Fre1p Cu2+ Reduction and Fet3p Cu1+ Oxidation Modulate Copper Toxicity in Saccharomyces cerevisiae*

Fre1p is a metalloreductase in the yeast plasma membrane that is essential to uptake of environmental Cu2+ and Fe3+. Fet3p is a multicopper oxidase in this membrane essential for high affinity iron uptake. In the uptake of Fe3+, Fre1p produces Fe2+ that is a substrate for Fet3p; the Fe3+ produced by Fet3p is a ligand for the iron permease, Ftr1p. Deletion of FET3 leads to iron deficiency; this deletion also causes a copper sensitivity not seen in wild type. Deletion of FTR1 leads to copper sensitivity also. Production in the ftr1Δ strain of an iron-uptake negative Ftr1p mutant, Ftr1p(RAGLA), suppressed this copper sensitivity. This Ftr1p mutant supported the plasma membrane targeting of active Fet3p that is blocked in the parental ftr1Δ strain. A ferroxidase-negative Fet3p did not suppress the copper sensitivity in a fet3Δ strain, although it supported the plasma membrane localization of the Fet3p·Ftr1p complex. Thus, loss of membrane-associated Fet3p oxidase activity correlated with copper sensitivity. Furthermore, in vitro Cu1+ was shown to be an excellent substrate for Fet3p. Last, the copper sensitivity of the fet3Δ strain was suppressed by co-deletion of FRE1, suggesting that the cytotoxic species was Cu1+. In contrast, deletion of CTR1 or of FET4 did not suppress the copper sensitivity in the fet3Δ strain; these genes encode the two major copper transporters in laboratory yeast strains. This result indicated that the apparent cuprous ion toxicity was not due to excess intracellular copper. These biochemical and physiologic results indicate that at least with respect to cuprous and ferrous ions, Fet3p can be considered a metallo-oxidase and appears to play an essential role in both iron and copper homeostasis in yeast. Its functional homologs, e.g. ceruloplasmin and hephaestin, could play a similar role in mammals.

Fre1p is a metalloreductase in the yeast plasma membrane that is essential to uptake of environmental Cu 2؉ and Fe 3؉ . Fet3p is a multicopper oxidase in this membrane essential for high affinity iron uptake. In the uptake of Fe 3؉ , Fre1p produces Fe 2؉ that is a substrate for Fet3p; the Fe 3؉ produced by Fet3p is a ligand for the iron permease, Ftr1p. Deletion of FET3 leads to iron deficiency; this deletion also causes a copper sensitivity not seen in wild type. Deletion of FTR1 leads to copper sensitivity also. Production in the ftr1⌬ strain of an iron-uptake negative Ftr1p mutant, Ftr1p(RAGLA), suppressed this copper sensitivity. This Ftr1p mutant supported the plasma membrane targeting of active Fet3p that is blocked in the parental ftr1⌬ strain. A ferroxidase-negative Fet3p did not suppress the copper sensitivity in a fet3⌬ strain, although it supported the plasma membrane localization of the Fet3p⅐Ftr1p complex. Thus, loss of membrane-associated Fet3p oxidase activity correlated with copper sensitivity. Furthermore, in vitro Cu 1؉ was shown to be an excellent substrate for Fet3p. Last, the copper sensitivity of the fet3⌬ strain was suppressed by co-deletion of FRE1, suggesting that the cytotoxic species was Cu 1؉ . In contrast, deletion of CTR1 or of FET4 did not suppress the copper sensitivity in the fet3⌬ strain; these genes encode the two major copper transporters in laboratory yeast strains. This result indicated that the apparent cuprous ion toxicity was not due to excess intracellular copper. These biochemical and physiologic results indicate that at least with respect to cuprous and ferrous ions, Fet3p can be considered a metallo-oxidase and appears to play an essential role in both iron and copper homeostasis in yeast. Its functional homologs, e.g. ceruloplasmin and hephaestin, could play a similar role in mammals.
There is a well recognized link between copper and iron metabolism. This link is illustrated by the copper dependence of high affinity iron uptake in the yeast Saccharomyces cerevisiae (1). This dependence is due to the role that copper plays as cofactor for the ferroxidase, Fet3p (2). Fet3p and the iron permease, Ftr1p, form a complex in the yeast plasma membrane (3). Genetic and biochemical studies show that in high affinity iron uptake Fet3p catalyzes the oxidation of Fe 2ϩ to Fe 3ϩ ; Ftr1p then transports the Fe 3ϩ into the cell (3)(4)(5)(6)(7). In humans, the Fet3p ortholog, ceruloplasmin, also catalyzes the oxidation of Fe 2ϩ for iron release into circulation (see Ref. 8 and references therein). Hephaestin, a membrane-bound ceruloplasmin paralog, is required for iron efflux from the placenta and enterocytes in mammals (9). Thus, the copper-dependent ferroxidase activity characteristic of these members of the multicopper oxidase family has a conserved role in iron metabolism.
Studies of FET3 suggest that this ferroxidase might also play a role in cellular defense against copper toxicity; that is, a yeast mutant carrying a disruption of the FET3 gene was found to be sensitive to high concentrations of Cu 2ϩ added to the medium. One explanation for this observation was that Fet3p functioned in copper detoxification by binding free copper, thus suppressing the pro-oxidant activity that this metal ion exhibits in the generation of oxygen free radicals (10). This copper-sensitive growth phenotype of fet3⌬ cells did correlate with a higher intracellular copper concentration than that observed in wild type cells (11).
In light of these observations we considered whether changes in iron status affected overall copper homeostasis. Thus, we investigated the molecular etiology of the copper sensitivity of a fet3⌬ mutant. We now have found that deletion of FTR1 causes a comparable copper-sensitive phenotype. Expression in an ftr1⌬ strain of an Ftr1p mutant protein that supports Fet3p⅐Ftr1p processing, including the activation of Fet3p and its plasma membrane localization but which is iron uptakenegative, suppressed the copper sensitivity of this mutant strain. On the other hand, localization of a ferroxidase-negative Fet3p to the plasma membrane failed to suppress this copper sensitivity. Biochemical studies using purified Fet3p showed that Fet3p catalyzes the oxidation of Cu 1ϩ to Cu 2ϩ . Last, deletion of the plasma membrane Cu 2ϩ reductase FRE1 suppressed the copper sensitivity exhibited by the fet3⌬ strain, whereas suppression of copper uptake activities did not. These results show that the role Fet3p plays in the cellular defense against copper toxicity is most likely associated with a novel cuprous oxidase activity acting in the extracellular space. Reasonably, the other multicopper oxidases that are ferroxidases, such as ceruloplasmin, hephaestin, and the bacterial CueO protein (12)(13)(14), share this additional metalloxidase activity and thereby support a comparable Cu 1ϩ detoxification pathway in their respective organismal milieu. DEY1457fre1⌬ (MAT␣ can1 his3 leu2 trp1 ura3 ade6 fre1::URA3), DEY1394 fre1⌬ (MAT␣ can1 his3 leu2 trp1 ura3 ade6 fre1::URA3 fet3::HIS3) were used for analyzing the suppressive effect of FRE1 deletion on the copper sensitivity of the fet3⌬ mutant. Strains ZY126 (MAT␣ can1 his3 leu2 trp1 ura3 ade6 fet4::LEU2 fet3::HIS3 ⌬ctr1::URA3) and ZY127 (MAT␣ can1 his3 leu2 trp1 ura3 ade6 fet4::LEU2 fet3::HIS3 ⌬ctr1::URA3) were used for analyzing the effect of FET4 and CTR1 deletions on suppressing copper sensitivity of the fet3⌬ strain. ZY126 and ZY127 were derived by one-step homologous recombination at the CTR1 locus in DEY1422 and DEY1422fet3⌬, respectively. Strain AMR76 was derived from DEY1394; it contained ftr1::TRP1 as well as fet3::HIS3 and was used in the in situ confocal fluorescence co-localization studies. The parental strains, DEY1457 and DEY1394, were provided by D. Eide (16). The CTR3 locus in all of these strains was non-functional due to a Ty2 transposon insertion (17). Plasmids YEp351, YEp351FTR1, and YEp351(RAGLA) were generously provided by A. Dancis (3). pDY133FET3 and pDY133FET3(T1D) were derived from pDY134, a low copy plasmid that contains the FET3 transcription unit (6).
Analysis of Copper Resistance-The various wild type and mutant strains were streaked onto agar containing synthetic complete (SC) 1 medium without or with CuSO 4 or with both CuSO 4 and (NH 4 )Fe(SO 4 ) 2 . The cells were incubated at 30°C for 3 days and photographed. Transformants of strain 42C carrying YEp351, YEp351FTR1, or YEp351FTR1(RAGLA) were streaked onto agar containing the appropriate drop-out synthetic medium (ϪLeu). The copper resistance of these cells was determined as above.
Kinetic Analysis of Metallo-oxidation by Fet3p, 59 Fe Uptake, and Cell Localization-Soluble forms of wild type Fet3p and of Fet3p(T1D) were purified as described (6,18,19). Cuprous ion oxidation was determined by oxygen consumption using the Oxygraph from Hansatech (www.hansatech-instruments.co.uk) as described (6,19). Stock solutions of cuprous chloride (Alfa) (in acetonitrile) and ferrous ammonium sulfate (Sigma) were freshly prepared in nitrogen-purged buffers; all transfers from this stock solution were done using gas-tight syringes. The buffer used for O 2 uptake measurements was air-saturated, pH 5.0, 0.1 M MES containing 5% (v/v) acetonitrile (final volume percent). 59 Fe uptake was performed in strain DEY1394 (fet3::HIS3) transformed with empty vector, pDY133FET3, or pDY133FET3(T1D) as described (19). Localization of Fet3p:GFP in live cells (transformants of AMR76) producing wild type Ftr1p and the Ftr1p(RAGLA) mutant or of Ftr1p:YFP in cells producing wild type Fet3p or Fet3p(T1D) was imaged using a Bio-Rad MRC 1024 confocal system equipped with a 15-megawatt krypton/argon laser and operating on a Nikon Optiphot upright microscope and an oil immersion 60ϫ 1.4NA objective. Optical sections were acquired at 0.5 m, and the XY resolution was set at Ն0.2 m using the Bio-Rad Lasersharp V3.0 software and later processed using Confocal Assistant V4.02. Fet3p:GFP was produced from plasmid pDY133FET3:GFP, and Ftr1p:YFP was produced from plasmid p703FTR1:YFP (19).

FET3 or FTR1 Deletion Induces a Copper Sensitivity in
Yeast-Deletion of FET3 leads to two discernible phenotypes, at least, iron deficiency (2, 3) and copper sensitivity (10,11). Reasonably, the copper sensitivity could arise from either the lack of Fet3p activity specifically or from the iron deficiency. To test if iron deficiency was the cause, we examined the copper sensitivity of an ftr1⌬ strain, since this deletion causes an iron deficiency equivalent to that observed in the fet3⌬ one (3). We screened first for this iron deficiency by defective growth on medium containing a non-fermentable carbon source (ethanol, YPE), an iron-demanding growth condition (Fig. 1, panel A). As expected, both mutant strains exhibited little if any respiratory growth, although they and the wild type grew equally well on agar containing synthetic complete medium with glucose as SC (panel B). However, on the same medium containing CuSO 4 , the ftr1⌬ strain, like the fet3⌬ mutant, exhibited a coppersensitive growth phenotype; the mutant cells, but not the wild type ones, exhibited a complete growth arrest at 800 M Cu 2ϩ as shown in Fig. 1, panel C.
These data not only confirmed the previous reports with respect the fet3⌬ strain (10,11) but demonstrated that deletion of either of the two genes encoding the components of the high affinity iron uptake complex induced a comparable copper sensitivity. Of course this result did not resolve the question of the molecular basis of this sensitivity. For example, if iron deficiency was the cause, iron application should rescue the growth of these two mutant strains in the presence of otherwise cytotoxic levels of copper. Indeed, we found that iron did rescue this growth defect (Fig. 1, panel D). These data parallel a report that showed that iron addition increased the growth rate of fet3⌬ cells in a liquid medium containing a high concentration of copper (11). On the other hand, these results did not exclude the possibility that the absence of active Fet3p in the two mutant strains contributed to the copper sensitivity. Note that the activation of Fet3p by the addition of its prosthetic group copper atoms requires the co-production and membrane targeting of an Ftr1p protein (3). Thus, both mutant strains were equivalently deficient in Fet3p activity.
Copper Sensitivity Correlates with the Loss of Fet3p Activity-Thus, we directly investigated whether loss of Fet3p activity in the plasma membrane was a contributing factor in the copper sensitivity exhibited by fet3⌬ strains. As noted above, Stearman et al. (3) demonstrate that deletion of FTR1 resulted also in a loss of Fet3p oxidase activity. That is, the ftr1⌬ strain, like the fet3⌬ mutant, is iron-deficient and devoid of Fet3p oxidase activity (3). Investigating the copper sensitivity in the ftr1⌬ strain gave us the means to assess the relative contributions to this sensitivity of iron deficiency versus the absence of plasma membrane Fet3p activity.
(3) demonstrate that high affinity iron uptake was virtually absent in strain 42C when it produced the Ftr1p(RAGLA) mutant as its sole Ftr1p protein, whereas Fet3p oxidase activity was unaffected. This defect in iron uptake is shown here in Fig. 2, panel A, as judged by the extremely slow growth of strain 42C carrying mutant Ftr1p(RAGLA) on the non-fermentable carbon source (ethanol). This respiratory defect was suppressed by iron supplementation as expected (Fig. 2, panel  B). In contrast to this failure of Ftr1p(RAGLA) to rescue the iron uptake deficiency in the ftr1⌬ strain, this iron uptakenegative Ftr1p mutant did suppress the copper sensitivity of this strain (Fig. 2 pattern indicates that iron deficiency per se does not correlate with the copper sensitivity exhibited by the ftr1⌬ mutant. This pattern is consistent, however, with the inference that the copper sensitivity is linked to the lack of active Fet3p in the yeast plasma membrane in this ftr1⌬ mutant strain. Stearman et al. (3) demonstrate that the Fet3p produced in a strain producing Ftr1p(RAGLA) was enzymatically active. That this Fet3p was trafficked normally to the plasma membrane was demonstrated directly by in situ confocal fluorescence microscopy. In this experiment, wild type Ftr1p or the Ftr1p(RAGLA) mutant was co-produced with Fet3p:GFP. As the image in Fig. 3, panel C demonstrates, this mutant Ftr1p supports the plasma membrane localization of Fet3p comparable with that of wild type Ftr1p (panel B); in the absence of any Ftr1p, Fet3p:GFP fails to localize to the plasma membrane (panel A). These results indicate that the copper sensitivity of fet3⌬ and ftr1⌬ mutants correlates most strongly with the absence of Fet3p in the yeast plasma membrane and not with the iron deficiency of these strains.
Fet3p Catalyzes Oxidation of Cu 1ϩ to Cu 2ϩ Ions-Currently, Fet3p is known only for its role in the oxidative conversion of Fe 2ϩ to Fe 3ϩ for iron uptake (1, 4, 5, 7). The results above suggested that we attempt to determine if Fet3p was catalytically active with Cu 1ϩ as substrate as well. Oxidation of Cu 1ϩ would naturally diminish copper toxicity because Cu 1ϩ is known to be more toxic than Cu 2ϩ (20). Indeed, wild type Fet3p is an active "cuprous" oxidase as demonstrated by the Fet3pcatalyzed O 2 uptake supported by Cu 1ϩ as substrate (Fig. 4,  panel A). In a standard velocity versus [Cu 1ϩ ] analysis employing an O 2 electrode to measure oxygen consumption (6,19), the kinetic constants for Cu 1ϩ turnover have been determined: K m ϭ 37.9 Ϯ 3.6 M, and k cat ϭ 79.2 Ϯ 2.7 min Ϫ1 . 2 These values can be compared with those for Fe 2ϩ oxidation under the same conditions: K m ϭ 5.4 M, and k cat ϭ 63.9 min Ϫ1 . 2 Importantly, a ferroxidase-negative Fet3p mutant, Fet3p(T1D), failed to catalyze this oxidation of Cu 1ϩ (Fig. 4, panel A). This mutant contains a C484S substitution and so lacks the type 1 copper required for the initial electron transfer from the reducing substrate; it retains the remaining three copper atoms found in wild type Fet3p, however (18). These results suggest that Fet3p most reasonably could suppress copper toxicity by catalyzing the conversion of Cu 1ϩ to the less toxic Cu 2ϩ rather than by serving as a "sink" for excess cell copper.
A Cuprous Oxidase-negative Fet3p Does Not Protect against Copper Sensitivity-This conclusion was tested by evaluating the ability of the cuprous oxidase-negative Fet3p(T1D) to suppress the copper sensitivity of a fet3⌬ strain. We first demonstrated that this mutant Fet3p failed to support iron uptake in this background (Fig. 4, panel B); whereas wild type Fet3p conferred 59 Fe uptake to this strain (closed circles), the Fet3p(T1D) mutant did not (open circles). Uptake by this transformant was not different from the vector-only control (ϫ's). On the other hand, this protein did support the trafficking of the Ftr1p⅐Fet3p(T1D) complex to the plasma membrane. This was determined by co-expression of either wild type Fet3p or Fet3p(T1D) and Ftr1p:YFP (Fig. 3, panels D--F). In the absence of any Fet3p protein, the Ftr1p:YFP fusion remained intracellular (panel D); in contrast, in the presence of either wild type Fet3p or Fet3p(T1D), this fusion protein was strongly localized to the plasma membrane (panels E and F, respectively). We can conclude from these results that a strain producing Fet3p(T1D) has a structurally intact but functionally inactive Fet3p⅐Ftr1p complex in the plasma membrane.
This mutant Fet3p did not suppress the copper sensitivity observed for the fet3⌬ strain (Fig. 5). Whereas transforming this deletion strain with pDY133FET3(WT) suppressed the copper sensitivity, transformation with pDY133FET3(T1D) did not (Fig. 5, panel B). This result is consistent with the hypothesis that the catalytic activity of Fet3p correlates with this protein's link to copper resistance in S. cerevisiae, most reasonably as a result of the cuprous oxidase activity demonstrated above.
Deletion of the Cu 2ϩ Reductase FRE1 Suppresses the Coppersensitive Growth Phenotype of fet3⌬ Cells-The cell surface metalloreductase Fre1p catalyzes reduction of Cu 2ϩ to Cu 1ϩ for 2 C. Stoj and D. Kosman, manuscript submitted for publication. Note that in the absence of the partner protein, the fluorescent fusion remained intracellular (panels A and D); with co-production of the partner protein, wild type or mutant, the fusion was trafficked to the plasma membrane (panels B and E and C and F, respectively). copper uptake as well as reducing Fe 3ϩ to Fe 2ϩ for iron uptake (6,7). If the role of Fet3p in copper sensitivity is linked to its activity toward Cu 1ϩ , then as the source of this Cu 1ϩ , Fre1p should be epistatic to Fet3p in this copper sensitivity. We tested this inference by comparing the copper sensitivity of wild type (WT), fet3⌬, fre1⌬, and fet3⌬ fre1⌬ cells. These isogenic strains grew equally well on SC plates (Fig. 6, panel A). In the presence of exogenous copper, the fet3⌬ mutant exhibited, as expected, a severe growth defect; in contrast, under the same conditions, the fet3⌬ fre1⌬ double deletion mutant was able to grow as well as the wild type and fre1⌬ strains, exhibiting little if any growth defect in this assay (Fig. 6, panel B). This growth difference between the fet3⌬ and fet3⌬ fre1⌬ strains demonstrated that deletion of FRE1 significantly suppressed the copper sensitivity of the fet3⌬ cells. This result supports the model that Fet3p suppresses copper toxicity by catalyzing the conversion of Fre1p-produced Cu 1ϩ to Cu 2ϩ .
Overproduction of Fre1p Induces Copper Sensitivity; Deletion of Genes Encoding Copper Uptake Proteins Does Not-The distinguishing aspect of this model is that the copper sensitivity in the fet3⌬ strain results from a defect in the handling of copper on the extracellular surface of the plasma membrane rather than as a consequence of aberrant copper accumulation. This model predicts that overproduction of Fre1p, and, therefore, of Cu 1ϩ at this surface would exaggerate copper sensitivity, whereas deletion of genes encoding copper transport proteins would not. We tested these two interrelated predictions by taking advantage of the MAC1 up1 allele (21,22). Mac1p is a transcriptional activator of the CTR1 and FRE1 genes that is repressed by copper in the growth medium (21,(23)(24)(25); Mac1p, therefore, is the key player in control of copper homeostasis in yeast. The Mac1 up1 protein carries a single point mutation that confers to the protein a copper-independent, constitutive transcriptional activity. Thus, FRE1 is strongly expressed in a strain carrying this allele (21,22); this strain exhibits a strongly elevated copper (and iron) reductase activity (22) and copper sensitivity (21). This latter phenotype is illustrated in Fig. 7, panel B, and is consistent with the prediction our model affords. The model is further supported by the fact that deletion of FRE1 in the MAC1 up1 background suppressed this strain's copper sensitivity (Fig. 7, panel B).
However, this result does not exclude the possibility that the copper toxicity observed in either fet3⌬ or ftr1⌬ cells or in the MAC1 up1 strain was linked also to an increase in copper accumulation. Increased copper accumulation has been reported in fet3⌬ cells (11) and in yeast carrying the MAC1 up1 allele (22). As noted above, CTR1 is strongly up-regulated in the MAC1 up1 background. To assess the contribution that increased copper uptake might make to the copper-sensitive phenotypes documented above, we carried out two experiments. First, we assessed whether deletion of CTR1 suppressed the copper sensitivity linked to the constitutively active Mac1 up1 . Our results show clearly that, unlike the suppressive effect of FRE1 deletion (Fig. 7, panels A and B), deletion of CTR1 had little or no impact on the copper sensitivity associated with the gain-offunction Mac1 protein (Fig. 7, panels C and D); this result has been noted previously (27). Second, we determined whether deletion of CTR1 or of the low affinity copper transporter, FET4 (26), could suppress the copper sensitivity in the fet3⌬  strain. The results of this experiment are shown in Fig. 7, panels E and F. As demonstrated by the plate shown in Fig. 7, panel F, deletion of FET4 alone, of CTR1 alone, or of both loci was without effect on the copper sensitivity exhibited by the fet3⌬ strain. These two sets of results strongly support our model in two ways. They indicate first that it is the increase in Fre1p metalloreduction and not the increase in Ctr1p-dependent copper uptake that leads to the copper sensitivity of the MAC1 up1 strain. Second, they demonstrate that FRE1 and not CTR1 (nor FET4) interact physiologically with FET3 in yeast copper homeostasis. Note that in all of the backgrounds used for the experiments in this work, the CTR3 locus was inactivated by a Ty2 insertion found in most laboratory strains of S. cerevisiae (17). Ctr3p, when produced, exhibits a copper uptake activity comparable with that supported by Ctr1p (17). DISCUSSION High affinity copper uptake through Ctr1p (and Ctr3p) in S. cerevisiae is preceded by reduction of Cu 2ϩ to Cu 1ϩ , a process mediated by the cell surface metalloreductase , Fre1p (7, 22).
Here, we have described the reverse of that process in which Cu 1ϩ is apparently re-oxidized to Cu 2ϩ , a process catalyzed by the cell surface ferrous oxidase (ferroxidase) Fet3p (2,(5)(6)(7). That is, we suggest that the cuprous oxidase activity of Fet3p we have demonstrated here suppresses the toxicity of copper present in the growth medium. The suppression of the coppersensitive growth phenotype of the fet3⌬ cells by the simultaneous deletion of FRE1 is consistent with this suggestion. In summary, our data suggest the more general conclusion that copper homeostasis is established in part by the physiologic interplay between the cupric reductase, Fre1p, and the cuprous oxidase, Fet3p, as depicted in Fig. 8.
These inferences are supported by the following data. First, we suppressed the copper sensitivity of an ftr1⌬ strain by production in this strain of the Ftr1p(RAGLA) mutant. Stearman et al. (3) show that the Ftr1p(RAGLA) mutant was inactive in iron uptake. We confirmed this by demonstrating the respiration deficiency exhibited by the ftr1⌬ strain producing this mutant protein, a growth deficiency corrected by the addition of iron salt to the medium. Stearman et al. (3) demonstrate also that this mutant Ftr1p did support the processing and activation of Fet3p. We now have shown that this Fet3p is localized to the plasma membrane. Taken together these data indicate that the loss of FET3 function in the plasma membrane, but not the iron deficiency that results from this loss, is causal in the copper sensitivity in the fet3⌬ and ftr1⌬ strains.
We have supported this conclusion by the demonstration that purified Fet3p catalyzes the oxidation of Cu 1ϩ to Cu 2ϩ ions in vitro. Our O 2 uptake data show that Fet3p is an excellent cuprous oxidase with an efficiency with Cu 1ϩ as substrate that is only slightly less than that exhibited with Fe 2ϩ as substrate. This fact is of particular significance because Cu 1ϩ is known to be more toxic to cells than Cu 2ϩ (20). Furthermore, the suppression of the copper-sensitive growth phenotype of fet3⌬ cells by FRE1 deletion offers additional support to the model that Fet3p protects cells from the deleterious effects of copper by converting Cu 1ϩ generated by Fre1p to the less toxic Cu 2ϩ redox state of this metal ion. The apparent genetic interaction between FRE1 and FET3 that we have inferred provides direct evidence that copper homeostasis is based partly on the modulation of the redox balance of this metal ion.
Oxidation of Cu 1ϩ by Fet3p would diminish the potential toxic effects of copper ions. In addition, the oxidation could also deprive Ctr1p of its substrate, Cu 1ϩ . Therefore, the oxidation of Cu 1ϩ might also function as a mechanism for limiting copper uptake by Ctr1p. Consistent with this notion, the Kaplan group found that a fet3⌬ mutant accumulated more copper than the wild type (11). On the other hand our genetic data indicate that FIG. 7. Copper sensitivity in the MAC1 up1 and fet3⌬ backgrounds is suppressed by deletion of FRE1 but not by deletion of genes encoding copper transport proteins. Strains carrying targeted gene deletions as indicated were streaked onto agar containing SC growth medium without (panels A, C, and E) or with CuSO 4 (700 M, panels B, D, and F). The cells were incubated at 30°C for 3 days and photographed. WT, wild type. FIG. 8. Model of copper and iron metabolism in S. cerevisiae. Note the dual roles of Fre1p and Fet3p. Fre1p generates Fe 2ϩ and Cu 1ϩ for high affinity uptake of both metal ions; Fet3p targets the Fe 2ϩ for iron uptake through ferroxidation and buffers the Cu 1ϩ /Cu 2ϩ ratio through cuprous oxidation. copper uptake per se is not responsible for the copper-sensitive phenotype of the fet3⌬ strain since deletion of the genes encoding the primary copper transporters found in most laboratory strains of S. cerevisiae, CTR1, and FET4 had no mitigating effects. Therefore, the data are consistent with the inference that the increased accumulation of copper in the fet3⌬ background reflects an increase in the steady-state level of Cu 1ϩ available to Ctr1p. However, our genetic and biochemical data indicate that, most reasonably, it is the increased extracellular Cu 1ϩ that is linked to the copper sensitivity exhibited by this strain, not the increased copper accumulation, since deletion of CTR1, in particular, was without protective effect.
Our data cannot fully exclude the possibility that Fet3p also detoxifies copper by scavenging copper ions, as proposed by the Thiele group (10). Note, however, that the cuprous oxidasenegative Fet3p mutant, Fet3p(T1D), did not complement in a fet3⌬ strain. The Fet3p(T1D) protein retains three of the four copper atoms found in wild type Fet3p (6,18); there is no evidence to suggest that the type 1 copper that is missing in this protein might have some role in copper metabolism not shared by the type 2 and type 3 copper atoms. Rather, this result is most easily understood in terms of a catalytic rather than copper binding mechanism of Fet3p protection against environmental copper.
Because both Fre1p and Fet3p are known to be crucial for iron uptake, the current work, thus, places these two proteins at the interface between copper and iron homeostasis (Fig. 8).
In response to changes in iron conditions, FRE1 and FET3 are regulated similarly, with their transcription enhanced under iron deficiency and repressed when iron is in excess. FRE1 expression is also altered when copper concentration changes; similar to CTR1 expression, FRE1 transcription is up-regulated under copper deficiency and repressed when copper is replete (22,28,29). More recent reports from Thiele and coworkers (30) and Winge and co-workers (31) have describe that FET3 expression is modulated in response to changes in copper conditions as well, but in a fashion opposite to the response exhibited by the CTR1 and FRE1 loci; copper deficiency represses FET3 transcription, whereas copper excess enhances it (30,31). This transcriptional behavior suggests that there is regulatory coordination between FET3 and FRE1/CTR1 in response to a changing copper environment. In fact, we have found that CTR1 expression is modulated by changes in iron status. 3 These several results may reflect some coordination between copper and iron homeostasis; we are currently investigating the molecular bases of this apparent coordination.
Eukaryotes produce quantities of Cu 1ϩ in the course of normal copper metabolism via the action of metalloreductases. In S. cerevisiae this reductase is the Fre1p protein (7,32); in the mammalian epithelial cell, it is the Dcytb protein (33). Obviously, as an enzyme product, this Cu 1ϩ has the potential to equilibrate with bulk solvent; that is, it can be considered "free." This certainly is the case in the yeast Cu 2ϩ reduction/ copper uptake system where cuprous copper for uptake can come directly from the bulk solvent and not necessarily from Fre1p (22). In short, eukaryotic cells normally produce a fairly aggressive pro-oxidant. In this context, our model predicts generally that an imbalance between Cu 1ϩ production and turnover would lead to an exaggerated copper sensitivity. This prediction appears born out by the yeast mutant carrying a gain-of-function mutation in the MAC1 gene (21)(22)(23)(24)(25). Mac1p is a transcriptional activator of expression of CTR1 and FRE1, whose activity is repressed by Ͻ1 M copper in the growth medium. The MAC1 up1 allele encodes a single point mutation that makes the Mac1 up1 protein insensitive to copper; therefore, Mac1 up1 constitutively activates transcription of the CTR1 and FRE1 genes, thus, strongly up-regulating plasma membrane reductase activity and copper uptake (22). In the context of our model we have shown here that the copper sensitivity exhibited by strains expressing the MAC1 up1 allele is not, apparently, due to an increase in intracellular copper. If it were, then deletion of CTR1 would be expected to suppress the copper sensitivity to a significant extent. In fact, such deletion had no such effect. Of note is the fact that in addition to activating transcription from CTR1 and FRE1, the Mac1 up1 protein appears to repress transcription from the FET3 locus (30). Thus, the MAC1 up1 carrying strain may be doubly at risk to environmental copper; that is, increased Cu 1ϩ production together with decreased Cu 1ϩ turnover. Note, however, that the repression by Mac1 up1 of FET3 expression has not been consistently observed (31).
Our data indicate that iron deficiency is not the sole or primary cause of the copper sensitivity observed in fet3⌬ and ftr1⌬ strains. This inference follows most directly from the fact that the copper sensitivity is suppressed by the trafficking of active Fet3p to the plasma membrane along with the iron uptake-negative Ftr1p(RAGLA) mutant. What then is the explanation for the fact that iron supplementation suppressed the copper sensitivity of the iron uptake-deficient yeast mutants? One explanation of this latter observation is that iron protects from copper toxicity at the level of Fre1p; Fe 3ϩ competes with Cu 2ϩ for reduction by this cell-surface reductase. In fact, the reactivity of Fe 3ϩ with Fre1p is ϳ70-fold greater than the reactivity of Cu 2ϩ when both are present at 40 M (22). To the extent that this difference reflects a greater affinity by Fre1p for Fe 3ϩ , a significant inhibition by Fe 3ϩ of the concurrent reduction of Cu 2ϩ can be predicted. In relation to the relative cytotoxicity of the products of this Fre1p reaction, Cu 1ϩ and Fe 2ϩ , it is informative that the Cu 2ϩ /Cu 1ϩ cycle is far more reactive under O 2 . As an example of the one-electron oxidative cycle that copper supports more efficiently in comparison to iron, dihydroascorbate is essentially stable in the presence of 10 M Fe 3ϩ , whereas this one-electron reductant is completely turned over in the presence of 10 M Cu 2ϩ (34). Cytotoxicity derives from this reaction by the production of the one-electron reduction products of O 2 that this redox cycle generates, primarily the superoxide and hydroxyl radicals.
The suggestion that Fet3p plays a specific physiologic role in preventing this one-electron cycle via its four-electron reduction of O 2 to water using 4Cu 1ϩ as substrate is predicated on the assumption that there is a significant steady-state level of reduced metal species at the cell surface due to the Fre1p reaction. This situation would require that the relative velocities of the two processes involved in copper uptake, reduction and transport, and the three involved in iron uptake, reduction, ferroxidation, and transport, were V R Ͼ V F Ͼ V T , with the ferroxidation step included for the case of iron. In fact, for iron these relative velocities are ϳ500 (22), 4 (5), and 0.4 (22) nmol of iron/min/mg of cell protein, respectively. The difference between the reduction and uptake velocities for copper is smaller due to the less robust reduction of Cu 2ϩ by Fre1p noted above; the reduction and uptake values for copper are 7 and 0.7 nmol of copper/min/mg of cell protein, respectively (22). Thus, the rate of formation of the lower valence metal ion is severalfold greater than the rate of this metal ion uptake (Cu 1ϩ ) or catalytic re-oxidation (Fe 2ϩ ). One could infer that the extracellular steady-state level of Cu 1ϩ increases in the MAC1 up1 strain based on the copper sensitivity of this strain. In fact, in this background the velocities of Cu 2ϩ reduction and Cu 1ϩ uptake are each increased ϳ7-fold to 46.2 and 5.1 nmol of copper/ 3 X. Shi and Z. Zhu, manuscript in preparation. min/mg of protein, respectively (22). Thus, the molar difference between uptake and reduction is increased from ϳ6 nmol of copper/min/mg of protein in wild type yeast to 40 nmol of copper/min/mg of protein in the MAC1 up1 strain, an increase of nearly 700%. In short, the data available are completely consistent with the notion that there is a steady-state level of both Cu 1ϩ (and Fe 2ϩ ) at the yeast plasma membrane. The results here indicate that the level of this plasma membrane Cu 1ϩ is modulated by the action of Fet3p and that this effect contributes to the copper resistance of S. cerevisiae.
The CueO protein, a multicopper oxidase from Escherichia coli that is encoded within a copper resistance operon in the genome of this bacterium, is a ferroxidase that exhibits a pattern of substrate reactivity similar to that recorded for ceruloplasmin (12)(13)(14). Three groups working with this protein and its encoding cue system (for Cu efflux) have suggested that CueO (previously known as the yakK gene product) could contribute to copper resistance by acting as a cuprous oxidase, thus shifting the Cu 2ϩ /Cu 1ϩ ratio toward Cu 2ϩ . In this model, the cupric ion would be less readily taken into the cell in addition to its being a weaker pro-oxidant, thus explaining the cytoprotective effect of the CueO protein. However, one of these groups reported that despite a substrate reactivity profile comparable with that of ceruloplasmin and Fet3p, including its reactivity with Fe 2ϩ , CueO exhibited no oxidase activity with Cu 1ϩ as substrate (14). Therefore, the role of cuprous oxidase activity in copper homeostasis in prokaryotes remains unresolved.
Serum ceruloplasmin plays a critical role in iron release from the reticuloendothelial system by catalyzing the oxidation of Fe 2ϩ to Fe 3ϩ ions (8). The central nervous system also expresses ceruloplasmin as a glycophosphatidylinositol-anchored protein (35)(36)(37). This form of ceruloplasmin most likely plays a critical role in iron metabolism of the central nervous system since patients with aceruloplasminemia, a neurodegenerative disease caused by loss-of-function mutations in the ceruloplasmin gene, accumulate excess iron within the central nervous system (8). Ceruloplasmin shares the cuprous oxidase activity exhibited by Fet3p; 2 it is likely that this glycophosphatidylinositol-anchored ceruloplasmin also catalyzes the oxidation of Cu 1ϩ ions. As a consequence, mutations in the ceruloplasmin gene might also impair the capacity of the brain to detoxify copper ions. Indeed, copper toxicity might well contribute to the oxidative damage and neurodegeneration associated with aceruloplasminemia. Future studies must now evaluate further what role(s) this family of oxidases plays in both iron and copper homeostasis.