A Copper-sensing Transcription Factor Regulates Iron Uptake Genes in Schizosaccharomyces pombe *

Copper and iron serve essential functions as catalytic co-factors in a wide variety of critical cellular enzymes. Studies in yeast have demonstrated an absolute dependence upon copper acquisition for proper assembly and function of the iron transport machinery. We have cloned genes for a high affinity copper transporter (Ctr4) and copper-sensing transcription factor (Cuf1) fromSchizosaccharomyces pombe. Interestingly, the primary structure of Ctr4 and a putative human high affinity copper transport protein, hCtr1, suggests that they are derived from a fusion of the functionally redundant but structurally distinct Ctr1 and Ctr3 copper transporters from Saccharomyces cerevisiae. Furthermore, although Cuf1 activates ctr4 + gene expression under copper starvation conditions, under these same conditions Cuf1 directly represses expression of genes encoding components of the iron transport machinery. These studies have identified an evolutionary step in which copper transport modules have been fused, and describe a mechanism by which a copper-sensing factor directly represses expression of the iron uptake genes under conditions in which the essential copper co-factor is scarce.

Copper and iron serve essential functions as catalytic co-factors in a wide variety of critical cellular enzymes. Studies in yeast have demonstrated an absolute dependence upon copper acquisition for proper assembly and function of the iron transport machinery. We have cloned genes for a high affinity copper transporter (Ctr4) and copper-sensing transcription factor (Cuf1) from Schizosaccharomyces pombe. Interestingly, the primary structure of Ctr4 and a putative human high affinity copper transport protein, hCtr1, suggests that they are derived from a fusion of the functionally redundant but structurally distinct Ctr1 and Ctr3 copper transporters from Saccharomyces cerevisiae. Furthermore, although Cuf1 activates ctr4 ؉ gene expression under copper starvation conditions, under these same conditions Cuf1 directly represses expression of genes encoding components of the iron transport machinery. These studies have identified an evolutionary step in which copper transport modules have been fused, and describe a mechanism by which a copper-sensing factor directly represses expression of the iron uptake genes under conditions in which the essential copper co-factor is scarce.
The mechanisms by which organisms accumulate, sense and respond to nutrients is a critical aspect of cellular growth control. Copper and iron are two metal ions that are essential for the growth of all organisms, serving as redox active cofactors for a wide variety of cellular enzymes including mitochondrial cytochromes, multi-copper ferroxidases, and superoxide dismutases (1)(2)(3)(4). Furthermore, as a consequence of chemical reactions between copper or iron and oxygen, these same metals generate reactive oxygen species that damage nucleic acids, proteins and membranes (5). When copper and iron accumulate in excess or are abnormally mobilized, a number of disease states are manifest such as Menkes and Wilson diseases and anemia, disorders of copper and iron acquisition, respectively, and hemochromatosis, an iron overload condition that is one of the most common inherited disorders affecting people of Northern European genealogy (6 -9). Therefore, it is critical for cells to maintain sufficient levels of copper and iron to carry out essential biochemical reactions, but regulatory mechanisms must be in place to prevent these metals from accumulating in excess.
The bakers' yeast, Saccharomyces cerevisiae, is an excellent model system to understand the function and regulation of fundamental mechanisms for copper and iron uptake and distribution in eukaryotic cells. Biochemical and genetic experiments have shown that copper and iron ions are reduced by cell surface Fe 3ϩ /Cu 2ϩ reductases, encoded by the FRE genes (10 -13). Cu ϩ is transported into S. cerevisiae cells via two plasma membrane high affinity copper transporters, Ctr1 (14) and Ctr3 (15). The Ctr1 and Ctr3 proteins are functionally redundant; while cells expressing either Ctr1 or Ctr3 are fully competent for high affinity copper transport, ctr1⌬ ctr3⌬ cells exhibit copper starvation phenotypes including the inability to grow on nonfermentable carbon sources, lack of measurable high affinity iron uptake activity, and defective copper/zinc SOD activity (14,15). Interestingly, Ctr1 and Ctr3 are structurally distinct proteins, with Ctr1 harboring several potential copper binding motifs (M-X 2 -M-X-M) in the amino-terminal extracellular domain while Ctr3 lacks these motifs, but has an abundance of cysteine residues throughout the protein. Three yeast copper chaperone proteins have been identified, Atx1 (16,17), CCS (18), and Cox17 (19), that deliver copper from the cytosol to a late secretory compartment, cytosolic copper/zinc superoxide dismutase, and mitochondria, respectively. Copper is delivered into the lumen of the Golgi/endosome by the Ccc2 protein (20), a P-type ATPase similar to those known to be mutated in Menkes and Wilson disease (21)(22)(23)(24)(25), where it is incorporated into the Fet3 multi-copper ferroxidase (26). Together with the Ftr1 iron permease, copper-metallated Fet3 traverses the secretory pathway to generate an active iron transport complex at the plasma membrane (7,27).
Many of the genes involved in copper or iron transport are transcriptionally regulated by copper or iron availability (4). CTR1, CTR3, FRE1, and FRE7 are induced by copper starvation and repressed by copper repletion. This regulation involves cis-acting copper-responsive elements (CuREs), 1 found in each of these promoters, with the consensus sequence 5Ј-TTTGC(T/ G)C(A/G)-3Ј, and a copper-sensing transcription factor Mac1 (13, 28 -30). In vivo footprinting has demonstrated that Mac1 is bound to CuRE elements under conditions of copper scarcity and is not bound under conditions of copper sufficiency (28). Furthermore, Mac1 undergoes copper-responsive intra-molecular interactions that modulate trans-activation function (31). Mac1 has homology with the amino-terminal 40 amino acids of the minor groove DNA binding domain found in the Ace1/Amt1 copper-activated DNA-binding proteins that activate metallothionein gene expression, but little homology outside of this region (32,33). In the Mac1 carboxyl-terminal region reside two cysteine-rich repeats, REP-I and REP-II, that have been demonstrated to function in copper sensing (34,35). The MAC1 up1 allele, one of several dominant alleles with mutations in REP-I, encodes a protein unable to sense copper and, as a consequence, is constitutively bound to CuREs and strongly constitutively activates expression of CTR1, CTR3, and FRE1/7 (13,28). Like genes involved in copper transport, FET3, FTR1, and many other iron homeostatic genes are regulated as a function of iron availability through conserved cis-acting promoter elements by the Aft1 iron-sensing transcription factor (36,37). Indeed, the interdependence of the iron uptake pathway on copper availability is also underscored by the observations that FRE1 is regulated by copper and iron and the ATX1 and CCC2 genes, encoding copper trafficking proteins, are regulated by iron and Aft1, but not copper (10,16,32,37).
The fission yeast Schizosaccharomyces pombe is phylogenetically distinct from S. cerevisiae, exhibiting several features more similar to mammalian cells. For example, S. pombe utilizes pathways involved in stress-responsive signal transduction that are conserved in humans but not in S. cerevisiae (38 -40). Moreover, little is known about trace metal ion homeostasis in fission yeast. In S. pombe three components involved in the high affinity iron uptake system have been identified. Frp1, which shares amino acid sequence similarity with the Fre reductases from S. cerevisiae and the mammalian gp91 phox subunit of cytochrome b 558 , is required at the cell surface to reduce Fe 3ϩ (41,42). Once reduced, Fe 2ϩ is taken up by a permease/oxidase complex, Fip1/Fio1, and transported into the cell, as occurs with the orthologous Fet3/Ftr1 complex in S. cerevisiae (43). To gain further insight into the structure and regulation of proteins involved in copper homeostasis, we have identified S. pombe genes encoding a copper-responsive transcription factor and a high affinity copper transporter. Interestingly, the S. pombe Cuf1 copper-sensing transcription factor is composed of modules from both the Mac1 and Ace1/Amt1 families of copper metalloregulatory transcription factors (44,45). Likewise, the S. pombe high affinity copper transporter, Ctr4, has no functionally redundant counterpart and resembles a fused version of the S. cerevisiae Ctr1 and Ctr3 copper transport proteins. Surprisingly, we show that the Cuf1 protein, which normally serves to activate expression of the ctr4 ϩ gene under conditions of copper scarcity, directly represses the expression of iron transport genes under the same conditions. These studies in S. pombe provide insight into the evolution of copper transport protein structure in eukaryotes and into the mechanisms by which cells couple copper availability with the expression of a functional iron transport complex.
S. pombe Homologues of Bakers' Yeast Copper Uptake System-The S. pombe Cuf1 and Ctr4 proteins were identified from the Sanger Center S. pombe genome data base based upon sequence homologies with the S. cerevisiae Mac1 (32) and Ctr3 (15) amino acid sequences, respectively. The cuf1 ϩ gene was identified to the ORF SPAC31A2.11c found in S. pombe chromosome I cosmid c31A2. The ctr4 ϩ gene was discovered corresponding to the ORF SPCC1393.10 found in S. pombe chromosome III cosmid c1393. From the cosmid c31A2, we performed a series of subcloning experiments using SmaI and BamHI to obtain a 4.2-kilobase pair genomic fragment encompassing the cuf1 ϩ gene. This latter fragment was isolated and ligated into the pSP1 vector (47), giving rise to pSPcuf1 S-B/4.2. To generate pSPctr4 ϩ B-E/1.6, primers were made corresponding to the position -737 with respect to A of the ATG of the ctr4 ϩ gene and the end of that ORF, respectively. The expected 1.6-kilobase pair fragment was obtained by PCR using S. pombe FY254 genomic DNA.
RNA Analysis Methods-Northern blot analyses were carried out according to Schleicher & Schuell's protocol using random-primed 32 Plabeled DNA probes. Experiments regarding the mRNA steady-state levels of FET3 and ACT1 were carried out using the bakers' yeast strain and experimental conditions as described previously (28). To conduct the RNase protection analyses as described previously (48), two plasmids for making antisense RNA probes were created. The pSKctr4 ϩ plasmid was made by the insertion of the 189-bp fragment (positions ϩ413 to ϩ602) isolated by PCR from S. pombe (FY254) genomic DNA into the pBluescript SKϩ. To generate the pSKact1 ϩ plasmid, a 151-bp DNA fragment, which hybridizes to the S. pombe act1 ϩ gene positions ϩ507 to ϩ658, was also amplified and cloned into the pBluescript SKϩ vector. The plasmid pKSlacZ used was described previously (28).
SOD Activity Assay-The S. cerevisiae isogenic strains 50.L4 (MAT␣ gal1 trp1-1 his leu2-3,-112 ura3-50) (49) and DTY116 (MAT␣ gal1 trp1-1 his leu2-3,-112 ura3-50 sod1⌬::TRP1) (50) were grown in YPD, while the S. pombe isogenic strains FY254 and SPY1 in YES media (51). Copper treatment of yeast strains was conducted as described previously (15). The cells were washed with the lysis buffer (25 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA) in presence of protease inhibitors. Once harvested, the cells were resuspended in 100 l of lysis buffer, and then disrupted with an equivalent volume of glass beads. Typically, 1% of Triton X-100 was added to the lysate and the mixture incubated on ice for 10 min. Equal concentrations of protein extracts were analyzed on a 10% native polyacrylamide gel. To detect the SOD activity, the gel was soaked in 2.5 mM nitro blue tetrazolium and then subjected to the developing solution (36 mM potassium phosphate, 2.8 mM TEMED, 28 M riboflavin (pH 7.8)) for 20 min.
Protein Epitope Tagging and Microscopy-To generate the pSPcuf1 ϩ GFP, an insert delineated by the NruI site located at position Ϫ977 with respect to A of the ATG of the cuf1 ϩ gene and the stop codon including a NotI site was isolated. The fragment was swapped for an identical DNA region, except for the NotI site, comprised into the pSPcuf1 ϩ S-B/4.2 plasmid, creating pSPcuf1 ϩ NotI S-B/4.2. Subsequently, the gfp gene on a 750-bp NotI fragment (52) was cloned into pSPcuf1 ϩ NotI S-B/4.2. A similar strategy was utilized for generating the pSPctr4 ϩ GFP plasmid. To observe GFP fusion proteins by microscopy, the cells were grown in EMM media (51) under conditions of low copper availability. A 4-l culture aliquot was prepared and examined as described previously (53). Calcofluor white M2R (Sigma) staining was carried out as described previously (54).
Electrophoretic Mobility Shift Assay-Different cultures of SPY1 (cuf1⌬), expressing either cuf1 ϩ or cuf1 ϩ GFP were grown to early log-phase and then control, copper-treated (100 M) or BCS-treated (100 M) cultures were incubated for 1 h. Cell extracts were prepared as described previously (55) except for the composition of the extraction buffer (20 mM HEPES (pH 7.9), 10% glycerol, 75 mM NaCl, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride). Electrophoretic mobility shift assay was conducted using 1ϫ binding buffer that contained 12.5 mM HEPES (pH 7.9), 50 mM KCl, 2 mM MgCl 2 , 1 mM EDTA, 2.5 M ZnSO 4 , 0.6 mM DTT, 0.05% Nonidet P-40, and 6% glycerol. Typically, 15 g of protein of whole extract was incubated for 15 min at 25°C with 1 ng of 32 P-end-labeled fio1 ϩ DNA fragment and 0.5 g of salmon sperm DNA. When indicated, the reactions were pre-incubated in presence of anti-GFP antibodies (1 g) for 15 min at 25°C in absence of the probe. The samples were analyzed on a 1.5% agarose gel as described previously (55).

The S. pombe Copper Metalloregulatory Transcription Factor
Cuf1-Although there is a large evolutionary gap between S. pombe and S. cerevisiae, almost as large as that separating both yeasts from animals, in general, S. pombe genes and proteins exhibit greater similarity to those from mammalian cells than do S. cerevisiae genes (56,57). Therefore, we sought to identify genes encoding components of the copper homeostasis machinery in S. pombe to gain insight about structurefunction relationships of copper balancing proteins in higher organisms. Analysis of genomic DNA sequences from the S. pombe Genome Project revealed an open reading frame (ORF) (SPAC31A2.11c) related to two previously identified groups of copper metalloregulatory transcription factors: the Ace1 and Amt1 proteins from S. cerevisiae and the Candida glabrata, respectively, involved in copper detoxification via induction of metallothionein gene transcription (58 -60), and the Mac1 protein from S. cerevisiae that regulates transcription of genes involved in copper acquisition (13,28,32). We denote this S. pombe protein as Cuf1 (Cu factor 1) (Fig. 1). Interestingly, the amino-terminal 61 amino acids of Cuf1 display 51% and 45% identity to the amino-terminal 63 and 62 amino acids of the S. cerevisiae Ace1 and C. glabrata Amt1 proteins, respectively. Within this region of Cuf1 lies a zinc coordination domain that constitutes a minor groove DNA binding domain also found in Mac1 (33); Cuf1 also harbors extended homology (amino acid residues 41-61) to a region similar to Ace1, known to be involved in copper cluster formation and copper-inducible DNA binding to metal response elements in S. cerevisiae metallothionein gene promoters (44,45). In support of the notion that the Cuf1 amino-terminal domain is more similar to that found in Ace1 and Amt1 than Mac1, we have demonstrated that Cuf1 can partially substitute for Ace1, but not Mac1 function in S. cerevisiae. 2 Although residues 1-61 of Cuf1 display lower identity to this same region of Mac1 (39%), Cuf1 possesses a cysteine-rich domain near its carboxyl terminus with strong identity to a similar domain (residues 321-345) that plays a critical role in copper ion sensing in Mac1, conserving the five Cys and one His residue as well as additional flanking and intervening amino acids (31,35). Therefore, Cuf1 bears strong similarity to domains found in the Ace1/Amt1 class, as well as the Mac1 class of copper metalloregulatory transcription factors.
To understand the role of Cuf1 in S. pombe, the cuf1 ϩ locus was insertionally inactivated by deletion and replacement with the S. pombe ura4 ϩ gene (61). cuf1 ϩ mRNA was found to be moderately abundant in wild type cells and not regulated by copper addition or starvation, and absent in cuf1⌬ cells as determined by RNA blotting (Fig. 2A). Although we detected no obvious defect in copper resistance in cuf1⌬ cells as compared with isogenic cuf1 ϩ wild type cells (data not shown), cuf1⌬ cells are completely defective in utilizing the respiratory carbon source glycerol for growth, with only slight growth defects observed when cells are grown on glucose (Fig. 2B). The glycerol growth defect could be corrected either by returning a wild type copy of the cuf1 ϩ gene expressed from a plasmid, or by the addition of CuSO 4 to the growth medium at concentrations of at least 25 M (Fig. 2B). To test whether cuf1⌬ cells display other phenotypes that are known to be associated with copper deficiency in S. cerevisiae cells, copper/zinc SOD activity was assayed from whole cell extracts using native enzyme polyacrylamide gel electrophoresis from wild type and cuf1⌬ cells grown either under control conditions or in the presence of 100 M CuSO 4 . Although copper/zinc SOD activity was detected in cuf1 ϩ cells under both conditions, this activity was detected only in cuf1⌬ cells supplemented with CuSO 4 (Fig. 2C). Moreover, cuf1 ϩ deletion cells are defective in growth under iron starvation conditions (Fig. 2D), presumably due to their inability to provide copper to the multi-copper ferroxidase Fio1, essential for iron transport in S. pombe (43). Furthermore, this defect in high affinity iron uptake is associated with the integrity of the copper transport machinery since an identical phenotype is observed when cells harbor an inactivated ctr4 ϩ gene (ctr4⌬), encoding a high affinity copper permease in fission yeast, as described below (Fig. 2D). Taken together, the observed defects in respiration, copper/zinc SOD activity, and iron acquisition indicate that cuf1⌬ cells are copper-and ironstarved, and suggest that Cuf1 plays a crucial role in copper and iron transport, rather than copper detoxification in S. pombe.
Since Cuf1 resembles two distinct classes of copper metalloregulatory transcription factors and plays an important role in copper acquisition, we localized the protein in S. pombe cells to understand how it functions in copper homeostasis. A cuf1 ϩ -GFP fusion allele was constructed that fully complements the cuf1⌬ glycerol growth defect (Fig. 3A). As shown in Fig. 3B, the Cuf1-GFP fusion protein localizes to the S. pombe nucleus as 2 S. Labbé and D. J. Thiele, unpublished data. shown by staining with 4,6-diamidino-2-phenylindole, a marker for nuclear and mitochondrial DNA. Furthermore, no change in the localization of Cuf1-GFP fusion protein was apparent based on cellular copper status (data not shown). Thus, the results described here are consistent with Cuf1 serving as a nuclear transcription factor required for expression of genes that are critical for copper acquisition.
An S. pombe Copper Transporter with Properties of Both Ctr1 and Ctr3-Based on the properties of Cuf1 and the hypothesis that it activates target genes involved in copper acquisition, we searched the unassigned S. pombe genome data base for ORFs with homology to either the Ctr1 or Ctr3 high affinity copper transporters of S. cerevisiae. A single ORF (SPCC1393.10) of 289 codons, which we have denoted Ctr4, was identified that bears homology to Ctr1 at its amino terminus (38% identity) and Ctr3 at the carboxyl terminus (41% identity) (Fig. 4A). Like both Ctr1 and Ctr3, Ctr4 is predicted (using the TopPred II analysis) to possess two or three transmembrane domains, an extracellular hydrophilic amino terminus and a cytosolic tail. The amino terminus of Ctr4 harbors five copies of the sequence Met-X 2 -Met-X-Met, a putative extracellular copper binding motif found eight times in the amino-terminal region of Ctr1 and absent in Ctr3. Furthermore, Ctr4 exhibits strong identity to Ctr3 from residues 111 to 248, which encompasses the carboxyl-terminal 65% of the protein, with very high identity within the two regions which encompass putative transmembrane domains (Fig. 4A). Based on these observations, it appears that the S. pombe Ctr4 protein represents a fusion between the S. cerevisiae Ctr1 and Ctr3 proteins. Indeed, inspection of the sequence of a putative human copper transporter, hCtr1 (62), strongly suggests that it is also structurally related in its amino terminus to S. cerevisiae Ctr1, but in its carboxyl terminus to Ctr3. Furthermore, both the S. pombe Ctr4 and human Ctr1 proteins exhibit more overall sequence homology to S. cerevisiae Ctr3 than to Ctr1.
A role for S. pombe Ctr4 in high affinity copper transport was confirmed by four experimental results. First, as observed for Cuf1, deletion of the ctr4 ϩ gene (ctr4⌬) renders S. pombe cells defective for growth on respiratory carbon sources that can be rescued with either copper supplementation or by providing a plasmid-borne copy of the ctr4 ϩ gene (Fig. 4B). Second, a functional Ctr4-GFP fusion protein is localized to the S. pombe plasma membrane (Fig. 4, C and D). Third, ctr4⌬ strain cells are defective in iron transport (Fig. 2D). Fourth, as found for the CTR1 and CTR3 genes in S. cerevisiae, the S. pombe ctr4 ϩ gene expression is repressed by copper and induced by copper limitation using the Cu ϩ chelator BCS (Fig. 5A). Taken together, these data are consistent with Ctr4 functioning as a  (25 M) or a plasmid-borne copy (cuf1⌬ ϩ pSP1cuf1 ϩ ) of the wild type cuf1 ϩ gene. Strains spotted onto YES-glucose and YES-glycerol media were incubated at 30°C for 3 and 7 days, respectively. C, an S. pombe strain bearing the disrupted cuf1⌬ allele displays a deficiency in copper/zinc SOD1 activity (right side), unless exogenous CuSO 4 (100 M) (ϩ) is added. As a control (left side) a wild-type strain of S. cerevisiae exhibits high levels of copper/zinc SOD1 activity, whereas the sod1⌬ disruption strain lacks such activity, even in presence of exogenous copper. Manganese SOD2 activity detected in bakers' yeast is indicated (Mn SOD2). D, strains harboring either an inactivated cuf1 ϩ or ctr4 ϩ gene are defective in iron accumulation as shown by their inability to grow on medium containing the iron chelator ferrozine (250 and 500 M).

FIG. 3. Cuf1 is a nuclear protein.
A, the Cuf1-GFP fusion protein complements the cuf1⌬ growth defect on nonfermentable carbon sources (glycerol) in a manner indistinguishable from the unadulterated Cuf1 protein. Cultures grown to identical optical density were spotted (3000 cells/5 l) onto YES-glucose and YES-glycerol media and incubated at 30°C for 3 and 7 days, respectively. B, representative cells from Cuf1-GFP cultures grown at 30°C for 16 h in the presence of BCS (100 M) were visualized using the appropriate filter, and then 4,6diamidino-2-phenylindole staining to determine the location of the nucleus. The cells were also viewed by Nomarski microscopy to examine cell morphology (on the right).
high affinity copper transport protein in S. pombe. Unlike the redundancy of copper transport functions provided by Ctr1 and Ctr3, encoded by separate genes in S. cerevisiae, deletion of the ctr4 ϩ gene is sufficient to confer a copper starvation phenotype to S. pombe cells.
Based on our assignment of Cuf1 as a nutritional coppersensing transcription factor, and the findings that deletion of either the cuf1 ϩ or ctr4 ϩ genes leads to defects observed in copper-deficient cells, we ascertained if ctr4 ϩ expression is under the control of copper and Cuf1. As shown in Fig. 5B, wild type cells express detectable basal levels of ctr4 ϩ mRNA that are repressed by copper and induced by copper starvation with the Cu ϩ chelator BCS, but not under iron starvation conditions (with iron chelator BPS). Moreover, ctr4 ϩ mRNA is absent in cuf1⌬ cells under all conditions (Fig. 5B). Taken together, the convergence of cuf1⌬ and ctr4⌬ phenotypes, and the absolute dependence of ctr4 ϩ expression on Cuf1, strongly argue that Cuf1 and Ctr4 represent a copper-sensing and uptake system in S. pombe.

Cuf1 Represses Expression of Iron Uptake Genes under Copper Starvation-In bakers' yeast a clear mechanistic link exists
between copper availability and iron acquisition and mobilization in at least two respects. The Fet3 protein is a multi-copper ferroxidase essential for assembly of an iron transport complex at the plasma membrane with the Ftr1 iron permease. Accordingly, S. cerevisiae cells that are copper-deficient or unable to deliver copper to the lumen of the Golgi/endosome are unable to assemble an active iron-transport complex at the plasma membrane and, as a consequence, are iron-starved. Furthermore, the Fre1 protein is an Fe 3ϩ and Cu 2ϩ reductase that functions in both copper and iron uptake. While FET3 and FTR1 gene expression is regulated by iron availability through the ironsensing transcription factor Aft1, the FRE1 gene is regulated by copper and iron availability (7,13,28). To ascertain if any of the genes known to be involved in iron uptake in S. pombe are regulated by copper availability, we carried out RNA blotting experiments with isogenic wild type and cuf1⌬ cells grown under copper or iron starvation or replete conditions. As shown in Fig. 6A for wild type strains, under conditions of iron adequacy but low copper availability, very low levels of frp1 ϩ metalloreductase mRNA are observed. In response to 100 M CuSO 4 , a dramatic increase in frp1 ϩ mRNA was observed relative to basal levels; however, copper starvation with BCS treatment did not alter frp1 ϩ mRNA abundance. This same observation was true in two independent wild type S. pombe strains, GYP4 and FY254. Moreover, as shown in Fig. 6B (left panel) fip1 ϩ iron permease and fio1 ϩ multi-copper ferroxidase mRNA is also dramatically induced by 100 M CuSO 4 and, as expected, induced by treatment with the iron chelator BPS. Remarkably, deletion of the cuf1 ϩ gene (cuf1⌬) significantly elevated basal levels of fip1 ϩ and fio1 ϩ mRNA and eliminated further induction by 100 M CuSO 4 (Fig. 6B, right panel). A slight (ϳ2-fold) reduction of fip1 ϩ and fio1 ϩ mRNA levels was observed in the cuf1⌬ strain in response to copper administra- FIG. 4. Primary structural comparison of bakers' yeast and human high affinity copper transporters with S. pombe copper transporter. A, the amino-terminal region (45-106) of Ctr4 exhibits 38% identity to Ctr1. Similarly, hCtr1  shows 28% identity to Ctr4. The motif Met-X 2 -Met-X-Met is depicted by the filled rectangles (also indicated with capital letters). The carboxyl terminus of Ctr4 (111-248) resembles the S. cerevisiae Ctr3, as shown by the score of 41% sequence identity. TM, putative transmembrane domain. B, growth defect phenotype on nonfermentable carbon source associated with the insertionally inactivated ctr4⌬ allele was analyzed as described in Fig. 2B. C, ctr4⌬ cells express a functional Ctr4-GFP fusion protein as assessed by complementation of a ctr4⌬ allele (ctr4⌬ ϩ pSP1ctr4 ϩ -GFP) in a similar manner as specified for the Cuf1-GFP fused protein in Fig. 3A except that the cultures were spotted at a density of 300 cells/5 l. D, representative cells from ctr4 ϩ -GFP cultures grown and examined under the same conditions as detailed in Fig. 3B for the Cuf1-GFP fused protein except that calcofluor staining was used to delineate the cell periphery. tion, perhaps due to stimulation of intracellular iron levels sensed by an Aft1 ortholog in S. pombe. As a control, fio1 ϩ and fip1 ϩ mRNA levels were assayed from cells grown under ironsufficient (iron) and iron starvation (BPS) conditions and as observed previously, the expression of these genes was induced by iron starvation and repressed by iron adequacy in both cuf1 ϩ and cuf1⌬ backgrounds (43).
We tested whether Cuf1 directly represses expression of iron uptake genes in S. pombe, or whether this is an indirect response due to copper starvation in cuf1⌬ strains and, hence, failure to provide copper to the iron transport complex resulting in iron starvation. The fio1 ϩ -fip1 ϩ genes are divergently transcribed in S. pombe (43). Within this shared promoter region, we identified six copies of a repeated sequence, 5Ј-TTTGTC-3Ј, that is similar, although not identical, to the CuREs found in the CTR1, CTR3, FRE1, and FRE7 genes in S. cerevisiae and bound by the Mac1 protein (13,28) (Fig. 7A). To ascertain whether these elements play a direct role in fio1 ϩ -fip1 ϩ regulation by Cuf1 and copper, we constructed CYC1 minimal promoter-lacZ fusion genes harboring either a 271-bp fragment from the fio1 ϩ -fip1 ϩ promoter that contains three TTTGTC elements, or the same fragment in which two of the three TTTGTC elements have been extensively mutagenized (GGGAGA instead of TTTGTC). As shown in Fig. 7B by a representative RNase protection assay, the wild type fio1 ϩ -fip1 ϩ -CYC1-lacZ fusion gene was induced approximately 4-fold by copper treatment, whereas the same fusion in which two of the three TTTGTC boxes have been mutated exhibits high constitutive basal levels of expression that are not further elevated by copper treatment. These data directly implicate the TTTGTC elements in repression of the fio1 ϩ and fip1 ϩ genes and in copper-mediated gene induction. Consistent with this notion, two inverted TTTGTC repeats are also found within the frp1 ϩ promoter (data not shown).
Interestingly, the promoter of FET3 and other iron homeostatic genes in S. cerevisiae harbor similar repeated TTTGTC elements in their regulatory regions. To test whether this Cuf1mediated repression mechanism may operate in S. cerevisiae, FET3 mRNA levels were measured in isogenic wild type cells, mac1⌬ cells, and cells bearing a dominant MAC1 up1 mutation in which the Mac1 protein is unable to sense copper, constitutively binds to DNA instead of releasing in response to copper, and strongly activates expression of the CTR1, CTR3, and FRE1 copper transport genes even in the presence of copper. As shown in Fig. 7C, in response to 1 or 10 M CuSO 4 FET3 mRNA levels are modestly, but significantly induced (approximately 4-fold). In marked contrast, FET3 mRNA levels are dramatically elevated in the mac1⌬ background under all conditions (by approximately 18-fold), and are super-repressed under all conditions in the MAC1 up1 strain, when Mac1 protein is constitutively bound to CuREs (3-fold lower than the basal levels detected in the wild type strain). Therefore, both S. pombe and S. cerevisiae regulate expression of iron uptake genes in response to alterations in the availability of copper, a critical co-factor for the iron transport complex, through orthologous copper-sensing transcription factors. Although direct repression of FET3 by Mac1 has not been established, based on our results with Cuf1, this is a likely mechanism.
To test whether Cuf1 directly represses fio1 ϩ and fip1 ϩ expression through the TTTGTC elements in these promoters, FIG. 5. The cuf1 ؉ gene is required for copper-dependent regulation of ctr4 ؉ gene expression. A, strains FY254 (ctr4 ϩ ) and MPY2 (ctr4⌬) were treated with CuSO 4 (0, 10, and 100 M) or BCS (100 M) for 1 h followed by total RNA isolation and RNase protection analysis of ctr4 ϩ and act1 ϩ mRNA steady-state levels (indicated by arrows). B, FY254 (cuf1 ϩ ) and SPY1 (cuf1⌬) cultures were incubated in the absence (Ϫ) or presence of CuSO 4 (10 and 100 M), the copper chelator BCS (100 M), or the iron chelator BPS (100 M) for 1 h at 30°C, and then total RNA was isolated. Shown is an RNA blot of ctr4 ϩ and act1 ϩ mRNA steady-state levels indicated by arrows, respectively. DNA binding experiments were conducted using extracts derived from S. pombe cuf1 ϩ wild type and cuf1⌬ cells, and from cuf1⌬ cells expressing a functional Cuf1-GFP fusion protein. As shown in Fig. 8A by a representative electrophoretic mobility shift assay, the wild type 271-bp fio1 ϩ -fip1 ϩ promoter fragment forms a strong protein-DNA complex when extracts are prepared from cuf1 ϩ wild type cells grown under copper starvation conditions (BCS) and this complex is diminished to background levels when cells were grown in the presence of 100 M CuSO 4 prior to extract preparation. Furthermore, a complex of slightly slower electrophoretic mobility formed on this DNA fragment when extracts were prepared from copperstarved cells expressing the Cuf1-GFP fusion protein, the levels of which were substantially diminished by growth in the presence of copper. Additionally, a protein-DNA complex of slower electrophoretic mobility was observed when anti-GFP anti-serum was added to the binding reactions in supershift experiments. As shown in Fig. 8B, the fio1 ϩ -fip1 ϩ promoter fragment in which two of the three TTTGTC elements were mutated did not form significant protein-DNA complexes with these same extracts under any of the conditions tested. Taken together, these data strongly suggest that the S. pombe Cuf1 protein, which activates copper transporter gene expression under conditions of low copper availability, directly represses the expression of the fio1 ϩ , fip1 ϩ , and frp1 ϩ iron uptake genes under these same conditions through the TTTGTC elements found within these promoters. DISCUSSION Our results in S. pombe have provided important information on the structural modularity of both copper ion-sensing transcription factors and copper ion transporters. Indeed, our studies clearly demonstrate that Cuf1 is functionally similar to Mac1 from S. cerevisiae in terms of its regulation of copper transporter gene expression and repression of iron transport genes. However, structurally Cuf1 has hallmarks of the Ace1/ Amt1 proteins at its amino terminus that extend beyond the similarity observed between Mac1 and the Ace1/Amt1 amino- FIG. 7. The fip1 ؉ -fio1 ؉ promoter TTTGTC elements confer copper starvation-mediated repression to the minimal promoter CYC1-lacZ. A, schematic representation of the fip1 ϩ -fio1 ϩ intergenic promoter region within which lie six copies of the repeated sequence, TTTGTC. The nucleotide numbers refer to the position relative to the A of the start codon of the fio1 ϩ ORF. The four upper boxes (Ⅺ) shown in the common promoter region indicate that the TTTGTC elements are oriented toward the fio1 ϩ gene, while the two lower boxes depict the elements oriented toward the fip1 ϩ gene. Vertical arrows represent the edges of the 271-bp promoter DNA fragment encompassing three cisacting elements that were inserted in its natural orientation into the minimal promoter of the CYC1 gene fused to lacZ. B, CYC-lacZ fusion genes containing the wild type 270-bp fragment (271 WT, Ⅺ) and the same fragment in which two of the TTTGTC elements have been mutated (271 Mut, f) were transformed in FY254 (cuf1 ϩ ) and analyzed for the lacZ mRNA steady-state levels status. act1 ϩ transcripts steadystate levels were also measured as control. Total RNA was isolated from untreated cultures (Ϫ) or cultures treated with copper (100 M), and BCS (100 M). C, the isogenic wild type (MAC1), mac1⌬, and MAC1 up1 strains were treated for 1 h in the presence of CuSO 4 (0, 1, and 10 M) or BCS (100 M). Shown is an RNA blot of FET3 and ACT1 mRNA steady-state levels indicated by arrows, respectively.

FIG. 8. Cuf1 and Cuf1-GFP proteins expressed in fission yeast bind to the repeated TTTGTC elements.
A, representative electrophoretic mobility shift gel of the wild type 32 P-labeled 271-bp fio1 ϩ promoter fragment used with whole cell extracts prepared from SPY1 cells (cuf1⌬) expressing either the pSP1 vector alone (Ϫ), no Cuf1 (cuf1⌬), wild-type Cuf1 (cuf1 ϩ ), or Cuf1-GFP (cuf1 ϩ -GFP) as their sole source of Cuf1. BCS-treated cells expressing either cuf1 ϩ or cuf1 ϩ -GFP alleles exhibit a specific TTTGTC-protein complex (B), which is absent in extracts from cells treated with copper (100 M). In the presence of anti-GFP either pre-incubated or added once the probe is mixed in the presence of the cell extracts, the protein-bound complex is shifted (S). FP, free probe DNA. B, representative electrophoretic mobility shift assay gel using the mutated fio1 ϩ -fip1 ϩ 271-bp fragment. As shown, no DNA-protein complexes were observed when two of the three TTTGTC elements harbor multiple point mutations (GGGAGA instead of TTTGTC).
terminal DNA binding domains. This similarity is further supported by the observations that Cuf1 will partially suppress an ace1⌬ mutation, but not when the major target gene of Ace1, CUP1-encoded metallothionein, is deleted. Furthermore, in ace1⌬ cells expressing Cuf1, the robust basal levels of CUP1 mRNA are repressed approximately 3-fold in response to copper, consistent with the manner by which Cuf1 regulates ctr4 ϩ gene expression in S. pombe and the plausibility that Cuf1 directly binds to the CUP1 promoter. 2 The possibility that the unique amino terminus of Cuf1 displays distinct DNA binding specificity from that of Mac1 is supported by the fact that the Cuf1 binding sites identified in the fio1 ϩ -fip1 ϩ promoter region (5Ј-TTTGTC-3Ј) differ from the CuRE element consensus sequence found in the CTR1, CTR3, FRE1, and FRE7 genes in S. cerevisiae (5Ј-TTTGC(T/G)C(A/G)-3Ј). Therefore, the network of hydrogen bonds made by the conserved Cuf1 amino-terminal minor groove DNA binding domain and the T-rich DNA sequence, while likely similar to those made by Mac1 and Ace1/ Amt1, are likely to be highly specific for each protein within the more G-C rich core of their respective cis-acting DNA sequences. We also note that we have shown Cuf1 DNA binding in cell extracts to be responsive to copper levels, a feature associated with Ace1 and Amt1 in cell extracts, but one that has only been demonstrated for Mac1 in vivo (28).
The Cuf1 carboxyl terminus harbors a Cys-His repeat element (REP element) absent in Ace1/Amt1 but found duplicated in both Mac1 and Grisea of Podospora anserina (63,64). Because a specific mutation of the first repeated element in Mac1 (MAC1 up1 ) generates a protein that is unable to repress target gene expression in response to copper, this element likely serves a copper-sensing function to homeostatically modulate expression of the ctr4 ϩ gene, and perhaps other genes, by Cuf1. Taken together, S. pombe has utilized copper-sensing domains from two distinct classes of copper metalloregulatory transcription factors for Cuf1, creating a chimeric protein from the phylogenetic standpoint.
The S. pombe Ctr4 protein appears to be a chimera of the S. cerevisiae Ctr1 and Ctr3 high affinity copper transport proteins. This is supported by the observation that the Ctr4 amino terminus harbors five copies of a putative extracellular copper binding domain (M-X 2 -M-X-M) that is found in the amino termini of both S. cerevisiae and human Ctr1, but not in S. cerevisiae Ctr3. Furthermore, the carboxyl-terminal ϳ65% of Ctr4 has strong identity to Ctr3, especially within the transmembrane spanning domains, but little identity to Ctr1 in this segment. Since the transmembrane domains are likely to play crucial roles in guiding copper across the plasma membrane, this homology to Ctr3 may be very important in copper transport. Indeed, the analysis of human Ctr1, by domain structure, sequence identity, and phylogenetic tree formation using the Clustal method, demonstrates that like Ctr4, the human Ctr1 protein has homology to S. cerevisiae Ctr1 in the amino terminus (29%) and Ctr3 in the carboxyl terminus (31%) (Fig. 4A). Furthermore, hCtr1 is more closely related to S. cerevisiae Ctr3 over the entire length of the protein, than it is to Ctr1. Based on these structural features of Ctr4, we speculate that S. pombe may represent an early evolutionary phase at which the Ctr1 and Ctr3 proteins, encoded by distinct and functionally redundant genes in S. cerevisiae, may have fused into one protein. This idea is supported by our observation that, unlike the functionally redundant situation in S. cerevisiae with Ctr1 and Ctr3, deletion of the S. pombe ctr4 ϩ gene (ctr4⌬) is sufficient to confer a copper starvation phenotype. This would predict that inactivation of the mouse homologue of Ctr1, if a single copy gene, would be likely to generate a copper deficiency.
It has been well established that iron uptake and mobiliza-tion in mammals is tightly linked to copper availability (65,66). Indeed, swine suffering from copper malnutrition exhibit several abnormalities in iron metabolism including the inability to absorb, mobilize, and utilize administered iron, and were diagnosed in an early study as anemic (67,68). Moreover, failure to incorporate copper ions into the copper-containing ferroxidase ceruloplasmin gives rise in an unstable protein and a loss of the ferroxidase activity, which is essential for iron mobilization (69). Furthermore, a recent study has revealed the existence of a putative copper-dependent ferroxidase, hephaestin, which would transfer iron from the intestinal mucosal cells through the baso-lateral membrane into the blood stream in concert with other component(s), such as a putative iron transporter (9). The requirement for copper in iron transport has also been biochemically characterized in some mammalian cell lines (70). Recent studies in S. cerevisiae have uncovered the mechanisms that underlie this dependence, through the isolation and analysis of genes encoding high affinity copper transporters (Ctr1 and Ctr3), the intracellular copper chaperone Atx1, a TGN localized P-type ATPase copper pump (Ccc2), a chloride channel that facilitates copper loading onto Fet3 (Gef1) (71)(72)(73), and two subunits of the high affinity iron uptake complex (the Fet3 multi-copper ferroxidase and the Ftr1 iron permease). The high affinity copper transporters are essential to mediate the passage of copper ions across the plasma membrane, with mobilization of copper to the Ccc2 P-type ATPase by the Atx1 copper chaperone, whereby it is delivered to the lumen of the secretory compartment through the action of Ccc2. Within the late secretory pathway, four copper atoms are assembled with Fet3 (74,75) with the assistance of the Gef1 chloride channel. The failure to incorporate copper into Fet3 is thought to lead to a misfolded and enzymatically inactive ferroxidase, which prevents the Fet3-Ftr1 high affinity iron transport complex from assembling at the plasma membrane, resulting in iron deficiency. Therefore, it is clear that copper must be provided for the iron transport/mobilization machinery to function. Previously, it was established that the ATX1 copper chaperone gene is regulated by intracellular iron availability by the Aft1 iron sensor. In this work, we have demonstrated that at least three genes involved in iron uptake in S. pombe, and the FET3 gene in S. cerevisiae, are transcriptionally regulated by the coppersensing transcription factors Cuf1 and Mac1, respectively. Our data support a direct role for Cuf1 in repression of these genes since either inactivation of the cuf1 ϩ gene, or mutation of Cuf1 binding sites in the fio1 ϩ -fip1 ϩ promoter region, gave a marked de-repression of the fio1 ϩ -fip1 ϩ gene expression. We suggest based on these data and on in vitro DNA binding experiments, that Cuf1 represses frp1 ϩ , fio1 ϩ , fip1 ϩ , and possibly other genes involved in iron uptake/mobilization, under conditions of copper scarcity, by binding to TTTGTC elements with their respective promoters. When copper is present at higher levels, Cuf1 de-represses gene expression through copper-sensing mechanisms that inactivate DNA binding and perhaps via other intramolecular conformational changes. This model is consistent with the demonstrated copper-responsive DNA binding of Cuf1 in vitro and Mac1 in vivo (28), and the marked repression of FET3 expression by the S. cerevisiae Mac1 up1 protein, known to bind DNA in vivo even in the presence of copper. Since the iron uptake genes are inducibly expressed in response to the iron chelator BPS, even under conditions of copper scarcity, we speculate that there may be functional interactions between the copper-and iron-sensing transcription factors. Further studies will elucidate how Cuf1 and Mac1 repress expression of the iron uptake genes and how these copper-responsive transcription factors may interact with ironsensing transcription factors.
The regulation of iron uptake genes in both S. cerevisiae and S. pombe by copper-sensing transcription factors provides a means of tightly coordinating and fine tuning iron transport with respect to copper availability. Since the Fet3 multi-copper ferroxidase is a key component of the high affinity iron transport complex at the plasma membrane, futile expression of the iron transport genes, in the absence of sufficient concentrations of the copper co-factor, would be undesirable. Yeast cells have therefore evolved both repressive and activator functions for copper-sensing transcription factors to maintain the delicate balance of these two essential, yet toxic metal ions.