INTRODUCTION

The biosynthesis pathway that is required to transform lanosterol into the major plasma membrane sterol, namely ergosterol and cholesterol, respectively, diverges in fungi and in mammals. Nevertheless this pathway constantly involves the isomerization of the B-ring double bond from the 8(9) to the 7(8) position in the sterol molecule. The enzyme responsible for this activity, namely 8-7 sterol isomerase, is membrane-bound and located in the endoplasmic reticulum. In mammals, this enzyme catalyzes the conversion of 5-cholesta-8,24-dien-3-ol (zymosterol) and 5-cholesta-8-en-3-ol (zymostenol, 8-cholestenol) to their corresponding 7-isomers (1, 2, 3). Recent studies have provided evidence in support of a regulatory role for this enzyme in controlling the overall rate of cholesterol biosynthesis (3). The molecular mass of sterol isomerase purified from rat liver microsomes was estimated at 80 kDa (1, 2, 3). In SDS-polyacrylamide gel electrophoresis, this purified enzyme migrates as one polypeptide exhibiting an apparent molecular mass of 21 kDa, suggesting that rat liver sterol isomerase is composed of four identical subunits. In spite of these results, no sequence data of any high eukaryote sterol isomerase have been available so far.

In contrast, the sterol isomerase-encoding ERG2 genes from the bakers' yeast Saccharomyces cerevisiae (4, 5), the rice blast fungus Magnaportae grisea (6), and the maize smut pathogen Ustilago maydis (7) have been isolated. Each of these three fungal genes encodes a polypeptide with a molecular mass ranging from 24 to 27 kDa depending on the species. Genetic analysis of S. cerevisiae erg2 mutants revealed several cases of intragenic complementation, suggesting that yeast sterol isomerase is also made of at least two identical subunits (8). All the three fungal ERG2 gene products share significantly similar sequences and present three common hydrophobic domains (6). A particularly high degree of identity could be observed within the central hydrophobic region in all three proteins, and it was suggested that this domain could correspond to the catalytic site of the isomerase. We had tried to take advantage of this possible sterol isomerase signature to isolate a similar sequence from a mammalian source by polymerase chain reaction or reverse transcription-polymerase chain reaction but without any success so far. We therefore undertook a completely different selection approach based on the complementation of the erg2 gene defect in yeast.

Successful functional complementation by expressing sterol biosynthesis enzyme-encoding cDNA in yeast mutants has already been achieved by others. For instance, rat squalene epoxidase and human lanosterol synthase complement the corresponding defects in Schizosaccharomyces pombe (9) and in S. cerevisiae (10), respectively. However, these genes correspond to enzymes of the sterol pathway section that is conserved from fungi to animals, and the corresponding enzymes share significant similarities in both phyla (9, 10). Our report is the first one on successfully applying such a strategy to isolate a mammalian sequence that encodes an enzyme of the post-lanosterol sterol biosynthesis pathway. We show in this report that this mammalian sequence corresponds to EBP,¹ a high affinity binding protein for the anti-ischemic phenylalkylamine Ca²⁺ antagonist [³H]emopamil and the photoaffinity label [³H]azidopamil (11, 12, 13, 14). EBP is a protein of unknown function that is sublocalized in the endoplasmic reticulum (11). EBP is expressed in many tissues and specially abundant in liver (14). A wide series of structurally diverse compounds with anti-ischemic effects in animal models of cerebral ischemia inhibit [³H]emopamil binding to EBP with high affinity. EBP might therefore represent a molecular target of anti-ischemic drug action (14). Although the primary structure of EBP is strikingly dissimilar to those of the fungal ERG2 gene products, our results indicate that EBP and yeast sterol isomerase (SI)¹ are functionally interchangeable in yeast and strongly suggest that EBP and SI are the same protein in mammals.

EXPERIMENTAL PROCEDURES

EMY47 (MAT, ura3, trp1-4, fen1::LEU2, erg2::TRP1) and EMY43 (MAT, ura3, trp1-4, erg2::TRP1) were isogenic derivatives of S. cerevisiae wt FL100 (ATCC 28383). Yeast culture media were YPD medium and synthetic minimal (SD) medium containing 2% glucose and adequately supplemented to fulfill the strain's auxotrophic requirement (32). Anaerobic conditions were obtained using an anaerobic glove box (La Calhène, Vélizy, France) under N₂-H₂-CO₂ (85%-10%-5%) atmosphere. For anaerobic growth, media were supplemented with 0.1% Tween 80 and 50 mg/liter ergosterol. The effect of drugs on transformed cell proliferation was tested as follows. Transformants were selected in the absence of oxygen and grown for 24 h in ergosterol-free selective medium in the presence of oxygen. 5 µl of each cell suspension (approximately 10⁵ cells) were plated onto medium containing drugs at different concentrations, as indicated, and plates were incubated at 30 °C for 24 h. Haloperidol and trifluoperazine were obtained from Sigma Chimie (Paris, France). Fenpropimorph was a gift from F. Karst.

Genetic engineering techniques and yeast DNA extraction were as described (32, 34). The mouse cDNA library was cloned into the BamHI-PstI sites of pEMR1023 (15), using size selected (>1 kilobase pair) poly(A)⁺ RNA derived cDNA from the mouse myeloblast cell line M1 (ATCC TIB 192). Library DNA was used to transform cells of S. cerevisiae EMY47 to Ura⁺, as described (33). 25,000 Ura⁺ transformants were pooled. About 10⁷ cells were spread onto YPD medium plates (32) containing 10 mg/liter cycloheximide; 60 cells were able to form colonies on that medium. Plasmid DNA was extracted from all these clones, amplified in Escherichia coli RR1 cells (Life Technologies, Inc.), and used to transform aerobically growing EMY47 cells. Plasmid DNA was extracted from 2 Ura⁺ cycloheximide-resistant transformants and analyzed by restriction mapping directly on double-stranded DNA. One insert was sequenced using the dideoxynucleotide chain termination method directly on denatured double-stranded DNA. The sequence was later on verified on DNA obtained by reverse transcription-polymerase chain reaction amplification. The human cDNA library was constructed into the pTZ18R vector (Pharmacia Biotech Inc.), using poly(A)⁺ mRNA derived cDNA from human cell line U937 (ATCC CRL 1593). Sequence analyses were performed using the University of Wisconsin Genetics Computer Group package (35). The percentage of similarity and pairwise homologies were assessed using the BestFit software. Data bases searches were conducted against GenBank^TM and European Molecular Biology Laboratory releases.

pEMR1023 is a multicopy plasmid, the structure and construction of which are fully described elsewhere (15). In summary, pEMR1023 is a S. cerevisiae-E. coli shuttle vector that contains URA3 as the selectable marker, a yeast 2-µm plasmid fragment encompassing the ARS and STB sequences, and an empty expression cassette containing the yeast PGK gene promoter and terminator. Plasmids pEMR1223, pEMR1235, and pEMR1336 are derived from pEMR1023 by the integration of a sterol isomerase-encoding sequence into the expression cassette, as indicated in Table I. Plasmids pEMR1349, pEMR1348, and pEMR1200 are centromeric plasmids that derive from pFL38 (ATCC77203) by the integration of an expression cassette into the polylinker, as indicated in Table I. This cassette is composed of the ERG2 promoter region (positions 295 to 1 of M74037 in GenBank^TM) followed by the ERG2 terminator region cloned as already described (15). Plasmid pEMR1292 is a pFL38-derived centromeric vector containing the mSI-expression cassette under the control of the yeast PGK gene promoter and terminator. Centromeric plasmids are normally maintained at about 1 copy per haploid genome in yeast; this copy number may increase slightly under selective conditions. The mSI sequence as cloned in pEMR1235, pEMR1292, and pEMR1349 corresponds to positions 3 to +703 with respect to the first base of the initiation triplet. The hSI open reading frame as cloned in pEMR1348 and pEMR1336 corresponds to positions +1 to +692 of sequence Z37986 in GenBank^TM. pEMR1223 contains a full-length mSI cDNA clone as described in the text. The mSI cDNA sequence has been submitted to the EBI data base (accession number X97755[GenBank]). All the cDNA fragments cloned in this study were sequenced to ensure the absence of mutations.

Table I. Colony formation efficiency of EMY43 cells transformed by various plasmids in the presence and the absence of oxygen For each transformant, the same quantity of cells was plated onto two dishes containing ergosterol-supplemented growth medium and incubated at 30 °C for 48 h in the presence and the absence of oxygen, respectively. Colonies were numbered to estimate the percentage of cells able to form colonies aerobically versus anaerobically. Experiments 1 and 2 (Exp1 and Exp2) are two independent experiments. Plasmids are as described under ``Experimental Procedures.'' 2-µm and CEN mean 2-µm plasmid-derived and centromeric vectors, respectively. Plasmid (vector/promoter and terminator/SI coding sequence) Number of colonies Cells able to form colonies in the presence of oxygen +O2  O2 % pEMR1023 (2-µm/PGK/none) Exp1 0 1494 0 Exp2 0 >1500 0 pEMR1223 (2-µm/PGK/original mSI cDNA) Exp1 891 1876 47 Exp2 389 997 39 pEMR1235 (2-µm/PGK/mSI) Exp1 976 1205 81 Exp2 707 765 93 pEMR1349 (CEN/ERG2/mSI) Exp1 269 1800 15 Exp2 70 1848 4 pEMR1336 (2-µm/PGK/hEBP) Exp1 1155 1212 95 Exp2 1835 2018 91 pEMR1348 (CEN/ERG2/hEBP) Exp1 1855 1982 94 pEMR1200 (CEN/ERG2/ySI) Exp1 968 955 101 Exp2 777 874 89

Sterols were extracted from lyophilized cells in the presence of a constant amount of cholesterol (50 µg/50 mg dry weight) as already described (18). Sterols were analyzed by gas chromatography with a Varian 3300 chromatograph, using a 30-m DB-1 column (inner diameter, 0.312 mm), a Ross injector, carrier gas helium 3 ml/min, column oven temperature programmed from 200 to 250 to 300 °C at 10 °C/min and 2 °C/mn, respectively. Detection was obtained by a flame ionization detector with a detector oven temperature of 300 °C. Sterols were compared with cholesterol by the area method (18, 15)

8-7 Sterol Isomerase Assay

Microsomes were prepared and incubated for 3 h in the presence of 75 µM cholest-8-en-3-ol as already described (20, 15). Sterols were extracted and purified from the reaction mixture (15), an aliquot was injected into an analytical C18 ultrasphere 5-µm column with a mobile phase of methanol:water (99.7:0.3) (v/v) at a flow rate of 0.7 ml/min. at room temperature.

RESULTS

In the presence of oxygen, all yeast erg2 gene disruptants produce 8-sterol molecules that characterize SI deficiency (5, 15). However, depending on the genetic background (15), some of these strains, like strain EMY47, are capable of proliferating under standard conditions of growth, whereas others, like strain EMY43, proliferate only anaerobically on ergosterol-supplemented medium. EMY43 cells being not easily transformable, we rather used EMY47 as the recipient strain to isolate the sterol isomerase-encoding cDNA. The accumulation of abnormal 8-sterol molecules and the absence of ergosterol in plasma membranes render EMY47 cells hypersensitive to various drugs like cycloheximide (5). Active sterol isomerase production obtained by expressing a foreign cDNA in EMY47 cells was thus expected to confer a relative cycloheximide resistance by restoring ergosterol biosynthesis. A cDNA library prepared from mRNA extracted from murine myeloblast M1 cells was cloned under the control of the strong PGK promoter into a yeast multicopy expression vector. Library DNA was used to transform EMY47 cells to uracil prototrophy and relative cycloheximide resistance. One plasmid, denoted pEMR1223, was found to confer both traits. The cDNA insert (1106 base pairs) of pEMR1223 was subcloned and entirely sequenced. This cDNA contained a large open reading frame corresponding to a membrane polypeptide of 230 amino acids (Fig. 1). A polyadenylation signal was present 260 nucleotides downstream from the stop codon. The start codon was preceded by a 135-base pair-long sequence that notably contained an out-of-frame ATG at position 31, a feature expected to exert negative effects on gene expression in yeast (16). Northern blot analyses of RNAs extracted from M1 cells using the cDNA insert as a probe revealed the presence of an RNA of the expected size (approximately 1.1 kilobases, data not shown). The cDNA translation product ends with a C-terminal sequence, KSKHN, which fits the consensus retrieval signal (K(X)KXX, where X is any amino acid) known to anchor type I integral membrane protein into the endoplasmic reticulum (17). Surprisingly, a search for similarities in protein and DNA data bases (SwissProt and GenBank^TM) revealed that the murine gene product (mSI) was 78% identical to human EBP (hEBP) (Fig. 1B), a phenylalkylamine Ca²⁺ antagonist binding protein, but failed to reveal any striking similarities with other known sequences, including yeast sterol isomerase (ySI, 13% identity, 40% similarity) (Fig. 1A). The hydropathy profile of the murine protein corresponded to that of EBP (14), exhibiting four putative transmembrane domains, but did not fit the three-hydrophobic domain model proposed for fungal SI (6).

EMY43 is an ERG2 gene disruptant that is unable to proliferate in ergosterol-free medium. In the presence of oxygen, EMY43 cells cannot utilize exogenously supplied ergosterol, so this strain proliferates only anaerobically on ergosterol-supplemented medium. pEMR1223, the original cDNA expressing vector, was used to transform EMY43 cells to Ura⁺ under strictly anaerobic conditions. The presence of the plasmid restored aerobic viability in EMY43 (Table I). GC analyses confirmed that ergosterol synthesis was indeed restored in EMY43 (pEMR1223) cells (Fig. 2). However, ergosterol was produced at a lower level in transformed cells than in wt, representing 28% of total sterols, whereas the 8-sterols that characterized sterol isomerase deficiency were still present at high levels. We constructed pEMR1235, a plasmid that differed from pEMR1223 only by the absence of the untranslated sequences at both sides of the cDNA open reading frame. EMY43 cells transformed to Ura⁺ by this optimized construct were able to proliferate aerobically and produced ergosterol and 8-sterols at wt levels (Fig. 2). The optimized expression cassette was cloned into a centromeric vector. EMY43 cells transformed with the resulting plasmid, denoted pEMR1292, produced ergosterol at a level 2-fold lower than with pEMR1235. Using cholest-8-en-3-ol as the substrate, SI activity was detected in microsomes prepared from EMY43 (pEMR1235) cells, whereas this activity lacked in the negative control, as expected (Fig. 3).

High pressure liquid chromatography assay for 8-7-sterol isomerase.

Because the primary sequence of the mSI was 78% identical to hEBP, it was therefore reasonable to assume that these proteins were functionally homologous. hEBP cDNA was cloned under the control of the PGK promoter in a high copy number plasmid to yield pEMR1336. EMY43 cells transformed by this vector fully recovered the capability of forming colonies in the presence of oxygen (Table I). Moreover, these transformed cells produced ergosterol and 8-sterols at approximately the same respective levels as found in the yeast sterol isomerase-producing control (pEMR1200, see Fig. 2). To ensure that the sterol isomerase activity exhibited by hEBP was really significant and not an artifact linked to high level production, we cloned the hEBP coding sequence between the ERG2 promoter and terminator in a low copy centromeric plasmid to yield pEMR1348. EMY43 cells transformed by pEMR1348 produced ergosterol at a level corresponding to approximately 30% of total sterol (Fig. 2). This ergosterol amount was sufficient to restore an aerobic viability phenotype close to wt in EMY43 cells (Table I). In contrast, substituting the hEBP coding sequence by that of mSI (pEMR1349) in this centromeric vector resulted in a dramatic decrease in ergosterol production, which was not high enough to ensure correct aerobic viability (Table I). Approximately 10% of EMY43 (pEMR1349) cells could form colonies in the presence of oxygen, possibly as the result of a discrete amplification of the plasmid copy number leading to higher cDNA expression. In these transformed cells, ergosterol production was approximately 10-fold lower than that of the wt control. Western blot analyses of mSI and hEBP expressed in yeast indicated an approximately 5-fold overproduction of the human enzyme as compared with the murine one (not shown). Such a difference in production could explain why complementation was more readily achieved with hEBP than with mSI.

Because EMY43 cells are strictly dependent on recombinant sterol isomerase activity to proliferate aerobically, these cells can be used to detect drugs that inhibit sterol isomerase (Fig. 4). As compared with the wt control EMY43 (pEMR1200), the mammalian SI-producing strains were found to be hypersensitive to the high affinity EBP ligand trifluoperazine but resistant to the low affinity EBP ligand haloperidol and to the fungal sterol biosynthesis inhibitor fenpropimorph. Sterol analyses of drug-treated cells indicated that trifluoperazine inhibited sterol biosynthesis at the SI step both in the yeast cells that were specifically producing hEBP or mSI (Fig. 5) and in animal cells,² but not in the yeast sterol isomerase producing cells (Fig. 5). The three sterol isomerases were assayed in the absence or in the presence of trifluoroperazine (Fig. 6). As expected, only the mammalian enzymes were inhibited by this compound. On the contrary, haloperidol was found to inhibit SI activity in ySI-producing cells, but not in hEBP- or mSI-producing cells, which did not exhibit any significant decrease in ergosterol production in the presence of the drug (Fig. 5). It is worth noting that haloperidol was nevertheless found to enhance the ergosta-8,14-dienol (ignosterol) peak in all the strains tested, which pointed to an inhibitory effect exerted on the yeast ERG24 gene product, namely C-14 sterol reductase (18, 19). Because ERG24 is also an essential gene for aerobic proliferation in yeast (19), C-14 sterol reductase inhibition is likely responsible for the anti-proliferation effect observed with relatively high concentrations of haloperidol on mammalian SI-producing cells. Fenpropimorph is a powerful inhibitor of both sterol isomerase and C-14 sterol reductase in yeast (18). Although this antifungal agent was found to provoke the accumulation of 8-sterols in all the strains tested, this effect was still much more efficient when exerted on the yeast enzyme (Fig. 5).

DISCUSSION

We have isolated a murine cDNA, the expression of which was found to restore the 8-7 isomerization step required for ergosterol biosynthesis in a yeast mutant impaired in the sterol isomerase-encoding ERG2 gene. This cDNA was found to encode the murine homologue of hEBP. Several arguments strongly suggest that EBP and SI are identical proteins in mammals. First, exchanging the yeast ERG2 coding sequence cloned under the control of its own promoter in a centromeric plasmid with that of hEBP was sufficient to restore a phenotype close to wt in EMY43 cells with regard to both sterol production and formation of aerobically grown colonies. Thus, the sterol isomerase activity associated with EBP can be considered as really significant. Second, the high affinity EBP ligand trifluoperazine selectively inhibits the EBP-induced sterol isomerization reaction, both in vivo and in vitro, whereas fenpropimorph, a powerful yeast SI competitive inhibitor (18, 20) is not very efficient in inhibiting the EBP-induced isomerization reaction. It is worth noting that the sterol isomerization reaction is inhibited by trifluoperazine in animal cells as well. These two observations strongly suggest that the sterol isomerization reaction is performed by EBP itself and not by an hypothetical endogenous cryptic catalytic subunit that would need EBP to function in erg2 mutants. The subcellular localization of EBP is the endoplasmic reticulum membrane (11, 14), as expected for SI. Moreover, the tissue distribution of EBP mRNAs, being ubiquitous but specially abundant in liver, matches that expected for a cholesterol biosynthesis enzyme. Finally, rat liver SI migrates in SDS-polyacrylamide gel electrophoresis with an apparent molecular mass (21 kDa, see Ref. 3) similar to that estimated by others for EBP from guinea pig liver and recombinant hEBP (22.5 kDa, see Ref. 14) as well as for mSI expressed in yeast (22 kDa, data not shown).

Neither EBP nor fungal SI share significant similarities with any other known protein of the data bases. The lack of similarity between EBP and ySI suggests the absence of phylogenetic relationship between both genes, which is not completely surprising because the segment of the sterol biosynthesis pathway that goes from lanosterol to cholesterol and ergosterol, respectively, is divergent in mammals and fungi.

EBP was first described as a high affinity binding protein for the phenylalkylamine Ca²⁺ antagonist [³H]emopamil and the photoaffinity label [³H]azidopamil. A series of structurally diverse sigma ligands that exerted anti-ischemic effects in animal models of cerebral ischemia inhibited [³H]emopamil binding to EBP with high affinity, which suggested that EBP was a sigma site-carrying protein that represented a molecular target of anti-ischemic drug action (11, 12, 13, 14). The so-called sigma receptors of unknown function bind a series of psychotropic drugs and are present in high density in the central nervous system and in peripheral organs (21). The role of sigma ligands as neuroprotective agents has been extensively investigated for the past 2 years (22, 23, 24, 25, 26, 27). These ligands appear to indirectly attenuate the receptor gated N-methyl-D-aspartic acid (commonly NMDA) calcium channels. This effect is thought to be exerted through sigma binding sites, although the so-called sigma receptors and the calcium channels are not co-localized subcellularly. It is worth noting that fenpropimorph itself was recently described as a sigma ligand exerting long term neuroprotective properties in glutamate neurotoxicity models (25). On the other hand, fenpropimorph was found to inhibit cholesterol biosynthesis in 3T3 cells (28), its main target appearing to be lanosterol demethylase. This inhibitory effect on an upstream enzyme might expectedly have masked the one on SI that we have clearly observed using recombinant yeast strains devoid of endogenous SI (see Fig. 5). Haloperidol, a typical sigma ligand, has never been reported to inhibit cholesterol biosynthesis so far. However, our results indicate that this compound is indeed a sterol biosynthesis inhibitor, at least in yeast. Therefore, one has to consider the possibility that inhibiting sterol biosynthesis at a post-lanosterol step might facilitate long term neuroprotection against glutamate toxicity. For instance, the accumulation of atypical sterols into the membrane lipid bilayer might indirectly attenuate glutamate-induced changes in calcium dynamics through altering membrane physicochemical properties. However, it is not yet clear if all the EBP ligands that display long term neuroprotective activities are indeed sterol biosynthesis inhibitors in mammals and, conversely, if all the SI inhibitors are neuroprotective drugs. The recently postulated role of EBP in neuroprotection (14), the suggestion that SI, alias EBP, could be the first identified member of the enigmatic sigma receptor family, and the hypolipidemiant properties of SI inhibitors (29, 30, 31) are all observations that point to SI as an enzyme whose medical importance might have been largely underestimated.