The Metalloreductase Fre6p in Fe-Efflux from the Yeast Vacuole*

The yeast vacuole is the storage depot for cellular iron. In this report we quantify the import-export balance in the vacuole because of the import of iron by Ccc1p and to export by the combined activity of Smf3p and the ferroxidase, permease pair of proteins, Fet5p and Fth1p. Our data indicate that the two efflux pathways are equally efficient in trafficking iron out of the vacuole. A major focus of this work was to identify the ferrireductase(s) that supplies the FeII for efflux whether by Smf3p or the Fet5p-Fth1p complex. Using a combination of flameless atomic absorption spectrophotometry to quantify vacuolar and whole cell iron content and a reporter assay for cytoplasmic iron we demonstrate that Fre6p supplies FeII to both efflux systems, while Fre7p plays no role in Fe-efflux from the vacuole. Enzymatic assay shows the two fusions to have similar reductase activity, however. Confocal fluorescence microscopy demonstrates that Fre6:GFP localizes to the vacuolar membrane; in contrast, Fre7:GFP fusions exhibit a variable and diffuse cellular distribution. Demonstrating a role for a vacuolar metalloreductase in Fe-efflux supports the model that iron is stored in the vacuole in the ferric state.

The yeast vacuole is the storage depot for cellular iron. In this report we quantify the import-export balance in the vacuole because of the import of iron by Ccc1p and to export by the combined activity of Smf3p and the ferroxidase, permease pair of proteins, Fet5p and Fth1p. Our data indicate that the two efflux pathways are equally efficient in trafficking iron out of the vacuole. A major focus of this work was to identify the ferrireductase(s) that supplies the Fe II for efflux whether by Smf3p or the Fet5p-Fth1p complex. Using a combination of flameless atomic absorption spectrophotometry to quantify vacuolar and whole cell iron content and a reporter assay for cytoplasmic iron we demonstrate that Fre6p supplies Fe II to both efflux systems, while Fre7p plays no role in Fe-efflux from the vacuole. Enzymatic assay shows the two fusions to have similar reductase activity, however. Confocal fluorescence microscopy demonstrates that Fre6:GFP localizes to the vacuolar membrane; in contrast, Fre7:GFP fusions exhibit a variable and diffuse cellular distribution. Demonstrating a role for a vacuolar metalloreductase in Fe-efflux supports the model that iron is stored in the vacuole in the ferric state.
Across all superkingdoms, transition metal ion homeostasis is strongly dependent on the regulation of nutrient accumulation rather than on excess nutrient excretion (1)(2)(3)(4)(5). Metal ion storage appears to be a complement of this regulatory pattern, storage that is illustrated by ferritin, the metallothioneins, and compartmentalization into cell organelles such as the mitochondria and vacuole. Although in some cases, this storage provides a detoxification mechanism (e.g. of heavy metals by the thioneins), it is as likely to play a dynamic role in cell and/or organismal metal ion homeostasis. This is certainly the case with ferritin in regards to vertebrate cell iron handling (2), and is very likely to be the case with the vacuoles found in the cells of many diverse eukaryotic genera; the vacuole in fungi and plants is a dynamic organelle that plays a significant role in the overall nutritional status of the organism (3)(4)(5). Vacuoles provide a storage depot for newly arrived nutrients as well as being the site of macromolecular degradation and nutrient recycling (6,7). In regards to metal metabolism the vacuoles in plants and fungi have been associated with the handling of copper, iron, manganese, and zinc, in addition to magnesium and calcium; the yeast vacuole is essential for the detoxification of several first row transition metals (8,9).
Iron metabolism in the yeast Saccharomyces cerevisiae has served as a paradigm for iron handling in many other fungi, plants and even humans (1,4,5,10). Under normal nutrient conditions ([Fe] environmental Յ 1 M) iron uptake is mediated by the high affinity iron uptake complex consisting of the multicopper ferroxidase Fet3p and iron permease Ftr1p. Homologous iron uptake complexes are likely to be found in most fungi and plants (10 -12). Uptake by yeast of environmental Fe III is preceded by ferric iron reduction to Fe II by plasma membrane reductase activity supplied principally by Fre1p; reduction of environmental Fe III (and Cu II ) is paradigmatic of the uptake of these metal ions by all aerobic organisms (3,11,13,14). Given the reduction potential of the cytoplasm, iron is undoubtedly present there as Fe II ; based on the GSH/GSSG ratio quantified in a wild-type yeast cell, this potential can be as low as Ϫ300 mV (15)(16)(17). This is likely to be the case in the cytoplasm where a major fraction of this GSH/GSSG is localized; in contrast, because of the overall non-equilibrium distribution of GSH in the cell, the reduction potential for the endoplasmic reticulum has been estimated at around Ϫ200 mV (17). Also, the GSH/ GSSG reduction potential is strongly pH-dependent: ⌬E/⌬pH is ϩ60 mV for a decrease in pH ϭ 1.0 (17). Thus, relatively acidic compartments like endosomes and the vacuole will be oxidizing in comparison to the cytoplasm; the presence of iron in the yeast vacuole as a ferric (poly)phosphate is consistent with this electrochemical condition (18).
One hallmark of iron metabolism in yeast is, therefore, redox cycling between Fe III (exocytoplasmic, vesicular) and Fe II (substrate for uptake, cytoplasmic) (1, 6, 11). As noted, Fre1p supports this cycling at the plasma membrane where exocytoplasmic ferric iron needs to be reduced prior to ferroxidation and uptake by the Fet3p, Ftr1p complex. The Fe II /Fe III ratio within the cell is likely determined by the overall reduction potential which, as noted, is strongly dependent on the [glutathione] total and pH. In addition, the GSH/GSSG ratio will be sensitive to the level of dioxygen; indeed, this ratio is a measure of cellular oxidative stress (17,19).
The differential redox speciation of iron in the cytoplasm and vacuole provides the context for the mechanism of the cycling of iron between these two compartments. Some of the proteins involved in this cycling in S. cerevisiae have been identified. Ccc1p is localized to the vacuolar membrane and appears to function as a divalent metal ion transporter; Ccc1p has been associated with calcium, manganese, and iron homeostasis in yeast (20 -22). Smf3p is a member of the Nramp divalent metal ion transporter family; Smf3p also exhibits relatively little selectivity among first row transition ions (23,24). Smf3p localizes to the vacuolar membrane (24). Both transporters are thought to support vectorial iron flux with Ccc1p supporting import (21) and Smf3p supporting export of iron from the vacuolar lumen (24) although neither activity has been directly demonstrated. A complex of Fet5p and Fth1p is likely to support export of vacuolar iron also; this complex is localized to the vacuolar membrane with an orientation that is identical to that of its homologues, Fet3p and Ftr1p (25). In other words, both complexes transport iron into the cytoplasmic compartment: Fet3p, Ftr1p from the exocytoplasmic space, and Fet5p, Fth1p from the vacuolar one.
As is the case for Fet3p, Ftr1p permeation, the substrate for Fet5p, Fth1p transport is most reasonably Fe II . However, as noted above, the valence state of vacuolar iron is likely Fe III . In parallel with plasma membrane iron import that starts with ferrireduction, so, too, does Fet5p, Fth1p export begin with the same reduction step. Of the seven yeast genes that encode reductase proteins, FRE6 is one of those up-regulated via the activity of the iron-responsive transcription factor, Aft1p (26). Transcription from the CCC1, FET5, FTH1, and SMF3 loci is also activated under conditions of iron deprivation (27). Although the expression of the latter gene is under control of the paralogous Aft2p transcription factor (28), the fact that all five loci are members of what has been referred to as the iron or Aft regulon suggests that together they could be part of the iron trafficking pathway that links cytoplasmic and vacuolar iron metabolism. In this pathway, we propose that Fre6p supplies the luminal reductase activity required to mobilize Fe II from vacuolar ferric (poly)phosphate stores. The data presented here support this proposition.

MATERIALS AND METHODS
Strain Construction-The yeast strains employed in this study were derived from DEY1457(MAT ade6 can1 his3 leu2 trp1 ura3) and are listed in Table 1. The fet5⌬ strain was a kind gift from D. Eide. The smf3⌬ strain was generated by using the HindIII-linearized smf3⌬::LEU2 plasmid pJS409, a kind gift from V. Culotta. Transformation of yeast with this linear plasmid resulted in deletion of the chromosomal SMF3 sequence from ϩ116 to ϩ1313, which was verified by PCR. The ccc1⌬ strain was generated by using the SacI-digested ccc1⌬::HIS3 plasmid (also a kind gift from V. Culotta) to transform DEY1457. The fre6⌬ strain was constructed by amplifying the KANMX4 sequence from the plasmid pFAKanMX4 (29); primers were used that contained sequences from the 5Ј-and 3Ј-flanking regions from the FRE6 locus. The resulting PCR fragment was transformed into DEY1457. Genomic DNA was analyzed by PCR for the gene disruptions using a primer flanking the FRE6 ORF and a primer in the integrated KAN gene. The strains for quantification of FET3 promoter activity (for Aft1p activation) were generated by transforming DEY1457 and various deletion strains derived from it, with the ApaI-digested FET3:lacZ plasmid (kindly provided by A. Dancis), which was integrated at the URA3 locus.
Growth Media-Stocks of yeast were maintained on standard yeast extract-peptone-dextrose media. The cultures for experimental analysis were grown in a synthetic minimal media (6.67 g/liter yeast nitrogen base without amino acids, 2% glucose, and appropriate mix of amino acids). The experiments involving the MET3 promoter were performed using the synthetic complete (SC) 2 media supplemented with methionine (3.75 mM) and cysteine (5 mM) for repression and a Met-and Cys-free medium for promoter activation.
Plasmid Construction-The methionine promoter plasmids were constructed by amplifying the 450-bp MET3 promoter inserting ApaI/SalI sites and cloning this fragment into pRS423, pRS424, pRS425, and pR426 (30). The open reading frames of SMF3, CCC1, FTH1, and FRE6 including 400 base pairs of 3Ј-UTR were amplified from yeast chromosomal DNA while the FET5 ORF and 3Ј-UTR was amplified from the YIp-FET5HA plasmid, a kind gift from D. Eide. These amplified genes were cloned into SalI/NotI sites in the MET3 promoter pRS series plasmids. The FRE6-GFP plasmid was constructed by amplifying FRE6 from chromosomal DNA and cloning it into the pMET3-424 plasmid at the SalI/NotI sites. The GFP ORF was inserted following amino acid residue 236 in the pMET3-FRE6 plasmid modified to include XbaI/AatII sites for this purpose. This placed the GFP in a loop predicted topologically to be in the cytoplasm. In the case of pMET3-FET7, the GFP ORF was inserted at four locations: in three predicted loop regions after residues 101, 150, and 227, respectively, and just prior to the FET7 termination codon to generate a C-terminal GFP fusion. Vacuole Isolation-Vacuoles were prepared from 600 ml of cells (OD 660 Х 1-1.5) grown in SC media. The cells were collected by centrifugation at 1800 ϫ g for 5 min, resuspended in 30 ml of 0.1 M Tris-SO 4 (pH 9.3) and 10 mM dithiothreitol and incubated for 10 min at 30°C. The cells were washed once with spheroplasting buffer (1.2 M sorbitol, 20 mM potassium phosphate, pH 7.4) and incubated with 20 g/ml lyticase for 45 min at 30°C. Spheroplasts were collected by centrifugation at 3500 ϫ g for 5 min and resuspended in 3.5 ml of 15% Ficoll buffer (15% Ficoll in 0.2 M sorbitol, 10 mM PIPES-KOH, pH 6.8). DEAE-Dextran (50 g/ml) was added to the spheroplasts, and the sample incubated for 3 min on ice and for 5 min at 30°C. The lysate (3.5 ml) was transferred to two SW41 tubes (Beckman Instruments) and overlaid sequentially with 3 ml of 8% Ficoll, 4 ml of 4% Ficoll, and 1 ml of buffer (0.2 M sorbitol, 10 mM PIPES-KOH, pH 6.8). The tubes were centrifuged at 110,000 ϫ g for 90 min. The vacuolar fraction was collected from the 0 -4% interface. Microscopic examination (see below) showed this fraction to contain Ͼ90% intact, sealed vacuoles with little contamination with other vesicular bodies or membrane fragments. The iron content of acid digested vacuoles was quantified by flameless atomic absorption spectrophotometry (fAAS) on a Perkin Elmer graphite furnace instrument. All analyses were performed in triplicate on samples from two independent experiments. The means Ϯ S.D. indicated in the figures were calculated from these combined data.
␤-Galactosidase and INT Reductase Assays-Strains carrying the pFET3::lacZ reporter were grown in SC medium. Midlog phase cells (1 ml) were resuspended in 650 l of Z buffer (0.1 M sodium phosphate buffer, pH 7.0, 10 mM KCl, 1 mM MgSO 4 ) containing 100 l of Y-PER reagent (Pierce) and 2 l of ␤-mercaptoethanol. The mixture was vortexed for 5 min at room temperature and 150 l of ONPG (4 mg/ml) was added. The reaction was stopped by addition of 400 l of 1 M Na 2 CO 3 . The reaction mixture was centrifuged for 10 min at high speed at 4°C. The absorbance of the supernatants was read at 420 nm and ␤-galactosidase activity was calculated in Miller units (31). Each condition was measured in triplicate using three separate cultures derived from separate transformants for each plasmid.
The reductase activity of GFP-tagged fusions of Fre1p, Fre6p, or Fre7p in membrane fractions was determined as described by Shatwell et al. (32). These proteins were expressed as above under pMET3 control in a fre1⌬ strain. Transformants were grown in the presence of 10 M copper and 100 M iron to repress endogenous reductase expression. A membrane fraction was isolated using differential centrifugation. The membrane was solubilized in the presence of 1% Triton X-100 and 8 g of total protein was incubated with 43 M 2-(4-iodophenyl-3(4-nitrophenyl)-5-phenyltetrazo-lium chloride (INT), 180 units/ml superoxide dismutase, 333 nM FAD, and 160 M NADPH in relaxation buffer (100 mM KCl, 3 mM NaCl, 3.5 mM MgCl 2 , 10 mM PIPES, 1 mM ATP, pH 7.3) in a total volume of 0.35 ml. The absorbance was measured at 490 nm in a Bio-Rad microtiter plate reader. Duplicate assays were carried out in the absence of added FAD or added NADPH; the reductase activi-ties quantified were dependent on FAD cofactor and NADPH substrate in all cases.
Confocal Fluorescence Microscopy-FM4 -64, DCFDA, and concanavalin A-Alexa 633 were purchased from Molecular Probes. For FM4 -64 labeling, the yeast cells were grown in SC media followed by incubation with a 30 M final concentration of FM4 -64 for 2 h. DCFDA labeling was used to confirm the integrity of the isolated vacuoles. For this, vacuoles were incubated with 10 M reagent for 30 min and then washed prior to microscopic examination. For concanavalin A-Alexa 633 staining, the cells were grown in SC medium and incubated with 100 M of stain for 1 h. The cells were then washed and resuspended in PBS. For double staining with FM4 -64 and ConA-Alexa 633, the yeast cells were grown in SC medium to OD 660 Х 0.4, 10 M of FM4 -64 was added to the growing cells and allowed to grow to OD Х 1.0. The cells were then collected and suspended in PBS with 100 M ConA-Alexa 633 and incubated for 30 min. The unbound dyes were then removed by washing two times in PBS. For GFP fusions, the yeast cells were grown in SC media to a density of ϳ2 ϫ 10 7 cells/ml. The pellet from ϳ1 ml cells was washed once with PBS and resuspended in 0.1 ml of PBS. The images were captured using a Bio-Rad MRC 1024 confocal imaging system attached to a Nikon Optiphot microscope using a ϫ60 objective. Images were recorded by excitation at 488 nm using a krypton:argon source.

Iron Accumulates in Fre6p-deficient Vacuoles-Vacuoles
lacking the Fe-importer, Ccc1p, contain less iron in the steady state than wild type (21). In contrast, vacuoles lacking the Feexporter, Smf3p, more strongly sequester iron from the cytoplasm (24). This pattern of vacuolar iron content is illustrated by the first three bars in Fig. 1. In addition, vacuoles lacking Fre6p accumulate even more iron in the steady state than vacuoles from the smf3⌬ strain (Fig. 1, last bar). One interpretation of this difference is that Fre6p supplies Fe II for both Smf3p and the alternative vacuolar Fe-export system comprised of the Fet5p ferroxidase and Fth1p iron permease (25).
The steady-state relationship between import, ferrireduction and export was examined in more detail by use of the regulated overproduction of Ccc1p and Smf3p in wild type and fre6⌬ cells. Using pMET3 as the promoter in the presence (repressing) and absence of methionine (activating), we examined the role of Fre6p in the balance between import and export of iron in the vacuole. Overproduction of Ccc1p in wild-type cells resulted in a 4-fold increase in steady-state vacuolar iron accumulation whereas overproduction of Smf3p alone reduced this iron accumulation by a factor of two (Fig. 2, open bars, compare with Fig. 1). Overproduction of Smf3p along with Ccc1p reduced by 50% the excess iron accumulation because of Ccc1p alone. In all cases, vacuolar Fe-accumulation in the fre6⌬ strain was increased (Fig. 2, shaded bars); the most dramatic increases (Ͼ2-fold) were in cells overproducing Smf3p, which in Fre6p wild type had the effect of counteracting the accumulation because of Ccc1p. The data indicate Fre6p is upstream of Smf3p in the Fe-export pathway from the vacuole.
As noted, the Fet5p, Fth1p ferroxidase-permease pair offers a second vacuolar Fe-efflux pathway. The role of Fre6p in this pathway was assessed in a similar fashion (Fig. 3). Although overproduction of Fet5p and Fth1p had little effect on iron accumulation in parental, wild-type cells (ϳ0.2 g of Fe/mg protein, Figs. 1 and 3), it reduced by 50% the excess iron accumulation because of Ccc1p overproduction (Fig. 3, open bars). This reduction was absent in the fre6⌬ strain indicating that Fre6p was upstream of Fet5p, Fth1p-mediated export as well (Fig. 3, shaded bars).
Cytosolic Iron as a Monitor of the Vacuolar Iron Import-Export Cycle-The balance between cytosolic and compartmentalized iron in yeast has commonly been monitored by use of a reporter plasmid in which the lacZ gene is placed under control of the Fe-responsive Aft1p transcription factor (21,24,25). When cytosolic iron is (relatively) low, Aft1p traffics to the nucleus where it activates transcription of genes in the Aft1p regulon; in (relatively) high cytosolic Fe, Aft1p returns to the cytoplasm (33). Using this reporter, the dependence of vacuolar iron accumulation on the overproduction of Ccc1p and Smf3p in wild type and fre6⌬ could be quantified. These data are shown in Fig. 4. In 4-h post-induction, wild-type cells overproducing Ccc1p exhibit an 8-fold increase in Aft1p activity; this increase is unaffected by the absence of Fre6p (Fig. 4 compare first two sets of open and  shaded bars). This parallels the vacuolar iron data shown in Fig. 3. Co-production of Smf3p reduces by 40% the increase in Aft1p transcriptional activity because of vacuolar Fe-ac-  Wild type (open bars) and fre6⌬ (shaded bars) strains were transformed with the plasmids pMET3:CCC1, pMET3:SMF3, or pMET3:CCC1 and pMET3:SMF3. The transformants were grown in 1.2 liters of minimal media supplemented with 7 mM methionine, 5 mM cysteine to an OD 660 Х 0.8. One-half the culture was harvested and vacuoles isolated. The remaining cells were washed twice with PBS and resuspended in minimal media without methionine. The cells were grown for additional 4 h and vacuoles isolated. Replicate numbers, acid digest, fAAS quantification of iron content and statistical analyses were as in Fig. 1. cumulation via Ccc1p consistent with the iron data shown in Fig. 3 (fourth open bar). Corresponding to the effect of the deletion of FRE6 on Smf3p-mediated Fe-efflux shown in Fig.  3, the ability of Smf3p to reverse the Aft1p activation due to the Fe-import supported by Ccc1p was absent in the fre6⌬ strain (fourth shaded bar). A similar correspondence was observed between the reporter and vacuolar iron assays in regards to the over-expression of the Fet5p, Fth1p ferroxidase, permease Fe-exporter. This export complex reversed the depletion of cytoplasmic iron because of overexpression of Ccc1p in wild type, but this reversal was attenuated by ϳ40% in the fre6⌬ strain (last two sets of bars). These data support further the inference that Fre6p acts upstream of both of the two Fe-export systems in the yeast vacuolar membrane.
Fre7p Does Not Support Vacuolar Fe-efflux-Fre7p differs from Fre6p in that is up-regulated by copper via the Cu-dependent transactivator, Mac1p. Also, Fre7p exhibits the least sequence homology to the other six yeast Fre reductases although it retains all of the sequence motifs involved in cofactor (heme) and substrate (pyridine nucleotide) binding (26). Nonetheless, we compared the effect of FRE7 deletion to that because of FRE6 deletion using the Aft1p-reporter assay to determine if it played any role in vacuolar Fe-efflux. The results of this experiment are shown in Fig. 5 in which the lacZ reporter assay was conducted in a fre7⌬ strain. Whereas deletion of FRE6 blocked Fe-efflux via Smf3p and the Fet5p, Fth1p ferroxidase, permease pathways (Figs. 2-4), deletion of FRE7 had no effect on Fe-export via either Smf3p or Fet5p, Fth1p (Fig. 5, compare all bars to open bars in Fig. 4).

Coordination of Plasma and Vacuolar Membrane Iron
Trafficking-We recognized that of the two measures of vacuolar membrane iron trafficking we used, vacuolar iron content, and the trans-activation of genes regulated by cytosolic iron, the latter assay was compromised by the extent to which plasma membrane iron trafficking contributed to the steady-state level of cell iron. For example, an increased vacuolar Fe-uptake because of overproduction of Ccc1p would decrease cytoplasmic iron and thereby stimulate Aft1p-dependent transcription that, in the end, would result in an increased Fe-uptake into the cytoplasm thus suppressing the signal obtained from our reporter assay. In a complementary fashion, the signal from a Smf3p overproducing culture would over-report on the increase in vacuolar iron efflux (seen as a decrease in reporter plasmid read-out) since the increase in cytoplasmic iron would decrease uptake of iron into the cell as well. We tested these predictions by measuring whole cell iron levels in the various cultures used in these studies. The data did, in fact, illustrate this balancing of plasma membrane Fe-uptake and cycling of iron out of and into the vacuole (Fig. 6).
Thus, in comparison to parental wild-type cells (indicated by the arrow) total cell iron was increased in a fre6⌬ but not in a fre7⌬ strain (first two open and shaded bars, respectively); total cell iron in the fre7⌬ strain and wild type was the same. This result is consistent with a model in which the reduced Fe-efflux into the cytoplasm in the fre6⌬ background in comparison to the fre7⌬ one resulted in a stronger activation of Fe-uptake at the PM. This relative difference was seen with overexpression of either vacuolar Fe-uptake (Ccc1p) or efflux (Smf3p or Fet5, Fth1p) activities (last three sets of bars), data that support the conclusion that Fre6p but not Fre7p is upstream from both pathways for vacuolar Fe-export. strains carrying a integrated copy of the pFET3:lacZ reporter were transformed with pMET3-regulated CCC1, SMF3, FET5 and FTH1containing plasmids either singly or together. The cells were grown to an OD 660 Х 0.8 in minimal media supplemented with 7.5 mM methionine ϩ 5 mM cysteine. Samples were removed for ␤-galactosidase assay, and the remaining cells were washed two times with PBS and suspended in minimal media without methionine. The cells were allowed to grow for 4 h and a sample taken out again for ␤-galactosidase activity. Statistical analyses were on triplicate samples from two independent experiments as above.

Fre6p and Fre7p Exhibit Reductase Activity Comparable to
Fre1p-To demonstrate if either Fre6p or Fre7p had a reductase activity, we prepared membrane extracts from a host fre1⌬ yeast strain producing GFP fusions of Fre1, Fre6, or Fre7 from a high copy vector under control of the MET3 promoter. The cells were grown in the presence of 10 M copper and 100 M iron to repress expression of endogenous reductase genes. In these cells under these growth conditions we expected that the major fraction of the reductase activity measured would result from episomal expression of a specific reductase gene. With such extracts it was not possible to perform metalloreductase assays with Fe II or Cu I as substrate (because of nonspecific metalloreduction) but we were able detect reductase activity using as substrate a standard one-electron acceptor for Fre proteins, INT (34). The results of this experiment are presented in Fig. 7, panel A and demonstrate that in this protocol an equivalent amount of INT reductase activity is produced irrespective of the reductase gene fusion being expressed. Although not quantitative, the Western blot shown in Fig. 7, panel B indicates that the GFP fusions were produced at comparable levels in the three cell samples. On this basis, Fre7p is indistinguishable as an INT reductase (at the least) in comparison to either Fre1p or Fre6p.
Fre6p but Not Fre7p Localizes to the Vacuolar Membrane-Neither Fre6p nor Fre7p has been localized in the yeast cell. We examined the locale of the GFP fusions noted above by confocal fluorescence microscopy. By this criterion, Fre6:GFP localized exclusively to the vacuolar membrane (Fig. 8). This was confirmed by co-staining with a dye specific for the vacuolar mem-brane, FM4 -64. The lack of any fluorescence from the vacuolar lumen indicated that little if any of the fusion protein was being internalized and degraded within the vacuole indicating that the localization observed was not a result of the overproduction; pMET3 was chosen expressly to avoid this type of false positive result (35).
In contrast to the distinct and reproducible localization of the Fet6:GFP fusion to the vacuolar membrane, none of the four   Fre7:GFP fusions constructed (see "Materials and Methods") exhibited a specific localization; fluorescence from these protein species was diffuse and primarily cytoplasmic with little if any accumulation in any cellular membrane, e.g. the vacuolar or plasma membranes (supplemental Fig. S1). Because all four Fre7p fusions, including the GFP insertion at residue 227 (homologous to the Fre6:GFP fusion), gave this result we infer that it was not due to the insertion of the GFP domain; all of these fusions were detectable in the cell by Western blot analysis (Fig. 7, panel B). When our Fre6p and Fre7p fusions were produced under control of their own promoters (under inducing low [Fe] and [Cu] conditions) they were undetectable by Western blot using anti-GFP antibody (data not shown).

DISCUSSION
Vacuolar iron in yeast is metabolically active; 70% of iron stores are mobilized from vacuoles in yeast switched from glucose to ethanol as carbon source, i.e. from fermentation to respiration (18). Yeast strains lacking a putative vacuolar iron import or export complex lag behind wild-type in this growth adaptation to a non-fermentable carbon source (25). There are two activities in the yeast vacuolar membrane shown to be involved in vacuolar iron mobilization. These include the Fet5p-Fth1p complex (25) and Smf3p, one of the three yeast homologs of the Nramp mammalian divalent metal iron transporters (24). Because the redox state of iron present in the cell is strongly dependent upon the GSH/GSSH ratio, which in turn is strongly pH-dependent (17), the relatively acidic vacuole is likely to be oxidizing in comparison to the cytoplasm; thus vacuolar iron is likely stored as a ferric (hydr)oxide and/or phosphate (18). In as much as Fe II is the substrate for the Fet5p-Fth1p complex and for Smf3p, ferrireduction is required for the mobilization of iron from the vacuole. Of the seven genes that encode reductase proteins in yeast, FRE6 is up-regulated by iron deprivation along with FET5, FTH1, and SMF3 (28). Our data indicate that Fre6p supplies (some of) the luminal reductase activity required to support the mobilization of Fe II from the vacuolar ferric iron stores and the export of iron from the vacuole via Smf3p and the Fet5p-Fth1p complex.
We have localized a GFP fusion of Fre6p specifically to the vacuolar membrane. In the background we used for our experiments Fre7:GFP does not co-localize to the vacuole; at the level of expression of the fusion because of pMET3 no specific cellular localization could be discerned for this reductase. On the other hand, at this level of expression Fre6p and Fre7p exhibited similar reductase activity, one that was equivalent to that measured for the plasma membrane metalloreductase, Fre1p.
In a recent, complementary report, Rees and Thiele (36) demonstrated a plasma membrane localization for a Fre7:GFP fusion protein, but only in an rsp5-1 background. RSP5 encodes a ubiquitin ligase that is involved in marking endosomal membrane proteins for degradation; rsp5-1 is a temperature sensitive allele that is inactivated at 37°C (37). In this background, plasma membrane localization of Fre7:GFP was observed clearly but only at the non-permissive temperature. This protein supported a PM metalloreductase activity under those conditions like that exhibited by Fre1p. Whether this PM localization resulted from suppression of ER quality control in the rsp5-1 background and the default trafficking of the fusion to the PM, or to a stabilization of the protein in the PM remains an open question.
Endogenous Fre6p and Fre7p differ in their regulation with the former produced under control of the Fe-sensitive transcriptional system of Aft1p/Aft2p and the latter under control of the Cu-sensitive, Mac1 protein. With their common regulation, the reductase/permease group of proteins that includes Fre6p, Smf3p, Fet5p, and Fth1p might be expected to represent a redundant metabolic pathway directed toward the mobilization of ferric iron from the vacuole. In this simple model, one could suggest that being Cu-regulated, Fre7p might play a role in the redox management of vacuolar copper, presumably reducing Cu II to Cu I as the plasma membrane reductases, Fre1p and Fre2p, do in support of cell Cu-uptake via the Cu-transporter, Ctr1p (38). Ctr2p is the vacuolar counterpart to Ctr1p, transporting copper out of the vacuole (39,40).
The complementary work by Rees and Thiele (36), however, indicates that Ctr2p-dependent Cu-efflux from the vacuole is, like Fe-efflux, dependent on Fre6p. Thus, Fre6p is similar to Fre1p in that both proteins function upstream from the iron and copper transport components in the vacuolar and plasma membranes of yeast, respectively. The two metalloreductases do, however, differ in their regulation since FRE1 and not FRE6 is under control of both the Fe-dependent Aft1p and Cu-dependent Mac1p transcription factors (28). Thus, whereas Fre1p activity is increased when either iron or copper is limiting Fre6p activity is increased only when cytoplasmic iron is low. This regulatory pattern suggests that Cu-efflux from the vacuole is stimulated only when an increase in Fe-uptake is signaled. This pattern can be rationalized in that the copper recycled from the vacuole can be used to activate the Cu-dependent Fet3p ferroxidase needed in support of Fe-uptake at the plasma membrane (11).
Last, we suggest our observations indicate the role of Fre7p as a metalloreductase will be found in some pathway other than membrane transport. One possibility is that as the endosomal, Steap3 reductase in mammals is required for efficient release of iron from diferric transferrin (41,42), Fre7p could be involved in the release of iron, as Fe II , from an intracellular ferric iron chelate. In as much as S. cerevisiae can accumulate environmental iron from siderophores (43), it is a reasonable hypothesis that these chelates could be the substrates for this reductase that as of now remains without a function.