Characterization of Recombinant Adrenodoxin Reductase Homologue (Arh1p) from Yeast

The mammalian electron transfer chain of mitochondrial cytochrome P450 forms involved in steroidogenesis includes very specific proteins, namely adrenodoxin reductase and adrenodoxin. Adrenodoxin reductase transfers electrons from NADPH to adrenodoxin, which subsequently donates them to the cytochrome P450 forms. The Saccharomyces cerevisiae ARH1 gene product (Arh1p) presents homology to mammalian adrenodoxin reductase. We demonstrate the capacity of recombinant Arh1p, made inEscherichia coli, to substitute for its mammalian homologue in ferricyanide, cytochrome c reduction, and, more importantly, in vitro 11β-hydroxylase assays. Electrons could be transferred from NADPH and NADH as measured in the cytochrome c reduction assay. ApparentK m values were determined to be 0.5, 0.6, and 0.1 μm for NADPH, NADH, and bovine adrenodoxin, respectively. These values differ slightly from those of mammalian adrenodoxin reductase, except for NADH, which is a very poor electron donor to the mammalian protein. Subcellular fractionation studies have localized Arh1p to the inner membrane of yeast mitochondria. The biological function of Arh1p remains unknown, and to date, no mitochondrial cytochrome P450 has been identified. ARH1 is, however, essential for yeast viability because an ARH1 gene disruption is lethal not only in aerobic growth conditions but also, surprisingly enough, during fermentation.

In mammalian cells, cytochrome P450-containing monooxygenase systems are usually classified in two main categories: the mitochondrial and the microsomal types (1,2). In both cases, electron transfer chains are involved in which electrons flow from an electron donor, NADPH, to a terminal electron acceptor, cytochrome P450. In the microsomal system, NADPH-cytochrome P450 reductase transfers electrons from NADPH as the reducing equivalent of microsomal cytochrome P450 forms. This protein of about 80 kDa contains FAD 1 and flavin mononucleotide prosthetic groups (3, 4). The electron transport machinery of mitochondrial cytochrome P450 forms is different from that of microsomal cytochrome P450 because it consists of two proteins, NADPH-ferredoxin reductase, a 50-kDa flavoprotein attached to the inner membrane of mitochondria, and a ferredoxin, a 12-14-kDa soluble protein of the mitochondrial matrix. These two proteins contain one prosthetic group per molecule, FAD and an iron-sulfur cluster (2Fe-2S), respectively (2,5). Electrons flow from NADPH to NADPH-ferredoxin reductase and are transferred to the ferredoxin. The latter donates electrons to the cytochrome P450 that catalyzes the enzymatic reaction.
Mitochondrial cytochrome P450-containing monooxygenase systems are primarily involved in the hydroxylation of bile acids and in vitamin D and steroid biosynthesis. In the latter case, NADPH-adrenodoxin reductase (ADR) and adrenodoxin (ADX), corresponding to NADPH-ferredoxin reductase and ferredoxin, respectively, form the electron transfer chain of mitochondrial cytochrome P450 forms. Unlike the microsomal NADPH-cytochrome P450 reductase, which is able to transfer electrons in a wide variety of reduction reactions, including cytochrome P450 forms, cytochrome b5, and lipid peroxidations, ADR and ADX constitute a very specific electron transfer chain, supporting the activity of only three mitochondrial cytochrome P450 forms: CYP11A1 (P450SCC), CYP11B1 (P45011␤), and CYP11B2 (P450aldo) (6). All of these proteins, ADR, ADX, P450SCC, P450aldo, and P45011␤, are produced mainly in adrenals, gonads, kidneys, liver, and brain of mammals. The existence of mitochondrial cytochrome P450 forms and the related electron transfer chain has not been detected in lower eucaryotes, such as Saccharomyces cerevisiae, either by direct mitochondrial detection or by protein sequence comparison.
Recently, we successfully produced a recombinant version of the bovine cytochrome P45011␤ in yeast mitochondria and demonstrated that co-expression of bovine cDNAs encoding P45011␤ and ADX in yeast mitochondria was sufficient to observe in vivo conversion of 11-deoxycortisol into cortisol (7). This observation strongly suggests the existence of a yeast protein capable of transferring electrons to recombinant bovine ADX in vivo, thus supporting the activity of the cytochrome P45011␤. A yeast gene, namely ARH1, encoding a protein homologous to mammalian ADR was identified and cloned (8,9). The encoded protein, Arh1p, shares 36 and 37% of identical amino acids with human and bovine ADR, respectively, and presents the putative FAD and NADPH binding domains of mammalian ADR (10). The similarity between mammalian ADR and yeast Arh1p is particularly significant in the aminoterminal region, where 45% of the amino acids are identical (9). Strikingly, this region is believed to contain the ADX binding domain together with FAD and NADPH binding sites (11). We postulated that Arh1p is a protein capable of forming an electron transfer chain with bovine ADX to support the in vivo activity of the cytochrome P45011␤ produced in yeast mitochondria.
In this article, several characteristics of Arh1p are presented. In order to investigate the biochemical properties of Arh1p, an active recombinant form of the protein, r-Arh1p, was generated in Escherichia coli using the classical glutathione S-transferase module. The ability of Arh1p to transfer electrons to bovine ADX was challenged. In addition, the functional importance of Arh1p and its subcellular localization were assessed in S. cerevisiae by gene disruption and subcellular fractionation followed by immunochemical detection, respectively.
The yeast expression vector pTG10954 was obtained by the introduction of ARH1 as a SalI-BglII DNA fragment into pTG10385 (12). In this plasmid, expression of ARH1 is controlled by the yeast transcription elongation factor 1 promoter (TEF1), and the phosphoglycerate kinase terminator (PGK1); a hygromycine resistance gene serves as selection marker.
The E. coli expression vector pTG10986 was obtained by the insertion of a SalI-NotI DNA fragment containing the ARH1 coding part and PGK1 terminator in pGEX4T3 (Amersham Pharmacia Biotech). The 5Ј-end of the ARH1 gene was previously modified in order to replace the DNA sequence encoding the amino-terminal region of Arh1p (Met 1 to Ile 9 ) with the specific recognition site for restriction protease Xa (9,13). This modification was done with a SalI-MluI PCR fragment using oligonucleotides OTG7705 (5Ј-TTCGTCGACAAAATCATCGAAGGCCG-TTCTTCACAAATAAACCGT-3Ј) and ORU126 (5Ј-CCTTTTGTTCCG-AGTTCGCCAG-3Ј). Thus, the encoded Arh1p (Ser 10 to Ile 492 ) is linked in frame to the carboxyl-terminal region of glutathione S-transferase through the factor Xa cleavage site (Fig. 1).
The integration plasmid pTG10932 was constructed with a SalI-BglII DNA fragment of pTG10930 containing ARH1. This fragment was inserted into pPOLYIII (14) to give pTG10931. A HindIII filled in DNA fragment of the S. cerevisiae LEU2 gene from pTG10159 (12), was inserted into the unique MluI restriction site of ARH1 (550 base pairs downstream from the ATG codon) in the plasmid pTG10931. The MluI site of this plasmid was previously filled in using the E. coli Klenow fragment of DNA polymerase I. The resulting plasmid, pTG10932, does not allow for replication in yeast.
The integration plasmid pTG10988 derives from pTG10053 (12). The SpeI-SpeI DNA fragment presenting the yeast 2-m replicon was deleted in pTG10053 to give pTG10395. A SalI-BglII DNA fragment encoding Arh1p was cloned into pTG10985 to give pTG10988. The latter does not contain any yeast replicon, and the expression of ARH1 is under the control of the inducible GAL10/CYC1 promoter (15).
Cell transformation was achieved according to classical lithium chloride and calcium chloride methods for yeast and E. coli, respectively (19).
Protein Purification-Expression and purification of a recombinant form of Arh1p (r-Arh1p) was achieved in E. coli BL21 transformed with the plasmid pTG10986. The latter allowed the production of Arh1p fused to the carboxyl-terminal region of GST. The cDNA encoding the fusion protein GST-Arh1p is controlled by the Tac promoter (20). r-Arh1p was purified according to published methods (13,20) with minor modifications. An overnight culture of E. coli BL21/pTG10986 was diluted 1:20 in 4 liters of fresh medium and grown at 37°C; 1 mM isopropyl-thiogalactoside was added when the absorbance at 600 nm reached 0.8. After 4 h of induction, cells were pelleted and resuspended in 40 ml of phosphate-buffered saline, pH 7.5, 1 mM EDTA. Cells were kept at 4°C, and mild sonication was carried out with a Branson Ultrasonic 450 sonicator. The bacterial lysate was slowly homogenized for 30 min at 20°C in the presence of 0.8% Triton X-100 and 10 M FAD. The homogenate was centrifuged at 12,000 ϫ g and 4°C for 10 min. The supernatant was collected and centrifuged twice under the same conditions. A 2.5-ml glutathione-Sepharose 4B column was prepared with a bulk GST purification kit (Amersham Pharmacia Biotech) according to the manufacturer's recommendation. The supernatant was loaded onto the column with a flow of 0.8 ml/min at 6°C. The column was washed with 40 ml of phosphate-buffered saline, pH 7.5, 10 M FAD, and 5 ml of 50 mM Tris-HCl, pH 8.0. GST-Arh1p was eluted with 50 mM Tris-HCl, pH 8.0, containing 10 mM reduced glutathione. During this step, the column flow was reduced to 0.2 ml/min. The eluate (3 ml) was dialyzed for 5 h at 6°C against 50 mM Tris-HCl, pH 7.5, 75 mM NaCl, 10 M FAD, 1 mM CaCl 2 . After dialysis, factor Xa was added to the eluate at a concentration equivalent to 1:25 of the total protein amount. Proteolytic cleavage of GST-Arh1p was carried out for 1 h at 4°C. The resulting product was again loaded onto the glutathione-Sepharose 4B column previously regenerated with 20 volumes of phosphate-buffered saline, pH 7.5; 3 ml of the flow-through containing r-Arh1p were collected and stored at Ϫ80°C in the presence of 20% glycerol. The purified proteins were analyzed by SDS-PAGE and microsequenced. As a control, a bacterial extract of the strain E. coli BL21 harboring the plasmid pGEX4T3 was prepared according to the same protocol. Cytochrome P45011␤, ADX, and ADR were prepared from bovine adrenal cortex according to published methods (21,22).
Enzyme Assays-Ferricyanide reduction was assayed in 50 mM potassium phosphate buffer, pH 7.4, at room temperature. The reaction mixture contained 200 M potassium ferricyanide, 50 M NADPH, 10 g E. coli protein extract presenting (or not) r-Arh1p in a 1-ml final volume. As a positive control, r-Arh1p was replaced with 40 nM purified bovine ADR. The reduction of ferricyanide to ferrocyanide was followed by the absorption decrease at 420 nm using ⑀ ϭ 1.021 mM Ϫ1 ⅐cm Ϫ1 .
Cytochrome c reduction was assayed in 50 mM potassium phosphate buffer, pH 7.4, in the presence of 20 M cytochrome c, 50 M NADPH, 30 nM ADX, 10 g of E. coli protein extract with (or without) r-Arh1p. In the positive control, r-Arh1p was replaced with 30 nM purified ADR. The reduction of cytochrome c was followed by the absorption increase at 550 nm, with ⑀ ϭ 19.1 mM Ϫ1 ⅐cm Ϫ1 . The Michaelis constants (K m ) of Arh1p for NADPH, NADH, and ADX were measured under similar conditions. For NADPH and NADH, 20 M cytochrome c, 1.6 M purified ADX, 8 g of r-Arh1p E. coli extract, and variable amounts of NADPH or NADH were present in the mixture. In the case of ADX, 50 M NADPH was used as electron donor, and variable amounts of ADX were added. Apparent K m values and standard deviations were calculated according to the representation of Lineweaver-Burk using a nonlinear regression program (24); six or seven different concentrations of each studied substrate (ADX, NADPH, and NADH) were used. Each experiment was repeated twice.

Subcellular and Submitochondrial Fractionation of Yeast Cells-
Subcellular and submitochondrial fractions were prepared according to the protocol of Zinser and Daum (25). Fractions resembling "contact sites" and "zwischen bande" in term of their position in the gradient were analyzed for the presence of the Arh1p antigen but were not characterized further.
Western Blot Analysis-Proteins were separated by SDS-PAGE and transferred onto nitrocellulose membranes according to classical methods using a semidry protocol (19). The nitrocellulose membrane was saturated with 0.5% nonfat dried milk (Bio-Rad) in phosphate-buffered saline containing 0.5% Tween 20. r-Arh1p was detected using a rabbit polyclonal anti-bovine ADR antiserum (Oxygene, Dallas, TX) together with a polyclonal donkey anti-rabbit serum (Dako A/S) as a second antibody. The presence of immune complexes were analyzed by chemiluminescence with an ECL kit (Amersham Pharmacia Biotech).
Gene Disruption and Southern Blot Analysis-The genomic ARH1 was disrupted by the insertion of a copy of the yeast LEU2 gene. A SalI-SalI DNA fragment of plasmid pTG10932 bearing the disrupted gene was introduced by homologous recombination into the genome of the diploid strain S. cerevisiae TGY106. Genomic DNA of the resulting strain TGY107 was prepared and analyzed by Southern blot according to classical methods (19).
Sporulation and Tetrad Analysis-Diploid cells of strain TGY107 were inoculated in a presporulation and sporulation solid media as described (19). Preparation of tetrads was done with zymolyase according to the classical method (19), and spore isolation was performed with a Singer MSM micromanipulator (Singer Instruments). Spores were grown for 72 h at 30°C in a medium containing 8 g/liter yeast extract, 20 g/liter agar, and 2% glycerol, 2% ethanol under aerobic conditions (oxidative metabolism) or 2% glucose under anaerobic conditions (fermentation metabolism). In the latter case, plates were placed in a jar containing generator Gas-PaK H 2 -CO 2 (BioMerieux) and palladium grains.
In the case of the strain TGY109, sporulation was carried out as described above. Spores were incubated at 30°C for 72 h in selective medium containing 1.7 g/liter yeast nitrogen base, 50 g/ml uracil, tryptophan, adenine, leucine, histidine, 0.1% glucose, and 2% galactose for GAL10/CYC1 promoter induction.

E. coli Expression and Partial Purification of r-Arh1p-To
study the biochemical properties of a recombinant form of Arh1p, an expression system in which the protein could be rapidly isolated from a bacterial lysate was developed. The E. coli expression vector pTG10986 contains the coding sequence of Arh1p fused in frame to the 3Ј-end of the Schistosoma japonicum glutathione S-transferase gene cloned into plasmid pGEX4T3. Between both genes, the sequence encoding the recognition site for specific cleavage of the protease Xa was introduced immediately upstream of the ARH1 coding region (Fig. 1). This construct does not present the sequence encoding the 9 amino-terminal amino acids of Arh1p. Indeed, when analyzed with the G.O.R. algorithm (26), this region of the protein could form a mitochondrial targeting sequence (data not shown), which likely is cleaved off upon mitochondrial import. Thus, the modification of the 5Ј-end ARH1 gene permits one to produce the corresponding protein starting from Ser 10 in the original amino acid sequence (9).
The expression of the glutathione S-transferase/Arh1p fusion protein (GST-Arh1p) was established in E. coli BL21 transformed with pTG10986 (18). The presence of the GST-Arh1p fusion protein was analyzed by SDS-PAGE (Fig. 2). A protein with an apparent molecular mass of 85 kDa was observed in the soluble fraction of the bacterial lysate. It fits well with the expected size of GST (27 kDa) (20) fused to r-Arh1p (55 kDa). The fusion protein was enriched by affinity column chromatography and was analyzed by SDS-PAGE (Fig. 2, lanes 2  and 3). This extract mainly contained two proteins presenting apparent molecular masses of 85 and 67 kDa, respectively. The amino-terminal sequences of these proteins were determined. The 85-kDa protein presents a unique polypeptide, MSPILGYKI, corresponding to the amino-terminal region of GST. This sequence, together with the molecular mass of 85 kDa, indicates that this protein corresponds to the GST-Arh1p fusion protein. The amino-terminal sequence of the 67-kDa protein shows a unique polypeptide, AAKDVKFGNDARV, corresponding to the bacterial chaperonin GroEL. The prepurified extract, containing the 85-kDa protein, was treated with factor Xa in order to cleave off the GST protein (Fig. 2, lane 4). In that case, the amino-terminal sequence of the major 64-kDa protein, IELDKFENMGA, allowed us to identify a truncated form of the chaperonin GroEL starting at amino acid 57. Following the protease treatment, the protein extract was loaded onto the glutathione-Sepharose 4B column. GST was retained on the column, whereas r-Arh1p was collected in the flow-through. Two major proteins were observed in this final extract (Fig. 2, lane 5): a 67-kDa protein, of which the amino-terminal sequence, AAKDVKFGNDARV, corresponds to the amino terminus of the bacterial chaperonin GroEL, and a 63-kDa protein, of which the amino-terminal sequence, XXQINRXXVXXVG, corresponds to the amino-terminal region of r-Arh1p. Using experimental data derived from amino-terminal sequences, the final purified product contained around 15% Arh1p and 85% GroEL. Thus, the final yield of purified product was about 0.1 mg of r-Arh1p per liter. In this purification system, r-Arh1p and the bacterial chaperonin GroEL co-purified. It was not possible to get rid of this chaperonin, even in the presence of ATP and GroES, as recently described (27). Nevertheless, the partially purified r-Arh1p contained in this bacterial extract was used for biochemical characterization.
Biochemical Characterization of r-Arh1p-Properties of the partially purified r-Arh1p were analyzed in vitro in comparison with purified bovine ADR. The latter is known to transfer electrons from NADPH to ferricyanide. It also reduces ADX in mitochondrial cytochrome P450 systems in a physiological re- The carboxyl-terminal region of GST is linked to amino-terminal region of Arh1p through the protease Xa recognition site. In this construct, the DNA sequence encoding amino acids Met 1 to Ile 9 of Arh1p, proposed as a mitochondrial transit peptide, was deleted. Consequently, proteolytic cleavage with factor Xa should occur immediately before Ser 10 , the first amino acid of Arh1p in the fusion protein. action (P45011␤ and P450SCC) or cytochrome c reduction, a nonphysiological reaction. The ability of r-Arh1p to replace bovine ADR in these electron transport systems was challenged. As a control, an E. coli extract of BL21/pGEX4T3 expressing GST was prepared according to the r-Arh1p purification protocol.
The reduction of ferricyanide was observed in the presence of the partially purified r-Arh1p and NADPH (Fig. 3A). On the contrary, a bacterial extract of the strain E. coli BL21/ pGEX4T3 that did not contain r-Arh1p was unable to reduce ferricyanide. This strongly suggests that Arh1p is the protein in the bacterial extract capable of reducing ferricyanide.
The reduction of cytochrome c was observed in the presence of partially purified r-Arh1p, purified bovine ADX, and NADPH (Fig. 3B). On the contrary, the control extract, in which r-Arh1p is absent, did not support cytochrome c reduction. Again, r-Arh1p in the bacterial extract appears to be the protein supporting cytochrome c reduction. Cytochrome c is not reduced in the presence of NADPH and r-Arh1p. Thus, r-Arh1p is unable to transfer directly electrons to cytochrome c. Similarly, bovine ADX cannot reduce cytochrome c in the absence of r-Arh1p. Consequently, Arh1p and bovine ADX seem to form an electron transfer chain allowing the cytochrome c reduction. In comparison with the mammalian electron transport system, in which electrons flow from ADR to cytochromes via ADX (21), r-ARH1p is assumed to transfer electrons from NADPH to ADX, which in turn reduces the cytochrome c.
The hydroxylation of 11-deoxycortisol to cortisol in the in vitro 11␤ hydroxylation reaction was also analyzed (Fig. 3C). The production of cortisol was observed in the presence of the bovine cytochrome P45011␤, ADX, and NADPH together with the partially purified r-Arh1p. As a control, a bacterial extract that did not contain r-Arh1p did not support the production of cortisol. This phenomenon points out the implication of r-Arh1p in the hydroxylation reaction. Cortisol production was completely dependent on the presence of r-Arh1p and ADX together with P45011␤ and NADPH. As already observed in the cytochrome c reduction, r-Arh1p and bovine ADX seem to form an electron transfer chain to support the activity of cytochrome P45011␤. Arh1p is assumed to transfer electrons from NADPH to ADX.
Interaction of r-Arh1p with NADPH, NADH, and ADX-The interaction of the partially purified r-Arh1p with different substrates was determined using the cytochrome c reduction. The Michaelis constant of r-Arh1p with NADPH, NADH, and ADX was calculated according to the representation of Lineweaver-Burk. The K m values for NADPH and ADX were 0.5 and 0.1 M, respectively (Table I). The reduction of cytochrome c in the presence of NADH as electron source demonstrates that r-Arh1p is also able to retrieve electrons from NADH, and the K m value for NADH, 0.6 M, is not very different from the one for NADPH (Table I).
Subcellular Localization of Arh1p in Yeast-In order to analyze the localization of Arh1p in yeast cells, the ability to detect it with the specific bovine ADR immunoglobulins was challenged. Various E. coli extracts from the r-Arh1p purification were analyzed by Western blotting (Fig. 4). An 85-kDa signal appears specifically in the extract containing the fusion protein GST-Arh1p (Fig. 4, lane 1). This 85-kDa signal fits very well with the expected molecular mass of GST-Arh1p (see Fig.  1). Furthermore, it disappears when the extract is treated with protease Xa, which cleaves specifically GST-Arh1p to liberate r-Arh1p (Fig. 4, lane 2). In the latter case, a 63-kDa signal is visualized corresponding to the apparent molecular mass of r-Arh1p (Fig. 2, lane 5). No signal was detected in the control extract from the strain E. coli BL21/pGEX4T3 in which r-Arh1p is absent (Fig. 4, lane 3). This experiment demonstrates that bovine ADR immunoglobulins are able to recognize r-Arh1p, specifically. This property was used to localize Arh1p in yeast cells. Thus, the subcellular fractions of the yeast S. cerevisiae strain TGY73.4 were analyzed by Western blot. The mitochondrial fraction presents a 63-kDa signal that is indis- tinguishable from the 63-kDa r-Arh1p signal described above (Fig. 5, lane 1). On the contrary, no protein was detected either in the cytosolic or in the microsomal fractions. To ensure that this 63-kDa protein was really Arh1p, the mitochondrial fraction of the strain TGY73.4/pTG10954 overexpressing Arh1p was analyzed by Western blot and in vitro reconstituted P45011␤ hydroxylase activity. In comparison with the mitochondrial fraction of TGY73.4, three clones of the strain TGY73.4/pTG10954 present a 63-kDa protein with a slightly increased signal intensity (Fig. 6). The P45011␤ hydroxylase activity corroborates this phenomenon, and cortisol production can only be detected in mitochondria coming from a strain overexpressing Arh1p (data not shown). These experiments demonstrate that the 63-kDa protein corresponds to Arh1p. Consequently, Arh1p is localized in the mitochondrial fraction of yeast cells.
Submitochondrial Localization of Arh1p-We assumed that Arh1p should have properties resembling mammalian ADR not only in terms of its function but also in terms of its subcellular and, more precisely, submitochondrial localization. Mitochondrial membrane fractions were prepared as described and were analyzed for their Arh1p content. The ATP/ADP carrier and porin served as markers for the inner and outer mitochondrial membrane fractions, respectively. The Arh1p antigen was found to co-fractionate almost exclusively with the inner mitochondrial membrane (Fig. 7). This localization would be in agreement with a functional interaction with recombinant ADX, a soluble component of the mitochondrial matrix of a yeast strain expressing the corresponding bovine cDNA (data not shown).
Gene Disruption of ARH1-To approach the biological function of Arh1p, the inactivation of the corresponding gene was studied in the diploid yeast strain TGY107. This strain was obtained from diploid TGY106 modified by homologous recombination with a DNA fragment bearing a copy of the yeast LEU2 gene inserted in the MluI site of the ARH1 gene as described under "Materials and Methods" (Fig. 8A). Genomic DNA from 10 clones, TGY107.1 to TGY107.10, treated with SalI restriction enzyme, was analyzed by Southern blot, using a probe corresponding to the 5Ј part of ARH1. The expected DNA fragment of 3.8 kilobases, corresponding to the functional ARH1 gene, was observed for the diploid TGY106 (control) and several clones of TGY107 (Fig. 8B). In addition, the latter clones presented a second DNA fragment with an estimated length of 5.8 kilobases, which corresponds to the expected size, considering the insertion of the LEU2 gene in ARH1. Thus, the diploid strain TGY107 is bearing two copies of ARH1: a functional copy and one disrupted by insertion of the LEU2 gene. The clone TGY107.10 was used for further characterization. Interactions of Arh1p with NADPH, NADH, and bovine ADX were studied with cytochrome c reduction at 550 nm as described under "Materials and Methods." Apparent K m values and S.D. were calculated from two independent experiments, including six or seven measurements with variable concentrations of substrates. Spore Analysis of the Strain TGY107-Sporulation of a single cell of TGY107 would generate two spores bearing the functional ARH1 gene and two spores harboring the disrupted one (Fig. 9A). The latter two should not grow if the ARH1 gene is essential for yeast survival. Precisely 40 spores corresponding to 10 diploid cells of the clone TGY107.10 were separated and grown in a rich medium containing glucose as a carbon source under aerobic conditions. As shown in Fig. 9B, 20 spores (2 of 4 per diploid cell) formed a colony after 4 days of incubation at 30°C. All clones were unable to grow without leucine and presented the parental phenotype leu Ϫ , corresponding to the nondisrupted ARH1 gene (data not shown). As a control, the mating type of these clones was determined; 8 of 20 were MATa, 8 were found to be MAT␣, and 4 were unable to grow, probably because cell fusion with the tester strains did not succeed (data not shown).
All other spores (20 of 40) were unable to form visible colonies. However, microscopic observation revealed microcolonies of approximately 64 daughter cells, corresponding to six division cycles (Fig. 9C). During the first 2 days of incubation, these spores presented a normal growth in comparison with the leu Ϫ spores described above. After that, cell division ceased very suddenly, and extended incubation did not permit further growth. In order to demonstrate that growth of the latter spores depends on the presence of a functional ARH1 gene, we tried to rescue these spores by the introduction of a functional copy of ARH1. To do so, TGY107.10 was transformed with pTG10988, allowing the introduction of ARH1 in the URA3 locus to give the strain TGY109. In that case, the segregation of ARH1 gene should lead to three types of tetrad, as shown in Fig. 10A. For the tetrad 1, three spores out of four from the diploid strain TGY109 are expected to have at least one functional copy of ARH1 gene (Fig. 10A). Thus, these spores should present normal growth, whereas the fourth, presenting the disrupted ARH1 gene, should not. Fig. 10B presents 40 spores obtained from 10 diploid TGY109.1 cells. For most of them, the segregation of ARH1 gene corresponds to tetrad 1. According to the expected result, three spores out of four per diploid cell were able to form colonies after 4 days of incubation at 30°C in complete medium with galactose (as a carbon source) and oxygen (Fig. 10B). The last spore again grew only to form a microcolony (about 64 daughter cells). The genomic structure of cells from spores belonging to the same tetrad, analyzed by Southern blot, corresponds to expected spores of tetrad 1 described in Fig. 10A (data not shown). Analyzing a larger number of spores, we were able to isolate spores from which the segregation of ARH1 corresponds to tetrads 2 and 3 ( Fig. 10A) with more or less the expected frequency (data not shown). In all cases, spores without any functional ARH1 were unable to grow. Taking together the results of tetrad analysis of strains TGY107.10 and TGY109.1, inactivation of ARH1 appears to be lethal. Thus, this study supports the conclusion that the biological function of Arh1p is essential for yeast survival. For further characterization, the spores of TGY107 were analyzed in a rich medium with glucose as a carbon source in the absence of oxygen. In those conditions, yeast produces ATP through the fermentation pathway. Again, two spores out of four per diploid cell were able to form a colony, and the other two did not (data not shown). The same result was observed in the presence of ethanol-glycerol as a carbon source and oxygen to allow yeast to produce ATP through the respiration pathway (data not shown).

DISCUSSION
Arh1p was analyzed in order to demonstrate its ability to form an electron transfer chain with mammalian ADX and to study its physiological importance in yeast. Also, attention has been paid to partially purifying Arh1p that was made in E. coli and to localize this protein in yeast cells.
Arh1p could be produced in E. coli as a fusion protein with glutathione S-transferase and was partially purified using the GST module. This purification system permits one to produce very rapidly a large amount of protein.
The reduction of ferricyanide in the presence of NADPH and r-Arh1p indicates that the protein Arh1p is capable of transferring electrons from a 2-electron donor to a 1-electron acceptor, ferricyanide. The cytochrome c reduction, together with 11␤-hydroxylase activity of purified bovine P45011␤, demonstrates that r-Arh1p can interact with bovine ADX, forming an electron transfer chain from NADPH to ADX. Catalyzing such redox reactions, Arh1p should present two characteristics. First, the redox potential of Arh1p should be between Ϫ320 mV, the redox potential of NADPH or NADH, and Ϫ270 mV, the redox potential of oxidized ADX (28,29). Secondly, the protein should have a prosthetic group assuming the electron transfer capability. The chemical nature of this potential prosthetic group is not known, but the FAD binding domain of mammalian ADR, Gly 13 -Gly 15 -Gly 18 -Ala 22 (10) was found in the amino acid sequence of Arh1p (9). This FAD binding site is very particular to ADR and differs from the consensus sequence of Ferredoxin NADP ϩ reductases, Arg 93 -Tyr 95 -Gly 130 -Met 167 -Gly 171 -Thr 172 -Gly 173 -Tyr 246 -Cys 272 -Gly 273 , according to the crystal structure of spinach ferredoxin NADP ϩ reductase (30). Furthermore, the consensus sequence of the flavin mononucleotide binding domain, Ser 10 -Thr 11 -Thr 12 -Asn 14 -Thr 15 , according to the crystal structure of the Desulfovibrio vulgaris flavodoxin, is not present in Arh1p (31,32). Thus, in parallel with mammalian ADR, it is rather likely that Arh1p has an FAD molecule as a prosthetic group.
The interaction of r-Arh1p with bovine ADX, NADPH, and NADH was studied through the cytochrome c reduction. The apparent K m values for NADPH and bovine ADX, 0.5 and 0.1 M, respectively, are slightly different from the values obtained for mammalian proteins: 0.017 M for bovine ADX to bovine ADR (33), 0.1 M for human ADX to human ADR (34), and 1.8 and 0.7 M for NADPH to bovine and human ADR, respectively (34,35). On the contrary, the apparent K m of r-Arh1p for NADH, 0.6 M, is completely different from that of the bovine ADR, 5400 M (5). Unlike mammalian ADR, which primarily transfers electrons from NADPH, Arh1p could receive electrons from NADPH or NADH because the apparent K m is roughly the same for both substrates in vitro.
The biological function of ARH1p in yeast cells remains FIG. 9. Spore analysis of the diploid strain S. cerevisiae TGY107. A, sporulation of the strain TGY107 bearing one functional ARH1 gene and one disrupted gene provides four spores; two of these, harboring the disrupted gene, should not grow if Arh1p is essential for yeast survival. B, 40 spores corresponding to 10 diploid cells TGY107 were dissected and cultured in a rich medium in the presence of glucose as described under "Materials and Methods"; spores were incubated at 30°C for 72 h. Lane 1, four spores corresponding to the diploid cell TGY107 clone 1 were dissected; lanes 2-10, spores corresponding to TGY107 clones 2-10, respectively, were dissected. Only two spores per diploid cell could be visualized because they are capable of growth. C, microscopic observation of a microcolony (magnification, ϫ 400). 50% of the spores of the strain TGY107 presented abnormal growth after 72 h of incubation at 30°C in a rich medium as described under "Material and Methods." These were unable to grow beyond the state of a microcolony of about 64 cells, corresponding to six division cycles.
unknown. Tetrad analysis of the diploid strains TGY107.10 and TGY109.1 demonstrate that inactivation of ARH1 is lethal. It is in agreement with a notion of ARH1 carrying an essential function described elsewhere (36). The microcolonies observed, which originated from spores carrying a deleted allele, indicate that only a minimal quantity of Arh1p is needed to support growth. Our observation was noticed for TGY107 under conditions stimulating the respiratory metabolism (glycerol/ethanol as a carbon source and oxygen), as well as the fermentation metabolism (glucose as a carbon source). Many mitochondrial electron transfer proteins are involved in the respiratory chain for ATP biosynthesis. Because Arh1p is essential even in fermentation conditions, this protein may not be involved in the respiratory chain. In comparison with mammalian ADR, which is involved in sterol biosynthesis, Arh1p was suspected to interfere in ergosterol biosynthesis. Arh1p was also found to be essential under conditions complementing ergosterol as well as heme biosynthesis (data not shown). Analyzing the yeast protein data base, only 12 genes encode mitochondrial proteins, presenting an essential function even in fermentation metabolism like Arh1p. Surprisingly, all of these proteins, namely Tom40, Tom22, Tim44, Tim23, Tim22, Tim17, Mge1p, Hsp10, Hsp60, Hsp70, Mas1, and Mas2, are components of the mitochondrial protein import machinery (37)(38)(39)(40)(41). Furthermore, mRNA production corresponding to these proteins, together with ARH1 mRNA, is enhanced during the diauxic shift (42). These phenotypic similarities may be indications of the function of Arh1p.
Regarding Arh1p interactor proteins resembling mammalian ADX and mitochondrial P450 in the yeast genome data base, it is possible to identify an open reading frame encoding a protein, YPL252C, that presents homologies with mamma- FIG. 10. Spore analysis of the diploid strain S. cerevisiae TGY109. A, a single diploid cell, TGY109, resulting from the integration in the URA3 locus of TGY107, provides three kinds of tetrad; 0.5, 0.25, and 0.25 represent the probability of appearance for tetrads 1, 2, and 3, respectively. Spores covered by a cross in tetrads 1 and 2 could not produce Arh1p. These should not grow if Arh1p is essential for yeast survival. B, 40 spores, corresponding to 10 diploid TGY109 cells, were separated and cultured in minimal medium in the presence of galactose as described under "Materials and Methods"; spores were incubated at 30°C for 72 h. Lane 1, four spores corresponding to the diploid cell TGY109 clone 1 were separated; lanes 2-10, spores corresponding to TGY109 clones 2-10, respectively, were separated. Most of the tetrads corresponded to tetrad 1 (A), for which only three spores per diploid cell could be visualized because they are capable of growth. lian ADX. Considering the mature form of bovine ADX, 40% of identical amino acids could be found in the yeast protein. More importantly, the cysteine cluster (Cys 46 , Cys 52 , Cys 55 , and Cys 92 ) involved in the fixation of the iron-sulfur center could be found in YPL252C as well as the interaction domain with mitochondrial P450 and ADR (Glu 75 to Thr 85 ) (43)(44)(45)(46). In addition, different amino acids described as being important for activity of bovine ADX (Thr 54 and Pro 108 ) are found in identical positions in the yeast protein (29,47). The similarities of YPL252C with mammalian ADX makes us think that it is a possible interactor of the yeast Arh1p.
On the other hand, looking for a mitochondrial cytochrome P450 in the yeast protein data base using the P450 heme binding domain of P450cam, Thr 349 to Glu 366 (48), it is possible to detect CYP56, CYP51, and CYP61. These cytochrome P450 forms are characterized and described as microsomal proteins. A possible targeting of these microsomal enzymes into mitochondria could be considered, as recently described for mammalian P4501A1 (49). But the typical reduced CO spectrum characterizing P450 proteins was not observed with isolated yeast mitochondria, suggesting the absence of mitochondrial P450 in yeast. Thus, the existence in S. cerevisiae of an electron transport chain equivalent to the mitochondrial ADR-ADX-P450 chain of mammalian adrenals seems to be unlikely. In that sense, the biological function of Arh1p in yeast may diverge from that of its mammalian homologue.
Finally, we have recently established the in vivo activity of recombinant bovine P45011␤ expressed in yeast (7). The coexpression of P45011␤ and bovine ADX cDNAs in yeast mitochondria was sufficient to observe the P45011␤ activity. It suggested the existence of a yeast protein transferring electrons to ADX. It was not possible to modulate the 11␤-hydroxylase activity in vivo to determine the implication of Arh1p because ARH1 disruption is lethal and P45011␤ seems to be limiting in our system. Nevertheless, regarding experiments presented in this report, mitochondrial Arh1p, which is able to form an electron transfer chain with bovine ADX in vitro, should support the in vivo P45011␤ activity. To our knowledge, Arh1p is the first protein of lower eucaryotes presenting homology with mammalian ADR.