Homeostatic Regulation of Copper Uptake in Yeast via Direct Binding of MAC1 Protein to Upstream Regulatory Sequences ofFRE1 and CTR1 *

Copper deprivation of Saccharomyces cerevisiae induces transcription of the FRE1 andCTR1 genes. FRE1 encodes a surface reductase capable of reducing and mobilizing copper chelates outside the cell, and CTR1 encodes a protein mediating copper uptake at the plasma membrane. In this paper, the protein encoded by MAC1is identified as the factor mediating this homeostatic control. A novel dominant allele of MAC1, MAC1 up2 , is mutated in a Cys-rich domain that may function in copper sensing (a G to A change of nucleotide 812 resulting in a Cys-271 to Tyr substitution). This mutant is functionally similar to theMAC1 up1 allele in which His-279 in the same domain has been replaced by Gln. Both mutations confer constitutive copper-independent expression of FRE1 and CTR1. A sequence including the palindrome TTTGCTCA … TGAGCAAA, appearing within the 5′-flanking region of the CTR1promoter, is necessary and sufficient for the copper- andMAC1-dependent CTR1 transcriptional regulation. An identical sequence appears as a direct repeat in theFRE1 promoter. The data indicate that the signal resulting from copper deprivation is transduced via the Cys-rich motif of MAC1 encompassing residues 264–279. MAC1 then binds directly and specifically to the CTR1 and FRE1 promoter elements, inducing transcription of those target genes. This model defines the homeostatic mechanism by which yeast regulates the cell acquisition of copper in response to copper scarcity or excess.

Copper is essential but toxic to cells. Copper is an essential prosthetic group of proteins such as cytosolic superoxide dismutase that is required for detoxification of oxygen free radi-cals and thus for the fitness of aerobic organisms (1). The copper sites in cytochrome oxidase are essential for the activity of this enzyme and thus for sustained cellular respiration (2). More recently, a role for copper enzymes in iron metabolism has been defined (3). Multi-copper oxidases, ceruloplasmin (mammals) (4,5) and FET3 (yeast) (6,7), have been implicated in iron transport. On the other hand, excess copper is toxic. The toxicity of copper may derive from the reaction of Cu(I) with hydrogen peroxide yielding the highly reactive hydroxyl radical that in turn may damage lipids, proteins, or DNA in cells (8). One way that cells solve the problem of acquisition of sufficient copper while avoiding toxic excess is by homeostatic control of copper uptake (2,9).
In aerated media, copper, like iron, is complexed to media components in its higher valence state. Since data indicate that the metals enter the cell uncomplexed to any extracellular ligand, an initial ligand displacement step must precede the movement of the metal ion across the plasma membrane (9). Ligand coordination to lower valence ions is typically more labile than to the corresponding high valence species (9), and therefore reduction of copper in the environment facilitates cellular uptake. The major surface reductase of Saccharomyces cerevisiae is the FRE1 gene product (10); FRE1 is capable of reducing extracellular copper or iron complexes (11,12). Consequently, this reductase plays a significant role in cellular copper acquisition (12).
In most laboratory strains of S. cerevisiae, cellular copper acquisition requires CTR1 (13,14). CTR1 protein exists in the cell as a multimer in the plasma membrane (15), although in some settings it may be internalized into the endocytotic pathway (16). CTR1 contains an unusual amino-terminal domain oriented toward the cell exterior (13). This domain, rich in methionine and serine, is similar to copper-binding domains present in bacterial proteins mediating copper resistance (17,18). The homology of the methionine-rich domain with copper binding proteins suggests that it will form a pocket with affinity for Cu(I). The subsequent events involved in internalization and translocation of the metal across the cell membrane and into the cytosol have not been characterized.
Expression of FRE1 (11,12) and CTR1 (15) are homeostatically regulated by copper availability, consistent with the roles of the two gene products in copper acquisition. The regulation of FRE1 (12), like the regulation of CTR1 (15), is mediated at the level of copper-dependent transcription. Copper deprivation induces and copper loading represses transcription of both of these genes. Thus, cellular component(s) must exist through which the signal for available copper levels is transduced into regulated transcription of these genes. ACE1 must be considered as a candidate for this transducing function (19). ACE1 encodes a yeast protein that binds copper (20) and activates the transcription of target genes in a copperdependent fashion (21). However, strains deleted for the ACE1 gene maintain homeostatic regulation of the surface reductase and copper uptake (15). Furthermore, ACE1 protein mediates the induction of genes involved in copper detoxification, including CUP1, the yeast metallothionein (19), and SOD1, the copper-zinc superoxide dismutase (22). Because these genes function in cellular protection, the threshold copper concentrations at which these genes are expressed are relatively high (M concentrations in defined media). In contrast, the copper exposure at which regulation of the copper uptake system occurs is much lower (less than 20 nM in defined media) (7,15). Also, the copper acquisition system is repressed by copper availability (15), and thus is regulated in a direction opposite from the detoxification system controlled by ACE1 (19). These several factors make ACE1 an unlikely candidate for the copper sensor controlling copper uptake.
A protein with homology to ACE1, called MAC1, must also be considered. The sequence of the MAC1 protein is notable for an amino-terminal domain (residues 1-42) with significant homology to ACE1. The activator for the metallothionein genes in Candida glabrata, AMT1, is also homologous in this region (23). In addition, MAC1 contains a motif conforming to the pattern CysXCysX 4 CysXCysX 2 CysX 2 His. This motif occurs twice in the carboxyl-terminal portion of the protein. It is within the first of these repeats, residues 264 -279, that the mutation in the dominant MAC1 up1 allele occurs, resulting in a His to Gln substitution at residue 279 (23). FRE1 expression and copper uptake are not repressed by copper in strains containing this dominant MAC1 allele (12,23). The MAC1 protein, fused to the lacZ gene product for purposes of tracking, has been localized to the nucleus consistent with a role for MAC1 in transcriptional regulation (23).
Here we report evidence that strongly indicates that the MAC1 protein functions as the copper sensor-regulator that controls the expression of surface reductase and copper uptake activity in yeast and thus provides homeostatic control of copper acquisition. We characterize the direct interaction of MAC1 protein with specific sequence elements in the promoters of the CTR1 and FRE1 genes and show that these elements alone support copper-and MAC1-dependent transcriptional activation. The data suggest that MAC1 binding to these elements plays a role in the transduction of environmental copper levels into regulated gene transcription within yeast cells.
Growth Conditions-Copper concentrations were manipulated in YPD medium (2% yeast extract, 1% peptone, 2% glucose) by the addition of the copper chelator BCS or by the addition of copper sulfate. In some experiments, the strains were grown in defined media containing yeast nitrogen base without amino acids modified to omit iron and copper. This ingredient (6.7 g/liter) was added to glucose (2%), MES buffer (25 mM, pH 6.1), and BCS for copper starvation or copper sulfate for copper loading. Iron was added back as ferric ammonium sulfate in a concentration of 10 M. A 1-mm loopful of yeast grown on a fresh YPD plate was inoculated into 10 ml of liquid medium and grown overnight with aeration and agitation at 30°C to late log phase (A 600 approximately 2.0). The following day, the culture was diluted into medium of the same composition at an A 600 of 0.2 and grown for an additional 5 h prior to harvesting. Methods for crossing, sporulation, spore dissection, and transformation of yeast were as described (25).
Demonstration That MA20 Carried an Allele of MAC1-MA20 (MAC1 up2 leu2-3,112) was crossed with WY10 (⌬mac1::LEU2 leu2-3,112) and the diploid was sporulated. Spore clones derived from 12 tetrads were evaluated for leucine prototrophy and non-repressing reductase activity. This phenotype has been associated with the MAC1 up1 mutation (23). Clones with non-repressing reductase were in all cases found to be incapable of growing in the absence of leucine, indicating lack of recombination between the marked MAC1 allele and the nonrepressing reductase phenotype of the MA20 mutant.
The MAC1 up2 allele was also rescued by gap repair (24). A 1.6kilobase pair NarI fragment containing the entire MAC1 open reading frame was removed from the 3.0-kilobase pair HindIII MAC1 clone in pSEY8 (23). The resulting gapped pSEY8 that now contained regions 5Ј and 3Ј to the MAC1 open reading frame as sticky ends was transformed into MA20 (MAC1 up2 ), and recircularized plasmid was recovered. This plasmid mixture was amplified in Escherichia coli and the recovered MAC1 up2 allele in pSEY8 was sequenced from Ϫ162 to ϩ1585 (numbering relative to MAC1 translation start, with termination at ϩ1251) by the standard chain termination method (Sequenase 2.0, U. S. Biochemical Corp.).
CTR1 mRNA and Protein Analysis-CTR1 (and FRE1) mRNA levels in wild type, MAC1 up1 , and MAC1 up2 containing strains were determined by Northern blot analysis following standard protocols (26). For determining CTR1 protein levels a myc-tagged protein fusion was constructed. To do this, the HindIII fragment was excised from plasmid 352-myc (13) and inserted into the HindIII sites of pRS416, creating plasmid 416-CTR1myc. This construct contains a myc epitope tag inserted at the unique EcoRI site of the genomic CTR1 clone and is carried on a centromere-linked vector.
CTR1 protein visualization was as follows. Cultures of transformants with the 352-myc plasmid (myc epitope labeled CTR1 genomic clone) were grown to log phase. Approximately 2 ϫ 10 7 cells were harvested and lysed by pelleting and resuspending in 150 l of 1.85 M NaOH, 2% 2-mercaptoethanol. Proteins were precipitated by the addition of 150 l of 50% trichloroacetic acid and incubation on ice for 30 min. The precipitate was then recovered by centrifuging at 16,000 ϫ g for 10 min prior to resuspending in 45 l of 2 ϫ Laemmli buffer and 5 l of 1 M Tris base. Ten microliters or 4 ϫ 10 6 cell equivalents were loaded per lane of an SDS-polyacrylamide gel. The proteins were blotted to nitrocellulose, and the myc-tagged CTR1 protein was visualized using 9E10 ascites anti-myc monoclonal antibody and peroxidase-conjugated rabbit antimouse secondary antibody. The enhanced chemiluminescence (Amersham Corp.) was used to develop the signal from the epitope labeled protein.
Deletion Analysis of the CTR1 Promoter-Plasmids containing portions of the CTR1 promoter fused to the lacZ gene were constructed as follows. Sequences from the CTR1 promoter, translation start, and codons 1-3 were amplified by PCR between unique SalI and BamHI sites and fused to the ␤-galactosidase coding region in plasmid pYE-Gal4. 5Ј deletions of the CTR1 sequences (numbered with respect to the translation start) were made by PCR, creating plasmids pCTRlac1-413, -334, -333, -311, and -226. For pGC-CTR(Ϫ337/Ϫ301), the oligonucleo-tides depicted in Fig. 6A and flanked by SalI half-sites were inserted into the XhoI site of pLG699-Z (27). Correctness of the inserted sequence and orientation of the insert were verified by direct sequencing of the plasmid. Sequences encoding the FLAG epitope were introduced 3Ј of the MAC1 coding sequence and subcloned into pBluescript IISK(Ϫ), creating plasmid pT7MAC1.
DNA Binding Assays-Electrophoretic mobility shift assays (EMSA) were performed with MAC1 protein that was prepared by in vitro transcription and translation. This assay was carried out essentially as described for the AFT1 EMSA (28). MAC1 mRNA was synthesized from the pT7MAC1 plasmid with T7 polymerase and translated in a wheat germ lysate (Promega). The binding reaction was carried out in 20 l of binding buffer (10 mM HEPES-KOH, pH 7.5, 150 mM KCl, 1 mM dithiothreitol, 0.5 mM EDTA, 0.05% Nonidet P-40, 7.5% glycerol) containing 2 ng of 32 P-end-labeled probe, salmon sperm DNA, and in vitro translated MAC1 protein. The binding reaction mixtures were incubated for 10 min at room temperature and then electrophoresed in a 4% native polyacrylamide gel in 6.6 mM Tris-HCl, pH 7.9, 3.3 mM sodium acetate, and 1 mM EDTA at 4°C.
Enzyme and Copper Uptake Assays-To evaluate reductase activity in yeast on solid media, spore clones were streaked onto YPD plates containing 50 M copper sulfate and incubated overnight at 30°C. The reductase activity was then evaluated by filter lift assay as described (29). The MAC1 up1 or MAC1 up2 clones gave a strong signal in this assay while the wild-type MAC1 clones were negative. Quantitative Fe(III) and Cu(II) reductase activity was determined in cell suspensions as described using bathophenanthroline disulfonate and bathocuproine disulfonate as indicators of Fe(II) and Cu(I) production, respectively (12). ␤-Galactosidase activity in yeast transformants was determined by a standard technique (27). For 67 Cu uptake measurement, strains were grown in YPD to log phase, washed in assay buffer consisting of 50 mM sodium citrate, pH 6.5, 5% glucose, and incubated at a density of approximately 2 ϫ 10 8 cells/ml in the presence of 1 M 67 CuSO 4 of specific activity 820 dpm/pmol copper. After 1 h incubation at 30 or 4°C, washed cells were collected on glass fiber filters and counted in a scintillation counter. Copper uptake, reported as pmol/million cells/h, was calculated after subtracting cell-associated counts obtained at 4°C from the cell-associated counts obtained at 30°C.

RESULTS
The promoter of the FRE1 gene mediates a wide range of transcriptional changes in response to copper or iron (10,12). Thus, mutants selected for FRE1 dysregulation included strains with abnormalities of copper or iron metabolism (13). The selection scheme, which has been described previously (13), involved the fusion of the FRE1 promoter element to a HIS3 selectable marker and integration into a his3 deleted strain. The mutants were selected for the inability to repress FRE1 transcription under conditions of abundant copper and iron. A mutant strain identified by this method, MA20, exhibited surface Fe(III) and Cu(II) reductase activity that was poorly repressed by copper (Fig. 1). This phenotype resembled that of the previously described MAC1 up1 mutant (12,23). The MA20 mutant was crossed with the parental strain 61, and the diploid, like the haploid, was found to possess reductase activity that was not repressible by copper indicating that the mutation behaved as a dominant (not shown). This dominance also resembled the previously described MAC1 up1 mutant (23). The MA20 strain was then mated with WY10, a strain carrying a LEU2-marked MAC1 allele, and the diploid was sporulated. Spore clones analyzed from 12 independent meioses revealed no recombination between the LEU2 marker and the nonrepressing reductase phenotype, suggesting that the mutation in MA20 was allelic to MAC1.
Direct rescue of the MAC1 mutant allele was then accomplished by means of gap repair (24), and the entire MAC1 coding region in MA20 was sequenced. A single mutation, a G to A missense mutation at nucleotide 812 (SCMAC1 accession no. X74551), was identified. This mutation was predicted to alter the residue at position 271 in the primary sequence from a cysteine to a tyrosine. Thus, this mutation causes an amino acid substitution within the same Cys-rich domain, residues 264 -279, as the His 279 to Gln substitution due to the MAC1 up1 allele (23) (Fig. 2).
To determine if the MAC1 up2 allele conferred the same elevated copper uptake observed in MAC1 up1 containing strains (12,23), high affinity copper uptake was evaluated in a congenic panel of strains containing the MAC1 up2 allele (MA20), a MAC1 wild-type allele (CM3262), or a mac1⌬ allele (3262⌬mac). The results revealed that the MAC1 up2 allele conferred increased copper uptake (Fig. 3). Evaluation of a ctr1⌬containing strain revealed negligible high affinity copper uptake similar to the mac1⌬ phenotype (13) (Fig. 3). The resemblance between ctr1⌬ and mac1⌬ strains suggested a model in which the nuclear MAC1 protein mediates the transcriptional activation of CTR1 in response to copper deprivation. According to this model, CTR1 protein, functioning downstream of MAC1, acts as the structural mediator of copper uptake at the plasma membrane.
A prediction of this model is that different MAC1 alleles should support different patterns of copper-dependent CTR1 expression. This was tested first by evaluating the steady-state FRE1 and CTR1 transcript levels in strains BR10 (MAC1), BR10⌬mac1, UPC31(MAC1 up1 ), CM3262(MAC1), and MA20 (MAC1 up2 ). These levels were determined by Northern blot analysis and are shown in Fig. 4A. In both MAC1 wild-type backgrounds, copper deprivation (BCS strongly inhibits copper uptake) led to strong induction of FRE1 and CTR1 mRNA species, and copper addition strongly repressed these levels. In the MAC1 up1 and MAC1 up2 mutants, the copper-dependent repression was abrogated. Finally, in the mac1⌬ strain, the CTR1 transcript was undetectable, and the FRE1 transcript was present in very low abundance (Fig. 4A), indicating a requirement for MAC1 in the expression of these two genes.
The level of CTR1 protein was also determined in a similar panel of allelic MAC1 strains grown in different concentrations of copper. CTR1 protein was assessed by transforming the strains with a myc-tagged version of the CTR1 gene, visualizing the myc-CTR1 fusion by Western blot analysis. The protein results precisely followed the Northern data. Like CTR1 mRNA, CTR1 protein expression was regulated by copper in prior to assay of surface reductase activity using Fe(III) as substrate (12). Assay for Cu(II) reduction using Cu(II) as substrate yielded similar relative activities between strains and conditions. the wild-type strain, with maximal expression in the copperdeprived cultures (Fig. 4B). The lower molecular weight band visualized with the anti-myc epitope antibody may represent a more rapidly migrating form of CTR1 protein. This CTR1 species may have resulted from improper glycosylation, since CTR1 protein has been observed to be heavily glycosylated via O-linkages (12). In contrast, CTR1 protein expression was constitutive in the MAC1 up2 strain and virtually undetectable in the mac1⌬ strain, regardless of manipulation of the copper in the growth medium.
These effects of MAC1 alleles on CTR1 expression suggested that MAC1 was likely to be a regulatory protein. To evaluate whether this regulation was occurring at the level of gene transcription or some other step in gene regulation, a promoter construct that included 431 base pairs of contiguous CTR1 upstream DNA fused to the lacZ gene was placed into MAC1, MAC1 up2 , and mac1⌬-containing strains. This region attracted attention because it included a perfect palindrome of the sequence TTTGCTCA, a likely candidate for a regulatory sequence (cf. Fig. 6A). Furthermore, this same sequence was found in the FRE1 promoter as a direct repeat in a region shown to confer metal-dependent expression of that locus (10). These three transformants were grown in the presence of BCS (ϪCu) or copper sulfate (ϩCu), and ␤-galactosidase activity was determined as a measure of CTR1 promoter activity. As the data in Fig. 5 show, in MAC1 wild-type strain CM3262 this CTR1 promoter fragment did confer copper-and MAC1-dependent expression of the reporter gene (pCTR lacZ-413, first entry). In contrast, in the MAC1 up2 strain, MA20, this expression was copper-independent, whereas it was essentially absent in the mac1⌬ strain.
To more closely define this copper-responsive element, progressive 5Ј deletions of this fragment were constructed and analyzed. Upon deletion to Ϫ334 (leaving the 5Ј TTTGCTCA intact, pCTR lacZ-334, second entry) the expression of the reporter gene was qualitatively unaltered in that both copper and MAC1 dependence were proportionately unaffected. However, deletion of even the 5Ј-terminal T of this sequence (pCTR1 lacZ-333, third entry) significantly reduced the copper dependence of this expression, as did deletion of the entire 5Ј-half of the palindrome (pCTR lacZ-311, fourth entry). In these two constructs, expression was less MAC1-dependent also as indicated by the increased ␤-galactosidase activity in the mac1⌬containing strain. Removal of the palindrome entirely (pCTR lacZ-226) abolished lacZ expression, suggesting that this re-  gion contained elements needed for even basal transcription. To further test the apparent enhancer activity of the CTR1 sequences Ϫ337 to Ϫ301, including the palindrome, they were cloned in front of a minimal promoter from the CYC1 gene including the transcription and translation start sites but lacking any upstream activating sequences (pGC-CTR). As the data show, these CTR1 sequences alone conferred MAC1-and copperdependent expression of lacZ in this construct (Fig. 5, last  entry).
A cis-acting palindromic element in the CTR1 promoter was thus identified as both necessary and sufficient for MAC1-and copper-dependent transcriptional activity. MAC1 action could be mediated through direct interaction with the DNA sequence or indirectly via intermediate proteins. To distinguish between these possibilities we examined the ability of an oligonucleotide containing the palindromic CTR1 target sequence (Fig. 6A) to specifically interact with MAC1 protein in an electrophoretic mobility shift assay (EMSA). In the first set of experiments, shown in Fig. 6B, the MAC1 protein was modified by insertion of a carboxyl-terminal FLAG epitope tag. This fusion protein was synthesized by means of an in vitro transcription-translation system. The MAC1-FLAG protein specifically retarded migration of the labeled oligonucleotide (Fig. 6B, lane 2). Two specific bands appeared with the addition of MAC1 protein to the probe. Various controls demonstrated the specificity of these signals. In the absence of added MAC1 neither band was observed (Fig. 6B, lane 1). Addition of 100-fold molar excess of cold competitor DNA with the same sequence as the labeled oligonucleotide eliminated the shifted signals (Fig. 6B, lane 5). In contrast, MAC1 did not bind to the cis element from the FET3 locus in S. cerevisiae recognized by the iron-regulated transcription factor, AFT1 (28), since an oligonucleotide representing this sequence did not compete with the CTR1 probe oligonucleotide for MAC1 (Fig. 6B, lane 4). Finally, antibody to the FLAG epitope appended to the MAC1 protein resulted in a decrease in the more slowly migrating signal and the appearance of a supershifted band. This was presumed due to delayed migration of the FLAG antibody complex in association with the DNA-MAC1 complex (Fig. 6B, lane 3, indicated by *). The effect of the antibody addition on the intensity of the lower band was minimal. In sum, a specific signal due to MAC1-DNA interaction were observed.
We then used the EMSA to evaluate more precisely the nucleotide sequence constraints on the protein-DNA interaction. This was done by adding as cold competitors at 100-fold excess mutated forms of this CTR1-derived oligonucleotide probe. Mutation of the first portion of the palindrome (M1, GCT to CGA in the 5Ј-half of the palindrome, Fig. 6A) partially inhibited the interaction (lane 6, compare with competition by the wild-type CTR1 promoter fragment, lane 5), whereas mu-tations in both the first and second portions of the palindrome (M2, in which an additional GCT to CGA substitution has been made in the 3Ј-inverted repeat, Fig. 6A) abolished the interaction (lane 7, no competition). These results indicate first that MAC1 does bind to this sequence in vitro and that both the GCT of the 5Ј and 3Ј portions of the palindrome are important in this regard.
The role of the direct repeat sequence in the 5Ј-flanking region of FRE1 (Fig. 6A) was also evaluated (Fig. 6B, lanes  8 -11). An oligonucleotide including this direct repeat from FRE1 (Fig. 6A) was able to compete (again at 100-fold excess) with the CTR1 palindrome for binding to MAC1 protein (lane 8). Subsequently, a directly labeled oligonucleotide with the same sequence was tested, and a specific MAC1-dependent shift was observed in the EMSA (lane 10) which was reduced upon addition of the FLAG antibody (lane 11).

DISCUSSION
In this paper, the MAC1 protein was identified as a regulatory DNA-binding protein through which cellular copper levels are transduced into the regulated transcription of genes involved in copper acquisition. Specifically, copper-regulated expression of FRE1 and CTR1 was found to be altered in MAC1 mutant strains, with copper-independent expression in the MAC1 up1 and MAC1 up2 strains and negligible expression in the mac1⌬ strains. A homeostatic feedback loop for the control of cellular copper levels can thus be defined. Copper uptake into the cell, requiring reduction of copper chelates (FRE1-mediated) and translocation across the plasma membrane (CTR1mediated), subsequently inhibits transcription of FRE1 and CTR1 (MAC1-mediated).
To define further this feedback loop, we identified a regulatory sequence from within the CTR1 promoter that was sufficient for conferring copper-dependent and MAC1-dependent expression to a reporter gene. MAC1 was shown in vitro to bind specifically to a motif within this sequence. This sequence element (A/T)TTTGCTCA appears as a palindrome in the native CTR1 promoter region and is capable of direct interaction with MAC1 protein. A direct repeat of an identical sequence appears in the FRE1 promoter region and is also capable of interacting with MAC1 protein. As indicated by the promoter deletion analysis, direct and specific binding of MAC1 to these sequence elements appears required for transcriptional activation of the target genes involved in copper acquisition.
Examination of the MAC1 primary sequence provides some hints as to how this protein-DNA interaction might occur. The amino-terminal domain of MAC1 contains a subdomain (residues 1-42) with strong similarity to the amino-terminal domains of ACE1 and AMT1, proteins implicated in DNA binding and copper-dependent transcriptional activation in S. cerevi- 5. Localization of a copper-and MAC1-responsive element in the CTR1 promoter. Progressive 5Ј deletions of the CTR1 promoter constructed by PCR and cloned in front of a ␤-galactosidase reporter gene are depicted by horizontal bars indicating the CTR1 sequences included in each construct. The plasmids are named according to the number of the initial nucleotide of the CTR1 region included in the construct (i.e. pCTR lacZ-413 contains CTR1 sequences from Ϫ413 to ϩ3, numbering with respect to the translation start). The last entry, pGC-CTR (Ϫ337/Ϫ301), is a hybrid construct created by cloning an oligonucleotide with the indicated CTR1 sequences into a CYC1-lacZ fusion plasmid lacking its own upstream activating sequences. The CTR1 promoter-lacZ fusion constructs were introduced into strains CM3262 (MAC1), MA20 (MAC1 up2 ), or 3262⌬mac (⌬mac1). Cells were grown in defined medium in the presence of 10 M BCS (ϪCu) or 50 M CuSO 4 . ␤-Galactosidase activities represent the average of three independent assays (Ϯ 1 S.D.). siae and C. glabrata, respectively. The GRP motif (residues 37-39 in MAC1), conserved among MAC1, AMT1, and ACE1, is also shared with a larger family of transcription factors (PAXI, HMGI, and Drosophila prd and Hin recombinase) (30). This motif, and in particular the R residue, has been implicated in the AMT1 interaction with the AT-rich minor groove of its binding site (30). The TyrXCysX 2 CysX 3 HisX 4 Cys motif (residues 9 -23 in MAC1) is likewise shared with AMT1 and ACE1. In AMT1, this motif appears to be a zinc-binding element (31,32), and there is evidence for its involvement in the DNAbinding activity of both ACE1 and AMT1. Thus, a potential DNA-binding domain can be identified in MAC1 that resembles the apparent DNA-binding domains in ACE1 and AMT1.
The DNA sequences recognized by ACE1/AMT1 or MAC1 proteins are also strikingly similar to each other (Fig. 7). The AMT1 metal response element possesses a critical T thought to lie in the minor groove of the DNA helix and to interact with the GRP motif in the AMT1 protein. This T residue of the metal response element site in AMT1 (nucleotide Ϫ195) is conserved in three of four MAC1 interaction sites (Fig. 7). Only the 5Ј CTR1 site has an A in this position instead. The other residues of the core region identified by methylation interference as contacts for AMT1 are all conserved with the MAC1 binding site with the exception of the G at position Ϫ189 (Fig. 7). The MRE element from the CUP1 (copper thionein) promoter that interacts with ACE1 regulatory protein includes the sequence TTTTCCG*CTG*A (the asterisk-marked bases indicate methylation protection in the ACE1-bound state) (33). The TTTc-CgCT constituents were also shown to be critical by base substitution analysis (34). This pattern of base selectivity is similar to that shown for the MAC1 binding site in the CTR1 promoter, in which the GCT of TTTTGCTCA was identified as critical for DNA binding.
Thus, ACE1 and MAC1 share primary sequence homology within an amino-terminal domain thought to be important for DNA interaction, and they share features between their DNA recognition sequences. Together these observations could suggest that the protein-DNA complexes may be similar. However, ACE1 and AMT1 activate copper detoxification activity (thionein genes) and are active as trans-factors in a copper-bound state. In contrast, MAC1 activates copper acquisition and the FIG. 7. Homology between trans-factor binding sites in the AMT1-and MAC1-responsive promoters. The cis elements identified for DNA-protein interaction in the AMT1 and CTR1/FRE1 promoters are shown. Core regions are boxed. An asterisk indicates a base that when methylated interferes with AMT1 binding to its own promoter (30). The vertical arrow indicates the T that is conserved in all but one of the four MAC1 binding sites indicated by the EMSA.
FIG. 6. MAC1 binding to the CTR1 promoter. A, oligonucleotides used for evaluation of specific MAC1-DNA interaction by EMSA are shown. CTR1 is an oligonucleotide representing the CTR1 promoter element (Ϫ337 to Ϫ301) identified as critical for MAC1 function. CTR M1 and CTR M2 contain base substitutions as indicated. In CTR M2, the 5Ј-TGC-3Ј on the strand shown encodes a 5Ј-GCA-3Ј in the inverted repeat, e.g. a T to A transversion. FRE1 represents sequences from the FRE1 promoter element (Ϫ291 to Ϫ259). The horizontal arrows indicate the inverted and direct TTTGCTCA repeats found in these two promoter elements, respectively. B, EMSA was performed by mixing FLAG epitope-tagged MAC1 synthesized by in vitro transcription/translation with either 32 P-end-labeled CTR1 (Ϫ337 to Ϫ301) or FRE1 (Ϫ291 to Ϫ259) oligonucleotide. In lanes 3 and 11 anti-FLAG antibody was added. In lanes 4 -8 a cold competitor oligonucleotide as indicated was added in 100-fold molar excess in comparison to probe concentration. The cold competitor oligonucleotide AFT1 was based on the FET3 binding site for the AFT1 protein, AAAGTGCACCCATTTG. The arrow indicates the major shifted band, and the asterisk marks the position of the supershifted band that appears in the presence of the anti-FLAG antibody. data indicate is most likely active in DNA binding and transactivation when in a copper-depleted or copper-free form. With this difference in mind, the evolutionary relationship between these copper regulatory proteins is interesting to consider. Coordinated but opposite regulation of iron uptake (transferrin receptor) and iron detoxification (ferritin) by a common regulatory molecule (IRP1) has been described in mammals (35). Perhaps the regulatory molecules MAC1 and ACE1 in S. cerevisiae, controlling copper uptake and copper detoxification, respectively, diverged from a common ancestral regulator.
The copper-regulated activity of MAC1 could occur at least in part as a consequence of copper-regulated DNA binding. However, we were unable to detect a copper dependence of this binding in the EMSA. This result was similar to what has been observed with in vitro AFT1-DNA binding in which this interaction was independent of iron, although in vivo AFT1 trans activity is iron-dependent (28). Despite this negative result for MAC1, a suggestion about how copper might regulate MAC1 is provided by the amino acid sequence. That is, cysteine and histidine residues are known to coordinate copper in proteins, and the Cys-His element in the carboxyl-terminal domain of MAC1 (Fig. 2) might be able to reversibly interact with copper, generating a biological signal. In this model, a copper-depleted MAC1 protein form would bind to DNA followed by transcriptional activation. The gain-of-function alleles of MAC1 fit this model in that they contain mutations in potential copper-coordinating residues in this domain (H279Q, C271Y). Thus, by interfering with the copper binding ability of the Cys-His repeat domain of MAC1, these mutations result in a constitutively active protein.
How MAC1 may function to activate transcription is not known. However, the distribution of charged amino acid residues in MAC1 is asymmetric, with the amino-terminal domain (residues 1-201) strongly basic in nature (31 Arg and Lys, 21 Asp and Glu) and the carboxyl-terminal domain (residues 202-417) predominantly acidic (27 Asp and Glu, 16 Arg and Lys). Also, the putative nuclear targeting domain is found in the amino-terminal region (24). These features are consistent with a model in which the amino-terminal domain is associated with nuclear targeting and DNA binding, whereas the carboxylterminal domain is associated with copper sensing and transactivation. This model remains to be tested. For example, we have not rigorously excluded the possibility that the copper dependence of MAC1 function requires another protein(s), with this other protein being the copper sensor. While this model is not excluded by our data, we believe the model that most simply explains the phenotype of the UP alleles is based on their having a diminished affinity for copper.
The interrelationships between iron and copper homeostasis in yeast are defined by a complex network of genes. Cellular iron uptake requires the activity of the plasma membrane iron transport complex composed of the FTR1 and FET3 gene products (36). FTR1 most likely encodes an iron permease function (36), whereas the FET3 gene encodes a multi-copper oxidase that is required for iron uptake (6). Thus, the cellular copper uptake and utilization pathway mediating delivery of copper to FET3 must be intact for high affinity iron uptake to occur. This copper delivery pathway includes CTR1, required for cellular copper uptake (13). We may now add MAC1 to this scheme, because MAC1 is required for CTR1 expression and consequently is required for iron uptake.
At the level of gene regulation, however, the pathways for responding to changes in iron or copper levels are distinct. Iron deprivation induces increased activity of the iron uptake system through a homeostatic feedback loop involving the AFT1 regulatory protein (28). In iron-deprived cells, AFT1 binds to a specific recognition sequence in the enhancer regions of genes involved in iron uptake, including FET3 and FTR1 (28). AFT1 induces transcription of the target genes under conditions of iron deprivation. A separate but analogous pathway mediates the cellular response to copper deprivation; MAC1 is induced to bind to the specific recognition sequences identified in this paper, leading to expression of the genes involved in copper acquisition (FRE1, CTR1). The two systems are distinct in that the iron regulatory protein, AFT1, is distinct from the copper regulatory protein, MAC1, and each recognizes different DNA sequence elements. Only the FRE1 gene possesses binding sites for both AFT1 and MAC1 (10) and thus constitutes a special case. The dual role of surface reductases in facilitating iron and copper acquisition by reduction of extracellular metal chelates may explain this dual regulation of the FRE1 gene although the MAC1-and copper-dependent regulation of FRE1 seems to predominate as demonstrated by the Northern analysis shown in Fig. 4A.
Humans, much like yeast, have Cu/Zn superoxide dismutase and copper-based cytochrome oxidase (2), and defects in these proteins have been implicated in human disease (37,38). Cytoplasmic thioneins exist in humans as in yeast and are involved in protection against copper toxicity (39). Recently, a human homologue of CTR1 has been identified that is capable of complementing the phenotype of the yeast ctr1⌬ mutant (40). This protein may be involved in copper uptake in human cells. Human FRE1 homologues are also likely to exist (41). Since the problems of acquiring copper and avoiding its toxicity must be confronted by virtually all organisms, copper-dependent, homeostatic gene regulation is likely to occur in humans, too. Whether a MAC1 homologue functions to mediate copper homeostasis in humans must await further study.