Sterol homeostasis in eukaryotic cells relies on the reciprocal interconversion of free sterols and steryl esters. Here we report the identification of a novel reversible sterol modification in yeast, the sterol acetylation/deacetylation cycle. Sterol acetylation requires the acetyltransferase ATF2, whereas deacetylation requires SAY1, a membrane‐anchored deacetylase with a putative active site in the ER lumen. Lack of SAY1 results in the secretion of acetylated sterols into the culture medium, indicating that the substrate specificity of SAY1 determines whether acetylated sterols are secreted from the cells or whether they are deacetylated and retained. Consistent with this proposition, we find that acetylation and export of the steroid hormone precursor pregnenolone depends on its acetylation by ATF2, but is independent of SAY1‐mediated deacetylation. Cells lacking Say1 or Atf2 are sensitive against the plant‐derived allylbenzene eugenol and both Say1 and Atf2 affect pregnenolone toxicity, indicating that lipid acetylation acts as a detoxification pathway. The fact that homologues of SAY1 are present in the mammalian genome and functionally substitute for SAY1 in yeast indicates that part of this pathway has been evolutionarily conserved.
Sterols are essential lipids of most eukaryotic cells and serve both important structural and signaling functions. Sterols are synthesized by enzymes located in the ER membrane and are enriched in the plasma membrane where they increase the permeability barrier of the membrane and thus are important to maintain the membrane potential (Haines, 2001). Steryl esters, on the other hand, serve to store metabolic energy in form of fatty acids and are deposited in intracellular lipid droplets. The synthesis and hydrolysis of steryl esters is important to maintain sterol homeostasis as the steryl ester pool conceptually serves to buffer both excess and lack of free sterols (Chang et al, 2006).
Sterols are generally regarded as being stable and long‐lived lipids. In metazoans, they are converted to steroids, ecdysone, vitamin D, oxysterols, and bile acids, and serve as lipid anchor for signaling proteins of the hedgehog family (Russell, 2000, 2003; Chang et al, 2006). However, sterol‐like molecules can also have adverse effects and act cytotoxic, interfere with endocrine signaling through nuclear hormone receptors, and even elicit epigenetic transgenerational phenotypes (Anway et al, 2005).
Yeast contains two ER‐localized acyl‐CoA:sterol acyltransferases, Are1 and Are2, that catalyze the synthesis of steryl esters. Mutants lacking steryl esters, however, are viable, indicating that steryl ester synthesis is not essential under standard growth conditions (Yang et al, 1996; Yu et al, 1996). The hydrolysis of steryl esters, on the other hand, is catalyzed by a family of three membrane‐anchored lipases, Yeh1, Yeh2, and Tgl1. Triple mutants lacking steryl ester hydrolase activity are again viable under standard growth conditions, indicating that similar to their synthesis, hydrolysis of steryl ester is dispensable for growth under standard conditions (Köffel et al, 2005).
In the course of a systematic search for steryl ester hydrolase activities, we uncovered an unidentified [14C]cholesterol‐labeled derivative in yeast cells lacking the putative esterase/lipase YGR263c, hereafter referred to as SAY1 (steryl deacetylase). This cholesterol derivative was identified as cholesteryl acetate, indicating that SAY1 is required for the deacetylation of acetylated sterols. Formation of the cholesteryl acetate requires ATF2, an acetyltransferase. ATF2 and SAY1 thus form two components that control a novel sterol acetylation/deacetylation cycle. Remarkably, lack of SAY1 results in the export of acetylated cholesterol into the culture media. This lipid export requires a functional secretory pathway, indicating that the transport of sterol acetate is vesicle mediated. These observations suggest that the substrate specificity of Say1 controls which acetylated sterols are deacetylated and hence retained, and which acetylated sterols are exported and hence potentially detoxified. In agreement with this proposition, we find that the steroid precursor pregnenolone is rapidly acetylated in a reaction that depends on Atf2, but that pregnenolone acetate is not subject to deacetylation by Say1 and hence is exported into the culture media.
Accumulation of cholesteryl acetate in a yeast mutant lacking SAY1
In a screen to identify enzymes that catalyze steryl ester hydrolysis, heme‐deficient strains bearing deletions of candidate esterase/lipase genes were labeled with [14C]cholesterol and mobilization of steryl esters was monitored (Köffel et al, 2005). Heme deficiency is required for cells to take up exogenous sterols under aerobic conditions (Lewis et al, 1985). The relative content of free cholesterol and steryl esters was then examined by thin‐layer chromatography (TLC). This analysis revealed the presence of a novel [14C]cholesterol‐labeled band in cells lacking YGR263c/SAY1 (Figure 1A). This previously unidentified sterol derivative comprised approximately 25% of the total [14C]cholesterol‐labeled lipids present in the say1Δ mutant, but was hardly detectable in wild‐type cells, thus indicating that SAY1 is required to prevent its accumulation in wild‐type cells. Mild base treatment of this sterol derivative showed that it was susceptible to hydrolysis of an ester bond, indicating that it contains [14C]cholesterol esterified to an unidentified substituent. This substituent confers lower relative mobility and thus is less hydrophobic than long‐chain fatty acids in steryl esters (Figure 1B). This novel cholesterol derivative has been identified as cholesterol acetate by comigration analysis with commercial cholesterol acetate (Figure 1C), and confirmed independently by mass spectrometry (Supplementary Figure S1). These results thus show that say1Δ‐mutant cells accumulate cholesterol acetate.
Since say1Δ‐mutant cells accumulate cholesterol acetate, we tested if Say1 could function as a deacetylase. Cells overexpressing Say1 from a galactose‐inducible promoter exhibited a ∼2.5‐fold higher in vitro activity to hydrolyze p‐nitrophenyl acetate compared with wild‐type cells (Figure 1D). Furthermore, this increased activity was specific for short p‐nitrophenyl esters, as activity against p‐nitrophenyl palmitate of say1Δ‐mutant cells was comparable to that of wild‐type cells. These observations are consistent with Say1 being required for deacetylation of cholesterol acetate in vivo.
Say1 has homology to mammalian deacetylases
Say1 contains a putative esterase/lipase domain from amino acid 156 to 375 that is predicted to form a canonical α/β‐hydrolase fold (Derewenda and Derewenda, 1991). The protein contains a possible N‐terminal transmembrane domain and three potential N‐linked glycosylation sites. PSI‐BLAST analysis revealed high homology of Say1 with two predicted ORFs from Caenorhabditis elegans (T09B9.1 and F27C8.6) and with human arylacetamide deacetylase (AADAC) and its homologues AADACL1, AADACL2, and AADACL4 (Probst et al, 1994). Multiple alignment of these sequences with Say1 revealed that the lipase consensus motif GXSXG is conserved in Say1 and its metazoan homologue as GDSAG, with a central nucleophilic Ser at position 250 and conservation of the HGGG oxyanion hole motif at position 176 of Say1 (Figure 2A). These two sequence motifs are signatures for the prokaryotic lipolytic enzymes classified as belonging to family IV, which also shows high similarity to mammalian hormone‐sensitive lipases (Hemila et al, 1994; Arpigny and Jaeger, 1999). Mutations of the putative active site Ser at position 250 of Say1 to Ala resulted in the accumulation of acetylated sterols, indicating that this Ser is required for Say1 function (Figure 2B). Loss of Say1p activity due to the exchange of Ser 250 for Ala does not affect protein abundance as indicated by western blot analysis of an myc‐epitope tagged version of Say1p, consistent with the proposition that Ser 250 forms the active site residues of Say1p (Figure 2C). Assignment of Ser 250 as the active‐site nucleophile is consistent with the assignment of active‐site residues in the mouse AADACL1 that is based on homology modeling (Nomura et al, 2006).
Say1 is an integral membrane protein with the putative active site in the ER lumen
To determine whether Say1 is membrane‐anchored and to examine its membrane topology and subcellular localization, N‐terminally GFP‐tagged Say1 under the control of the GAL1/10 promoter and C‐terminally myc‐tagged versions of Say1 were generated. Both fusion proteins were functional, as cells expressing these proteins displayed no detectable accumulation of cholesteryl acetate when labeled with [14C]cholesterol (Figure 3A). Differential fractionation revealed that GFP‐Say1 and Say1‐myc were enriched in the 13 k membrane fraction, indicating that Say1 is membrane associated (Figure 3B). Say1 was released from membranes only upon treatment with 1% SDS or 1% Triton X‐100, indicating that Say1 is an integral membrane protein (Figure 3C). Accessibility of the N‐ and C‐terminal tags to degradation by proteinase K revealed that the N‐terminal GFP was readily hydrolyzed. At the same time, the ER lumenal Kar2 was protease protected, indicating intactness of the membrane seal. The myc‐tagged C‐terminus of Say1, on the other hand, resisted protease treatment, but became protease sensitive in the presence of Triton X‐100, consistent with a lumenal localization of the C‐terminus of Say1 (Figure 3D). Say1 contains three asparagines that could potentially be subjected to N‐linked glycosylation. To determine whether Say1 is N‐glycosylated, we treated membranes with endoglycosidase H. This analysis revealed that none of these consensus residues for N‐linked glycosylation is actually glycosylated in Say1, which does neither confirm nor invalidate the topology model of Say1, as glycosylation of Asn–Xaa–Ser/Thr sequons is not obligatory (Figure 3E; Nilsson and von Heijne, 2000).
Analysis of the subcellular localization of Say1 by fluorescence microscopy showed colocalization of GFP‐Say1 with the ER marker Kar2‐mRFP‐HDEL (Gao et al, 2005). Immunofluorescence microscopy revealed colocalization of Say1‐myc with Sec61, a subunit of the ER translocon, indicating that Say1 is an ER‐localized protein (Figure 3F). ER localization of Say1 is consistent with its fractionation properties on an Accudenz density gradient on which the protein cofractionates with the ER lumenal chaperone Kar2 but not with the plasma membrane localized H+‐ATPase, Pma1 (Figure 3G). Taken together, these observations indicate that Say1 is a type II integral membrane protein with its putative active site Ser exposed to the ER lumen (Figure 3H). Deacetylation of cholesterol acetate thus occurs in the ER lumen.
ATF2 is required for the formation of acetylated sterols
As acetylated cholesterol is detectable only in cells lacking SAY1, we looked at the acetyltransferase(s) required for its formation. Yeast contains two genes encoding alcohol O‐acetyltransferases, ATF1 and ATF2 (Mason and Dufour, 2000). Atf1 and Atf2 control the production of a broad range of volatile esters that are important for the fruity flavor of fermented beverages (Verstrepen et al, 2003). Atf2 has previously been shown to acetylate the steroid precursor, pregnenolone, and it has been proposed that this acetylation is important for detoxification and excretion of steroids (Cauet et al, 1999). To determine whether ATF1 and/or ATF2 are required for the formation of cholesteryl acetate in vivo, heme‐deficient atf1Δ say1Δ and atf2Δ say1Δ double mutant cells were generated. These were labeled with [14C]cholesterol and lipids were analyzed for the presence of cholesteryl acetate. While cholesteryl acetate was present in cells lacking ATF1, no acetylated cholesterol was detectable in cells lacking ATF2, indicating that ATF2 is required for the formation of acetylated sterols (Figure 4A).
Atf2 contains two weakly predicted transmembrane domains located between amino acids 313–339 and 478–493. To determine the cellular site of sterol acetylation, the localization of Atf2 was examined in more detail, using N‐terminal GFP fused to Atf2 under control of a GAL1/10 promoter. Expression of the GFP‐tagged Atf2 resulted in a functional protein as indicated by the presence of acetylated cholesterol upon induction of GFP‐Atf2 expression (Figure 4B). Furthermore, overexpression of GFP‐Atf2 resulted in the appearance of acetylated sterols even when Say1 was functional, indicating that the activities of Atf2 and Say1 in wild‐type cells must be balanced to prevent the accumulation of acetylated sterols (Figure 4B).
Differential fractionation indicated that GFP‐Atf2 is membrane associated as it was enriched in both the 13 and 30 k membrane pellets (Figure 4C). The protein was solubilized by detergent treatment only, indicating that Atf2 is an integral membrane protein (Figure 4D). Proteinase K treatment of N‐ and C‐terminally tagged versions of Atf2 indicates that both termini of the protein are protease protected in the absence of detergent, but protease sensitive in the presence of detergent, consistent with a lumenal orientation of these termini. The appearance of N‐ and C‐terminal cleavage fragment is furthermore consistent with a protease attack within an exposed cytosolic loop (Figure 4E). Endoglycosidase H treatment revealed no altered mobility of GFP‐Atf2, indicating that the protein is not glycosylated (Figure 4F). Fluorescence microscopy is consistent with an ER localization of GFP‐Atf2 (Figure 4G). Taken together, this analysis indicates that Atf2 is required for acetylation of sterols and that Atf2 is an integral membrane protein of the ER with at least two transmembrane domains with both termini oriented toward the lumenal compartment (Figure 4H). This topology places the heptapeptide WRLICLP that is conserved between Atf1 and Atf2 and that has been proposed to be part of the active site of the two acetyltransferases into the ER lumen (Mason and Dufour, 2000), suggesting that sterol acetylation occurs in the lumen of the ER.
Acetylated sterols are exported
Given that Atf2 and Say1 catalyze the acetylation and deacetylation respectively of exogenously supplied cholesterol, we next asked what the physiological function of this acetylation might be. Since both acetylation and deacetylation of sterols occurs in the ER and say1Δ‐mutant cells accumulate sterol acetate but do not display any ER‐related phenotype (data not shown), we examined the possibility that acetylation of sterols could be a signal for export of the lipid. Therefore, we examined whether say1Δ cells secrete acetylated cholesterol into the culture media. Following labeling of say1Δ with [14C]cholesterol, lipids from the cell pellet and the growth media was examined. This analysis revealed that the acetylated cholesterol was present in the culture supernatant of say1Δ‐mutant cells, indicating that the acetylation of sterols may determine their excretion (Figure 5A). The fact that the medium is devoid of long‐chain steryl esters indicates that the presence of acetylated cholesterol is not simply due to cell lysis. Excretion of cholesterol acetate is selective as indicated by the fact that more than 50% of the total cholesterol acetate, but only 3% of steryl esters is found in the culture supernatant (Figure 5B). Export of acetylated cholesterol is stimulated by the presence of ergosterol in the media (Supplementary Figure S2) and requires ATP, but is independent of ongoing protein synthesis (data not shown), indicating that it is an active process rather than occurring by passive diffusion of acetylated cholesterol through the cell membrane.
Export of the acetylated form of the steroid precursor pregnenolone has previously been suggested to depend on two plasma membrane‐localized ABC transporters, Pdr5p and Snq2p, since pregnenolone toxicity is increased in pdr5Δ and snq2Δ‐mutant cells (Kralli et al, 1995; Kolaczkowski et al, 1996; Mahe et al, 1996; Cauet et al, 1999). Direct examination of a possible role of the two ABC transporters, in export of sterol acetate revealed that a pdr5Δ snq2Δ say1Δ triple mutant, exported cholesterol acetate into the culture supernatant similar to a say1Δ single mutant, indicating that these two ABC transporters are not required for the export of cholesterol acetate (Figure 5C).
Since cholesterol acetate is formed in the ER lumen, we next tested whether export of this lipid requires ongoing vesicular transport from the ER. Therefore, vesicle formation was conditionally blocked using a temperature‐sensitive allele of SEC12, sec12ts. SEC12 encodes the guanine nucleotide exchange factor for Sar1p and is required for the assembly of the COPII coat on the ER membrane (Barlowe and Schekman, 1993). sec12ts say1Δ‐mutant cells secreted cholesterol acetate into the culture supernatant at the permissive temperature. When shifted to the non‐permissive temperature of 37°C, however, cholesterol acetate was absent from the supernatant, indicating that export of cholesterol acetate depends on ongoing membrane transport out of the ER (Figure 5D). Lack of export of cholesterol acetate is not simply due to lack of its synthesis, as cholesterol acetate accumulates intracellularly under conditions that block secretion. Taken together, these results indicate that cholesterol acetylation/deacetylation cycle controls the export of acetylated sterols through the secretory pathway.
Formation of cholesterol acetate was not impaired in cells lacking the two acyl‐CoA:sterol acyltransferases, Are1 and Are2 (Yang et al, 1996; Yu et al, 1996), indicating that formation of long‐chain steryl esters is not a prerequisite for acetylation (Supplementary Figure S3A). are1Δ are2Δ say1Δ triple mutant cells, however, exported cholesterol acetate less efficiently than say1Δ‐mutant cells (Supplementary Figure S3B). The fact that deletion of either Say1 or Atf2 in an are1Δ are2Δ double mutant background results in viable triple mutants indicates that acetylation and export of sterols does not become essential in cells that lack the capacity to buffer sterols by forming long‐chain steryl esters (Supplementary Figure S4).
Export of pregnenolone requires Atf2 but is independent of Say1
Given that acetylated cholesterol but not non‐acetylated cholesterol is secreted from the cells, substrate acetylation and/or deacetylation is likely to control sterol export. To examine whether substrate recognition could play a role in discriminating between sterols that need to be secreted from those that are retained, we tested whether the steroid precursor pregnenolone or aberrant endogenously synthesized sterols are substrates of the acetylation/deacetylation cycle. Therefore, we first examined pregnenolone acetylation and export in wild‐type, say1Δ‐ and atf2Δ‐mutant cells. Cells were incubated with radiolabeled pregnenolone for 2 h, lipids were extracted from the cell pellet and culture supernatant, and examined by TLC. This analysis revealed that pregnenolone was rapidly acetylated in a reaction that depends on Atf2 and exported from the cells (Figure 6A). These observations are consistent with the report that Aft2 is an acetyl‐CoA:pregnenolone acetyltransferase and that pregnenolone acetate is exported from cells (Cauet et al, 1999). The fact that wild‐type and say1Δ‐mutant cells had comparable levels of pregnenolone acetate indicates that pregnenolone acetate is not subject to deacetylation by Say1. This could be either because pregnenolone acetate has no access to Say1 or because it is not a substrate for Say1. In the absence of Atf2, pregnenolone is acylated with long‐chain acyl‐CoA in a reaction that depends on the presence of Are1 and Are2, and accumulates intracellularly (Figure 6A). Also, export of pregnenolone acetate is mechanistically distinct from that of cholesterol acetate, as it does not depend on ongoing vesicular transport, as shown by its excretion from sec12ts mutants under both permissive and non‐permissive conditions, nor is it blocked by the deletion of eight ABC transporters, including Pdr5 and Snq2 (Nakamura et al, 2001; Figure 6B). These results are thus consistent with a proofreading function of Say1 in which substrate access or recognition by Say1 determines whether the acetylated sterol is deacetylated and hence retained, or whether it is not deacetylated and thereby exported from the cell.
Further analysis of the in vivo substrate specificity of the sterol acetylation and export pathway indicated that sitosterol, progesterone, 7‐ketocholesterol, and lanosterol are not subject to acetylation, but that 25‐hydroxycholesterol, which inhibits cholesterol biosynthesis in mammalian cells and does not support growth of yeast (Rodriguez et al, 1983; Adams et al, 2004), behaves like pregnenolone, that is, is subject to Atf2‐dependent acetylation (Supplementary Figure S5).
Endogenously synthesized sterols are subject to the acetylation/deacetylation cycle
To examine whether the acetylation/deacetylation cycle is also operating on endogenously made sterols, we blocked intermediate steps in the biosynthetic pathway for ergosterol and analyzed the formation of acetylated sterol intermediates and their export. Among the viable mutants in the ergosterol biosynthetic pathway tested, we observed that erg4Δ mutants displayed accumulation of an aberrant sterol in a say1Δ‐mutant background, as revealed by labeling endogenously synthesized sterols with tritiated methionine, which labels the methyl group at position C‐24 in the side chain of ergosterol. ERG4 encodes the sterol C‐24 reductase that catalyzes the final step in ergosterol biosynthesis (Daum et al, 1998). erg4Δ mutants thus accumulate ergosta‐5,7,22,24(28)‐tetraenol, which is modified in a reaction that depends on ATF2, and the aberrant sterol formed migrates similar to cholesterol acetate when analyzed by TLC (Figure 7A). The acetylated sterol intermediate produced in erg4Δ‐mutant cells is exported in the absence of SAY1, indicating that Say1 deacetylates it and thus prevents its secretion (Figure 7B). Export of aberrant sterols that are produced under aerobic conditions, however, is much less efficient compared with the export of cholesterol under anaerobic conditions. Taken together, these observations indicate that the sterol acetylation/deacetylation cycle also operates on the aberrant sterols that accumulate endogenously in an erg4Δ mutant, and results in the secretion of the acetylated sterol intermediate.
Sterol deacetylation is conserved
We next wondered whether this sterol acetylation/deacetylation cycle that controls sterol export in yeast is conserved in metazoans. Say1 shows homology to the arylacetamide deacetylase AADAC and its homologues AADACL1, AADACL2, and AADACL4 in humans. We first tested whether expression of the human homologues in yeast would complement for Say1. Therefore, expression of AADAC and AADACL1 was placed under control of a GAL1/10 promoter and the proteins were expressed in a say1Δ hem1Δ‐mutant background. Cells were then labeled with [14C]cholesterol and lipids were analyzed by TLC. This analysis revealed that expression of AADAC but not of AADACL1 prevented the accumulation of cholesterol acetate, indicating that AADAC provides deacetylase activity against cholesterol acetate when expressed in yeast (Figure 8).
Say1 and Atf2 are required for growth in the presence of eugenol and affect pregnenolone toxicity
To examine whether the sterol acetylation cycle might have a function in detoxification of cells from potentially toxic steroid‐like compounds, we examined the growth of say1Δ and atf2Δ‐mutant cells in the presence of various naturally occurring hydrophobic and steroid‐like compounds. This analysis revealed a strong growth advantage of cells expressing Say1 and Atf2 on media containing eugenol (Figure 9A). Eugenol is a member of the allylbenzene class of compounds that is present in clove oil, nutmeg, cinnamon, and bay leaf, and is used as local antiseptic and anesthetic. The fact that both say1Δ and atf2Δ‐mutant cells are sensitive against eugenol indicates that the full acetylation/deacetylation cycle must be operating for growth in the presence of eugenol, which is consistent with a possible function of this cycle in lipid detoxification.
Export of the acetylated form of the steroid precursor pregnenolone has previously been suggested to depend on two plasma membrane‐localized ABC transporters, Pdr5p and Snq2p, since pregnenolone toxicity is increased in pdr5Δ‐ and snq2Δ‐mutant cells (Kralli et al, 1995; Kolaczkowski et al, 1996; Mahe et al, 1996; Cauet et al, 1999). Pregnenolone toxicity in these ABC transporter mutants, however, depends on a tryptophan auxotrophy of the genetic background, as pdr5Δ snq2Δ TRP1 double mutant cells do not show any steroid‐dependent growth phenotype (Vico et al, 2002). To examine whether Say1 and Atf2 affect the sensitivity of cells to grow in the presence of pregnenolone, we generated an isogenic set of strains that are either tryptophan prototrophic (TRP1) or auxotrophic (trp1Δ) and tested their growth in the presence of pregnenolone. Consistent with the notion that pregnenolone toxicity depends on tryptophan auxotrophy, we find that loss of ATF2 rendered cells sensitive to pregnenolone in a tryptophan auxotrophic but not in a prototrophic background (Figure 9B). Cells lacking long‐chain steryl esters (are1Δ are2Δ double mutants), on the other hand, are not more pregnenolone sensitive than a wild type, indicating that esterification with long‐chain fatty acids and intracellular storage of the resulting steryl esters does not act to buffer the toxic effect of pregnenolone. The tryptophan dependence of pregnenolone toxicity is likely linked to the fact that surface transport of the tryptophan permease, Tat2, is dependent on membrane sterols, which renders tryptophan auxotrophic strains more sensitive to sterol alterations (Umebayashi and Nakano, 2003). Overexpression of Atf2 from a GAL1 promoter in a tryptophan auxotrophic background, on the other hand, rendered cells more resistant against pregnenolone, whereas overexpression of Say1 rendered them sensitive against pregnenolone (Figure 9C). Taken together, these data show that a functional acetylation cycle affects growth in the presence of toxic lipids, and indicates that the lipid acetylation cycle might function as a detoxification pathway.
In the course of a systematic analysis of the role of candidate lipases in steryl ester hydrolysis, we here uncovered a novel cholesterol‐derived lipid, cholesterol acetate, in cells lacking SAY1, a hydrolase that converts cholesterol acetate into free cholesterol. Even though SAY1 is not essential, it is conserved in fungi and has homologues in metazoans, suggesting that its metabolic function has been conserved. Say1 is a type II integral membrane protein of the ER with the presumed catalytic site in the ER lumen. The acetyltransferase required for synthesis of acetylated sterols is encoded by ATF2, a sterol O‐acetyltransferase that is bound to the ER membrane and acetylates the steroid precursor pregnenolone (Cauet et al, 1999; Mason and Dufour, 2000). SAY1 and ATF2 are thus two components of a novel sterol acetylation/deacetylation cycle. This cycle operates on endogenously synthesized ergosterol precursors as well as on exogenously supplied steroids, and could serve to detoxify the cells of steroid‐like compounds and hydrophobic phytochemicals such as flavonoids that are present in plants and fruits, one of the natural environments of yeast, as indicated by the sensitivity of say1Δ‐ and atf2Δ‐mutant cells against eugenol (Figures 9 and 10). Examination of the expression of Say1 and Atf2 under aerobic and anaerobic conditions, however, revealed no oxygen‐dependent regulation of the steady‐state levels of the two enzymes, suggesting that this putative detoxification cycle is operating under both growth conditions (Supplementary Figure S6).
Atf1p and Atf2p have homology to another ER protein, Sli1p, which N‐acetylates myriocin and confers resistance against this sphingolipid inhibitor. In this case, however, acetylation serves to inactivate the drug rather than to excrete it, because the acetylated form of ISP‐1/myriocin fails to bind to the target enzyme (Momoi et al, 2004). In vertebrates, on the other hand, sulfonation and glucuronoidation of sterols and steroids is important for detoxification and excretion of these otherwise poorly soluble compounds (Tukey and Strassburg, 2000; Strott, 2002).
The observation that expression of the human homologue of SAY1, AADAC, in yeast rescues the sterol acetate accumulation phenotype of say1Δ‐mutant cells indicates that the human aryl acetamide deacetylase acts on cholesterol acetate and thus has overlapping substrate specificity with Say1 in vivo, even though the enzyme has been identified as an N‐deacetylase based on its in vitro activity against aryl acetamide (Probst et al, 1994). The mouse AADAC is expressed in liver, intestinal mucosa, the pancreas, and also the adrenal gland, and has been proposed to play a role in promoting the mobilization of lipids from internal stores and in the liver for assembling VLDL (Trickett et al, 2001). One of the orthologues of AADAC, AADACL1, has recently been identified as a detoxifying enzyme for organophosphorus nerve poisons, and null mutant mice are viable and phenotypically normal under standard conditions (Nomura et al, 2005).
The human genome has no identifiable homologue of the acetyltransferase Atf2. We thus examined whether pregnenolone is acetylated by human cells. Incubation of HepG2 hepatoma cells with radiolabeled pregnenolone, however, did not reveal any conversion to pregnenolone acetate or export of modified pregnenolone into the culture media, indicating that the substrate specificity of a putative lipid acetylase, if present in mammals, is different from that of yeast cells (data not shown).
In Mycoplasma capricolum, cholesterol acetate can replace cholesterol for growth (Lala et al, 1979). Our observation that cholesterol acetate is selectively secreted from yeast cells lacking Say1, however, indicates that cells have evolved mechanism that allow to distinguish cholesterol from cholesterol acetate even though both lipids can form membranes in vitro (Kwong et al, 1971).
While steroids were thought to passively diffuse through cellular membranes as postulated by the free hormone hypothesis (Mendel, 1989), there is now increasing evidence that these poorly water soluble compounds are transported by plasma carriers and then taken up by receptor‐mediated endocytosis of loaded carriers similar to uptake of cholesterol by the LDL receptor (Hammes et al, 2005). Export of steroids from certain mammalian cells growing in culture, on the other hand, is temperature and energy dependent, and a saturable process that acts on selected steroids only (Gross et al, 1970). Our observation that export of cholesterol acetate requires ongoing vesicular transport from the ER, whereas export of pregnenolone acetate is independent of secretion indicates that yeast cells have at least two distinct mechanisms to export acetylated sterols and steroids. Further genetic and biochemical approaches are now required to define the mechanism that governs export of pregnenolone acetate from yeast cells in more detail.
In mammals, oxidation of cholesterol to either 7α‐ or 27‐hydroxycholesterol, followed by further oxidation to bile acids comprises a major pathway of cholesterol catabolism (Russell, 2003). In this case, cholesterol is rendered more hydrophilic and ultimately water soluble to allow its excretion via the bile. In contrast, the acetylation cycle described here for yeast appears to follow a different strategy as acetylation renders the lipid even more hydrophobic. This opens the question how cholesterol acetate is rendered soluble, either in the ER lumen or after it has been secreted into the culture supernatant. One possibility might be that cholesterol acetate is selectively recognized and binds to a soluble protein, possibly already within the ER lumen, and is then also rendered soluble through this protein interaction after secretion. Further studies are now under way to identify such a putative cholesterol acetate‐binding protein.
Materials and methods
Yeast strains and growth conditions
Yeast strains used in this study are listed in Supplementary Table SI. Strains bearing single deletions of non‐essential genes were obtained from EUROSCARF. Strains were cultivated in YPD rich or minimal media. Media supplemented with sterols and fatty acids contained 0.05 mg/ml Tween 80 and 20 μg/ml ergosterol or cholesterol (Sigma Chemical Co., St Louis, MO). Alternatively, hem1Δ‐mutant cells were supplemented with 10 μg/ml Δ‐aminolevulinic acid (ALA). Site‐directed mutagenesis was performed as previously described (Toulmay and Schneiter, 2006). Expression clones for AADAC and AADACL1 were obtained from the Deutsches Resourcenzentrum für Genomforschung (RZPD, Berlin‐Charlottenburg, Germany).
Subcellular fractionation, detergent and salt extractions, and proteinase K treatment were performed essentially as previously described (Köffel et al, 2005).
Deglycosylation by endoglycosidase H was performed by incubating total cell extracts with 200 mU of EndoH (Roche‐Diagnostics, Rotkreuz, Switzerland) at 37°C for 16 h. Proteins were precipitated with TCA and analyzed by western blot.
Fluorescence microscopy of living cells was performed using a Zeiss Axioplan 2 (Carl Zeiss, Oberkochen, Germany) equipped with an AxioCam CCD camera and AxioVision 3.1 software. Indirect immunofluorescence was performed using mouse anti‐myc and rabbit anti‐Sec61 antibodies. Secondary antibodies used were FITC‐conjugated anti‐mouse and TRITC‐conjugated anti‐rabbit (both 1:200; Sigma) antibodies. Nuclei were visualized by DAPI (4,6‐diamidino‐2‐phenylindole) staining.
Lipid labeling and analysis
Uptake of [14C]cholesterol was performed essentially as described (Reiner et al, 2006). hem1Δ‐mutant cells were cultured in cholesterol/Tween‐containing media and labeled with 0.025 μCi/ml [14C]cholesterol (American Radiolabeled Chemicals Inc., St Louis, MO) for 16 h at 24°C. Cells were diluted to OD 0.5 in fresh medium containing cold cholesterol and grown for 4 h. Cells were disrupted with glass beads in the presence of chloroform/methanol (1:1) and lipids were extracted. Lipids were separated by TLC on silica gel 60 plates (TLC; Merck, Darmstadt, Germany) with the solvent system petroleum ether/diethylether/acetic acid (70:30:2; per volume), and radiolabeled lipids were quantified by scanning with a Berthold Tracemaster 40 Automatic TLC‐Linear Analyzer (Berthold Technologies, Bad Wildbad, Germany). TLC plates were then exposed and visualized using a phosphorimager (Bio‐Rad Laboratories, Hercules, CA). Lipids were deacylated by treatment with 0.1 M NaOH for 60 min at 30°C.
Newly synthesized ergosterol was labeled by incubating cells with 10 μCi/ml l‐[methyl‐3H]methionine (85 Ci/mmol; American Radiolabeled Chemicals Inc.) for 16 h at 24°C.
To examine sterol export into the culture media, cells were labeled with [14C]cholesterol (0.025 μCi/ml), for 16 h at 24°C. Cells were collected by centrifugation, washed twice with YPD, diluted to OD600 of ∼1, and grown for the time indicated. Lipids were extracted from the cell pellet and the culture media. Samples were dried and analyzed by TLC.
To examine pregnenolone export, cells were labeled with 1 μCi/ml [3H]pregnenolone (American Radiolabeled Chemicals Inc.) for 2 h at 24°C. Cells were collected by centrifugation and lipids were extracted from the cell pellet and the culture media and analyzed by TLC.
In vitro deacetylase assay
Microsomal membranes were incubated with 0.3 mg/ml p‐nitrophenyl acetate or p‐nitrophenyl palmitate in 100 mM potassium phosphate, pH 6.9, 50 mM Mg2Cl2 for 30 min at 30°C. The reaction was stopped by the addition of Triton X‐100 to 2%; samples were placed on ice and cleared of precipitates by centrifugation. Conversion of the acylated p‐nitrophenyl to p‐nitrophenol was determined colorimetrically by reading absorbance at 410 nm (Gupta et al, 2002).
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
We thank O Aebischer for help with mass spectrometry, D Picard and V Sundaramurthy for comments on the manuscript, and the Swiss National Science Foundation (631‐065925 and PP00A‐110450) for financial support.
- Copyright © 2007 European Molecular Biology Organization