral-lipid acyltransferase expression.
A trivial explanation for these results is that the various neutral-lipid acyltransferases are all glycosylated, and their stability is sensitive to even small perturbations in glycosylation. To test whether Are1p or Lro1p is glycosylated and whether their glycosylation state is affected by CAX4, protein extracts from wild-type and cax4Δ cells were treated with endo H to remove N-linked oligosaccharides. The molecular weight of Are1p was unchanged whether in cax4Δ cells or in wild-type cells treated with endo H, which suggested that Are1p is not an N-linked glycoprotein (Fig. (Fig.4B).4B). This result also suggested that CAX4 does not affect Are1p stability through glycosylation. Although the molecular weight of Lro1p was reduced by endo H treatment, indicating that Lro1p is a glycoprotein, only the glycosylated form of Lro1p was detected in cax4Δ cells (Fig. (Fig.4B).4B). Given these results, CAX4 does not appear to affect neutral-lipid acyltransferase expression through their glycosylation.
In cax4Δ cells and in cells where N-linked glycosylation has been otherwise compromised, sphingolipid composition is significantly altered (52). Specifically, in cax4Δ cells there is a considerable reduction in inositolphosphorylceramides (IPCs), which represent a major class of sphingolipids (52). To test whether the effect of CAX4 on sphingolipid composition has a bearing on neutral lipid storage, lipid droplets in lcb1-100ts (YJN63), lcb2ts (YJN64), and wild-type (W303-1A) cells were visualized using Nile red fluorescence microscopy. The lcb1-100ts and lcb2ts mutants are defective in the first commitment step for the biosynthesis of all sphingolipids (43). After cultures were incubated at 37°C for 3 h, both the number and fluorescence intensity of Nile red-stained lipid droplets markedly increased in lcb1-100ts (average number of droplets per cell, 9.5; n = 101) and lcb2ts cells (average number of droplets per cell, 7.9; n = 251) as compared to the congenic wild-type control (average number of droplets per cell, 5.2; n = 228). When cultured at 23°C, all strains had a comparable number of lipid droplets (4.4 to 4.8 per cell). These findings indicated that the inhibition of all sphingolipid biosynthesis results in a concomitant proliferation in lipid droplets. These results were, however, opposite to those observed in cax4Δ cells, in which reduced IPC levels correlated with an absence of neutral lipids and lipid droplets.
In the genomic screen, several but not all deletion mutants corresponding to vacuolar H+-ATPase subunits (e.g., vma2Δ, vma9Δ, vma21Δ, and tfp1Δ) were susceptible to sterol inhibitors. To determine whether the vacuolar H+-ATPase is required for lipid storage, we inspected these deletion mutants for defects in neutral-lipid synthesis. In particular, a significant increase in the number of lipid droplets was observed in vma9Δ cells (Table (Table4).4). VMA9 encodes subunit e of the V0 vacuolar H+-ATPase (59). Deletion of VMA9 resulted in significant increases in ASAT activity (2.5-fold) and steryl ester levels (3.4-fold) compared to wild-type cells (Fig. (Fig.2A2A and Fig. Fig.3).3). These increases in steryl esters were specific since no change in triacylglycerol synthesis or levels was detected in vma9Δ cells (Fig. (Fig.2B2B and Fig. Fig.3).3). The results suggested that other vacuolar H+-ATPase deletion mutants might also have specific effects on steryl ester storage. In vma21Δ cells, a modest increase in lipid droplet number and increases in steryl ester synthesis and levels were detected (Fig. (Fig.2A2A and and3).3). VMA21 encodes an ER-localized protein required for the assembly of the vacuolar H+-ATPase complex (25, 36). None of the other vacuolar H+-ATPase deletion mutants had significant effects on lipid droplets, as determined by Nile red staining (Table (Table4).4). These results indicated that steryl ester storage is not dependent on vacuolar H+-ATPase function per se. However, in vma9Δ cells, and to a lesser degree vma21Δ cells, steryl ester storage and triacylglycerol storage are uncoupled. The proliferation of lipid droplets is consistent with the increased steryl ester synthesis measured in these particular vacuolar H+-ATPase mutants.
In contrast to vma9Δ cells, the profusion of lipid droplets in ume6Δ and cdc50Δ cells was coupled with a striking increase in Nile red fluorescence intensity (Table (Table4).4). Cdc50p regulates and physically interacts with the Drs2p P-type ATPase aminophospholipid translocase (10, 22, 45), which generates phospholipid asymmetry in membranes (53, 58). This suggested that the lipid droplet proliferation in cdc50Δ cells might be a consequence of defects in Drs2p phospholipid translocase activity. Consistent with this possibility, drs2Δ was identified in our sterol genomic screen and others have reported sterol defects in drs2Δ cells (56). However, we observed a very minor increase in lipid droplet numbers (albeit statistically significant [P = 0.04]) in drs2Δ cells stained with Nile red (Table (Table4),4), and insignificant lipid defects were detected in biochemical assays (Fig. (Fig.22 and and33 [see below]). Because the S. cerevisiae genome contains four other potential aminophospholipid translocases that, in some cases, have functional overlap with DRS2 (27), we tested whether these P-type ATPases (DNF1 to DNF3, NEO1) affected lipid droplets. As observed by Nile red staining, no appreciable changes in the number of lipid droplets were observed in neo1-1 (ZHY628-15B), neo1-2 (ZHY628-34A), dnf1Δ dnf2Δ dnf3Δ (PFY3273A), or dnf1Δ dnf2Δ dnf3Δ drs2-31ts (ZHY410-3A) mutants, regardless of temperature (unpublished data). These results suggested that CDC50 affects lipid droplets through a mechanism that is independent of the P-type aminophospholipid translocases.
To determine how lipid droplets are affected in cdc50Δ cells, lipid esterification activities and neutral lipid levels were analyzed. To test if increased ASAT and DGAT/PDAT activities caused the increase in lipid droplets, cdc50Δ cultures were pulse-labeled with [3H]oleate to measure the rate of steryl ester and triacylglycerol synthesis. Compared to the wild-type strain, the cdc50Δ strain had a 3-fold increase in ASAT activity and a 1.9-fold increase in DGAT activity (Fig. 2A and B). However, these increases in enzyme activities manifested only a modest 1.4-fold increase in both steady-state steryl ester and triacylglycerol levels (Fig. (Fig.3).3). Thus, the proliferation of lipid droplets in cdc50Δ cells as observed with Nile red is not entirely attributable to increases in neutral-lipid levels.
In ume6Δ cells, neutral-lipid levels were also unaffected (Fig. (Fig.3)3) despite the observed increase in the number and fluorescence intensity of Nile red-stained lipid droplets (Table (Table4).4). Similar to cdc50Δ cells, in ume6Δ cells ASAT activity was markedly induced (3-fold) relative to the wild-type control, while DGAT activity was elevated only by 1.3-fold (Fig. (Fig.2).2). Despite the induction of ASAT activity in ume6Δ cells, no meaningful changes in steady-state steryl ester levels were detected and triacylglycerol levels were normal. Based on these results, the lipid droplet defects observed by fluorescence microscopy in ume6Δ cells are not attributable to changes in neutral-lipid levels. Based on these findings, the biogenesis of lipid storage particles is affected by two distinct classes of mutants: one group that affects neutral-lipid synthesis (e.g., cax4Δ and vma9Δ), and another that is independent of lipid synthesis (e.g., cdc50Δ and ume6Δ).
In addition to the enzymes that synthesize neutral lipids, lipid storage also involves lipases that hydrolyze and release lipids from storage and structural factors for storage particle assembly (42). Because steady-state levels of neutral lipids were not grossly affected in cdc50Δ and ume6Δ cells, we examined lipid storage particles in these mutants for structural defects by using electron microscopy (Fig. (Fig.5).5). Consistent with Nile red staining, the number of lipid droplets in cdc50Δ cells was greater than wild type when viewed by electron microscopy. In wild-type cells, we observed 0.9 lipid droplet per cell section (n = 186), whereas in cdc50Δ cells there were 1.8 droplets per cell section (n = 84). In ume6Δ cells, the number of lipid droplets observed by electron microscopy was also greater than wild type (2.1 droplets per cell section; n = 40). Because optical sections of Nile red-stained cells represent a greater thickness through the cell than thin sections for electron microscopy, the average number of lipid droplets counted on electron micrographs was fewer than those observed by fluorescence microscopy. When observed by electron microscopy, the lipid droplet cortex in wild-type cells was surrounded by a discrete darkly stained “shell” (Fig. (Fig.5).5). In ume6Δ cells, the lipid droplet shell was exaggerated and clearly thicker than in wild-type cells (Fig. (Fig.5).5). In cdc50Δ cells, however, lipid droplets appeared less well defined and had no distinct border (Fig. (Fig.5).5). These results not only confirmed previous results showing an increase in Nile red-stained lipid droplets in cdc50Δ and ume6Δ cells, but indicated lipid storage particle assembly was defective in both mutants. These findings also indicated that CDC50 and UME6 have dramatically different effects on lipid droplet ultrastructure.
One of the deletion mutants identified (arv1Δ) has an established defect in ergosterol localization (6, 72). To determine if any of other deletion mutations we identified disrupt the normal ergosterol distribution, cells were fixed and treated with filipin complex to visualize unesterified sterol lipids. Filipin is a specific fluorescent probe for unesterified sterol localization in both mammalian cells and yeast (6, 62). In wild-type yeast, filipin fluorescence is observed at the plasma membrane (6) and this pattern of localization is consistent with previous studies that showed that ergosterol, the most abundant yeast sterol, is concentrated in the plasma membrane (87). In addition to the plasma membrane staining, 15.7% of wild-type cells exhibited small filipin-fluorescent cytoplasmic spots. In 8.5% of wild-type cells, membrane strands were also observed (Table (Table55 and Fig. Fig.6).6). In terms of morphology and localization, the membrane strands are consistent with peripheral ER. This could not be confirmed because of technical limitations of using filipin, which prevented costaining with ER markers. Nonetheless, these results affirmed that the plasma membrane is the primary repository of unesterified sterols in S. cerevisiae, but additional filipin-stained structures were also evident.
As listed in Table Table5,5, the normal pattern of filipin fluorescence was defective in several mutants other than the arv1Δ strain. Cytoplasmic spots were infrequently observed in wild-type cells, but a significant number were observed in erg3Δ, ptc1Δ, and plc1Δ cells (Table (Table55 and Fig. Fig.6).6). ERG3 encodes a sterol biosynthetic enzyme (2), PTC1 encodes a protein phosphatase type 2C (35, 57), and PLC1 encodes phospholipase C (20). Compared to wild-type cells, 3.2-fold more erg3Δ cells contained internal filipin-fluorescent spots, 2.6-fold more ptc1Δ cells and 2.2-fold more plc1Δ cells contained spots. In contrast, 3.7-fold fewer filipin/ergosterol spots and 4.4-fold more filipin-fluorescent membrane strands were observed in arv1Δ cells, compared to wild-type cells. Eleven deletion mutants accumulated filipin-fluorescent membrane strands, whereas strands were all but absent in erg3Δ cells (Table (Table55 and Fig. Fig.6).6). These results suggested that different deletion mutations had distinct effects on the intracellular distribution of sterols.
To determine whether changes in unesterified ergosterol levels were associated with abnormal sterol distributions, sterols were extracted from those deletion mutants with pronounced sterol/filipin localization defects. In ptc1Δ and plc1Δ cells, ergosterol levels were somewhat elevated (150 and 154%, respectively) compared to wild-type cells. These values were similar in magnitude to the 176% increase in arv1Δ cells, which is in agreement with previous studies (Fig. (Fig.7)7) (72). Thus, in ptc1Δ, plc1Δ, and arv1Δ cells the internal accumulation of sterols observed by filipin fluorescence microscopy was associated with modest increases in unesterified ergosterol within cells. As predicted, the erg3Δ control that does not synthesize ergosterol had no detectable ergosterol. These findings suggested that like ARV1, PTC1 and PLC1 play a role in sterol trafficking within yeast cells.
We also investigated whether any of the lipid storage mutations affected unesterified ergosterol levels. Compared to the wild-type control, the level of unesterified ergosterol in cdc50Δ cells was unchanged and the ergosterol levels in cax4Δ and ume6Δ strains were only a little higher (Fig. (Fig.7).7). These findings indicated that in the mutants with lipid droplet mutations, increased concentrations of unesterified ergosterol were not necessarily coupled with defects in sterol storage.
In mammalian cells, the late endosome represents a sorting compartment for internalized cholesterol (54). To determine whether the cytoplasmic filipin-fluorescent spots observed in yeast correspond to an endosomal compartment, cells were incubated with the endosome-specific dye FM4-64 (78) and then fixed and stained with filipin. FM4-64 is a fluorescent lipophilic dye that is internalized from the yeast plasma membrane, through endosomal compartments, to the vacuole (78). Colocalization of filipin and FM4-64 fluorescence was detected 25 min after FM4-64 internalization but not at earlier times of endocytosis (Fig. (Fig.8).8). This finding suggested that in wild-type cells, filipin stains both ergosterol in the plasma membrane and, in a minority of cells, sterols in late endosomes.
To determine whether the excess filipin-stained spots observed in plc1Δ and ptc1Δ cells corresponded to endosomes, cells were incubated with FM4-64 prior to fixation and then filipin stained. After 25 min (and not before), the colocalization of FM4-64- and filipin-stained spots was significantly greater in plc1Δ and ptc1Δ cells than in wild-type cells (Fig. (Fig.8).8). In wild-type cells (n = 239), filipin fluorescence was detected in 8% of FM4-64-stained late endosomes. In contrast, 41% of late endosomes costained with filipin in plc1Δ cells (n = 310) and 27% costained with filipin in ptc1Δ cells (n = 255). These results suggested that in the absence of PLC1 or PTC1 function, unesterified sterols accumulated in late endosomes, though not exclusively.
In a genome-wide screen, we identified 56 mutants from the yeast nonessential deletion collection that were susceptible to drug-induced perturbations in sterol homeostasis. Sterol homeostasis is maintained through the interplay of several processes: sterol transport between membranes, the regulation of sterol biosynthesis, and the storage of sterol esters in lipid droplets/lipid storage particles (Fig. (Fig.9).9). Some of the mutants we identified affected sterol homeostasis as a result of defects in lipid droplet generation (Fig. (Fig.9).9). Direct examination of specific mutants defined at least two distinct events in lipid storage particle biogenesis, namely neutral lipid synthesis and lipid droplet organelle assembly (Fig. (Fig.10).10). Further biochemical analysis of neutral-lipid synthesis in some of these mutants revealed an unanticipated link between neutral-lipid synthesis and secretory protein glycosylation. Yet another group of mutants we identified had mutations that affected sterol homeostasis through defects in the membrane localization of unesterified sterols (Fig. (Fig.9).9). These results implicated phospholipase C (Plc1p) and protein phosphatase type 2C (Ptc1p) in the intracellular trafficking of unesterifed sterols. These findings affirmed that multiple independent pathways contribute to the maintenance of cellular sterol-lipid homeostasis.
Previous genomic studies have analyzed the pharmacological effects of sterol-targeting drugs on yeast deletion strains (21, 34, 49), but the causal basis for the drug sensitivity was not explored. The nonessential deletion collection was also screened for mutants that cannot grow in anaerobic cultures in order to identify sterol uptake mutants (56). Oxygen is essential for sterol synthesis, and under anaerobic conditions yeast must import sterols from the medium to survive. Under these anaerobic conditions, however, sterol-esterification-defective mutants grow normally, which explains why lipid storage mutants were not represented in the list of 37 anaerobically sensitive mutants (56). In fact, only 2 of these 37 deletion mutants (tkl1Δ and rlr1Δ) were also identified by our approach. As such, our study complements previous genome-wide screens and identifies novel sterol-associated genes.
As a confirmation of the efficacy of our approach, we identified several deletion mutants that correspond to previously identified sterol-related genes. Deletion mutations that affect isoprenoid and sterol biosynthesis were identified, including bts1Δ, ram1Δ, erg3Δ, and erg6Δ (15), as well as arv1Δ, which affects the normal distribution of unesterified sterols (72). Many of the deletion mutations corresponded to general transcription factors that could affect the expression of genes required for sterol biosynthesis and homeostasis. However, the deletion of UME6, a transcriptional regulator in mitotic and meiotic cells (18, 69, 71, 77), had pronounced and specific effects on sterol-lipid storage. Thus, our unbiased analysis of yeast nonessential gene deletions identified both predicted targets as well as novel mutations not previously linked to sterol homeostasis.
In our mutant identification, there were some genes with direct and inferred connections to sterol homeostasis that were not detected. Because the screen was performed under aerobic conditions, most anaerobic mutations that affect sterol uptake were not identified (56), and of course essential genes or those with redundant/overlapping functions would also not be detected. We note that deletion mutations representing lipid droplet-localized proteins were not detected. From genome-wide protein localization studies (4, 27), we compiled a list of 29 potential lipid droplet proteins. Of the 29 corresponding deletion mutants, only 2 mutants had lipid droplet defects as observed by Nile red fluorescence microscopy. In tgl3Δ cells, an increase in Nile red/lipid droplet fluorescence was detected whereas ldb16/ycl005wΔ cells had decreased numbers of lipid droplets (unpublished results); TGL3 encodes a TAG lipase (4) and LDB16 has a potential role in protein glycosylation (11). These results suggested that the vast majority of lipid droplet proteins are either functionally redundant or not required for lipid droplet formation/maintenance. Since proteins on lipid droplets have such limited effects, the implication is that lipid droplet biogenesis is mainly regulated by nonresident proteins.
From the studies of many different cell types, a general model for lipid droplet biogenesis has been established (14, reviewed in reference 42). This model posits that neutral lipid synthesis occurs in specific ER microdomains wherein lipid droplets coalesce between the bilayer leaflets. Lipid storage particle maturation results after the neutral lipid core is sheathed with a phospholipid monolayer and buds from the ER surface into the cytoplasm. In Fig. Fig.10,10, our findings are integrated into this model to depict the distinct events in yeast lipid droplet biogenesis revealed by the defects in cax4Δ, cdc50Δ, ume6Δ, vma9Δ, and vma21Δ cells.
The first requisite step in lipid droplet biogenesis is the synthesis of its bulk neutral lipid components by the acyltransferases ASAT (Are1p and Are2p) and DGAT/PDAT (Dga1p and Lro1p). Remarkably, the expression of these enzymes was dependent on CAX4, which otherwise has a secondary role in secretory protein N glycosylation (19, 76). Lipid droplet biogenesis was also dependent on SEC53, which also affects N-linked glycosylation (30). However, direct glycosylation did not appear to play a role in regulating the expression or turnover of the neutral lipid acyltransferases. Alternatively, acyltransferase expression might be indirectly affected by signaling pathways that respond to unfolded and/or unglycosylated secretory proteins. However, none of the glycoprotein folding and/or quality control mutants we tested (i.e., ire1Δ, HAC1-238 [S238A] , and cne1Δ) had significant lipid droplet defects (unpublished results). Another less-understood consequence of N-glycosylation defects is a concomitant change in sphingolipid composition (52). Whereas in cax4Δ cells, the reduction in sphingolipid (IPC) levels was associated with fewer lipid droplets, we found that the number of lipid droplets dramatically increased in mutants that block sphingolipid biosynthesis. These findings suggest a potential link between sphingolipids, N glycosylation, and neutral-lipid storage, but the mechanism is anything but straightforward.
Other mutants were defective in another aspect of yeast lipid droplet biogenesis, the structural assembly of the lipid-storage particle (Fig. (Fig.10).10). Both CDC50 and UME6 deletion mutations had striking effects on lipid droplet structure. In addition to a marked increase in the number of lipid droplets and a modest increase in neutral lipids, the lipid storage particles in cdc50Δ cells exhibited a distinctive morphological defect. Lipid droplets in cdc50Δ cells lacked the discrete electron-dense cortex, suggesting that CDC50 affects the addition of protein(s) onto lipid droplets. In contrast, UME6 had the opposite effect on the structural assembly of lipid droplet organelles. In ume6Δ cells, the intensity of Nile red fluorescence was more intense and the electron-dense shell surrounding the lipid storage particles was more prominent than in wild-type cells. Since Ume6p is a Cys zinc binuclear transcription factor that coordinates mitotic and meiotic gene expression (69), these findings suggest a role for Ume6p in regulating the mitotic and meiotic proliferation of lipid storage particles. UME6 might affect lipid droplet maturation during meiosis and sporulation, when large increases in neutral-lipid content occur (28). Despite their different effects, both UME6 and CDC50 define new processes in lipid storage particle assembly.
In wild-type yeast, sterol lipids are concentrated in the plasma membrane (87), but some filipin/sterol fluorescence was observed in internal membranes. Based on the overlap of filipin/FM4-64 fluorescence, some of the internalized sterols correspond to the late endosome. Many deletion mutants affected the normal pattern of filipin/sterol staining, but plc1Δ and ptc1Δ cells exhibited the most striking defects. In these deletion mutants, a significant increase in filipin fluorescence was observed, most of which corresponded to endosomes. Since unesterified and esterified sterol levels in plc1Δ and ptc1Δ cells were only modestly higher than wild type, increases in internal filipin fluorescence were not caused by larger amounts of sterols but rather by sterol redistribution. A simple explanation for this redistribution is that endosomal sorting of sterols is defective in PLC1 and PTC1 mutants. Of the many endosomal mutants represented in the deletion collection, however, none were identified in our screen, suggesting that the role of PLC1 and PTC1 in sterol sorting is independent of established endosomal trafficking pathways.
Tentative connections have been reported that link PLC1 and PTC1 to endosomal function. PTC1 encodes a PP2C phosphatase that is best described as a negative regulator of the HOG mitogen-activated protein kinase pathway, which responds to osmotic stress by increasing cellular glycerol concentrations (85). Independent of this function, however, ptc1Δ genetically interacts with conditional alleles of the clathrin heavy chain gene (9), which affect Golgi and endocytic trafficking (51). Ptc1p purportedly binds the ASAT Are2p (26), although no lipid droplet defects were detected in our analysis of ptc1Δ cells. Plc1p, the phospholipase C homologue, hydrolyzes phosphatidylinositol bisphosphate to produce DAG (20), which in turn stimulates vacuolar membrane dynamics (29). In plc1Δ cells, fragmented vacuoles accumulate, suggesting a defect in vacuolar/endosomal trafficking or vacuole fusion (29). Both Ptc1p and Plc1p also share a link to osmoregulation (32), and both PTC1 and PLC1 are linked to calcium signaling (29, 63, 73). Regardless of whether sterol trafficking involves these or a novel function, Ptc1p and Plc1p appear to play an important role in how sterols are distributed within cells.
Another potential regulator of sterols that was identified is calcineurin. In the absence of the calcineurin regulatory subunit, encoded by CNB1, cells were lovastatin sensitive but nystatin resistant. However, no significant sterol biosynthesis, sterol transport, or sterol storage defects were detected in our analysis of cnb1Δ cells. This finding is noteworthy only because several of the other deletion mutations identified (i.e., bts1Δ, bck1Δ, cax4Δ, drs2Δ, ptc1Δ, van1Δ, vma2Δ, and vma21Δ) are lethal in combination with cnb1Δ (44, 45, 49, 63, 75). In both S. cerevisiae and Candida species, calcineurin has been previously implicated in promoting resistance to sterol biosynthetic inhibitors through an adaptive mechanism (12, 13, 17, 24, 48). Perhaps calcineurin plays a similar role in adaptation to the defects in sterol homeostasis caused by some nonessential deletion mutants.
We note that many of the sterol homeostasis genes that were identified by this yeast functional/chemical genomics approach are conserved in mammals. In particular, yeast genes that affect lipid droplet biogenesis might play a conserved role in humans. For instance, the link between neutral-lipid synthesis and secretory protein glycosylation might be applicable to human adipocytes, as in yeast. Thus, the study of sterol homeostasis in yeast might not only be pertinent to human cholesterol regulation, but also valuable for providing novel gene targets for treating obesity.
We give special thanks to Trisha Davis, Wanjin Hong, Joe Nickels, Randy Schekman, Janet Shaw, and Kazuma Tanaka for strains, antibodies, and plasmid constructs. We gratefully acknowledge Michel Leroux, Keith Kozminski, and Nancy Hawkins for suggestions on the manuscript and Aaron Chiam for technical assistance on the analysis of lipid droplet resident proteins.
C.T.B. was supported by grants from by a Natural Science and Engineering Research Council (NSERC) of Canada, an SFU President's Research Grant, and by joint contributions for microscopy equipment from the Canadian Foundation for Innovation (CFI) and the British Columbia Knowledge and Development Fund (BCKDF). H.Y. was supported by the Ministry of Education, National Medical and Biomedical Research Councils of Singapore.
This study was initiated at the University of California, Berkeley.
Published ahead of print on 21 December 2007.