Cloning by metabolic interference in yeast and enzymatic characterization of Arabidopsis thaliana sterol delta 7-reductase.

Reduction of the delta 7 double bond of sterols, a key biosynthetic step in higher eukaryotes, is lacking in lower eukaryotes like the yeast Saccharomyces cerevisiae, leading to terminal sterols with a delta 5,7-conjugated diene structure. Genes encoding two sterol reductases involved, respectively, in the reduction of sterol delta 14 and delta 24(28) double bonds have been cloned to date, but no sequence information was available on the enzyme responsible for delta 7-bond reduction. This study presents the cloning of the NADPH-sterol delta 7-reductase (delta 7-red) from Arabidopsis thaliana, based on a metabolic interference approach in yeast. The principle is the functional expression of a plant cDNA library in the yeast strain FY1679-28C tolerant to sterol modifications and the selection of clones resistant to the polyene fungicide nystatin. The toxicity of this compound is dependent on the presence of delta 5,7-unsaturated sterols in the yeast plasma membrane. One clone out of 10(5) transformants exhibits a cDNA-dependent alteration of cell sterol composition. The 1290-base pair cDNA open reading frame was isolated and sequenced. The corresponding protein presents a significant sequence similarity with yeast delta 14- and delta 24(28)-reductases and with human lamin B receptor. The coding sequence was extracted by polymerase chain reaction and inserted into a galactose-inducible yeast expression vector to optimize expression. Analysis using transformed wild type yeast or sterol altered mutants, indicated that delta 5,7-ergosta- and cholesta-sterols are efficiently reduced in vivo, regardless of the structural variations on the side chain. No reductase activity was observed toward the delta 14 or the delta 5 positions of sterols. In vivo extensive delta 7-reduction of the free and esterified pools of sterols was observed upon induction of the enzyme. Ergosterol present before induction was reduced into ergosta-5,22-dieneol, whereas ergosta-5-eneol is the new end product of sterol neosynthesis, indicating that the yeast delta 22 desaturase may be no longer active on C-7-saturated sterols. In vitro tests indicated that delta 7-reductase activity is preferentially associated with the endoplasmic reticulum membrane and confirmed the previous finding that NADPH is the reducing agent.

Sterols are major components of eukaryotic cell membranes. End products of the sterol biosynthetic pathway differ depending on species; cholesterol is encountered in animals, ergosterol is the most common sterol in fungi, while sitosterol, campesterol, and stigmasterol are typical plant sterols (1). Most of terminal sterols feature a similar four-ring structure including a C-5-C-6 unsaturation on ring B and a more or less branched side chain. Additional features depend on the origin; cholesterol has a C 8 side chain, whereas plant sterols bear an additional alkyl (methyl or ethyl) group on C-24. Ergosterol, in addition to a C-24 methyl group, contains two additional unsaturations on the C-7-C-8 bond of the B ring and on the C-22-C-23 bond of the side chain. These unsaturations are specific to fungal sterols and make them the target for antifungal polyene drugs like nystatin (2).
The enzymatic steps that allow conversion of lanosterol resulting from the oxidative cyclization of squalene to cholesterol have been documented (3)(4)(5)(6)(7). Most of steps are also found in Saccharomyces cerevisiae ergosterol biosynthesis (8,9). Few of the involved mammalian genes like the P450 lanosterol demethylase have been isolated (10,11). From zymosterol (cholesta-8,24-dieneol) to cholesterol (cholesta-5-eneol) or to ergosterol (ergosta-5,7,22-trieneol), there are multiple alternative pathways leading to the terminal sterol. In mammals, the main flow involves the following sequence: isomerization of the double bond from C-8 to C-7, introduction of a double bond at C-5, and two reduction steps at the C-24 and C-7 double bonds. These reduction steps are absent in yeast but are catalyzed by the sterol ⌬24-reductase and the ⌬7-red, 1 respectively, in higher eukaryotes. Very little is known about these two reactions excepted that they can be reproduced in vitro using a reconstituted system involving partially purified fractions from rat liver (12,13) or plant (14) microsomes. Indeed, inhibitors of the ⌬7-red have been tested as therapeutic drugs against hypercholesterolemia (15), and it was recently reported that a recessive autosomal disorder, the Smith-Lemli-Optiz syndrome responsible for multiple congenital anomalies, corresponds to a deficient activity of this enzyme (16). Lipoproteins of affected patients are enriched in ⌬7-cholesterol (cholesta-⌬5,7-dieneol) and are highly depleted in cholesterol (17). Despite its interest, identification of the gene encoding the ⌬7-red appeared to be a puzzling task due to the total lack of sequence information as well as of purified protein or related antibodies.
Here we present the cloning of the Arabidopsis thaliana ⌬7-red based on the observation that nystatin toxicity is highly dependent on the presence of sterol carrying a ⌬5,7-dienic structure. Selection for functional expression in yeast of a plant cDNA encoding ⌬7-red activity was based on the expectation that reduction of the ⌬5,7-endogenous yeast sterols would decrease nystatin toxicity. Nevertheless, the possibility to substitute ergosterol by its reduction product was a major concern since disruption of ergosterol biosynthesis is the classical mode of action of antifungal drugs like ketoconazole. In fact, the bulk (structural) function of the sterols can be fulfilled in yeast by different sterols such as cholestanol, cholesterol, lanosterol, or intermediates of sterol biosynthesis pathway, provided that some residual level of ergosterol be present (18,19). This is known as the "sparking effect," which is probably related to a cell cycle control mechanism in wild type strains (20). This requirement is abolished in cells harboring fen1 and/or fen2 gene mutations (21,22) suggesting to use strain FY1679-28C, a naturally occurring fen1 mutant, as a host for the ⌬7-red screening. We also chose a plant, A. thaliana as cDNA source because of the high ⌬7-red activity present in plant microsomes 2 and its low relative genome complexity (23,24).
Vectors-The plasmid pUC9-N was derived from pUC9 (Pharmacia Biotech Inc.) by insertion of a NotI restriction site in the filled EcoRI site (gift from F. Lacroute). pBlueScript (Stratagene) was used to subclone different fragments of the ⌬7-red gene. The Escherichia coli-S. cerevisiae shuttle vectors are pYeDP1/8-2 (named V8), which carries a yeast "2" origin of replication, the URA3 selection marker, an expression cassette based on the galactose inducible GAL10-CYC1 promoter, a multiple cloning site, and the phosphoglycerate kinase gene (PGK) terminator (28) and pFL61 (29).
Screening of the A. thaliana cDNA Expression Library in Yeast-The cDNA library from full seedling A. thaliana at the two-leaf stage was kindly provided by Dr. F. Lacroute (29). In this library, cDNAs are cloned as a NotI cassette placed under the transcriptional control of PGK transcription promoter and terminator sequences in a pFL61 E. coli-S. cerevisiae shuttle vector. A "2" origin of replication and a URA3 selection marker are used for propagation in yeast, whereas the E. coli propagation part is derived from pUC19. The FY1679-28C yeast strain was transformed with the cDNA library using the lithium acetate procedure (30). Cells were plated on synthetic medium lacking uracil (SGI with 2% agar (Difco)). Transformation yielded 10 5 primary transformants prototrophic for uracil. The cells were pooled and plated at 5 ϫ 10 4 cells per plate containing SGI solid medium and either 2 or 5 g/ml nystatin. After 3 days of incubation at 28°C, a hundred clones were growing in the presence of 2 g/ml nystatin. Individual clones were finally analyzed for their sterol composition by HPLC. Among them, one named F22 was resistant to 5 g/ml nystatin and exhibited a sterol composition with a lowered ⌬5,7 content based on the 280 nm HPLC traces.
E. coli was transformed by the plasmidic DNA extracted from clone F22 as described previously (31). The plasmidic DNA from individual E. coli transformants was digested by NotI. Two different classes of cDNA inserts were identified which correspond to 600-bp and 1.6-kb NotI inserts respectively. The plasmid carrying the 1.6-kb cDNA insert was found to be also rearranged at the pFL61 level as judged by the altered restriction pattern. The FY1679-28C yeast strain was retransformed with this plasmid and the sterol composition of transformants analyzed. All transformants exhibit the same anomalous sterol pattern as compared with the void pFL61 transformed strain.
Nucleotide Sequence Determination-The NotI cDNA insert in pFL61 was extracted and subcloned into the unique NotI site of pUC9-N. The nucleotide sequence was determined using the Sequenase kit (U. S. Biochemical Corp.), the direct and reverse primers of pUC9 and of pBlueScript (T3 and T7 primers) and specific oligonucleotide sequences belonging to the sequenced gene. After completing the full sequence of one strand, the complementary strand was fully sequenced using a series of specific oligonucleotides as primers.
Reformatting and Cloning ⌬7-red cDNA into Expression Vector pYeDP1/8-2-Deletion of the 5Ј-and 3Ј-non-coding regions of the ⌬7-Red cDNA was performed by PCR amplification using specific primers designed to introduce a BamHI restriction site immediately upstream of the initiation codon and a KpnI site immediately downstream of the stop codon.
Direct primer:

5Ј-cgcggatccATGGCGGAGACTGTACATTC-3Ј
Reverse primer: 5Ј-cagggtaccTCAATAAATTCCCGGAATG-3Ј Sequences identical or complementary to the cDNA are shown as uppercase, and restriction sites are underlined. The ⌬7-red cDNA was amplified using 33 thermal cycles with 2 units of Pfu DNA polymerase (Stratagene) in the presence of 10 pmol of each primer and 0.2 mM of each dNTP, in the recommended buffer. The temperature cycles were 10 s at 94°C, 50 s at 52°C, followed by 1 min 30 s at 74°C. The 1300-bp PCR product was BamHI/KpnI digested and inserted between the BamHI and KpnI sites of pYeDP1/8-2, resulting in plasmid ⌬7red/V8. The integrity of the PCR-amplified fragment was confirmed by sequencing.
Integration of Reformatted ⌬7-Red cDNA into the ADE2 Locus of FY1679-28C Strain-The BglII DNA fragment of the yeast ADE2 gene included in the plasmid pASZ11 (32) was first subcloned into the BamHI site of pBlue-Script, resulting in pBS-ADE2. Primers: 5Ј-agatct-TGAGAAGATGCGGCCAGCAAAAC-3Ј (hybridizing the 3Ј-ends of URA3) 5Ј-GATTACGCCAAGCTTTTCGAAAC-3Ј (hybridizing the 3Ј-end of the PGK terminator) were designed to amplify the full ⌬7-Red/V8 expression cassette including the GAL10-CYC1 promoter, the ⌬7-red cDNA coding sequence and the PGK transcription terminator: 80 ng of ⌬7red/V8 template, 0.5 M phosphorylated primers, 0.2 mM of each dNTP diluted in the commercial buffer were first denatured 1 min at 95°C, after which 1 unit of native Pfu DNA polymerase was added and the reaction mixture cycled for 35-fold using 5 s at 95°C, 30 s at 56°C, and 4 min 30 at 70°C. The amplified 2440-bp fragment was purified and subcloned blunt-end into the unique StuI site of pBS-ADE2, giving pAD⌬7. The 4720-bp NotI-PstI fragment of pAD⌬7 containing the disrupted ADE2 sequence was isolated and used to transform strain FY1679-28C according to previously described methods (30). The resulting yeast strain was called ELR01.
Southern Blot Analysis-Genomic DNAs from A. thaliana, yeast, and different species were restricted overnight in the appropriate buffers. DNA fragments were separated by electrophoresis on a 0.8% agarose gel. DNA was transferred onto nitrocellulose filters (BA85; Schleicher & Schuell). The open reading frame of the ⌬7-red cDNA was isolated from ⌬7-red/V8 on agarose gel and purified with Jetsorb extraction kit (Bioprobe). The hybridization of the 32 P labeled probe was performed at 42°C for 3 days using standard procedures excepted that formamide concentration was reduced from 50 to 30% (by volume) to reduce stringency. Filters were washed using as final conditions 0.1 ϫ SSC, 0.1% SDS, and 37°C before exposition in a PhosphorImager. 2 A. Rahier, personal communication.
Cell Culture and Subcellular Fractionation-Yeast cells transformed with pFL61 and p⌬7red/V8 were grown, respectively, into SGI and SGRI synthetic medium until stationary phase. SGRI is similar to SGI except that the glucose concentration was 5 g/liter. The saturated culture was diluted with an equal volume of YP medium complemented with ethanol (final concentration 1.5% by volume) as carbon source and grown until cell density reached a minimal value of 7 ϫ 10 7 cells/ml (A 600 ϭ 10). ⌬7-Red expression was induced by addition of D-galactose at a final concentration of 20 g/liter. The culture was stopped when density reached 1.5 to 2 ϫ 10 8 cells/ml (A 600 ϭ 20 -30). Alternatively, the gene expression was also induced by culture up to 2 ϫ 10 7 cells/ml (A 600 ϭ 3) in synthetic medium SLI. Harvested yeast cells were broken after spheroplasts preparation by enzymatic digestion of cell wall and subcellular fractionation as described previously (28). Alternatively, the subcellular fractionation was performed by ultracentrifugation. The cells were collected, washed twice in buffer TE-KCl (50 mM Tris-HCl, pH 7.4, 1 mM EDTA, 0.1 M KCl) and suspended in TE, 0.6 M sorbitol. Glass beads (0.45-0.5 mm diameter, Braun) were added until skimming the top of the cell suspension and cell walls were disrupted mechanically by handshaking for 5 min in cold room. The membrane debris, nuclei and mitochondria were pelleted by centrifugation at 20,000 ϫ g, 13 min at 4°C. The ultracentrifugation of the supernatant at 100,000 ϫ g, 1 h at 4°C allows the separation of microsomes, which are pelleted, and cytosol.
Sterol Analysis-Total sterols were prepared by alkaline saponification as described previously (33). The sterol analysis and purification was performed at 55°C on reverse phase HPLC operated on a 100 ϫ 2.1 mm (or 100 ϫ 4.6 mm for purification) reverse phase C 18 column (Applied Biosystems). The column was eluted with a linear gradient of aqueous methanol from 50 to 100% (by volume) at 1.0 ml/min (or 3.0 ml/min for purification). The sterol composition was also analyzed by gas chromatography with a SE30 capillary column (30 m ϫ 0.32 mm, Altech) and helium as carrier gas. The structure of purified sterols from transformed yeast strains was determined based on comparison of their relative retention times to standards in GC and HPLC and further confirmed by GC-mass spectrometry fragmentation and NMR analysis when appropriate.
Enzymatic Assays-For ⌬7-reductase, assays were performed according to a previously described method (14). Subcellular fractions (700 g of protein in 200 l) were incubated for 90 min at 37°C in 100 mM Tris/HCl buffer, pH 7.3, containing 150 M cholesta-⌬5,7-dieneol (7dehydrocholesterol), emulsified with Tween 80 (final concentration 1.5 g/liter) and 2 mM NADPH. The reaction was stopped by addition of an equal volume of methanol-dichloromethane (50:50, by volume). The lower phase was collected and air-dried, and products were dissolved in methanol prior to GC analysis. The authentic products cholesta-5,7dieneol and cholesterol were well separated on the column, and their retention times are quite different from the endogenous yeast sterols.

RESULTS
Isolation of the cDNA Encoding ⌬7-Red Activity-Effects of polyene antifungal like nystatin, amphotericin B, or filipin are critically dependent on the presence of a ⌬7-bond unsaturation (2) as demonstrated by the highly nystatin-resistant phenotype of erg2 mutants, which are deficient in the sterol ⌬8 -7 isomerase gene product. The selection is based on the assumption that such situation could be mimicked in a wild type strain expressing heterologous ⌬7-red activity. Yeast FY1679-28C was transformed with an A. thaliana cDNA library placed under the transcriptional control of the host phosphoglycerate kinase transcription promoter and terminator sequences. Selection of transformants was first performed on the basis of the ura3 mutation complementation by the plasmid-borne marker. A second selection of primary transformants was performed for an increased nystatin resistance. Resistant clones were finally screened individually by HPLC for an alteration of the sterol profile leading to a significant decrease in the proportion of 280 nm-absorbing hexane-extractable material (⌬5,7-dienic system signature) compared with 205 nm-absorbing compounds (any unsaturation) as detailed under "Experimental Procedures." Out of more than 10 5 primary transformants, only one clone F22, which is resistant up to 80 g/ml nystatin, was finally selected. The sterol profile of this strain was examined by HPLC and GC techniques; ergosterol was no longer the major component as observed in the parent strain transformed by a void plasmid (data not shown). Two new major sterols lacking any 280 nm absorption accumulate in similar amounts in F22, representing globally more than 90% (by weight) of the total sterol content based on the GC profile. The sterol profile alteration was found to be plasmid-dependent by transfer of the F22 plasmid into E. coli and back-transformation of the FY1679-28C strain.
Individual sterols from yeast strain back-transformed with pF22 were purified by preparative HPLC (about 8 mg each recovered) and their structure determined by GC-mass spectrometry and Fourier-transform NMR analysis. The more lipophilic sterol (second main peak in HPLC and GC) exhibits a molecular peak at m/e ϭ 400 in GC-MS and was identified as ergosta-5-eneol or dihydrobrassicasterol on the basis of NMR. The compound co-migrates in HPLC and GC with its commercial epimer: campesterol. Detailed comparison of NMR data with the spectra of authentic compounds in the literature is indicative of an "S" absolute configuration at the C-24 position of ergosta-5-eneol (data not shown). The more polar additional sterol (first to elute from HPLC and GC) has a GC-MS molecular peak at 398, and its fragmentation and 1 H NMR spectra are fully consistent with an ergosta-5,22-dieneol structure. This sterol is thus the direct reduction product of ergosterol (ergosta-5,7,22-trieneol) at the C-7 double bond level. Accumulation of these two compounds highly suggests that the plasmid borne cDNA is encoding a sterol ⌬7-red activity. Concomitant accumulation of ergosta-5-eneol with the ergosterol reduction product suggests that reduction of the ⌬-7 bond can also occur in vivo at the level of the biosynthesis intermediate ergosta-5,7-dieneol and/or of its precursor ergosta-5,7,24(28)-trieneol.
Nucleotide Sequence of the A. thaliana ⌬7-Red-The cDNA in pF22 was extracted and subcloned before sequencing on both strands. The cloned cDNA (Fig. 1) without its poly(A) tail is 1488 bp long and contains an open reading frame of 1290 bp, which is coding for a protein of 430 amino acids with a calculated molecular mass of 49,458 Da. The 5Ј-and the 3Ј-untranslated regions are 76 and 121 bp long, respectively. A possible consensus polyadenylation signal is located 34 bp upstream of the poly(A) tail. A computer search on a sequence data base reveals that the deduced amino acid sequence of ⌬7-red exhibits a significant similarity with the ones of ⌬14and ⌬24(28)sterol reductases (Fig. 2), suggesting that all known sterol reductases belong to a single sequence family. These reductases are the S. cerevisiae sterol C-14-reductase (34), the Neurospora crassa sterol C-14-reductase (accession no. X77955 in EMBL data base), the YGL022 open reading frame later identified as the S. cerevisiae sterol ⌬24(28)-reductase (35), and the Schizosaccharomyces pombe SST1 gene product (27). The latter protein product is likely to be a sterol ⌬24(28)-reductase too (36). In addition, the ⌬7-red shows a striking similarity with the 400 C-terminal amino acids of the lamin B receptors from chicken and human as already evidenced for other sterol reductases (37)(38)(39). The N-terminal end of these two proteins contains a typical DNA binding domain (40), absent in all identified sterol reductases including the ⌬7-red one.
Consensus sequences involved in the binding sites of NADPH or NADH and/or flavin have been described in different reductases family like P450 reductases or nitrate reductases (41). However, it was not possible to identify similar motif in the sterol reductase protein family, even with the newly included sequence. In addition detailed sequence comparison between ⌬7-, ⌬14-, and ⌬24(28)/sterol reductases (eight sequences) did not allow identification of a clear sequence signature corresponding to the different regio-specificities for sterol reduction (Fig. 2). Globally, sequence conservation within the family is high in the C-terminal half of the enzymes, with a clear LLXSGWWGXXRH signature almost perfect in all members. In contrast, a more limited sequence similarity is present on the N-terminal half. Particularly the EFGGXXG signature common to ⌬24(28), ⌬14, and lamin receptor is not present in ⌬7-red. Interestingly enough, the hydrophobic profiles remain very similar among all family members even within the Nterminal half (starting residue 440 for lamin B receptors). The lamin B receptor sequence cannot be distinguished from that of other family members either on sequence or on hydrophobic profile criteria.
Southern Blot Analysis of ⌬7-Red Locus-A. thaliana genomic DNA was probed with the open reading frame of ⌬7-red cDNA (Fig. 3). Based on PstI digestion (absent site from the cDNA) yielding a single hybridizing band in low stringency conditions, the presence of a single ⌬7-red gene can be deduced. Absence of overlapping bands was confirmed by double digestion with BamHI. Cleavage by BamHI alone (two bands: c and c1) or in combination with PstI (three fragments: a, a1, and a2) are indicative of the presence of at least one intron in the gene.
PCR applied to genomic DNA using a primer situated at both extremities of the open reading frame led to amplification of a single 3.6-kb fragment confirming the presence of a total of 2.5 kb of intronic sequences within the open reading frame of a unique gene. Digestion with PvuII, which cuts once within the cDNA, gave rise as expected to two hybridizing bands (e1 and e). In a second experiment, genomic DNAs of different origins (human, quail, Drosophila melanogaster, Xenopus laevis, maize, and yeast) were tested (Fig. 3B). EcoRI-restricted DNA from parental yeast FY1679-28C exhibits three weak bands at 4.2, 2.5, and 2.3 kb upon low stringency hybridization, which could correspond to endogenous ⌬14 and ⌬24(28)-reductase genes. As a control, strain ELR01 (see later), which contains an expression cassette for A. thaliana ⌬7-red integrated within the yeast genome, was tested. The strong hybridization signals corresponding to the two expected EcoRI fragments were observed in addition to two weak signals also found with the parental strain. Interestingly enough, a well defined hybridization signal was found with quail DNA.This hybridization signal was absent under high stringency conditions, suggesting de- tectable but limited interspecies sequence conservation. Weak but defined signals were also found with maize, but not with human, X. laevis, and D. melanogaster DNAs, thus illustrating the limits of interspecies cross-hybridization approaches.
Overexpression of ⌬7-Red and Time Course of in Vivo Sterol Conversion-To optimize expression, the cDNA encoding the ⌬7-red was reformatted and cloned into pYeDP1/8-2, placing the flanking sequence-free open reading frame under the transcriptional control of a galactose-inducible GAL10-CYC1 promoter. FY1679-28C cells were transformed by the resulting pV8/⌬7red. Cells were first grown on glucose-repressed conditions in which the plasmid-borne ⌬7-red composite gene is silent. Following the transfer in inducing culture conditions, the time course of the changes in the yeast cell sterol composition was followed (Fig. 4). As expected, during growth in glucose or in ethanol, the main sterol is ergosterol and ergosta-5,22-dieneol and 5-eneol are hardly (ethanol) or not at all (glucose) detectable. Following the addition of galactose, ergosterol content rapidly decreases with a concomitant increase in ergosta-5,22-dieneol content. A very limited formation of ergosta-5-eneol occurs during the first 2 h following induction. This compound nevertheless slowly accumulates with increas-ing induction times (up to 9 h), while ergosta-5,22-dieneol content remains constant and then slowly decreases. At the end of the culture, ergosterol accounts for 5% (w/w) of the total sterols, ergosta-5-eneol for 45% (w/w), and ergosta-5,22-dieneol for 50% (w/w). The proportion of the other sterol intermediates could be estimated to less than 10% of the total sterol based on GC analysis. This indicates that in vivo accumulation of ergosta-5,22-dieneol results from the direct reduction of previously accumulated ergosterol. In contrast, accumulation of ergosta-5-eneol requires de novo sterol biosynthesis, and likely results from reduction of the biosynthetic intermediate ergosta-5,7dieneol (Fig. 5). The decrease in ergosta-5,22-dieneol content during the late induction phase suggests that ergosta-5-eneol is no longer a good substrate for the yeast ⌬-22 desaturase enzyme.
In Vivo Analysis of ⌬7-Red Substrate Specificity-To investigate in more detail substrate specificity of the ⌬7-red, mutant strains PLC 1051, 1451, and 1061, which accumulate sterol biosynthesis intermediates were transformed with pV8/⌬7red. Main sterols accumulated by yeast mutants expressing or not the ⌬7-red activity are listed in Table I. In the ⌬22-desaturasedeficient strain PLC1051, ⌬7-red expression results mainly in FIG. 2. Sequence alignment of the sterol reductase family. Alignment was performed using the PILEUP program of the UGCG package run with the default parameters. Positions with a consensus residues present in five or more sequences are indicated in bold uppercase letters. 2428sc and sst1 stand, respectively, for the ⌬24(28)-sterol reductases from yeast S. cerevisiae and S. pombe; hlr440 and clr440 stand, respectively, for the human and chicken lamin receptor sequences starting at residues 440; 14str, 14strpb, and nc-14red stand, respectively, for the S. cerevisiae, the S. pombe, and the N. crassa sterol ⌬14-reductases. This last sequence is probably incomplete on the N-terminal side based on alignments. The ⌬7-red sequence is labeled d7red.
In Vitro Analysis of ⌬7-Red Enzymatic Properties-The subcellular location of the ⌬7-red was analyzed using the cholesta-5,7-dieneol as substrate since cholesterol, the expected reduction product, is absent from yeast and well resolved from endogenous sterols. GC profiles of sterols were examined after incubation of cholesta-5,7-dieneol and NAPDH with microsomal fractions from the FY1679-28C strain expressing ⌬7-red. Two peaks corresponding to the residual substrate (cholesta-5,7-dieneol) and to the cholesterol formed are present and well separated from endogenous sterols. The cholesterol peak was found to be absent when either cholesta-5,7-dieneol or NADPH were omitted or when microsomes from a yeast transformed with a void plasmid were used. In addition, a negative result was also obtained when NADH was substituted to NADPH. In our hands, the microsomal fractions exhibits the highest specific activity (versus protein content) but some activity was also found in lipid droplets and cytosol, suggesting a rather diffuse subcellular location of the enzyme. The activity of ⌬7-red toward sterol esters was tested using sterol acetate as a model.  PLC1051 strain transformed with V8/⌬7-red. Both esters were found to be efficiently reduced at C-7 upon incubation with the cytosolic fractions. 7-Dehydrocholesterol ester was clearly a better substrate than the ergosterol ester (45% versus 15% of conversion). Similar experiments with microsomal fractions led surprisingly to the fast hydrolysis of steryl acetate by some endogenous esterase, a reaction that was absent when cytosolic fractions were used. This indicates that a free hydroxyl group at the C-3 position is not required for activity of ⌬7-red and that fatty acid steryl esters might be physiological substrates of this enzyme. Genome Integration of the ⌬7-Red Expression Cassette and Effects on Cell Viability-Genomic integration of the ⌬7-red expression cassette opened the way to add additional heterologous activities in engineered cells and allowed us a better analysis of the physiological effects of ⌬7 reduction of sterols. The cassette containing the GAL10-CYC1 promoter, the ⌬7-red open reading frame, and the PGK terminator was extracted from pV8/⌬7-red by PCR, and the resulting fragment was inserted within a plasmid containing the yeast ADE2 gene (see "Experimental Procedures"). FY1679-28C cells were transformed by the interrupted ADE2 sequence, and homologous recombination events were selected on the basis of the generation of ade2 clones. PCR and Southern blotting analysis (Fig.  3B) of clone ELR01 confirmed that the full ⌬7-red expression cassette had been integrated within the ADE2 genomic locus. Galactose induction of the ⌬7-red expression induced a dramatic change in sterol composition as was observed with the pV8/⌬7-red transformed strain, except for an even lower level of residual ergosterol. A single integrated copy of the expression cassette is thus sufficient to completely reduce in vivo ⌬7-sterols. The ELR01 doubling time and the final cell density at saturation were found to be similar to those of the parental strain under similar culture conditions. Cell viability was analyzed and indicated in both cases that more that 90% of cells in exponential growth phase were viable. Together, these results indicate that ⌬7-reduced sterols support yeast growth and cell viability as well as unsaturated sterols under tested conditions. DISCUSSION The quantitative and qualitative sterol compositions and the presence of double bonds or branched groups are known to modulate membrane properties. This aspect has been well documented in fungi and especially in the yeast S. cerevisiae since several mutant strains are available (26,42). In a Candida albicans mutant with an inactive lanosterol C-14 demethylase, the accumulation of 14␣-methyl sterols is responsible for a more rigid membrane (43). Similarly, alteration of membrane properties in erg mutants is suggested by a constantly lowered efficiency of transformation by plasmids. In the case of the erg6, a defect in tryptophan transport and increased drug sensitivity was also documented (44). As a compensation mechanism, the fatty acid composition of membranes could be adapted to balance the changes in membrane properties induced by accumulation of abnormal sterols (45). In addition to the "bulk" effect, ergosterol also plays a role in the control of the yeast cell cycle. Sterols that can satisfy this "sparking" effect must possess a double bond at C-5. Unsaturation at C-22 facilitates the sparking better than the C-7 unsaturation, and the 24␤-methyl group seems to be of low importance (19).
The double bond at C-7, while not important for the sparking effect, is critical for polyene antibiotic sensitivity, which also depends on others factors, including phospholipid composition and presence of a suitable membrane potential (46). The particularly high level of resistance of the erg2 mutant (deficient in sterol ⌬8-⌬7 isomerase) confirms that the presence of ⌬5,7-conjugated double bond is determining. Consistently, FY1679-28C expressing ⌬7-red is highly nystatin-resistant (up to 80 g/ml compared with 2 g/ml for parental FY1679). This resistance correlates with accumulation of almost 95% of C-7saturated sterols. Upon induction, ⌬7-red rapidly converts the whole ergosterol pool, including steryl esters to the corresponding reduction products. This observation supports the finding that esterified sterols could be in vivo substrates for the enzyme. The low content of free sterols within the microsomal membranes in comparison to the value into the plasma membrane is generally explained by a vectorial transport process involving the synthesis of steryl esters in microsomes and their hydrolysis by a specific lipase at the plasma membrane (47). Steryl esters stored into the lipid droplets and the free sterols within the plasma membrane might rapidly interconvert thus explaining reduction of the whole ergosterol pool by ⌬7-red.
The deduced amino acid sequence of the A. thaliana ⌬7-red demonstrates that the enzyme belongs to the same sequence and consequently structural family as the ⌬14and ⌬24(28)reductases. Of particular interest is the confirmation than lamin B receptor, a nuclear membrane protein featuring a N-terminal DNA binding domain, is a member of the sterol reductase family. A close examination of the sequence alignment demonstrates that all highly conserved regions among ⌬7-, ⌬14-, and ⌬24(28)-reductases are also conserved in lamin B receptors. This puzzling observation strongly suggests that lamin B receptor may actually have a sterol reductase activity in addition to its DNA binding role. Chicken lamin B receptor was expressed in a sterol C-14-reductase-deficient yeast and did not complement the mutation. In addition, no equivalent of lamin B receptor has been found in yeast to date on the basis of the systematic sequence analysis. Yeast has no need for ⌬24reductase activity, a step specifically required for cholesterol biosynthesis. We thus propose that lamin B receptor might be the cholesta-5,24-dieneol ⌬24-reductase, the last missing member of the sterol reductase family.