The FRE1 ferric reductase of Saccharomyces cerevisiae is a cytochrome b similar to that of NADPH oxidase.

Plasma membrane preparations from strains of the yeast Saccharomyces cerevisiae gave a reduced minus oxidized spectrum characteristic of a b-type cytochrome and very similar to the spectrum of flavocytochrome b558 of human neutrophils. The magnitude of the signal correlated with the level of ferric reductase activity and the copy number of the FRE1 gene, indicating that the FRE1 protein is a cytochrome b. Sequence similarities with the flavin binding site of flavocytochrome b558 and other members of the ferredoxin-NADP reductase family, together with increased levels of noncovalently bound FAD and iodonitrotetrazolium violet reductase activity in membranes from a yeast strain overexpressing ferric reductase, suggested that the FRE1 protein may also carry a flavin group. Potentiometric titrations indicated that FRE1, like neutrophil NADPH oxidase, has an unusually low redox potential, in the region of -250 mV, and binds CO.

Plasma membrane preparations from strains of the yeast Saccharomyces cerevisiae gave a reduced minus oxidized spectrum characteristic of a b-type cytochrome and very similar to the spectrum of flavocytochrome b 558 of human neutrophils. The magnitude of the signal correlated with the level of ferric reductase activity and the copy number of the FRE1 gene, indicating that the FRE1 protein is a cytochrome b. Sequence similarities with the flavin binding site of flavocytochrome b 558 and other members of the ferredoxin-NADP reductase family, together with increased levels of noncovalently bound FAD and iodonitrotetrazolium violet reductase activity in membranes from a yeast strain overexpressing ferric reductase, suggested that the FRE1 protein may also carry a flavin group. Potentiometric titrations indicated that FRE1, like neutrophil NADPH oxidase, has an unusually low redox potential, in the region of ؊250 mV, and binds CO.
Iron uptake in Saccharomyces cerevisiae is a two-step process. An externally directed plasma membrane ferric reductase converts insoluble, environmental ferric (Fe 3ϩ ) iron to the soluble ferrous (Fe 2ϩ ) form which is transported across the membrane by an iron transport complex (Stearman et al., 1996). Reduction of Fe 3ϩ is primarily attributable to the FRE1 protein (Dancis et al., 1990(Dancis et al., , 1992, although in its absence low levels of residual activity are detectable, due largely to a second reductase, FRE2 (Georgatsou and Alexandraki, 1994).
The FRE1 gene encodes a protein 686 amino acids in length, with a calculated molecular mass of 78.8 kDa. It has an apparent 22-amino acid membrane insertion leader peptide and hydropathic analysis (Fig. 1B) reveals multiple hydrophobic regions consistent with membrane spanning domains, thus indicating that the FRE1 gene product is a membrane bound structural component of the reductase. This view is supported by its homology with the large ␤ subunit, gp91 phox , of NADPH oxidase from human phagocytic cells (Dancis et al., 1992;Roman et al., 1993). NADPH oxidase requires the assembly of gp91 phox with a smaller ␣ subunit, p22 phox , creating the flavocytochrome b 558 . This flavocytochrome is located in the plasma membrane and membrane of the specific granules, and becomes incorporated into the wall of the phagocytic vacuole. It takes electrons from NADPH in the cytoplasm and passes them across the membrane via FAD and heme to molecular oxygen, generating superoxide that is expelled into the lumen of the vacuole (Wientjes and Segal, 1995).
The C-terminal 402 amino acids of FRE1 show 18% identity and 62% similarity with gp91 phox . In addition there are several clusters of much higher identity. These include an HPFTXXS motif which is believed to function in FAD binding in the respiratory burst oxidase and a glycine-rich motif and cysteineglycine couplet, which represent peptide loops thought to be involved in NADPH binding ( Fig. 1A) (Taylor et al., 1993). The hydropathic profiles of the two proteins when aligned from the C terminus also show some resemblance (Fig. 1B). Given that both proteins are electron transporters, the similarity in structure suggested that FRE1 might also be a membrane bound flavocytochrome. Further evidence for this hypothesis came from Lesuisse and Labbe (1989) who reported that heme deficient yeast strains lack ferric reductase activity. In this report we present evidence that the yeast FRE1 protein is a cytochrome b and quite probably a flavocytochrome, with properties similar to those of flavocytochrome b 558 of the human NADPH oxidase.

MATERIALS AND METHODS
Strains and Media-A parental strain H1085 112) and two derivative strains of S. cerevisiae were used for these experiments. To create strain ⌬fre1::LEU2, the FRE1 locus of H1085 was replaced with a LEU2 marker gene by double homologous recombination. The replacement of the genomic sequences lying between flanking ClaI sites of the FRE1 locus was verified by Southern blotting. To generate strain 352-FRE1, the 4.2-kilobase pair BamHI-SacI genomic fragment of FRE1 was subcloned into the vector YEp352, and this high copy number plasmid (approximately 40 copies per cell) was used to transform strain H1085 to uracil prototrophy.
Cells were grown to a high density (A 600 approximately 1.5) in 6.7 g/liter yeast nitrogen base lacking iron and copper (BIO 101 Inc.), 20 g/liter dextrose and 20 g/ml uracil and/or 33 g/ml L-leucine as appropriate, at 30°C on an orbital shaker. They were then diluted back to an A 600 of 0.2 into YPD (1% yeast extract, 1% peptone, 2% dextrose) with 100 g/ml bathocuproine-disulfonic acid and grown for 5 h prior to harvesting (A 600 approximately 0.5-0.6).
Isolation of Plasma Membranes-Cultures were harvested and the cells washed once in 0.4 M sucrose in buffer A (2 mM EDTA, 25 mM imidazole, pH 7.0, with protease inhibitors, 1 mM phenylmethylsulfonyl fluoride, 100 mM N-tosyl-L-phenylalanine chloromethyl ketone, 2 g/ml pepstatin A). They were then disrupted by vortexing with glass beads, diluted 3-fold in 0.4 M sucrose in buffer A and spun at 530 ϫ g. The supernatant was centrifuged at 22,000 ϫ g, and the pellet, which included the plasma membranes and mitochondria, was resuspended in buffer A and loaded onto a discontinuous sucrose gradient comprising 2.25 M, 1.65 M, and 1.1 M sucrose in buffer A. After overnight centrifugation at 80,000 ϫ g, the essentially pure plasma membranes were removed from the 2.25 M/1.65 M interface, diluted 4-fold, and pelleted at 30,000 ϫ g. Membranes were resuspended in 0.1 mM EDTA, 25 mM imidazole-HCl, pH 7.0, 50% glycerol and stored at Ϫ20°C. The absence of significant mitochondrial contamination in membrane preparations produced by this technique was demonstrated by measuring the level of * This work was supported in part by the Wellcome Trust and by National Institutes of Health Grant AI24838. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Ferric Reductase Assay-Cells were assayed for ferric reductase activity at the time of harvesting as described previously (Dancis et al., 1990).
FAD Assay-FAD was determined by reconstitution of apo-glucose oxidase activity. Membrane preparations were diluted with 25 mM imidazole-HCl, 0.1 mM EDTA (pH 7.0), boiled for 3 min to extract the FAD, and microcentrifuged for 5 min at 13000 rpm. The supernatant was added to a reaction mixture comprising 11.1 mM sodium citrate (pH 6.5), 0.44 mM 4-aminoantipyrine, 2.2 mM 3,5-dichloro-2-hydroxybenzene-sulfonic acid, 20 mg/ml D-glucose, 3.3 nM apo-glucose oxidase (from Aspergillus niger; purchased from Sigma and prepared essentially by the method of Morris and Buckler (1983)), 1.1 mg/ml horseradish peroxidase (from Boehringer Mannheim and further purified by ion exchange chromatography on a Mono Q resin), in a microtiter plate. The absorbance at 520 nm was monitored using a Dynatech MR7000 microtiter plate reader fitted with an Advanced Applications program cartridge, and the FAD concentration was determined from the rate of reaction against a standard curve.
Superoxide Assay-Superoxide generation was determined from the rate of cytochrome c reduction inhibitable by superoxide dismutase. Assays were performed in a 150-l final volume in 96-well microtiter plates. Measurements were made on detergent solubilized membranes (8 g of protein) in relaxation buffer (100 mM KCl, 3 mM NaCl, 3.5 mM MgCl 2 , 10 mM PIPES, 1 1 mM ATP, pH 7.3) plus 108 M horse heart cytochrome c (Sigma), with and without 333 nM FAD, 160 M NADPH, and 180 units/ml superoxide dismutase (from bovine erythrocytes; Sigma). The absorbance was monitored at 550 nm and analyzed in a kinetic microtiter plate reader.
Iodonitrotetrazolium Violet (INT) Reductase Activity-INT reductase assays were performed on solubilized membranes (8 g of protein) in relaxation buffer plus 43 M 2-(4-iodophenyl-3-(4-nitrophenyl)-5-phenyltetrazolium chloride and 180 units/ml superoxide dismutase, in the presence and absence of 333 nM FAD and 160 M NADPH. Increasing absorbance was monitored at 490 nm in a microtiter plate Protein Assay-Protein was determined using the method of Schaffner and Weissmann (1973).
Purification of Flavocytochrome b 558 -Flavocytochrome b 558 was purified from the neutrophils of patients with chronic myeloid leukemia using a modification of the method of Harper et al. (1984).
Spectroscopy-Dithionite-reduced minus oxidized difference spectra were determined for the plasma membrane preparations using a Shimadzu UV-3000 double beam spectrophotometer. The concentration of heme was determined from the height of the Sorret peak in the reduced minus oxidized spectrum using an absorption coefficient of 121 mol cm Ϫ1 (Segal et al., 1992).
Determination of Extinction (Absorbance) Coefficients-Purified plasma membrane fractions from strain 352-FRE1 were solubilized in 1% (v/v) heptyl ␤-D-thioglucopyranoside (Calbiochem) by stirring at 4°C for 30 min. Insoluble material was removed by centrifugation at 100,000 ϫ g for 30 min. A portion of the solubilized membrane was dissolved in alkaline pyridine (final concentration, 100 mM NaOH, 20% v/v pyridine), and the dithionite-reduced minus air-oxidized difference spectrum of the pyridine hemochrome was recorded. A ⌬⑀ 557-541 of 20.7 mM Ϫ1 cm Ϫ1 (Porra and Jones, 1963) for the reduced minus oxidized protoheme pyridine hemochrome was used to calculate the concentration of protoheme in the solubilized membranes. A second portion of the solubilized extract was used to record the reduced minus oxidized difference spectrum of the hemoprotein in aqueous buffer and hence derive extinction coefficients.

RESULTS
Three strains of S. cerevisiae were used in this study; H1085, a wild-type strain, ⌬fre1::LEU2, a mutant derived from H1085 by deletion of the FRE1 gene and 352-FRE1, the wild-type strain transformed with a high copy number plasmid carrying the FRE1 gene. The cells were grown under conditions that facilitated a high level of ferric reductase activity. Reduction of Fe 3ϩ was measured at the time of harvesting and shown to be negligible in the deletion mutant and substantially raised in 352-FRE1 relative to the parental strain (Table I).
Plasma membranes were isolated from these cultures on a sucrose density gradient and their dithionite-reduced minus oxidized spectra determined over the wavelength range 650 -400 nm. These spectra are shown together with a spectrum for pure neutrophil flavocytochrome b 558 in Fig. 2. The similarities are striking. H1085 and 352-FRE1 both gave spectra characteristic of a b-type cytochrome. There is an ␣ peak at 558 nm, a ␤ peak at 528 nm, and a large ␥ or heme peak at 428 nm. Importantly the magnitude of the peaks increases with the level of ferric reductase activity in the yeast cells (Table I) and with the FRE1 copy number indicating that FRE1 is the plasma membrane cytochrome b. A very small heme peak can be seen in the spectrum for the deletion mutant, ⌬fre1::LEU2 (Fig. 2); this may be due either to another plasma membrane FIG. 1. Relatedness of the yeast ferric reductase, FRE1, and gp91 phox . A, shared amino acid motifs. Motifs correspond to the putative sites for 2, FAD-isoalloxazine binding; 4, NAD/P-ribose binding; and 5, NAD/P-adenine binding (identical residues are in bold, conserved residues are in italics). B, hydrophobicity plots for FRE1 and gp91 phox aligned from the C terminus (predicted amino acid sequences have been analyzed by the Kyte-Doolittle algorithm with a window size of 11 amino acids). Increasing hydrophobicity is shown above the x axis. The numbered scale reflects the amino acid positions for FRE1. The lines numbered 1-5 represent the positions of the motifs shown in part A. reductase, possibly FRE2 (the level of FRE2 expression varies according to the strain background and the growth phase of the cells, and was minimal under the conditions of these experiments), or to a very low level of mitochondrial crosscontamination.
To confirm the nature of the cytochrome, the dithionitereduced minus air-oxidized difference spectrum of the pyridine hemochrome was recorded for detergent-solubilized plasma membranes from strain 352-FRE1. The concentration of protoheme was calculated using ⌬⑀ 557-541 of 20.7 mM Ϫ1 cm Ϫ1 (Porra and Jones, 1963) for the reduced minus oxidized protoheme pyridine hemochrome, and this in turn was used to derive the extinction coefficients from the reduced minus oxidized difference spectrum of the hemoprotein in aqueous buffer. The calculated extinction coefficients for the ferric reductase hemoprotein are shown in Fig. 3. The calculated values of the principal spectral features are summarized in Table II. Of interest is the unusually low absorbance of the ␤-band in the reduced minus oxidized difference spectrum, a feature that is shared with neutrophil cytochrome b 558 . The relatively small extinction coefficient is primarily a result of the large absorbance of the oxidized cytochrome in this spectral region.
Attempts to determine the midpoint potential for FRE1 by potentiometric titration of the hemoprotein in solubilized plasma membrane fractions from strain 352-FRE1 did not yield optimal titrations, due to the apparent instability of the ferrous form of the heme which resulted in progressive loss of the absorbance spectrum. However, little reduction was observed at potentials above Ϫ200 mV and reduction was virtually complete at Ϫ300 mV (data not shown). Thus, the midpoint potential was estimated to be around Ϫ250 mV. This low redox potential is remarkably similar to that of flavocytochrome b 558 which has two heme centers with closely spaced midpoint potentials of Ϫ225 mV and Ϫ265 mV (Cross et al., 1995b). The low potential in the neutrophil system is necessary to catalyze the production of O 2 . from molecular O 2 at a kinetically competent rate and the low potential of the ferric reductase suggests that it too might be capable of generating O 2 . .
Neutrophil flavocytochrome b 558 forms a low affinity complex with CO (Cross et al., 1982). Although the cytochrome is thought to transfer electrons to O 2 from the heme edge rather than by direct ligation of O 2 to heme iron, the ability to bind CO is often taken as a sign of oxygen reactivity among hemoproteins. This ability is shared by the ferric reductase heme protein as shown in Fig. 4. Assuming the extinction coefficient of the ferrous-CO complex is similar to that of the ferric hemoprotein, the ferric reductase is fully complexed to CO after a 180-s exposure of the ferrous form to CO and thus has a somewhat higher affinity for CO than cytochrome b 558 . Approximately 40% of the latter forms a CO complex at room temperature and 1 atm CO (Cross et al., 1982).
One possibility is that S. cerevisiae exploits the rapid reaction of O 2 . with ferric iron as a mechanism for releasing environmental iron. Plasma membranes from strains ⌬fre1::LEU2 and 352-FRE1 were tested for superoxide generation by meas-  3. Extinction coefficients of the FRE1 hemoprotein. The concentration of the hemoprotein was derived from the pyridine hemochrome spectrum as described under "Materials and Methods." The spectrum shown is that of the dithionite-reduced minus air-oxidized difference spectrum.  (412) 157 Soret r-o (427.5) 129 Soret peak-trough (427.5-411) 202 uring the rate of reduction of cytochrome c inhibitable by superoxide dismutase. Both membrane preparations showed some cytochrome c reductase activity (data not shown), but this was superoxide dismutase-insensitive. FRE1, therefore, appears to be incapable of generating significant amounts of O 2 . , at least under the conditions used in the assay. Alignments of the predicted amino acid sequences for FRE1 and gp91 phox reveal a highly conserved region corresponding to the binding site for the FAD-isoalloxazine ring (Fig. 1A). Furthermore, in the reduced minus oxidized spectra for the yeast plasma membranes, a shallow flavin trough appears at a wavelength of roughly 450 nm, adjacent to the ␥ peak (Fig. 2). Extracts from the yeast membrane preparations were analyzed for FAD to determine whether, like flavocytochrome b 558 of NADPH oxidase, the FRE1 cytochrome carries a noncovalently bound flavin group. Membranes from strain 352-FRE1 were consistently found to contain 2-3 times more FAD than those of the deletion mutant and wild-type S. cerevisiae (Table I).
Plasma membranes from strains ⌬fre1::LEU2 and 352-FRE1 were tested for INT reductase activity. It has been shown that neutrophil NADPH oxidase is capable of reducing INT in a manner that is independent of O 2 . production (Cross et al., 1994) and there is mounting evidence that INT accepts electrons directly from the flavin center (Cross and Curnutte, 1995;Cross et al., 1995a). Membranes from strain 352-FRE1 demonstrated INT reductase activity that was 5-fold higher than that of the deletion mutant and independent of exogenous FAD (Table III), implying that, like NADPH oxidase, the yeast ferric reductase possesses diaphorase activity. This may be a further indication that FRE1 carries a flavin group.

DISCUSSION
The correlation between the magnitude of the reduced minus oxidized spectrum and the level of ferric reductase activity in the yeast strains provides strong evidence that FRE1 is a plasma membrane cytochrome b. Whether the protein also carries a flavin group is rather more equivocal. The homology between FRE1 and gp91 phox at the putative FAD-binding site, together with the raised levels of noncovalently bound FAD in the plasma membranes from strain 352-FRE1 suggest that FRE1 is likely to be a flavocytochrome, as does its apparent INT reductase activity. However, if the heme:FAD ratio is calculated for the data presented in Table I (the concentration of heme and FAD in the deletion mutant having first been subtracted as an indication of background levels), a seemingly implausible value of 18.6:1 is obtained. This compares with an apparent ratio of 2:1 for flavocytochrome b 558 (Segal et al., 1992), which correlates with the two electron transfer catalyzed by this protein. Higher heme to FAD ratios have been reported (Pealing et al., 1992), but a partial loss of the FAD cofactor during membrane purification may provide a more satisfactory explanation for the nonstoichiometric increase in FAD with heme in strain 352-FRE1. Alternatively the flavin group may reside within a separate, loosely associated membrane protein.
The very low midpoint potential and the apparent oxygen reactivity of the FRE1 protein suggested a mechanism whereby Fe 3ϩ is reduced to the ferrous form by O 2 . . A mechanism of iron reduction utilizing a small intermediate would explain the ability of the FRE1 reductase to reduce chemically varied substrates (ferric citrate, ferric-EDTA, ferricyanide, ferrioxamine B, Cu 2ϩ , cytochrome c, nitro blue tetrazolium, resazurin c) (Lesuisse and Labbe, 1994). However, O 2 . was not detected in the cell free assay. This may reflect the lability of a critical cofactor that was lost during membrane purification. Alternatively, the absence of O 2 . production by the yeast membranes could result from the lack of one or more essential cytosolic proteins. Generation of O 2 . by NADPH oxidase in a cell free assay requires three cytosolic factors, p47 phox , p67 phox , and p21 rac1 , together with an amphipathic activating reagent such as SDS or arachidonic acid. NADPH is the most probable electron donor for the FRE1 ferric reductase. It donated electrons in the INT reductase assay and in potentiometric titrations, was a good reductant at higher potentials although it failed to drive the potential below about Ϫ225 mV, apparently because of the inherent instability of the protein. Furthermore, motifs likely to be involved in NADP(H) binding have been identified at positions analogous to the NADPH binding sites of neutrophil NADPH oxidase (Fig. 1).
Although the human NADPH oxidase is a heterodimer comprising both ␣ and ␤ subunits, both the NADPH and FAD binding sites, together with at least one of the hemes, are accommodated entirely within the ␤ subunit, suggesting that the ␣ subunit may have a regulatory function. For the yeast FRE1 reductase, regulation occurs at the level of control of transcription of the FRE1 gene, and a regulatory subunit may not, therefore, be required. Evidence regarding the presence of a second protein subunit for the FRE1 reductase has been equivocal. Expression of the FRE1 genomic clone on a high copy number plasmid leads to increased surface ferric reductase, indicating that a second subunit, if present, is not limiting for reductase activity. Searches for reductase deficient mutants have led to repeated identification of mutant alleles of FRE1. FIG. 4. The reaction of ferrous cytochrome b with CO. Solubilized membranes were reduced with a few crystals of sodium dithionite and the reduced minus oxidized difference spectrum recorded (a). The reduced spectrum was stored in the spectrophotometer memory to obtain a new base line (reduced minus reduced) (b), and CO gas was passed through the sample at a rate of 1-2 bubbles s Ϫ1 for a total of 30 s (c), 90 s (d), and 180 s (e) before re-recording the spectrum. However, a single isolate of a mutant in the UTR1 gene (Swissprot accession no. P21373) was noted to be deficient in reductase. This gene product was not membrane associated and the sequence bore no resemblance to other sequences in the data base. Thus the role of the UTR1 protein in the FRE1 reductase system remains unclear. FRE1 shows homology not only to gp91 phox but also to FRE2 and to the plasma membrane ferric reductase of the evolutionarily distant yeast Schizosaccharomyces pombe, Frp1 (Roman et al., 1993). FRE1, FRE2, and Frp1 demonstrate functional and regulatory similarities and, furthermore, share with gp91 phox a similar hydropathic profile and clusters of amino acid identities at analogous positions. It is possible, therefore, that these three proteins represent a distinct family of membrane bound flavocytochromes capable of transporting electrons across the cell membrane.