The yeast Saccharomyces cerevisiae is a facultative anaerobic organism. Under anaerobiosis, sustained growth relies on the presence of exogenously supplied unsaturated fatty acids and ergosterol that yeast is unable to synthesize in the absence of oxygen or upon haem depletion. In the absence of exogenous supplementation with unsaturated fatty acid, a net accumulation of SFA (saturated fatty acid) is observed that induces significant modification of phospholipid profile [Ferreira, Régnacq, Alimardani, Moreau-Vauzelle and Bergès (2004) Biochem. J. 378, 899–908]. In the present paper, we focus on the role of SFH2/CSR1, a hypoxic gene related to SEC14 and its involvement in lipid metabolism upon haem depletion in the absence of oleic acid supplementation. We observed that inactivation of SFH2 results in enhanced accumulation of SFA and phospholipid metabolism alterations. It results in premature growth arrest and leads to an exacerbated sensitivity to exogenous SFA. This phenotype is suppressed in the presence of exogenous oleic acid, or by a controlled expression of FAS1, one of the two genes encoding FAS. We present several lines of evidence to suggest that Sfh2p and oleic acid regulate SFA synthase in yeast at different levels: whereas oleic acid acts on FAS2 at the transcriptional level, we show that Sfh2p inhibits fatty acid synthase activity in response to haem depletion.
- fatty acid synthase (FAS)
- neutral lipid
- oleic acid
- Saccharomyces cerevisiae
The budding yeast Saccharomyces cerevisiae is a facultative anaerobic organism, which can alternatively utilize respiration or fermentation for its energetic requirements. Physiological adaptation to aerobiosis and anaerobiosis is accompanied by the differential expression of a large number of genes in relation to oxygen availability [1–3]. Respiration deficiency resulting from oxygen limitation has several consequences, among which UFAs (unsaturated fatty acids) and ergosterol deprivation are potentially sufficient to compromise growth. Indeed, the biosynthetic pathways of these essential molecules are strictly aerobic. Transcriptional induction of hypoxic genes encoding oxygen-dependent enzymes, including ERG11 (for sterol biosynthesis) and OLE1 (for fatty acid desaturation), can be a temporary means for the cells to counteract oxygen limitation. However, long-term growth in strict anaerobiosis necessitates supplementation with ergosterol and oleic acid as a source of UFAs.
Since haem is a cofactor of several enzymes necessary for sterol and UFA synthesis, a major consequence of haem depletion in yeast is UFA and sterol biosynthesis preclusion. In addition, haem depletion also results in the activation of hypoxic genes through its effect on several transcriptional factors, including Rox1p and Hap1p [1,4]. Haem depletion can be easily achieved in a strain deleted for the HEM1 gene, which encodes δ-aminolevulinic acid synthase. In the presence of exogenous δ-aminolevulinic acid, a hem1Δ strain is able to synthesize UFA and ergosterol, and hypoxic genes are repressed; this corresponds to aerobic-like growth. In contrast, a hem1Δ mutant cultivated in complete medium lacking δ-aminolevulinic acid can mimic most of the anaerobiosis-induced effects of oxygen deprivation, including the ergosterol and oleic acid requirement for robust growth and hypoxic gene induction [5,6].
In this experimental system, UFA starvation in haem-depleted medium results in growth arrest after five generations. Interestingly, cells accumulate SFAs (saturated fatty acids) which are mainly stored not within TAGs (triacylglycerols) and SEs (steryl esters), but rather within specific phospholipid species, with a marked preference for PtdIns . In yeast, the balance of PtdCho (phosphatidylcholine) and PtdIns is controlled by Sec14p, an essential protein that is involved in protein transport from the yeast Golgi complex. More specifically, it is currently proposed that this protein regulates the interface between phospholipids and proteins that participate in the traffic machinery in the secretory pathway . Numerous data suggest that Sec14p controls PtdCho metabolism through the CDP-choline pathway. Indeed some of the mutants that relieve cells of Sec14p requirements (the so-called ‘by-pass Sec14p’ mutants) inactivate structural genes for enzymes of the CDP-choline pathway (e.g. CKI1, CPT1) for PtdCho biosynthesis. In addition, biochemical data indicate that Sec14p can inhibit Pct1p, an enzyme of the CDP-choline pathway .
Overexpression of SFH2 has been shown to suppress the thermosensitive phenotype of sec14-1 . Sfh2p belongs to a family of five yeast proteins which are structurally homologous with Sec14p, hence its name (Sec fourteen homologue 2). All, except for Sfh1p, exhibit PtdIns, but not PtdCho, transfer activity. Some experimental evidence supports a role for Sfh4p at contact sites between endoplasmic reticulum and Golgi membranes to facilitate PtdSer (phosphatidylserine) translocation. The phenotype of sfh3Δ strain, and the localization of Sfh3p in lipid particles, led to the proposal that SFH3 could facilitate sterol traffic between the endoplasmic reticulum and lipid particles. The function of the other Sfh proteins is largely unsolved. On the basis of a detailed analysis of sec14-1 suppression by SFH2, it was suggested that Sfh2p is the closest functional homologue of Sec14p . Interestingly, SFH2 is the only hypoxic gene within the Sec14p family [10,12]. In the present paper, we report on the possible role of SFH2 in lipid and fatty acid metabolism in response to haem depletion and oleic acid starvation.
Yeast strains, media, genetic techniques and culture conditions
The yeast strains used throughout the present study are listed in Table 1. Strains were grown in standard minimal medium [YNB (yeast nitrogen base)] supplemented with the appropriate amino acids and/or bases for plasmid selection. The medium used for anaerobic growth was supplemented with 1% (v/v) Tween 80 (polyethylene sorbitan mono-oleate as a source of oleic acid), 80 μg·ml−1 ergosterol and 1% (v/v) Tergitol NP-40/ethanol (1:1). In contrast with what is routinely used in the laboratory , the hem1Δ strains was supplemented with 120 μg·ml−1 δ-aminolevulinic acid, which achieved aerobic growth conditions until glucose depletion (i.e. stationary phase). In addition, unless otherwise stated, haem-induced lipid-starved conditions were obtained by inoculating 2×105 cells·ml−1 of stationary-phase cells in YPD medium. In these conditions, expression of hypoxic genes was routinely observed 6 h after the shift. Alternatively, YPD medium was supplemented with 80 μg·ml−1 ergosterol and/or 1% (v/v) Tween 80, used as the source of UFAs.
Plasmid pJU1 contains the SFH2 ORF (open reading frame) and 220 bp of upstream sequence inserted into Yep352 . The insert was isolated as a SacI fragment from pMR270 (one of the original plasmids suppressing the sec14-1 phenotype ) into Yep352; we have confirmed that this plasmid allows full suppression of sec14-1 mutation and complements SFH2 deletion. Plasmid pMR306 encoding the fusion protein MBP (maltose-binding protein)–Sfh2p was constructed as follows. First pMR265  was used to amplify the coding sequence of SFH2 using the two primers CSRBGL2 (5′-AAGATCTATTATGTCCTTTGATAGA-3′) and CSR3SAL (5′-AGTCGACTCAAACTTTTAGAGAACC-3′). The two oligonucleotides introduce a BglII and a SalI restriction site immediately upstream and downstream of the SFH2 coding sequence. The PCR product was cloned into pGEM®-T Easy (Promega) to generate pMR303. The 1.3 kb BglII/SalI fragment from pMR303 was inserted into pMAL™-c2 (New England Biolabs) digested with SalI and BamHI to generate pMR306. Plasmid pCF8 contains FAS1 ORF under the control of the tetO7 promoter; it was constructed as follows: FAS1 ORF was amplified by PCR from the plasmid pJS229 (kindly provided by Professor H. J. Schüller, Universität Greifswald, Greifswald, Germany) using the two primers 5FAS1 (5′-CGCTCATTATGGACGCTTAC-3′) and 3FAS1 (5′-TTAGGATTGTTCATACTTTTC-3′) and cloned into pGEM®-T Easy (Promega) to generate pGF4-S. It was digested by NotI and inserted into pCM185 (kindly provided by Dr François Doignon, Université de Bordeaux II, Bordeaux, France) to generate pCF8. This plasmid was used to transform a heterozygous FAS1/fas1Δ strain from which a haploid fas1Δ was isolated. Growth of this strain in the absence of doxycycline was indistinguishable from the wild-type strain, but was totally inhibited with doxycycline concentrations above 1 μg·ml−1 in the medium. This confirmed the functionality of the tetO7 promoter to control accurately FAS1. Plasmid pCG4GFP is an integrative vector based on Yip351 which expresses GFP (green fluorescent protein)–Sfh2p under the control of the SFH2 promoter. It was constructed as follows: SFH2 ORF was isolated as a NotI fragment and cloned into pNG188 (kindly provided by Dr Jean-Marc Galan, Université de Paris VI et VII, Paris, France) to generate pMR314; this plasmid contains an in-frame fusion between GFP and SFH2 with a c-Myc epitope in between that can be isolated as an SpeI restriction fragment. A 600 bp fragment containing the SFH2 promoter was amplified by PCR and inserted into Yip351 to generate Yip-PROCSR. The terminator of PMA1 was inserted as a BamHI/NotI fragment from plasmid pNEV-N ; this plasmid was named YpA. Finally, the SpeI fragment from pMR314 was inserted into YpA digested with SpeI to generate pCG4GFP. This plasmid was used to transform MRY73-6a to generate YPCG4-G9d. It was introduced into TDY36-1d using standard genetic methods.
Lipid analysis and labelling
Phospholipids and neutral lipids were analysed using the standard procedure in use in our laboratory . Fatty acid methyl esters obtained from bulk lipids, or each lipid class separated by TLC as described in , were analysed by GC using a 25 m×0.32 mm AT-1 capillary column (Alltech) with methylheptadecanoic acid as a standard. Fatty acids were quantified and used to calculate the amount of lipid in each class. The lipid content was expressed in nmol of fatty acid per 109 (broken) cells.
For labelling experiments with palmitic acid, yeast cells were grown in liquid medium supplemented with δ-aminolevulinic acid (80 μg·ml−1) to a cell density of 4×107 cells·ml−1 after which they were incubated with 1.5 μCi·ml−1 [9,10-3H]palmitic acid (1.11 TBq·mmol−1) (ARC Radiochemicals). After 16 h, cells were harvested, washed with fresh non-radioactive YPD medium, and chased for the indicated time (0, 4, 6, 8 or 16 h). At the end of the experiments, cells were recovered and mixed with 2×109 non-radioactive cells. Lipids were extracted as described above. During the course of the chase experiment, there was no significant loss of radioactivity per volume of the culture within the lipid pool. The tritiated spots of individual lipid species were scraped off from the silica gel plates, transferred into scintillation counting vials, and the 3H content was determined by liquid-scintillation counting.
Enzyme activities and MBP–Sfh2p purification
β-Galactosidase assays were performed as described previously  on hem1Δ cells containing a plasmid bearing the 1 kb DNA sequence immediately upstream of the DAN1 ORF cloned 5′ to the reporter gene lacZ (plasmid pMR278 ), grown in the appropriate medium, or with plasmid pBF16 (ARS CEN TRP1 vector)  which contains a FAS2 promoter and part of the coding sequence fused in-frame to the lacZ coding sequence. Expression of the reporter gene, given as nmol of o-nitrophenol (measured at 420 nm)/min per mg of protein, was calculated as follows: A=(A420×100×1700)/(4.5×V×c×t), where A420 is the absorbance at 420 nm, V is the volume of the protein extract (ml), c is the protein concentration (mg·ml−1) and t is the incubation time at 30 °C (min).
FAS activity was determined using the method of Lynen  by assaying β-ketoacyl reductase-dependent oxidation of NADPH at 334 nm in the presence of acetyl-CoA and malonyl-CoA. Activity is expressed as m-units·mg −1 protein (nmol of NADPH oxidized per min per mg of protein). Cells were broken with glass beads and cleared by a 10 min centrifugation at 5000 g. Depending on the strain used, protein amounts varied between 70 and 130 μg to ensure an enzyme activity below 10 m-units, which guaranteed linearity between activity and enzyme amount . MBP–Sfh2p was purified essentially as recommended by the manufacturer's instruction (New England Bioloabs), with minor modifications as described elsewhere .
Liquid cultures were grown overnight in complete medium containing δ-aminolevulinic acid. Cells were harvested, washed twice with distilled water and were cultured in complete medium with or without δ-aminolevulinic acid for 6 h. Cells were resuspended in PBS with 0.1% (w/v) BSA (Sigma, Fraction V). Staining of lipid particles was performed by incubating cells with 50 μg·ml−1 Nile Red (Molecular Probes) for 5 min, and washing in PBS/0.1% BSA. Indirect immunofluorescence was performed from TDY33-3a using the protocol described in the TRIPLES database , except that mounted solution contained 10% (v/v) Vectashield (Vector Laboratories) and that the antibodies that were used were HA.11 polyclonal anti-HA (haemagglutinin) primary antibody (1:100) (Covance) and Alexa Fluor® 488-conjugated goat anti-rabbit IgG [H+L (heavy and light chain)] (Invitrogen) (1:400). Confocal microscopy was performed with the use of a confocal Olympus FV 1000 spectral mode microscope. The excitation wavelengths for GFP and Alexa Fluor® 488 were set to 488 nm. The excitation wavelength for Nile Red was set to 543 nm. Image acquisition and conversion were carried out separately for red and green channels and processed with Olympus Fluoview version 1.4 software.
Lipid composition of hem1Δ sfh2Δ mutant in haem-depleted medium
To determine whether SFH2 is involved in lipid metabolism under haem-depleted conditions, a hem1Δ sfh2Δ mutant and the control strain (hem1Δ) were cultivated in complete medium supplemented with δ-aminolevulinic acid to stationary phase and inoculated into fresh medium in the absence of δ-aminolevulinic acid, ergosterol and oleate (i.e. haem-induced lipid starvation). Under these conditions, expression of hypoxic genes was detected 6 h after the shift (see below), and growth ceased 16 h after the shift owing to sterol and oleic acid starvation . The amount of phospholipids and neutral lipids was quantified in both strains before the shift and 16 h later. Aerobic-like cultures of mutant and control strains grown to stationary phase had very similar profiles of phospholipids (Figure 1A). In contrast, noticeable differences in phospholipid composition were observed following oleic acid starvation and haem depletion: whereas a 53% decrease in PtdCho was observed in the control strain, there was a 31% increase in the mutant strain. We also noted a slight decrease in PtdIns abundance 16 h after the shift in the mutant strain as compared with the control strain. The amount of PtdOH (phosphatidic acid), a proposed key component of phospholipid metabolism , was also affected since haem depletion resulted in a 2-fold increase in the amount of PtdOH in the mutant strain; in the control strain, it remained more or less unchanged. Finally, the amounts of PtdEtn (phosphatidylethanolamine) and PtdSer were similar between the two strains.
Next, we examined neutral lipid composition in the mutant strain following haem-induced lipid depletion. As reported in Figure 1(B), TAG and SE accumulated to the same extent in control and mutant strains cultivated to stationary phase in the presence of δ-aminolevulinic acid. We observed that SE was efficiently degraded in both strains following the shift to haem-induced lipid starvation. Residual amounts of SE were still detected 16 h after the shift. It corresponded to the amount of SE detected in normal aerobic growth during exponential phase. Surprisingly, whereas 90% of the pool of TAG was depleted 16 h after the shift in the hem1Δ strain, 40% of the TAG accumulated under aerobiosis was still observed in the hem1Δ sfh2Δ (Figure 1B). We also noted a 61% increase in the amount of DAG (diacylglycerol) in the mutant strain, whereas there was a 2.3-fold decrease in DAG in the control strain.
Sfh2Δ produces more SFAs upon haem-induced lipid starvation
We have reported previously that SFAs preferentially accumulate within specific phospholipid species in response to haem-induced fatty acid starvation . Thus we looked more closely at the ratio of SFA within phospholipids and neutral lipids in response to haem depletion. As reported in Table 2, the increase in the amount of SFAs did not affect all of the lipid species to the same extent. We have already shown (see Figure 1A) that haem depletion did not affect the total amount of PtdSer and PtdEtn; in Table 2, it is reported that the ratio of SFA in PtdEtn was not modified either. We noted a slight decrease in the SFA ratio in PtdSer in the hem1Δ sfh2Δ strain 16 h after the shift as compared with the control hem1Δ. The fatty acid composition of PtdCho was not modified 16 h after the shift between mutant and control (Table 2). However, bearing in mind that the amount of PtdCho had increased by 53% in the mutant and decreased by 35% in the control (see Figure 1A), it suggests that a substantial amount of SFA was contained in this phospholipid. In addition to a 2-fold increase in the amount of PtdOH (Figure 1A), the amount of SFA represented 72.4% of this lipid in the mutant and 46.3% in the control strain. Finally, although the amount of PtdIns had decreased in the mutant, but increased in the control strain, following the shift to haem-depleted medium (Figure 1A), the proportion of SFA was increased to 71.6% in the mutant and to 60.0% in the control. Thus similar amounts of SFA were contained in this phospholipid species. SFAs were also preferentially stored in TAG in the mutant after the shift (they constitute 64.5% of total TAG in the mutant 16 h after the shift, but only 29.2% in the control). The amount of DAG had increased to 61% in the mutant relative to the control (Figure 1B), but the proportion of SFAs in this species was not dramatically altered (84.0% in mutant and 81.6% in control). Finally, free SFA represented 41.4% of total NEFA (non-esterified fatty acid) in the mutant and 34.4% in the control.
To confirm that the hem1Δ sfh2Δ mutant accumulates more fatty acids than the hem1Δ control strain, the amount of fatty acids was compared between mutant and control, in response to haem-induced fatty acid depletion. As shown Figure 2(A), and as already described, the amount of fatty acid in the cells decreased rapidly following inoculation in complete medium , because the cells preferentially reallocate their endogenous pool of fatty acid accumulated in stationary phase within lipid particles: indeed, 4 h after inoculation from stationary phase, 2.5 generations had been accomplished, and the amount of fatty acids in the cells was 4-fold lower, suggesting that there is only a modest fatty acid biosynthesis following inoculation into fresh medium (see below).
Soon after the onset of hypoxic gene expression (6 h), but before growth arrest of the mutant (see below, Figure 3), total fatty acids represented roughly 37.7% of the initial amount in control and 35.8% in mutant, whereas 2 h later (i.e. 8 h after the shift), it was maintained at a similar level in the control strain, but it had increased to 53.4% in the mutant strain. It is interesting to observe that, in both strains, the amount of UFA per 109 cells was very similar and roughly constant during the experiment (Figure 2C). Because until 8 h after the shift, growth of both strains was not arrested, it follows that both strains were still capable of converting newly synthesized fatty acids into UFAs. This is most likely to be the primary consequence of hypoxic induction of OLE1, thereby allowing cells to desaturate fatty acid even though oxygen is limiting . As shown in Figures 2(B) and 2(C), the increase in total fatty acid content of the mutant described above can be accounted for solely by the steady increase in the amount of SFA.
SFH2 is necessary for growth under haem-depleted conditions
To address the consequences of the lipid pattern modifications in the mutant, we compared the growth curves of the mutant and the wild-type strains in response to haem deprivation. Control hem1Δ and mutant hem1Δ sfh2Δ were cultivated to stationary phase in the presence of δ-aminolevulinate. Growth rate was monitored after a shift to fresh YPD medium without supplementation. As shown in Figure 3 and as already reported , haem depletion in YPD medium results in growth arrest of the control strain after about five generations (i.e. 16 h). Hypoxic gene expression, monitored by using the lacZ reporter gene fused to the promoter of the hypoxic gene DAN1, was detected 6 h after the shift (see arrow on Figure 3) . This is probably the time period necessary to deplete the cells of their endogenous pool of δ-aminolevulinic acid. We observed that, within 5 h of the shift, no effect of SFH2 deletion could be observed. Longer incubation in the absence of lipid supplementation (oleic acid and ergosterol) results in growth arrest of the mutant: 8 h after the shift, cell growth was totally arrested (Figure 3). In contrast, the control strain was still able to sustain growth until 15 h after the shift. In addition, the aberrant phenotype of the hem1Δ sfh2Δ mutant is relieved by supplementation of the medium with oleic acid (Figure 3), but not by ergosterol (results not shown). Thus we conclude that oleate starvation, induced by haem depletion, results in rapid growth arrest in the absence of a functional copy of the SFH2 gene.
Neutral lipid accumulation is not responsible for the sfh2Δ phenotype
Our results indicate that SFH2 deletion has two major consequences for lipid metabolism in response to oleic acid starvation: (i) a global increase in the SFA content, and (ii) an alteration of the lipid composition, including an accumulation of TAG, DAG, PtdOH and PtdCho. The accumulation of TAG in lipid particles in response to SFH2 deletion could be interpreted in at least two ways. It could simply result from a partial failure to hydrolyse neutral lipids from lipid particles. Alternatively, neutral lipids might be efficiently degraded and used as acyl donors for phospholipid synthesis; SFH2 deletion would secondarily provoke improper storage into neutral lipids. We thus questioned the significance of TAG accumulation following haem depletion in the null SFH2 mutant.
Lipids from control hem1Δ and hem1Δ sfh2Δ mutant growing in the presence of δ-aminolevulinic acid were metabolically labelled by extensive incorporation of tritiated palmitic acid until stationary phase. Cells were then washed and inoculated in fresh YPD medium lacking δ-aminolevulinic acid and oleic acid. Radioactivity within neutral lipid species separated by TLC was quantified at various time points after the shift. As shown in Figures 4(A) and 4(B), 35.3±2.7% and 35.9±0.9% of the radioactivity incorporated into lipids before the shift was recovered in TAG in the control strain and the mutant strain respectively. SE represented 6.0±0.9% and 8.2±1.3% of total radioactivity in control and mutant respectively. We observed that 8 h after the shift, both types of molecules were effectively hydrolysed. In addition, we could observe a concomitant increase in the radioactivity associated with phospholipids, demonstrating that acyl chains from neutral lipids had been reallocated into newly synthesized phospholipids (results not shown). However, with longer chase experiments, an increase in the amount of radioactivity associated with TAG was observed in the mutant strain 16 h after the shift. In the mutant, radioactivity incorporation into TAG constituted 28.3±1.8% of total radioactivity 16 h after the shift; in the control strain, it represented only 13.5±2.4%. This led us to conclude that deletion of SFH2 leads to an improper synthesis of TAG in response to haem starvation.
To evaluate whether TAG accumulation is responsible for the premature growth arrest, the phenotype of hem1Δ sfh2Δ inactivation was investigated in a strain background inactivated for the genes encoding TAG and SE synthases . We observed that a lro1Δ are1Δ are2Δ dga1Δ hem1Δ strain grew approximately one-third as much as the hem1Δ strain (Figure 4C). This is not totally surprising since it lacked most of the fatty acids contained in the TAG and SE of the hem1Δ strain. Nevertheless, we observed that the growth defect resulting from SFH2 deletion in haem- and oleic acid-depleted medium was still observed in this genetic background (Figure 4C).
Moreover, deletion of the encoding genes TAG and SE synthases (i.e. the quadruple dga1Δ lro1Δ are1Δ are2Δ mutant), resulted in an enhanced production of some lipid species that already accumulate in the hem1Δ sfh2Δ mutant (i.e. PtdOH, PtdCho and DAG), without any noticeable effect on the lack of growth of this strain in haem-depleted conditions (results not shown). This suggests that TAG, PtdCho, DAG and PtdOH accumulation in the sfh2Δ strain is not primarily responsible for the lack of residual growth upon haem deprivation.
The sfh2Δ mutant is hypersensitive to SFAs
The results presented so far suggest that the substantial accumulation of SFA in the mutant in response to oleic acid starvation under haem-depleted growth could be responsible for the hem1Δ sfh2Δ phenotype. Thus one could predict that the mutant strain would be more sensitive to exogenous SFAs than the control strain under haem-depleted conditions. To address this possibility, hem1Δ sfh2Δ and hem1Δ strains were cultivated to stationary phase in the presence of δ-aminolevulinic acid and shifted to complete medium containing a limited amount of Tween 80 as a source of oleic acid and variable amounts of Tween 40 as a source of palmitic acid. The attenuance at 600 nm of the cultures was measured 16 h later. An example, representative of three independent experiments, is presented in Figure 5. A limited amount of Tween 80 (0.01%) could significantly increase the growth of control and mutant strains. As expected, the hem1Δ sfh2Δ was much more sensitive to the addition of SFAs than the hem1Δ strain: indeed, the presence of Tween 40 (0.005%) in the medium had no detectable effect on the hem1Δ strain, whereas it reduced growth of the hem1Δ sfh2Δ strain to 35% of what is observed in the absence of Tween 40. The amount of Tween 40 necessary to inhibit growth despite the supplementation with 0.01% oleate was roughly 3-fold less in the hem1Δ sfh2Δ than with the hem1Δ strain (0.02% and 0.06% respectively). Therefore the growth defect of the mutant is exacerbated by SFA supplementation, suggesting a deleterious effect. This confirms that growth inhibition of the hem1Δ sfh2Δ mutant results from endogenous accumulation of SFAs.
Enhanced activity of FAS (fatty acid synthase) in sfh2Δ in vivo
We therefore tried to understand the basis of SFA accumulation in the mutant in response to oleic acid deprivation. In yeast, SFA are synthesized by a heteromultimeric α6β6 complex encoded by the two genes FAS1 and FAS2 for the β and α subunits respectively . Both genes are regulated similarly and appear to be down-regulated by exogenous SFA in the culture medium . We wondered whether oleic acid could also control FAS expression. For this purpose, we used a lacZ gene reporter assay to monitor FAS2 expression under our growth conditions (plasmid pBF16  kindly provided by Professor H. J. Schüller). To minimize possible interference from the genetic background, the experiments were performed in a hem1Δ sfh2Δ mutant transformed either with the control plasmid (YEp352) or with a plasmid expressing SFH2 (pJU1). As shown in Figure 6, control and mutant strains behaved very similarly: addition of oleic acid to the medium in the presence of δ-aminolevulinate resulted in a 2.6-fold decrease in the reporter gene activity, a level of repression in the same range as with an SFA (palmitic acid: 2.9-fold decrease). In contrast, we observed a 1.8-fold induction of the reporter gene under haem and oleic acid starvation. Importantly, supplementation with oleic acid in the absence of haem reduced FAS2 expression 3.1-fold, to a level of expression below that observed in the presence of haem. Therefore oleic acid starvation in cells lacking haem results in enhanced expression of FAS2.
This last observation was somewhat surprising, since, although both strains regulate transcription of FAS2 identically, the mutant only showed an increase in the amount of SFA in response to haem depletion and oleic acid starvation. Thus we measured the activity of FAS in the hem1Δ sfh2Δ and in the hem1Δ strains in response to haem depletion. As shown in Figure 7, FAS activity in cells grown to stationary phase under aerobiosis was similar in both strains (28.8±2.9 and 23.4±3.2 m-units·mg−1 respectively). In the hem1Δ, FAS activity was maintained at a low level following oleic acid starvation in haem-depleted medium: 10 h after the shift, FAS activity was 31.9±0.1 m-units·mg−1. In contrast, in the hem1Δ sfh2Δ strain, haem depletion and oleic acid starvation were associated with an increase in FAS activity (57.87±4.7 m-units·mg−1 detected 6 h after the shift) which was maintained at a high level during the rest of the experiment.
Suppression of sfh2Δ phenotype by controlled expression of FAS1
It has been shown that the activity of FAS holoenzyme is primarily determined by the amount of Fas1p, since overexpression of the FAS1 gene results in a concomitant increase in FAS2 mRNA abundance and an enhanced activity of FAS . Thus it is possible to alter the amount of active FAS by controlling the transcriptional level of the sole FAS1 gene. This observation provided a convenient way to test whether increased FAS activity is responsible for the phenotype of the hem1Δ sfh2Δ strain. Plasmid pCF8 contains an engineered FAS1 gene whose expression is controlled by the tetO7-CYC1 promoter which is negatively regulated by doxycycline . It was used to transform a heterozygous FAS1/fas1Δ strain from which a viable haploid fas1Δ was isolated. In this strain, expression of the episomal FAS1 gene is driven by the tetO7-CYC1 promoter and consequently FAS activity is modulated by the amount of doxycycline added to the medium, as demonstrated for other genes : indeed, we observed that fas1Δ [pCF8] strain under aerobiosis is viable, and its growth is inhibited by doxycycline concentrations above 1 μg·ml−1 because expression of the engineered FAS1 gene is totally repressed and FAS activity is undetectable (results not shown).
Then, we constructed hem1Δ fas1Δ and hem1Δ fas1Δ sfh2Δ strains containing plasmid pCF8; both strains were cultivated in the presence of δ-aminolevulinic acid to stationary phase and inoculated further in complete medium with variable amounts of doxycycline in the medium in order to control FAS activity. The attenuance reached by the cultures was measured 40 h after the shift. As shown in Figure 8(A), in the absence of doxycycline in the medium, or with concentration of doxycycline above 0.4 μg·ml−1, the mean attenuance reached by the strains (0.186±0.046 and 0.184±0.004 for hem1Δ fas1Δ [pCF8] and hem1Δ fas1Δ sfh2Δ [pCF8] strains respectively) suggest that residual growth was limited, if not totally impaired. Indeed, the hem1Δ sfh2Δ mutant attained similar values under these growth conditions (results not shown). In contrast, with 0.15 μg·ml−1 doxycycline, a significant residual growth was attained with the hem1Δ fas1Δ: indeed, the attenuance achieved for this strain was 0.474±0.02, which is only slightly lower than for a hem1Δ strain [0.508±0.090 obtained with a similar inoculum (2×105 cells·ml−1)]. Interestingly, a higher concentration of doxycycline was needed for the hem1Δ fas1Δ sfh2Δ strain to achieve a significant residual growth (D600 of 0.391±0.050 with a doxycycline concentration of 0.30 μg·ml−1). FAS activity was also measured in both strains after the shift (Figure 8B). We observed that, for any concentration of doxycycline, FAS activity detected in the hem1Δ fas1Δ [pCF8] strain was always lower than in the hem1Δ fas1Δ sfh2Δ [pCF8] strain. This confirms the up-regulation of FAS activity in the sfh2-null mutant. To get a better insight into those activities, it is useful to compare with values obtained in hem1Δ sfh2Δ and hem1Δ strains, in which FAS activity depends on chromosomal expression of FAS genes, in response to haem depletion and oleic acid starvation (Figure 8C). FAS activity close to that obtained in the hem1Δ strain was observed with doxycycline concentrations compatible with significant residual growth (i.e. 0.15 and 0.30 μg·ml−1 for hem1Δ fas1Δ [pCF8] and hem1Δ fas1Δ sfh2Δ [pCF8] respectively). In contrast, with lower amounts of doxycycline in the medium, values of FAS activities were closer to (or even higher than) values obtained in the hem1Δ sfh2Δ: under these growth conditions, no residual growth was observed in the corresponding strains as a consequence of overexpression of episomal FAS1 in the hem1Δ fas1Δ [pCF8] strain. Thus these experiments demonstrate that down-regulating FAS1 transcription is sufficient to suppress the phenotype of hem1Δ sfh2Δ. In addition, the high amount of doxycycline necessary for the hem1Δ fas1Δ sfh2Δ [pCF8] strain to achieve residual growth is consistent with an already higher FAS activity in the triple mutant strain than in the hem1Δ fas1Δ strain and confirms that residual growth necessitates a tight regulation of FAS activity by Sfh2p.
Sfh2p inhibits FAS activity
The elevated FAS activity in vivo in the absence of oleic acid and SFH2 led us to investigate the possibility that Sfh2p inhibits the activity of FAS. For this purpose, we engineered an N-terminally tagged version of Sfh2p, namely MBP-Sfh2p. We measured the effect of adding purified recombinant Sfh2p on FAS activity in vitro. In this assay, cellular extracts were prepared from the hem1Δ sfh2Δ mutant incubated for 6 h in a δ-aminolevulinic and oleic acid-free complete medium (i.e. high level of FAS activity). Various amounts of recombinant MBP–Sfh2p purified from Escherichia coli were incubated in the assay medium 1 min before initiating the reaction. As shown in Figure 9, recombinant Sfh2p could substantially inhibit FAS activity (up to 50% inhibition). In contrast, addition of 150 ng of purified MBP in the assay did not affect FAS activity significantly (results not shown). We have already mentioned that, under aerobiosis, deletion of SFH2 resulted in a 30% increase in FAS activity. Addition of recombinant Sfh2p to cell extracts of the sfh2Δ mutant grown under aerobiosis restored the activity to a wild-type level, suggesting that under aerobic growth, Sfh2p could also be used to regulate FAS activity (results not shown).
Sfh2p is localized in lipid particles
The cellular localization of Sfh2p is still a matter of debate. It was reported to localize, at least partially, to endosomes . In contrast, Schnabl et al.  reported that Sfh2p localized mainly to the cytosol and microsomes. In order to address the subcellular localization of Sfh2p, we used an integrated version of SFH2 which expresses a GFP-tagged Sfh2p. The fluorescent tag was inserted at the N-terminus of Sfh2p and was functional, as judged by the ability of the recombinant GFP–Sfh2p to rescue the phenotype of hem1Δ sfh2Δ mutation (results not shown). The recombinant strain was grown to stationary phase in the presence of δ-aminolevulinic acid. GFP–Sfh2p localized to several punctate structures scattered in the cytoplasm (Figure 10, left-hand panels). Interestingly, this pattern was similar to the one observed for the fluorescent dye Nile Red, which is specific for neutral lipids, and is used as a marker of lipid particles. Computer merging of these images confirmed the localization of GFP–Sfh2p with Nile Red structures (results not shown). We estimated that more than 90% of the GFP–Sfh2p-positive structures were also positive with Nile Red. Following inoculation in δ-aminolevulinic acid-depleted medium and in the absence of oleic acid, the pattern of GFP–Sfh2p remained very similar, suggesting that the protein remained associated with lipid droplets (results not shown). In contrast, in the presence of δ-aminolevulinic acid in the medium, the signal promptly faded, and, 6 h after the shift, the protein was no longer detectable (results not shown), as expected for the product of a hypoxic gene whose level of expression in exponential phase under aerobiosis is low .
We also tried to localize Fas1p in the cell. For this purpose, we used a strain expressing a functional Fas1p–3×HA fusion protein that was kindly provided by Dr Snyder (TRIPLE database) from which a hem1Δ derivative was obtained. Indirect immunofluorescence using anti-HA monoclonal antibodies was performed from the strain grown in a medium supplemented with δ-aminolevulinic acid. In exponential growth, the signal was diffuse in the cell (results not shown), suggesting a cytosolic distribution, as described above for this protein. In contrast, a sample from stationary-phase cells revealed a very different localization of Fas1p–3×HA (Figure 10B, left-hand panel): the fluorescent signal was localized in discrete punctate structures, which again correspond to lipid particles as judged by the co-localization with Nile Red (Figure 10B, middle panel). Inoculation of this strain in complete medium lacking δ-aminolevulinic acid and oleic acid did not modify the localization of Fas1p–3×HA (Figure 10C) which was still associated with lipid particles 6 h after the shift. In contrast, in the presence of δ-aminolevulinic acid in the medium, the fluorescent signal was not associated with lipid droplets (Figure 10D), suggesting a cytosolic distribution of this protein. Nevertheless, under haem depletion and oleic acid starvation, both Sfh2p and Fas1p appeared to be co-localized in lipid particles. Interestingly, this localization was not modified in an sfh2Δ-null mutant (T. Desfougères, T. Bergès and M. Régnacq, unpublished work). All together, genetic data and localization data support the notion that Sfh2p negatively regulates FAS.
S. cerevisiae is only able to synthesize mono-unsaturated fatty acids containing a Δ9 double bond, primarily palmitoleic acid (C16:1) and oleic acid (C18:1), which represent respectively approx. 45 and 25% of total fatty acids respectively under aerobiosis. The Δ9 desaturase responsible for the synthesis of these UFAs is encoded by a single essential gene, OLE1. Under aerobiosis, transcriptional regulation of OLE1, and modulation of OLE1 mRNA stability, maintain an adequate balance of UFAs against SFAs . Since Ole1p is strictly dependent on haem availability, hence on oxygen status, life under hypoxic or anoxic conditions has important consequences for this ratio. Accordingly, when the oxygen status is low in the cell, induction of FAS genes and of OLE1 facilitates desaturation of both existing and newly synthesized fatty acids as long as oxygen is still available . However, under strict anaerobiosis or in haem-depleted conditions, i.e. when Ole1p is not functional, FAS genes transcription is still observed if fatty acids are not provided in the medium, and import of exogenous UFAs from the medium becomes strictly necessary for optimal growth.
In the present paper, we report a body of evidence which indicates that Sfh2p modulates FAS activity in response to haem depletion and oleate starvation and that, under these conditions, this regulation is essential to prevent SFA accumulation that eventually leads to premature growth arrest. Consistently, sfh2Δ displays enhanced sensitivity to SFA in the medium, whereas UFA supplementation restores the wild-type phenotype. In agreement with a detrimental effect of SFH2 deletion on FAS activity, we observed an increase in FAS activity in the mutant in response to haem depletion and UFA starvation. This is confirmed by the observation that sfh2Δ can be rescued by down-regulation of FAS1 gene expression, whereas enhanced expression of FAS1 gene generates a phenotype which is similar to sfh2Δ (Figure 8). However, the mechanism of FAS inhibition by Sfh2p is still unclear. We would tend to favour a model by which Sfh2p exerts a direct effect on FAS, because: (i) in vitro, addition of purified Sfh2p inhibits FAS activity, (ii) both proteins co-localize within lipid particles, and (iii) we have observed in vitro that Sfh2p can bind Fas1p (T. Desfougères, T. Bergès and M. Régnacq, unpublished work). However, we have not yet been able to confirm this interaction in vivo.
We also observed that residual growth of sfh2Δ mutant is impaired in genuine anaerobiosis in the absence of oleic acid supplementation (T. Desfougères, T. Bergès and M. Régnacq, unpublished work), suggesting that Sfh2p accomplishes the same function under genuine anaerobiosis as in response to haem depletion. A possible way to understand the physiological relevance of Sfh2p function in response to oxygen depletion is to examine the phenotype resulting from altered expression of FAS (Figure 8). We observed that neither high nor low FAS activity can restore residual growth in response to haem depletion and UFA starvation. This is very surprising, since there is a sharp transcriptional induction of FAS genes in the absence of oleic acid and haem (Figure 6). These apparent conflicting results can be reconciled bearing in mind that, when oxygen becomes limited, there is transcriptional derepression of hypoxic genes, including OLE1, and also FAS1 and FAS2 if fatty acids are not present in the medium. As long as Ole1p is active (i.e. oxygen is still present in the cell), the adequate balance between SFAs and UFAs is still guaranteed in the cell. However, as soon as haem synthesis ceases, consequent to oxygen depletion, fatty acid desaturation by Ole1p is precluded, whereas FAS is still abundant and active in the absence of fatty acid supplementation. We propose that the function of Sfh2p is to inhibit FAS under these conditions and that this fine-tuning of FAS by Sfh2p would allow the cell to prevent the deleterious accumulation of SFAs. As a Sec14p- related protein, Sfh2p contains a hydrophobic pocket that can potentially bind phospholipids. It is tempting to speculate that inhibition of FAS activity by Sfh2p might be controlled by a lipid-bound form. Among the lipids that accumulate in a hem1Δ sfh2Δ in response to haem depletion and UFA starvation, PtdOH fulfils several criteria to be a candidate to accomplish this function. Indeed, not only does this lipid accumulate in response to haem and UFA depletion, but also it is highly enriched in SFAs (Table 2). Moreover it is produced in lipid particles , i.e. a compartment where Sfh2p and Fas1p are mainly localized in response to haem depletion. Further work is needed in order to test this hypothesis.
SFH2 encodes a gene which was originally identified in a screen for multicopy suppressors of the double mutant chs5 spa2 defective in chitin synthesis and cellular morphology [27a]. It was also observed that it can physically interact with a thiol peroxidase type II isoform (cTPx II) in vitro , although the functional significance of this interaction is unclear considering the function of Sfh2p as described in the present paper. Most interestingly, in addition to the sequence homology between Sfh2p and Sec14p, a protein involved in the regulation of phospholipid metabolism, it was observed that SFH2 was also a multicopy suppressor of SEC14 loss of function [10,12]. A decrease in FAS activity might be relevant to understand the mechanism of suppression of sec14Δ by SFH2 overexpression. Indeed, we observed an increase in FAS activity in a sec14-1 mutant at the restrictive temperature which was restored in response to SFH2 overexpression (T. Desfougères, T. Bergès and M. Régnacq, unpublished work). Sec14p controls PtdCho metabolism by inhibiting Pct1p, the rate-limiting enzyme of the CDP-choline pathway for PtdCho synthesis. It has been also demonstrated that Sec14p is involved in PtdCho degradation by stimulating Nte1p, a phospholipase B specific for PtdCho [29,30]. Therefore PtdCho levels in yeast are controlled both at the level of its synthesis and at the level of its degradation. Interestingly, we observed that deletion of SFH2, which stimulates FAS activity, results in a significant accumulation of PtdCho in haem- and oleic acid-depleted medium (Figure 1). This suggests that FAS activity impinges on PtdCho biosynthesis and/or degradation. This view is supported by data obtained from CHO (Chinese-hamster ovary) cells, demonstrating that inhibition of FAS activity by the antibiotic cerulenin impairs PtdCho biosynthesis [31,32]. These authors observed that this effect is mediated by an inhibition of Pct1p. Although it is tempting to speculate that, in yeast, fatty acid synthesis inhibition, induced by Sfh2p, could lead to Pct1p inhibition and thus facilitate a bypass Sec14p phenotype, this view is not fully supported by experimental data. Indeed, SFH2 overexpression in the absence of SEC14 does not seem to affect PtdCho degradation . However, a significantly diminished rate of choline uptake, an accumulation of cytoplasmic choline, and an altered PtdCho/choline ratio in the cells towards a lower amount of PtdCho compared with choline were observed in response to SFH2 overexpression in a sec14Δ mutant. This suggests that the metabolic flux in the Kennedy pathway is diminished, but not through inhibition of Pct1p (which would then result in an increase in PtdCho at the expense of choline). One attractive possibility would be that this reduced flux is a consequence of the limited availability of the substrates and more specifically of reduced synthesis of DAG as a consequence of FAS inhibition. The observation that DAG also accumulates in the absence of SFH2, in haem- and oleic acid-depleted medium (see Figure 1), is consistent with this hypothesis.
This work was supported by the French MENRT (Ministère de l'Éducation Nationale, de la Recherche et de la Technologie) and CNRS (Centre National de la Recherche Scientifique). T. D. was supported by a grant from MENRT. We gratefully acknowledge Professor H. J. Schüller for generously providing us with plasmid pBF16 and pJS229, and members of the Snyder laboratory (Yale Genome Analysis Center, Yale University, New Haven, CT, U.S.A.) for providing yeast strain V83B2. François Doignon is kindly acknowledged for providing pCM185, and we thank Jean-Marc Galan for providing pNG188. Anne Cantereau is acknowledged for her technical assistance with confocal microscopy.
Abbreviations: DAG, diacylglycerol; FAS, fatty acid synthase; GFP, green fluorescent protein; HA, haemagglutinin; MBP, maltose-binding protein; NEFA, non-esterified fatty acid; ORF, open reading frame; PtdCho, phosphatidylcholine; PtdEtn, phosphatidylethanolamine; PtdIns, phosphatidylinositol; PtdOH, phosphatidic acid; PtdSer, phosphatidylserine; SE, steryl ester; SFA, saturated fatty acid; TAG, triacylglycerol; UFA, unsaturated fatty acid; YNB, yeast nitrogen base
- © The Authors Journal compilation © 2008 Biochemical Society