Regulation of Zinc Homeostasis in Yeast by Binding of the ZAP1 Transcriptional Activator to Zinc-responsive Promoter Elements*

Zinc homeostasis in yeast is controlled primarily through the regulation of zinc uptake. Transcription of theZRT1 and ZRT2 zinc transporters increases in zinc-limited cells, and this induction is dependent on theZAP1 gene. We hypothesized previously that ZAP1encodes a zinc-responsive transcriptional activator. Expression ofZAP1 itself increases in zinc-limited cells. This response is also dependent on ZAP1 function through a potential positive autoregulatory mechanism. In this report, we describe the characterization of zinc-responsive elements (ZREs) in the promoters of the ZRT1, ZRT2, and ZAP1 genes. A ZRE consensus sequence, 5′-ACCYYNAAGGT-3′, was identified and found to be both necessary and sufficient for zinc-responsive transcriptional regulation. We also demonstrate that ZREs are DNA binding sites for ZAP1. First, a dominant ZAP1 mutation,ZAP1–1 up, which causes increased expression of ZAP1-regulated genes in zinc-replete cells, exerted its effects specifically through the ZREs. Second, electrophoretic mobility shift assays and in vitro DNase I footprint analyses indicated that ZAP1 binds to ZREs in a sequence-specific fashion. These studies demonstrate that ZAP1 plays a direct role in controlling zinc-responsive gene expression in yeast by binding to zinc-responsive elements in the promoters of genes that it regulates.

The variety of roles that zinc plays in cellular processes is a prime example of the utility of metal ions in biology. Zinc is required for the activity of more than 300 enzymes, including alcohol dehydrogenase, Cu/Zn superoxide dismutase, carbonic anhydrase, and many proteases (1). Zinc is also important for the correct folding of specific domains in many proteins. The largest class of proteins that require zinc as a structural cofactor are transcription factors that contain zinc-dependent DNAbinding motifs, such as zinc fingers and zinc clusters (2). For example, almost 2% of the genes in the genome of the yeast Saccharomyces cerevisiae contain zinc-dependent DNA binding domains (3,4). When zinc-dependent enzymes are also considered, perhaps as many as 5% of all yeast proteins require zinc for their function.
Although an essential nutrient, zinc can be toxic if excess amounts are accumulated. The precise cause of zinc toxicity is unknown, but the metal may bind to inappropriate intracellu-lar ligands or compete with other metal ions for enzyme active sites, transporter proteins, and so forth. Therefore, in the face of fluctuating extracellular zinc availability, cells must maintain an adequate intracellular zinc level to meet cellular requirements while preventing metal ion overaccumulation. Mechanisms of regulating the amount or availability of intracellular zinc include binding of the metal by metallothioneins (5), storage in intracellular compartments (6), and transport of zinc out of the cell (7). In S. cerevisiae, the regulation of zinc homeostasis is mediated primarily through the control of zinc uptake across the plasma membrane (8 -10).
S. cerevisiae has two separate uptake systems to obtain zinc from its environment. One system has a high affinity for zinc, and the ZRT1 gene encodes the transporter protein of this system (9). The second system is encoded by the ZRT2 gene and has lower affinity for zinc (8). Both ZRT1 and ZRT2 are regulated by zinc availability; zinc limitation induces ZRT1 and ZRT2 transcription, whereas growth under zinc-replete conditions represses their expression (9,10). This zinc-responsive transcriptional regulation requires the activity of the ZAP1 gene (10). From our initial characterization of ZAP1, we proposed that this gene encodes a transcriptional activator, the function of which is repressed by zinc. This hypothesis was based on several observations. First, deletion of the ZAP1 gene resulted in zinc-deficient phenotypes and abolished zinc-responsive expression of the ZRT1 and ZRT2 genes. Second, a semidominant mutation in ZAP1, called ZAP1-1 up , caused expression of these target genes in both zinc-limited and zincreplete cells. Third, sequence analysis of ZAP1 indicated that the encoded protein shares similarity with many transcriptional activators; the C-terminal region contains five C2H2 zinc finger domains and the N terminus contains two acidic regions that could act as transcriptional activation domains. Finally, we found that ZAP1 was required to increase its own transcription in response to zinc, potentially through a positive autoregulatory mechanism.
These observations established that ZAP1 plays a central role in zinc ion homeostasis by regulating transcription of the zinc uptake systems. As compelling as these data were, however, they failed to show whether ZAP1 plays a direct or an indirect role in the response. For example, an equally plausible alternative model was that ZAP1 is part of a signal transduction pathway that communicates the cellular zinc status to another protein, which then regulates transcription. In this report, we demonstrate that ZAP1 is in fact the transcriptional activator that directly controls zinc-responsive gene expression in yeast and identify its binding sites in the promoters of the ZRT1, ZRT2, and ZAP1 genes.

MATERIALS AND METHODS
Yeast Strains and Growth Conditions-Strains used were DY1457 (MAT␣ ade6 can1 his3 leu2 trp1 ura3) and ZHY7 (MAT␣ ade6 can1 his3 * Funding for this research was provided by National Institutes of Health Grants GM48139 and GM58265 (to D. E.) and by the Howard Hughes Medical Institute (to J. R.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. leu2 trp1 ura3 ZAP1-1 up ) (10). Yeast were grown in standard culture media (synthetic defined and yeast extract-peptone media) (11) supplemented with 2% glucose and auxotrophic supplements. A low zinc medium (LZM) 1 was prepared in a similar manner as LIM (12) and had the following composition: 0.17% yeast nitrogen base without amino acids, (NH 4 ) 2 SO 4 , or zinc (BIO101); 0.5% (NH 4 ) 2 SO 4 ; 10 mM Na 3 citrate pH 4.2; 2% glucose; 1 mM Na 2 EDTA; and 0.01% each adenine, histidine, leucine, and tryptophan. The MnCl 2 and FeCl 3 concentrations in LZM were adjusted to final concentrations of 25 and 10 M, respectively. Cell number in liquid cultures was determined by measuring the optical density of cell suspension at 600 nm (A 600 ) and converting to cell number with a standard curve.
␤-Galactosidase Assays and Atomic Absorption Spectroscopy-Cells were grown for 15-20 h to mid-exponential phase (A 600 ϭ 0.5-1) in LZM or synthetic defined medium supplemented with different amounts of ZnCl 2 . ␤-Galactosidase activity was assayed as described by Guarente (13) and is expressed in Miller units calculated as follows: (⌬A 420 ϫ 1000)/(min ϫ ml of culture used ϫ absorbance of the culture at 600 nm). Control plasmids for ␤-galactosidase assays were pHIS4-UAS (14) and pLG⌬312 (15). Measurement of cell-associated zinc levels was performed by atomic absorption spectroscopy. Cells were washed twice with an equal volume of distilled deionized water, twice with an equal volume of 1 M EDTA, resuspended in one-fifth volume 6 M nitric acid, and incubated at 95°C for 24 h. The acid-digested samples were then assayed for zinc content on a Varian Spectra AA-30 atomic absorption spectrometer.
Plasmids and DNA Manipulations-Escherichia coli and yeast transformations were performed using standard methods (16,17). The fusion and deletion junctions for all plasmids constructed were confirmed using the TaqTrack DNA sequencing method (Promega). Insert DNA fragments used for the 5Ј deletion analysis of the ZRT1 promoter were generated by the polymerase chain reaction using primers containing either a BamHI site (upstream primer) or an EcoRI site (downstream primer) added to their 5Ј-ends. These fragments were digested with BamHI and EcoRI and inserted into similarly digested integrating vector YIp353 (18) to fuse the ZRT1 promoter region and translation initiation codon to the lacZ open reading frame. The numbering system used throughout this report to describe subcloned promoter regions is relative to the first base of the translation initiation codon, which is designated as ϩ1. The resulting plasmids were linearized with NcoI to allow integration of the plasmid at the chromosomal URA3 locus after transformation (19).
Insert fragments for the plasmids containing internal deletions were generated by two-step overlapping polymerase chain reaction (20), such that the 11-bp zinc-responsive element (ZRE) sequences were precisely removed. The resulting fragments were inserted into either YIp353 (for ZRT1 analysis) or YEp353 (18) (for ZRT2 and ZAP1 analyses). Cloning of oligonucleotide inserts was performed using the synthetic oligonucleotides listed in Table I. Each single-stranded oligonucleotide was dissolved in water (100 M) and phosphorylated using T4 DNA kinase (16). To anneal complementary oligonucleotides, 200 pmol of each fragment were mixed, 10ϫ SSC (16) was added to a final concentration of 0.32ϫ, and the mixtures were incubated for 15 min at 85°C and then incubated at 55°C overnight. The annealed products, which contain 5Ј overhangs suitable for cloning into EagI and SalI sites, were ethanolprecipitated and then subcloned into EagI-SalI-digested pNB404 (21).
Two plasmids were generated for in vitro protein synthesis. A fragment containing the full ZAP1 open reading frame was prepared by polymerase chain reaction using primers with either ClaI or SacI sites added to their 5Ј-ends. The resulting fragment was inserted into ClaI-SacI-digested pBluescript SK ϩ (Stratagene) to generate pSK ϩ ZAP1. A 540-bp EcoRI-BamHI fragment encoding ZAP1 amino acids 687-880 was inserted into pBluescript SK ϩ to generate pSK ϩ ZnF. Two plasmids were also generated for in vitro DNase I footprint analysis. Plasmid pSK ϩ FPF is pBluescript SK ϩ with an inserted 362-bp ZRT1 promoter fragment (from Ϫ361 to ϩ1) and pSK ϩ ZRT2 is pBluescript SK ϩ containing a 250-bp insert derived from the ZRT2 promoter (i.e. position Ϫ375 to Ϫ74).
In Vitro Translation and Electrophoretic Mobility Shift Assays-Plasmids pBluescript SK ϩ , pSK ϩ ZAP1, and pSK ϩ ZnF were linearized with SalI and used as templates for an in vitro transcription/translation system (TnT, Promega). Annealed oligonucleotides were end-labeled using Klenow fragment DNA polymerase and [␣-32 P]dCTP. Electro-phoretic mobility shift assay (EMSA) binding reactions were carried out in 15-l volumes containing 10% glycerol, 12 mM HEPES-NaOH (pH 7.9), 4 mM Tris-Cl, 1 mM dithiothreitol, 60 mM KCl, 10 mM MgCl 2 , 10 M ZnCl 2, 2 g of poly(dI-dC), 2 ng of annealed and 32 P-labeled oligonucleotides, and the stated volume of in vitro transcription/translation product. This mixture was incubated for 15 min at 30°C, resolved on 5% polyacrylamide gels run at 4°C, and visualized by autoradiography. The electrophoresis running buffer (50 mM Tris base, 400 mM glycine, 2 mM EDTA, pH 8.0) was supplemented with 20 M ZnCl 2 to ensure activity of the ZAP1 zinc finger domains.
In Vitro DNase I Footprint Analysis-Plasmid pSK ϩ FPF was linearized with EagI and end-labeled with Klenow DNA polymerase and [␣-32 P]dCTP. Subsequent digestion of the plasmid with BamHI generated a 360-bp fragment that was used as a probe for in vitro DNase I footprint analysis. Probes for the ZRT2 and ZAP1 promoters were generated in a similar fashion from pSK ϩ ZRT2 (8) and pKO-ZAP1 (10), respectively. These plasmids were linearized with EcoRI, end-labeled with [␣-32 P]dATP, and then digested with KpnI. Five l of in vitro transcription/translation product was mixed with the labeled probe in the same binding reaction mixture used for EMSAs. Two l of DNase I (1U/l) was then added to each binding reaction and incubated for 1 min at room temperature. Reactions were terminated by adding 30 l of stop buffer (20 mM EDTA pH 8.0, 1% SDS, 0.2 M NaCl, and 200 g/ml yeast tRNA). Samples were extracted once with phenol/chloroform/ isoamyl alcohol (25:24:1), precipitated with ethanol, resuspended in formamide loading buffer (80% formamide, 1 mg/ml xylene cyanol FF, 1 mg/ml bromphenol blue, 10 mM EDTA), and resolved on a 6% denaturing polyacrylamide gel prior to autoradiography.

Identification of ZREs in the ZRT1
Promoter-Our previous studies indicated that the promoter of ZRT1 was completely contained within the upstream region extending from Ϫ706 to the translation initiation codon (ϩ1) (9). To map the ZREs that control expression of the ZRT1 gene, a progressive series of 5Ј deletions was generated such that increasingly smaller fragments of the ZRT1 promoter were fused to a lacZ reporter gene on an integrating plasmid vector (Fig. 1A). These plasmids were transformed into a wild type strain and assayed for expression in cells grown under zinc-limiting and zinc-replete conditions. Deletion of 5Ј-flanking sequences from Ϫ706 to Ϫ521 had no effect on the zinc responsiveness of the reporter gene (data not shown). Deletion of the region from Ϫ521 to Ϫ361 reduced ␤-galactosidase activity in zinc-limited cells by approximately 20% with little effect on expression in zincreplete cells. When sequences from Ϫ332 to Ϫ305 were deleted, an additional loss of zinc-limited expression was observed, again with no change in the zinc-replete expression. Finally, deletion from Ϫ221 to Ϫ201 completely eliminated expression in both media conditions. Initiation of transcription of ZRT1 was found by primer extension analysis to occur at a single site, position Ϫ45 (data not shown), indicating that the ZRT1 TATA box is unlikely to be affected by these deletions.
The results obtained with the 5Ј deletion series described in Fig. 1A are consistent with there being at least three different ZREs in the ZRT1 promoter. Analysis of the DNA sequence of this promoter revealed the presence of a conserved 11-bp sequence (5Ј-ACC(C/T)(C/T)(A/G/C)AAGGT-3Ј) in each of the three regions implicated by the deletion series. These sequences are located at positions Ϫ319 to Ϫ309, Ϫ204 to Ϫ194, and, in the opposite orientation, Ϫ454 to Ϫ444, and we refer to them as ZRE1, ZRE2, and ZRE3, respectively. Deletion of any single ZRE from the full-length promoter had little or no effect on zinc-responsive gene expression, whereas deletion of two elements (e.g. ⌬ZRE1 ⌬ZRE2) greatly reduced expression. Deletion of all three elements completely eliminated expression. These data demonstrate that ZRE1, ZRE2, and ZRE3 are necessary to confer zinc-responsive gene expression on the ZRT1 promoter. Moreover, the results show that these elements are required for the low level of ZRT1 expression observed in the zinc-replete conditions used in these experiments. We suggest that this expression is due to incomplete repression by zinc under these growth conditions. Consistent with this hypothesis, higher concentrations of zinc (2 mM) completely repressed a full-length ZRT1-lacZ reporter gene (data not shown).
To determine whether ZRE1, ZRE2, and/or ZRE3 are sufficient to confer this regulation, oligonucleotides containing these elements or portions thereof (Table I) were inserted into the promoter of a lacZ reporter gene bearing the CYC1 TATA boxes but lacking an upstream activation sequence (i.e. a "minimal" CYC1 promoter). First, three overlapping clones containing all or portions of the ZRE1 sequence were tested for ZRE function. When assayed for ␤-galactosidase activity in zinclimited and zinc-replete cells, the oligonucleotide insert containing the intact 11-bp sequence (i.e. Dg2) conferred zincresponsive expression on the minimal CYC1 promoter, whereas the two flanking inserts (Dg0 and Dg3) did not. Mutant variants of the Dg2 oligonucleotide were also constructed such that the sequences of either the 5Ј flanking 8 bp, the central 11 bp (i.e. ZRE1), or the 3Ј flanking 11 bp (designated M1, M2, and M3, respectively) were altered by transversion mutations. Mutation of the conserved 11-bp sequence eliminated ZRE function, whereas mutation of DNA flanking either side of the sequence did not decrease the zinc responsiveness of the reporter gene. These results demonstrate that ZRE1 is sufficient as well as necessary to confer zinc-responsive regulation and indicate that sequences flanking the conserved 11-bp element are not required for this activity. These data also demonstrate that a single ZRE is active; multiple elements are not required for function. Oligonucleotide insertions containing the ZRE2 and ZRE3 elements were also sufficient to confer zinc-responsive gene expression. Moreover, ZRE3 was inserted in its normal orientation, i.e. opposite that of ZRE1 and ZRE2, indicating that these elements are functional in either orientation relative to the start site of transcription.
Identification of ZREs in the ZRT2 and ZAP1 Promoters-The promoter region of the ZRT2 gene from Ϫ1047 to ϩ1 was sufficient to confer wild type zinc-responsive expression on a lacZ reporter (10). An examination of this region for sequences similar to the ZREs of ZRT1 identified two potential elements located at positions Ϫ310 to Ϫ300 and, on the opposite strand, from Ϫ261 to Ϫ251. We refer to these sites as ZRE1 and ZRE2.
To determine whether these are functional ZREs, we deleted ZRE1 and/or ZRE2 from the ZRT2-lacZ reporter. Deletion of each single element reduced zinc responsiveness, and deletion of both ZREs abolished the regulation (Fig. 1B). In contrast with the ZRT1 promoter, the "ZRE-less" ZRT2-lacZ reporter retained an almost normal level of expression in zinc-replete cells, suggesting that other factors may activate transcription of this gene under these conditions. Oligonucleotides containing the ZRE1 and ZRE2 elements of ZRT2 also conferred zincresponsive gene expression on the minimal CYC1 promoter (Table I). Therefore, the two ZRT2 ZREs are both necessary and sufficient to confer zinc-responsive gene expression.
Analysis of the ZAP1 promoter sequence identified a single potential ZRE at Ϫ143 to Ϫ133. Deletion of this sequence from the ZAP1 promoter (Ϫ1111 to ϩ1) eliminated expression of a ZAP1-lacZ reporter (Fig. 1C), and insertion of an oligonucleotide containing the ZAP1 ZRE conferred zinc-responsive regulation on the minimal CYC1 promoter (Table I). The data in Fig. 1 and Table I establish the identity of six different ZREs in the promoters of the ZRT1, ZRT2, and ZAP1 genes. Alignment of the sequences of these elements allowed the derivation of a consensus ZRE sequence, 5Ј-ACCYYNAAGGT-3Ј (Fig. 2). No conserved bases were identified outside the central 11-bp element. This observation is consistent with our results from the mutational analysis of ZRT1 ZRE1, indicating that flanking sequences were not required for zinc responsiveness. Therefore, we propose that this 11-bp consensus sequence is the complete element responsible for zinc-responsive gene expression in yeast.
ZREs Are Responsive to the ZAP1-1 up Mutant Allele-The ZAP1 gene encodes a potential transcriptional activator protein. Previously, we hypothesized that ZAP1 directly controls the expression of the ZRT1, ZRT2, and ZAP1 genes in response to zinc. This hypothesis was based in part on the isolation of a semidominant mutation in the ZAP1 gene, called ZAP1-1 up , that causes high level expression of these genes under zincreplete conditions (10). If our hypothesis concerning the func-tion of ZAP1 is correct, the increased expression caused by the ZAP1-1 up allele in zinc-replete cells should be exerted through the ZRE sequences. Therefore, we measured the effect of the ZAP1-1 up allele on several of the ZRE-lacZ fusion constructs described in Table I. Prior to assay for ␤-galactosidase activity, cells were grown in a zinc-replete medium in which the ZAP1-1 up allele effects are observed. In each case, expression of the ZRE lacZ reporter was elevated in the ZAP1-1 up mutant relative to wild type cells (Table I). We also examined the effect of the ZAP1-1 up allele on the vector alone, on a HIS4-lacZ fusion (pUAS-HIS4), and on a fusion containing the entire CYC1 promoter fused to lacZ (pLG⌬312). ZAP1-1 up had no effect on expression of any of these promoters (Table I), indicating that this allele does not cause global changes in gene expression. Thus, the effects of the ZAP1-1 up allele are mediated specifically through the ZRE sequences.
ZAP1 Binds to ZREs in a Sequence-specific Manner-An additional prediction of the hypothesis that ZAP1 is the direct regulator of zinc-responsive gene expression is that the ZAP1 protein binds directly to ZREs. To test this prediction, the full-length ZAP1 protein or a truncated ZAP1 polypeptide containing the five C-terminal zinc fingers (ZnF 1-5 ) (Fig. 3A) was produced with an in vitro transcription/translation system and

TABLE I Oligonucleotides used and their ZRE activities
Two single-stranded oligonucleotides were annealed to yield the following double-stranded fragments with flanking four base overhangs (not shown) for cloning into EagI and SalI sites. The bases of the ZREs are underlined and the mutated bases are in lowercase letters. Dg2, Dg0, and Dg3 contain all or portions of ZRE1 from the ZRT1 promoter. M1, M2, and M3 are mutant variants derived from the Dg2 oligonucleotide. These oligonucleotides were inserted into pNB404 and transformed into wild type (DY1457) and ZAP1-1 up (ZHY7) cells. Zinc responsiveness was assayed in wild type cells grown in low (ϪZn) or high (ϩZn) media as described in Fig. 1 and the induction is the ratio of the two values. ZAP1-1 up allele effects were determined in ZHY7 (Up) and DY1457 wild type (WT) strains grown in SD glucose medium. U/W is the ratio of these two values. Each value is the average of three independent assays Ϯ 1 S.D. ND, not determined. used in EMSAs. The ZnF 1-5 polypeptide comprises amino acids 687-880 of the full-length protein, and the probe used contained the ZRE1 sequence of ZRT1 (i.e. oligonucleotide Dg2, Table I). Protein/DNA complexes were not observed in reactions containing either no protein or increasing amounts of vector-programmed transcription/translation product (Fig. 3B,   lanes 1-4). Protein/DNA complexes were observed when either the full-length ZAP1 (Fig. 3B, lanes 5-7) or the ZnF 1-5 fragment (Fig. 3B, lanes 8 -10) was used in the assay. Whereas ZnF 1-5 formed only a single complex, the full-length protein formed three distinct complexes. The multiple bands detected using full-length ZAP1 may be due to the formation of homo-  3. ZAP1 binds to ZRE1. A, schematic representation of full-length ZAP1 and the ZnF 1-5 fragment used. The filled boxes represent the five C-terminal zinc fingers, the hatched boxes represent the two putative activation domains, and the numbers refer to the amino acid positions of the N and C termini of each polypeptide. B, EMSA using a 32 P-labeled ZRT1 ZRE1 oligonucleotide (Dg2). No protein (lane 1) or 1-, 2-, or 4-l samples prepared by in vitro transcription/translation reactions using either the pBluescript SK ϩ vector (lanes 2-4), pSK ϩ ZAP1 (lanes 5-7), or pSK ϩ ZnF (lanes 8 -10) as the template were used in the binding reactions. The ZAP1-or ZnF 1-5 -specific DNA-protein complexes are indicated by arrows. multimeric ZAP1 complexes or are an artifact of the in vitro transcription/translation system. We favor the latter hypothesis because shift assays performed with yeast whole cell extracts showed only a single ZAP1-dependent complex. 2 Perhaps alternative translational start sites are being utilized during in vitro protein synthesis or heteromultimers are forming with components of the in vitro system.
The experiment described in Fig. 3 indicated that ZAP1 is a DNA-binding protein and that binding is conferred by one or more of the five C-terminal zinc fingers. To test whether binding is specific for ZREs, we performed a competitive EMSA analysis (Fig. 4). The probe used was the Dg2 fragment (Table  I); competitor oligonucleotides were added in 100-fold molar excess. As shown in Fig. 4, A and B, oligonucleotides that retained ZRE function (i.e. Dg2, M1, M3, and the other intact ZREs) were effective competitors of all three labeled complexes formed by the full-length ZAP1. Oligonucleotides that lacked functional ZREs did not compete for binding. These results demonstrate that ZAP1 binds to ZREs in a sequence-specific manner. Identical results were obtained when EMSAs were performed using the zinc finger ZnF 1-5 polypeptide (Fig. 4, C  and D), suggesting that the full DNA binding activity of the intact protein is contained within the C-terminal 194 amino acids.
An independent test of the sequence-specificity of ZAP1 DNA binding was provided by in vitro DNase I footprint analysis of the ZRT1, ZRT2, and ZAP1 promoters (Fig. 5). No difference in the DNase I digestion products was observed between the no protein and vector-only controls (data not shown). However, full-length ZAP1 and the ZnF 1-5 polypeptide gave clear protection from DNase I cleavage in the regions of the ZRT1, ZRT2, and ZAP1 promoters that correspond to the ZREs. These data further demonstrate that the ZAP1 protein binds specifically to ZREs. Moreover, no difference in the length of the protected regions was observed between the full-length and truncated ZAP1 proteins. This observation strongly supports the hypothesis the truncated protein contains the full DNA binding domain. Only weak binding of ZAP1 was observed in similar experiments using the ZRE3 region of the ZRT1 promoter as the probe, suggesting that this site has lower affinity for ZAP1 binding (data not shown).
Differential Regulation of the ZRT1, ZRT2, and ZAP1 Promoters-Our previous studies of zinc uptake in yeast (8,9) indicated that although the ZRT1 high affinity transporter was completely repressed in zinc-replete cells, the low affinity ZRT2 transporter remained active. Surprisingly, the analyses described in this and a previous report (10) demonstrated that expression of both ZRT1 and ZRT2 is zinc-responsive and regulated by ZAP1. These observations presented a paradox; how can ZRT2 be active in zinc-replete cells yet still be regulated by ZAP1? To address this question, we examined expression of the ZRT1, ZRT2, and ZAP1 promoters during growth in media containing a range of zinc concentrations. Previous results suggested that repression of these genes occurred in response to an intracellular pool of the metal (9). Because no methods are currently available to directly measure the concentration of this regulatory zinc pool, we estimated its level by measuring total cell-associated zinc levels. As zinc concentrations in the medium rose, so did the cell-associated zinc levels (Fig. 6). The ZAP1 and ZRT1 promoters were extremely zinc responsive. For example, approximately 50% repression of both promoters was observed at 5 M zinc in the medium, corresponding to a cell-associated zinc level of 75-100 pmol/10 6 cells. In contrast, the ZRT2 promoter was repressed to a similar degree only at a zinc concentration in the medium that was 50-fold higher, i.e. 250 M. This medium zinc concentration corresponded to a cell-associated zinc level of 175 pmol/10 6 cells. Therefore, although ZRT1, ZRT2, and ZAP1 are all ZAP1 target genes, the ZRT2 gene requires higher zinc levels to repress its expression. DISCUSSION The ZAP1 gene is required for zinc-responsive gene expression in yeast (10). Target genes regulated by ZAP1 include the ZRT1 and ZRT2 zinc transporter genes and the ZAP1 gene itself. We proposed previously that ZAP1 encodes a transcriptional activator that binds to the promoters of these genes and 2 D. R. Winge, personal communication. activates their transcription when intracellular zinc levels are low. The experiments described in this report directly support this hypothesis, i.e. ZAP1 binds to zinc-responsive elements in the promoters of these genes in a sequence-specific fashion. Moreover, the identification of a ZAP1 binding site within its own promoter further supports the hypothesis that ZAP1 regulates its own expression in response to zinc via a positive autoregulatory mechanism. Presumably, this autoregulation provides a rapid and amplified response to changes in ZAP1 activity in response to zinc.
Characterization of zinc-responsive elements in ZAP1-regulated promoters has identified an 11-bp consensus sequence. This sequence appears to contain all of the information required for sequence-specific ZAP1 binding and this conclusion is based on two observations. First, sequence conservation among the different ZREs was found only within this 11-bp element and not in flanking sequences. Second, mutation of the flanking nucleotides had no effect on the zinc responsiveness of the ZRT1 ZRE1 element. Our studies have also illuminated several characteristics of ZREs. First and foremost, ZREs are both necessary and sufficient for zinc-responsive gene expression. Second, ZREs can function in either orientation relative to the start site of transcription. We also found that single ZREs are active; multiple copies are not required for function. Moreover, when these elements are present in multiple copies in a promoter, they are additive rather than cooperative in their effects (Fig. 1).
Surprisingly, there was no clear correlation between the zinc responsiveness of a given ZRE-lacZ fusion and its responsiveness to the ZAP1-1 up allele. For example, the ZRE1 and ZRE2 elements identified in the ZRT1 promoter showed relatively similar responsiveness to the ZAP1-1 up allele, whereas ZRE1 had 9-fold higher zinc responsiveness than the ZRE2-containing promoter (Table I). Likewise, the ZRE from the ZAP1 promoter showed only moderate zinc responsiveness (24-fold), whereas it showed the greatest response (58-fold) to the ZAP1-1 up allele. These inconsistencies may reflect differences in the affinity of ZAP1 for different ZREs, or the ZAP1-1 up allele may alter the DNA binding specificity of the ZAP1 protein. We view this latter hypothesis as unlikely given that the ZAP1-1 up mutation, a cysteine-to-serine substitution at amino acid 203, is far removed from the DNA binding domain.
The additive effects of multiple ZREs in a promoter can explain the different levels of expression among ZAP1 target genes. On Northern blots of mRNA from zinc-limited cells, we observed previously that ZRT1 is expressed at the highest level, ZRT2 is expressed at an intermediate level, and ZAP1 is expressed at the lowest level (10). This pattern of relative expression levels correlates with the number of ZREs found in these promoters (three, two, and one, respectively) and fits well with our understanding of the different functions of these genes. ZRT1 is needed at high levels under zinc limitation because of its critical role in supplying zinc to the cell under these extreme conditions. ZRT2, the low affinity transporter, plays more of a housekeeping function, supplying zinc to the cell under zinc-replete conditions, and ZAP1 is expressed only at low levels because of its role as a transcriptional regulator.
Our studies have also demonstrated that there is differential zinc responsiveness among the ZRT1, ZRT2, and ZAP1 genes; i.e. significantly more zinc is required to repress the ZRT2 promoter than is required to repress the ZRT1 or ZAP1 promoters (Fig. 6). This differential sensitivity to zinc is also consistent with the different functions of these proteins and leads us to propose the following scenario: basal (i.e. ZAP1independent) expression of the ZRT2 low affinity transporter is sufficient to supply zinc to cells under zinc-replete conditions (8). As cells first become zinc-limited, their initial response is to increase the activity of the ZRT2 transporter. If zinc limitation becomes more severe, the ZRT1 high affinity transporter is induced to provide high affinity uptake activity for zinc accumulation. Increased expression of the ZAP1 gene, which would allow maximum expression of its target genes, would only be needed under conditions of extreme zinc limitation. The mechanism underlying the differential regulation of these ZAP1 target genes is not yet known. If zinc controls the affinity of ZAP1 for its ZRE binding sites, one possible model is that other proteins bind to the ZRT2 promoter and help stabilize binding of ZAP1 to the ZREs, thus increasing the affinity of ZAP1 for these sites. This and other possible models will be addressed in future studies.
Given the size of the ZRE sequence, we propose that this site is bound by a single ZAP1 polypeptide. Our current understanding of how zinc fingers bind to DNA comes largely from x-ray crystal structures of protein-DNA complexes (22)(23)(24)(25)(26). In all of these structures, there are contiguous zinc finger interactions with base pairs in the major groove. In Zif268, for example, each of three fingers binds to a 4-bp site that overlaps the site of the adjacent finger by a single bp. We predict that three consecutive zinc fingers of ZAP1 would bind to a 10-bp sequence, similar in size to the 11-bp ZRE. Given that there are five potential zinc fingers in the DNA-binding, C-terminal 194 amino acids of ZAP1, and only three may be required for site-specific binding, we propose that the two additional fingers play roles in nonspecific DNA binding interactions and/or in protein-protein interactions. We have also recently learned of the presence of two additional zinc finger motifs at amino acids 581-604 and 618 -641 (4), and our data indicate that these upstream fingers are not required for DNA binding. The functions of each of the zinc finger domains in ZAP1 are currently under investigation. It should be noted that our ZRE consensus contains a near 2-fold dyad symmetry, suggesting the alternative possibility that ZAP1 binds as a dimer rather than a monomer.
An intriguing question that remains to be answered is precisely how zinc regulates ZAP1 activity. The characterization of the ZAP1 binding site in its target promoters is a critical step toward understanding this regulation. For example, in vitro and in vivo studies are now possible to determine whether zinc alters ZAP1 DNA binding. The characterization of a ZRE consensus sequence also provides us with a powerful tool to identify other zinc-responsive genes in the yeast genome, made possible by the recent completion of the Saccharomyces genome sequence. Using the consensus ZRE sequence derived in this study, sequence data base analysis (PatMatch software; http:// genome-www.stanford.edu/Saccharomyces/) identified a total of 20 genes in the yeast genome that contain one or more ZRE-like sequences in their promoters. This list of potential ZAP1 target genes is an exciting resource for the future analysis of how eukaryotic cells respond to zinc limitation and maintain zinc homeostasis.