Functional characterization of the iron-regulatory transcription factor Fep1 from Schizosaccharomyces pombe.

In response to excess iron, Schizosaccharomyces pombe cells repress transcription of genes encoding components involved in iron uptake through the Fep1 transcription factor. Fep1 mediates this control by interacting with the consensus sequence 5'-(A/T)GATAA-3', found in iron-dependent promoters. In this report, we show that Fep1 localizes to the nucleus under both iron-replete and iron-starved conditions. The Fep1 DNA binding domain (amino acids 1-241) contains two GATA-type zinc finger motifs. Although we determine that the Fep1 C-terminal zinc finger (ZF2) is essential for DNA binding, we show that the N-terminal zinc finger (ZF1) enhances DNA binding affinity approximately 5-fold. Between the two zinc finger motifs of Fep1 resides an invariant amino acid sequence, denoted the Cys-rich region (amino acids 68-94), in which four highly conserved Cys residues are found. Cells harboring mutant alleles in which two or more of the conserved Cys residues were substituted by alanine exhibited elevated fio1(+) mRNA levels. We determine that the dissociation constant for the resulting complex between each of the Cys mutants and the sequence 5'-(A/T)GATAA-3' reflects a much lower affinity that correlates with failure to repress fio1(+) gene expression. Deletion analysis identified two heptad repeats (amino acids 522-536) within the C-terminal region of Fep1 that are necessary and sufficient to mediate Fep1 dimerization. Moreover, mutations that impair dimerization also negatively affect transcriptional repression. Together these findings reveal several novel features of Fep1, a non-canonical GATA factor required for iron homeostasis.

The transition metal iron is both essential and, at high levels, toxic to cells (1)(2)(3)(4). The essentiality of iron is derived from its ability to cycle between two redox states, oxidized Fe 3ϩ and reduced Fe 2ϩ . This redox property of iron makes it a suitable catalytic cofactor for several enzymes that carry out critical biological processes, including DNA synthesis, respira-tion, the citric acid cycle, and photosynthesis (5,6). In excess, iron can have deleterious effects by reacting with hydrogen peroxide or dioxygen to produce hydroxyl radical, a free radical species that is responsible for cellular damage such as lipid peroxidation, oxidation of proteins, and cleavage of DNA or RNA molecules (7,8). Cells therefore possess mechanisms to ensure that sufficient and non-toxic levels of iron are present in the cell at all times (2, 9 -14).
In the fission yeast Schizosaccharomyces pombe, two pathways have been identified for iron assimilation (15)(16)(17). The first system consists of the production and secretion of a hydroxamate-type siderophore (15). Once secreted, this ferric ionchelating compound, named ferrichrome, can be recaptured and imported into the fission yeast cell as siderophore-iron complexes. Although the fission yeast produces the extracellular hydroxamate siderophore ferrichrome, S. pombe can also take up another type of siderophore, ferrioxamine B, when secreted into the environment by other microorganisms (18). To date three genes, str1 ϩ , str2 ϩ , and potentially str3 ϩ genes have been identified whose products are involved in uptake of siderophore-bound iron (18). Of these three siderophore transporters, Str1 exhibits specificity for ferrichrome-iron, whereas Str2 is specific for ferrioxamine B-iron and to a lesser extent ferrichrome-iron (18). Although Str3 may participate in the mobilization of iron bound to siderophores, its substrate specificity has not been determined. In S. pombe, a second system for iron uptake involves reduction of iron from Fe 3ϩ to Fe 2ϩ by the Frp1 cell surface ferrireductase (17). Subsequent to reduction, iron is transported across the plasma membrane by a bipartite protein complex composed of the multicopper oxidase Fio1 and the transmembrane permease Fip1 (16). Within this two-component iron transporting complex, Fio1 can re-oxidize Fe 2ϩ to Fe 3ϩ , which is then transported into the cell by the high affinity carrier Fip1.
Under conditions of iron starvation, frp1 ϩ , fio1 ϩ , fip1 ϩ , str1 ϩ , str2 ϩ , and str3 ϩ mRNA levels are strongly induced, whereas under iron-replete conditions, expression of these genes is inactivated (16 -20). The promoter element necessary for iron-mediated repression of the frp1 ϩ , fio1 ϩ , fip1 ϩ , str1 ϩ , str2 ϩ , and str3 ϩ genes has been determined to be a GATA-type regulatory sequence 5Ј-(A/T)GATAA-3Ј (18,19). Consistently, the sensor for down-regulating expression of these genes in the presence of exogenous iron has been demonstrated to be Fep1, a member of the GATA factor protein family (18,19). S. pombe cells in which fep1 ϩ has been disrupted display elevated levels of expression of the fio1 ϩ gene (19). Furthermore, fep1⌬ mutant cells exhibit increased activity of the cell surface reductase Frp1 (19). Moreover, the fep1⌬ disruption strain is hypersensitive to phleomycin, an antibiotic that cleaves nucleic acids in the presence of excess iron (19). We have identified two proteins, Tup11 and Tup12, which act as corepressors for ironmediated transcriptional repression (19). We demonstrated that Tup11 and Fep1 physically interact with each other for down-regulation of iron uptake genes (21). The C-terminal amino acids from residues 405 to 541 are required for this interaction (21). The Fep1 N-terminal 241 amino acids have been shown to be important for the binding of Fep1 to DNA (19). Furthermore, this region of Fep1 is highly similar to the N-terminal regions of the Urbs1, SRE, SREA, and Sfu1 proteins that have been shown to play a role in the regulation of iron-responsive genes of Ustilago maydis, Neurospora crassa, Aspergillus nidulans, and Candida albicans, respectively (22)(23)(24)(25)(26)(27)(28)(29). Within their N termini, these regulators harbor two Cys 2 / Cys 2 -type zinc fingers and a highly conserved intervening Cysrich region located between the two zinc fingers (25,26,30). We refer to these domains as the N-terminal zinc finger (ZF1), 1 and the C-terminal zinc finger (ZF2), according to their relative positions in the GATA-type protein. Although Urbs1, SRE, and SREA share a common highly conserved N terminus, there are marked differences with respect to the amino acid sequences required for their respective functions. In U. maydis, Urbs1 requires only ZF2 to bind DNA and regulate siderophore biosynthesis as a function of iron availability (23). In N. crassa, although ZF2 plays a more important role in DNA binding in vitro, results suggest that both the N-and C-terminal zinc fingers of SRE are required for DNA binding (24). Mutations affecting either or both SRE zinc fingers block the ability of the transcription factor to negatively regulate siderophore gene expression under high iron concentrations (24). In comparison, in the context of a DNA fragment containing two GATA sites that had been shown to bind the GATA factor NREB (31), the DNA binding domain of SREA expressed as a fusion protein in Escherichia coli requires the presence of both zinc finger motifs for specific DNA binding (26). In SRE, the four highly conserved Cys residues found between the two zinc finger motifs have been mutated, resulting in constitutive repression of target genes in the presence or absence of iron. Consistently, these mutant versions of SRE retain the ability to bind to DNA (25).
To gain additional insight into functional features of ironregulatory GATA-type transcriptional repressors, we performed a detailed structure-function analysis of the Fep1 protein. We determined that a functional Fep1-GFP fusion protein localizes to the S. pombe nucleus. Furthermore, its nuclear localization remains unchanged in response to variations in iron levels. We found that the presence of both finger domains is required for iron-dependent down-regulation of fio1 ϩ gene expression. Precisely, we show that ZF2 is necessary for Fep1 to interact with the 5Ј-(A/T)GATAA-3Ј sequence in vitro. Furthermore, we determine that a truncated form of Fep1 in which ZF1 was removed exhibits a much lower DNA affinity (ϳ5-fold) that correlates with failure to down-regulate fio1 ϩ gene expression. We demonstrate that mutation of the first two, the last two, or all four invariant Cys residues found between the two zinc finger motifs blocks iron repression by Fep1 and results in lower binding affinities for GATA elements in vitro. We determined that a leucine zipper domain, residues 522-536, included within the C-terminal region of Fep1 is essential for Fep1-Fep1 dimer formation and critical for full transcriptional repression. Taken together, our data show that both zinc fingers in concert with the Cys-rich region are required to allow maximal Fep1 DNA binding, resulting in repressing fio1 ϩ gene expression under iron-replete conditions. Furthermore, Fep1 dimerization increases its potency as a transcriptional repressor.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-The S. pombe strains used in this study were the wild type FY435 (h ϩ his7-366 leu1-32 ura4-⌬18 ade6-M210) and the fep1⌬ mutant strain (h ϩ his7-366 leu1-32 ura4-⌬18 ade6-M210 fep1⌬::ura4 ϩ ) (19). Fission yeast cells were maintained in yeast extract plus supplements medium. Under selection, cells were grown in Edinburgh minimal medium with necessary auxotrophic requirements (32). Iron deprivation or iron repletion was carried out by adding the indicated amount of bathophenanthrolinedisulfonic acid (BPS) or FeCl 3 to cells grown to mid-logarithmic phase (A 600 nm of ϳ1.0). After treatments at 30°C for 90 min, 20-ml samples were withdrawn from the cultures for subsequent steady-state mRNA or protein analyses (33). To assess the phleomycin-sensitive growth phenotype, 20 g/ml phleomycin (Sigma) was added to yeast extract plus supplements as described previously (19).
DNA Constructs-The S. pombe fep1 ϩ promoter up to Ϫ1478 from the start codon of the fep1 ϩ gene was isolated by PCR and then inserted into the pJK148 vector (34) at the SacII and BamHI sites. The resulting plasmid was denoted pJK-1478fep1 ϩ prom. The fep1 ϩ open reading frame was isolated by PCR from genomic DNA of the wild type FY435 strain. The PCR product obtained was digested with XhoI, filled-in with Klenow, and digested with BamHI. The purified DNA fragment was cloned via XhoI (Klenow) and BamHI sites into the SmaI and BamHI sites of the pJK-1478fep1 ϩ prom plasmid to generate pJK-1478fep1 ϩ . Therefore, we ensured that the fep1 ϩ gene was under the control of its own promoter, once integrated in the genome. The gfp gene was isolated by PCR from the pSF-GP1 plasmid (35). Because the primers contained SstI and NsiI restriction sites, the purified DNA fragment was digested with these enzymes and cloned into the pGEM-7Zf(ϩ) vector (Promega, Madison, WI). Subsequently, a 1563-bp SphI-SstI PCR-amplified fragment containing the last 520 codons of the fep1 ϩ open reading frame was isolated. The SphI-SstI DNA fragment was inserted into the SphI-SstI-cut pGEMgfp plasmid, creating a chimeric plasmid that has the last 520 codons of fep1 ϩ fused to GFP. The gfp open reading frame, in which a XhoI restriction site was previously engineered by PCR and placed immediately after the stop codon, was used in combination with SphI to isolate the gfp gene in-frame with the C-terminal region of Fep1. The purified DNA fragment was cloned into pSP1-1478fep1 ϩ codons 1-44 to generate a plasmid that encodes a full-length Fep1-GFP fusion protein. Once generated, the fep1 ϩ -gfp fusion allele was isolated using XhoI (Klenow) and BamHI and subcloned into pJK-1478fep1 ϩ prom to create pJK-1478fep1 ϩ -GFP. To generate the fep1-R184A-R185A-D186A-E187A allele, the primers FEP1RRDEMUT-UP (5Ј-CAAATACTCCTTTGTGGCTCCTCGCCGTGTCGGGTAATCCGAT-CTG-3Ј) and FEP1RRDEMUT-LO (5Ј-CAGATCGGATTACCCGACAC-GGCGAGGAGCCACAAAGGAGTATTTG-3Ј) were made with mutations (underlined) and utilized in the overlap extension method as described previously (36) to generate the R184A, R185A, D186A, and E187A substitutions within the fep1 ϩ gene. Using the pJK-1478fep1 ϩ -GFP plasmid, a set of deletions (5-59⌬, 5-129⌬, and 5-261⌬) was made within the N-terminal portion of Fep1. Three primers containing a BamHI site and the first four codons of the fep1 ϩ gene were engineered to anneal pairwise to DNA at codons 60, 130, and 262 of fep1 ϩ , respectively. The antisense primer for amplification contained a BlpI restriction site. The DNA sequence from each respective PCR-amplified fragment was digested with BamHI and BlpI and cloned into the corresponding sites of pJK-1478fep1 ϩ -GFP. The fep1 ϩ allele was amplified by PCR using primers designed to generate BamHI and NotI sites at the upstream and downstream termini of the open reading frame. Once generated, the DNA fragment was inserted into the corresponding sites of pREP1-NTAP (37). The resulting plasmid, named pNTAPfep1 ϩ , was subsequently digested with PacI, treated with mung bean nuclease, and then digested with SpeI to produce a DNA fragment corresponding to the TAP tag fused to the first 149 codons of Fep1. This latter fragment was swapped for an identical DNA region into the pJK-1478fep1 ϩ plasmid using the BamHI (Klenow) and SpeI sites to create the pJK-1478-NTAPfep1 ϩ plasmid. To create truncations from the C-terminal end of NTAP-Fep1, the fep1 ϩ gene starting at ϩ446 after the start codon up to ϩ723, ϩ954, ϩ1077, ϩ1473, or ϩ1623 was amplified using primers designed to introduce SpeI and PstI at the termini of the upstream and downstream DNA fragments, respectively. The PCR products obtained were digested with SpeI and PstI and cloned into the corresponding sites of pJK-1478NTAPfep1 ϩ to generate plasmids pJK-1478NTAPfep1-M241, pJK-1478NTAPfep1M318, pJK-1478NTAPfep1M359, pJK-1478-NTAPfep1M491, and pJK-1478NTAPfep1M541. To create the fep1 mutant alleles C70A/C76A, C85A/C88A, and C70A/C76A/C85A/C88A, the plasmid pJK-1478fep1 ϩ -GFP was used in conjunction with the overlap extension method (36) and the oligonucleotides FEP1C70AC76A-UP (5Ј-GCAAGTTTGTAAAAACGGAACCGCTGCTGGAGATGGATTCGC-TAACGGTACGGGTGGCAG-3Ј), FEP1C70AC76A-LO (5Ј-CTGCCACC-CGTACCGTTAGCGAATCCATCTCCAGCAGCGGTTCCGTTTTTACA-AACTTGC-3Ј), FEP1C85AC88A-UP (5Ј-CGGGTGGCAGTGCTTCAGC-TACTGGTGCTCCTGCTTTAAATAATCG-3Ј), and FEP1C85AC88A-LO (5Ј-CGATTATTTAAAGCAGGAGCACCAGTAGCTGAAGCACTGCCA-CCCG-3Ј) (underlined letters represent nucleotide substitutions that gave rise to mutations). To generate the fep1MutLI 3 A allele, the primer FEP1MUTLI-A (5Ј-CATTGTCAGAGGACCTCGGCCATTCAT-CTGGTTGTTGAGGGTGCTGATTTGGATTGTGCGGCTCCTGAATC-A-3Ј) was made corresponding to the end of the fep1 ϩ gene with mutations (underlined) in the sequence that generated the L522A, L525A, I529A, I532A, and L536A substitutions. Furthermore, the primer was designed to ensure the presence of the PpuMI restriction site found naturally at the end of the gene. Using a second primer that hybridized downstream at the BlpI restriction site, a 804-bp BlpI-PpuMI fragment was amplified by PCR and then used to replace the equivalent wild type fep1 ϩ DNA segment in pJK-1478fep1 ϩ -GFP.
Assays-To perform the RNase protection analyses as described previously (19), plasmids pSKfio1 ϩ and pSKact1 ϩ were used for producing antisense RNA probes, allowing the detection of steady-state levels of fio1 ϩ and act1 ϩ mRNAs, respectively. Immunodetection of Fep1-GFP and PCNA proteins was conducted with monoclonal anti-GFP antibody B-2 (Santa Cruz Biotechnology, Santa Cruz, CA) and monoclonal anti-PCNA antibody PC10 (Sigma), respectively. Fluorescence and differential interference contrast images of the cells were obtained on an Eclipse E800 epifluorescent microscope (Nikon, Melville, NY) equipped with an ORCA ER digital cooled camera (Hamamatsu, Bridgewater, NJ). Cells were grown in yeast extract plus supplements medium to A 600 of 1.0 and then BPS (100 M) or FeCl 3 (100 M) was added. After 90 min incubation at 30°C, Hoechst 33342 stain (Sigma) was added to 1 ml of each cell culture to a final concentration of 5 g/ml for DNA staining. Cells were washed with medium, mixed with 1% low melting agarose, and mounted onto glass slides. The samples were subjected to microscopy analysis, using ϫ1000 magnification and the following filters: 465-495 (GFP) and 340 -380 (Hoechst). His-tagged Fep1 proteins were affinity purified from E. coli extracts using nickel-agarose chromatography and the purity of the proteins determined by Coomassie Blue staining of SDS-polyacrylamide gels. For EGS cross-linking experiments, purified His 6 -Fep1 fusion proteins were diluted into buffer A (50 mM HEPES, pH 7.9, 50 mM NaCl, 2 mM dithiothreitol, 10% glycerol) to a final concentration of 1 g. EGS was added to concentrations specified in Fig. 10, incubated for 30 min at 25°C, and quenched with 50 mM Tris-HCl, pH 8.0. The cross-linking proteins were resolved by electrophoresis through a 12% SDS-polyacrylamide gel and analyzed by immunoblotting with the monoclonal anti-His antibody 34460 (Qiagen, Mississauga, ON).
Analysis of DNA-Protein Interactions-The maltose-binding protein (MBP) was fused to the N terminus (residues 2-241) of the Fep1 protein or its mutant derivatives. To generate these fusions, each fragment was obtained by PCR amplification using as a template the plasmid pJK-1478fep1 ϩ -GFP or one of its mutant versions. The purified fragments were digested with BamHI and PstI and subsequently cloned into the corresponding sites in plasmid pMAL-c2x (New England BioLabs, Beverly, MA). Once produced in and purified from E. coli TB1 cells as described previously (19), each MBP-Fep1 derivative fusion protein was used for DNA binding analyses. Electrophoretic mobility shift assays were conducted as described previously (19), except that the binding reactions were carried out using 1ϫ binding buffer, which contained 12.5 mM HEPES, pH 7.9, 75 mM NaCl, 4 mM MgCl 2 , 1 mM EDTA, 10% glycerol, 4 mM Tris-HCl, pH 7.9, 0.6 mM dithiothreitol, 1 g of poly(dI-dC) 2 , 10 M ZnSO 4 , and 10 M FeCl 3 . For quantitative in vitro DNA binding studies, the amount of affinity purified MBP-Fep1 derivative fusion used for DNA binding was determined as described previously (33). Protein dilutions were carried out by sequential 2-fold serial dilutions into 1ϫ binding buffer. The probe concentration was maintained at 1 ng for apparent K d (K d(app) ) determinations. The data derived from the PhosphorImager quantitation were analyzed using GraphPad Prism software (GraphPad Software, San Diego, CA). The K d(app) for each affinity purified MBP-Fep1 derivative protein was determined by fitting the probe titration plots to the Hill equation.
Protein Pull-down Interaction Assays-Plasmid pET-28a-360 fep1 ϩ541 containing the C-terminal region of Fep1 fused downstream of and in-frame to the His 6 tag region was created by introducing a 546-bp NdeI-NotI PCR-amplified fragment containing the fep1 ϩ gene starting at codon 360 to 541 into the NdeI-NotI-digested pET-28a vector (EMD Biosciences, San Diego, CA). Using primers designed to introduce BamHI and SmaI at the termini of DNA fragments, the His6-360 fep1 ϩ541 fusion gene was isolated from pET-28a-360 fep1 ϩ541 and cloned into the pREP4X vector (43,44). The resulting plasmid was named pREP4-His6-360 fep1 ϩ541 . Plasmid pNTAP-360 fep1 ϩ541 was generated as follows. The DNA sequence encoding the C-terminal domain of Fep1 (residues 360 -541) was isolated by PCR, purified, and cloned into the BamHI and NotI sites of pREP1-NTAP (37). For protein pulldown interaction experiments, FY435 cells were co-transformed with pNTAP-360 fep1 ϩ541 or pREP1-NTAP (empty), and pREP4-His6-360 fep1 ϩ541 or pREP4X (empty). For protein expression, cells were grown to A 600 of 1.0 in the absence of thiamine. After incubation for 18 -20 h at 30°C, cells were collected by centrifugation and lysed at 4°C with glass beads in buffer A (20 mM HEPES, pH 7.9, 100 mM NaCl, 50 M EDTA, pH 8.0, 1 mM dithiothreitol, 10% glycerol, 1 mM phenylmethylsulfonyl fluoride, 10 mM imidazole) in the presence of a complete protease inhibitor mixture (P8340, from Sigma) followed by centrifugation at 3500 rpm at 4°C for 5 min. Protein extracts (ϳ3 mg of total protein) were incubated for 6 h with a 50-l bed volume of nickelnitrilotriacetic acid-agarose beads (Qiagen, Mississauga, ON). Beads were washed four times with 1 ml of buffer A, with the beads completely transferred to a fresh microtube before the last wash. The samples were resuspended in 100 l of SDS loading buffer, incubated for 5 min at 95°C, and resolved by electrophoresis on a 8% SDS-polyacrylamide gel. For protein detection of NTAP-360 Fep1 541 , His6-360 Fep1 541 , and PCNA, the following primary antisera were used: polyclonal anti-mouse IgG antibody (ICN Biomedicals, Aurora, OH) monoclonal anti-His antibody 34460 (Qiagen), and monoclonal anti-PCNA antibody PC10 (Sigma). After immunoblots were washed in TBS (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% bovine serum albumin), membranes were incubated with appropriate horseradish peroxidase-conjugated secondary antibodies (Amersham Biosciences) for 30 min, washed, and developed with enhanced chemiluminescent detection reagents. Pull-down experiments with TAP-tagged-319 Fep1 564 (wild type or mutant version) and GSTtagged Tup11 were carried out as described previously (21).

Fep1 Is Localized within the Nucleus of the S. pombe Cell under Both Iron-limiting and Iron-replete Conditions-Our
previous studies revealed that the S. pombe Fep1 protein mediates the iron-dependent repression of genes encoding components of the reductive (e.g. Frp1, Fio1, and Fip1) and nonreductive (e.g. Str1, Str2, and Str3) iron transport systems (18,19). As shown in Fig. 1A, fio1 ϩ mRNA levels in a strain expressing the wild type Fep1 protein exhibited down-regulation (ϳ4-fold) in the presence of both low (1 M) and elevated (100 M) iron concentrations. Conversely, in the presence of 100 M BPS, fio1 ϩ mRNA levels were induced (ϳ8-fold) over basal levels. In a fep1⌬ mutant strain, the fio1 ϩ mRNA levels were strongly increased (ϳ15-fold) compared with the basal levels observed in the wild type strain, and were virtually unregulated by iron or iron-deprivation. Importantly, iron-dependent regulation of fio1 ϩ gene expression was corrected by integrating a fep1 ϩ -GFP fusion allele expressed from its own promoter, indicating that the fusion retained wild type function (Fig. 1A). In a previous study, we demonstrated that the steady-state levels of fep1 ϩ mRNA are constitutive and unaffected by cellular iron status (19). To further investigate the mechanism by which Fep1 activity is regulated, we tested whether post-transcriptional control of Fep1 operated through iron-regulated changes in protein stability. Analysis of Fep1-GFP expressed in fep1⌬ cells grown under low or elevated iron concentrations for 3 h revealed that Fep1-GFP is stable and present in similar relative amounts in iron-deficient and iron-replete cells (Fig.  1B). Steady-state levels of another fission yeast nuclear protein, the polymerase ␦ auxiliary protein PCNA (45) localization (46 -48). For instance, masking of a nuclear localization signal by a conformational change, a sequestering protein or by phosphorylation may inhibit nuclear transport of the factor (48 -50). We examined the localization of Fep1-GFP as a function of iron deficiency or sufficiency. As shown in Fig. 1C, fluorescence microscopy demonstrated that Fep1-GFP is localized in the nucleus under both iron-starvation and iron-replete conditions. Indeed, Fep1-GFP fluorescence co-localized with the DNA staining dye Hoechst, which was used as a marker to stain the nucleus. Whereas no fluorescence was observed in empty vector-integrated cells, the fep1⌬ deletion strain expressing GFP alone exhibited a fluorescent signal throughout the cells. Taken together, these data reveal that regulation of Fep1-GFP by iron does not occur through iron-regulated changes in translation efficiency, protein stability, or altered protein localization.
Both Zinc Fingers of Fep1 Are Necessary for Iron-mediated Repression of fio1 ϩ , yet Only the Second Finger Is Sufficient for in Vitro DNA Binding-The N-terminal region of Fep1 contains two GATA-type zinc finger motifs, denoted ZF1 (residues 12-60) and ZF2 (residues 172-220) ( Fig. 2A). To investigate the role of the two fingers in Fep1, we converted the Arg 184 , Arg 185 , Asp 186 , and Glu 187 residues to Ala to generate the Fep1-GFP R184A/R185A/D186A/E187A mutant that nullifies the second zinc finger motif as reported for Urbs1 (23). Furthermore, N-terminal deletions were created to remove indi- vidually, and in combination, the sequences encoding the two zinc finger motifs, creating the Fep1-GFP-(5-59⌬), Fep1-GFP-(5-129⌬), and Fep1-GFP-(5-261⌬) mutants ( Fig. 2A). Because of the possibility that point mutations and deletions introduced by site-directed mutagenesis may result in a non-functional Fep1-GFP because of protein mislocalization, we ascertained the localization of each of these Fep1-GFP derivatives in a fep1⌬ strain by using fluorescence microscopy ( Fig. 2A). All of the fusion products were properly localized in the nucleus, except the Fep1-GFP-(5-261⌬) mutant that accumulated predominantly in the cytosol, and in most cells, was largely excluded from the nucleus ( Fig. 2A and data not shown). This may result from loss of a putative nuclear localization signal and consequently an inhibition of its nuclear import. For each mutant, we measured the steady-state levels of fio1 ϩ mRNA in response to iron versus basal or iron-starvation (BPS) conditions (Fig. 2B). The fio1 ϩ mRNA levels were strongly increased (ϳ18-fold) compared with the basal levels observed in the wild type strain. For each case, this resulted in sustained expression of the fio1 ϩ mRNA levels that were unresponsive to iron. Although the mutant proteins Fep1-GFP R184A/R185A/D186A/ E187A, Fep1-GFP-(5-59⌬), and Fep1-GFP-(5-129⌬) exhibited nuclear localization, they were inactive because no iron-mediated repression of fio1 ϩ was observed. Thus, these results suggested that the two zinc finger motifs in Fep1 play an important role(s) in the regulation of Fep1 activity to mediate iron-responsive gene expression of fio1 ϩ .
Because inactivation of either one or both zinc fingers affected the function of the Fep1-GFP protein in vivo, we hypothesized that their presence may be essential for Fep1 binding to DNA. To test this hypothesis, we separately expressed the wild type and mutated forms of the N-terminal region of Fep1 fused to MBP in E. coli cells. These polypeptides were purified to near homogeneity using two rounds of one-step affinity chromatography based on MBP affinity for maltose (Fig. 3A). To examine whether the N-terminal domain of Fep1 with ZF1, ZF2, or both zinc fingers interacts with GATA elements, DNA binding experiments were carried out with the purified fusion proteins. As shown in Fig. 3B by a representative electrophoretic mobility shift assay, the wild type 32 P-labeled 46-bp oligomer, which contains two copies of a repeated 5Ј-(A/T)GATAA-3Ј sequence, forms a DNA-protein complex in the presence of wild type MBP-2 Fep1 241 . Furthermore, using quantitative electrophoretic mobility shift assays (data not shown), we determined that the wild type MBP-2 Fep1 241 binds to the GATA elements with a K d(app) of 6.3 ϫ 10 Ϫ8 M. In contrast, the MBP-Fep1-(5-59⌬) and MBP-Fep1-(5-129⌬) mutant proteins, in which ZF1 was deleted, bind to the GATA sequences, but at a K d(app) of 3.3 ϫ 10 Ϫ7 and 4.0 ϫ 10 Ϫ7 M, respectively, which is 5.2 and 6.3 times weaker than the wild type MBP-2 Fep1 241 . Although we cannot eliminate the possibility that ZF1 may constitute an essential part of a protein-protein interaction domain to facilitate intermolecular repression, the quantitative in vitro DNA binding studies revealed that ZF1 plays an important role in increasing the affinity of the Fep1-DNA complex. The fact that the mutant proteins have a much lower affinity (5.2-and 6.3fold less) for 5Ј-(A/T)GATAA-3Ј motifs could explain why without ZF1, Fep1-(5-59⌬) and Fep1-(5-129⌬) were found to be insufficient to mount a normal transcriptional repression of the wild type fio1 ϩ promoter in vivo. On the other hand, the MBP-Fep1 R184A/R185A/D186A/E187A mutant protein in which the second zinc finger (ZF2) was mutated did not form any DNAprotein complexes (Fig. 3B). Importantly, this latter fusion protein produced in and purified from E. coli cells was not compromised for stability as shown in Fig. 3C. To show the specificity of complex formation with the N-terminal portion of the Fep1 protein, we carried out an experiment with a chimeric protein containing MBP fused to Fep1 C-terminal codons 262-564. Consistent with the lack of a potential motif that would be involved in DNA binding, no DNA-protein complexes were observed in the mobility shift assay. Furthermore, we carried out competition experiments with unlabeled oligomers using either wild type GATA or GATA with multiple point mutations within the 46-bp DNA fragment (19). Formations of the DNA-protein complexes were inhibited by incubation with excess wild type oligomer, but not by the mutant competitor, indicating that the complexes were formed by sequence-specific interactions (data not shown). Taken together, these data reveal that the presence of both zinc finger motifs of Fep1 are necessary for ironregulated expression of fio1 ϩ , whereas the C-terminal zinc finger motif alone is sufficient for in vitro DNA binding.
Mutation of Cys-rich Region Inactivates Fep1-A conserved amino acid sequence (residues 68 -94), denoted the Cys-rich region, contains four invariant Cys residues located between the two zinc fingers of Fep1 (25,26,30). To ascertain if these Cys residues play a role in fio1 ϩ regulation by iron, either one or both pairs of Cys were substituted by Ala (Fig. 4A). These mutations had no effect on Fep1 steady-state levels (Fig. 4B). Furthermore, these mutants properly localized in the nucleus. 2 fep1⌬ cells transformed with the vector alone exhibited a high level of fio1 ϩ gene expression without any change in response to iron repletion or iron starvation conditions (Fig. 4C). Expression of wild type Fep1-GFP fusion protein restored iron-mediated repression of fio1 ϩ . In contrast, fep1⌬ cells in which the mutant alleles (fep1-GFP C70A/C76A, fep1-GFP C85A/C88A, and fep1-GFP C70A/C76A/C85A-C88A) were re-integrated failed to regain the ability to repress fio1 ϩ when exogenous iron was present in the medium (Fig. 4C). In fact, a constitutive high level of fio1 ϩ gene expression was observed in these cells. Furthermore, fep1⌬ cells harboring the Cys mutant proteins exhibited increased activity of the cell surface reductase Frp1, as a consequence of lack of transcriptional repression of frp1 ϩ . 2 To gain insight how the mutated Cys residues alter regulation of fio1 ϩ by iron, experiments were carried out using purified Cys-mutant MBP-2 Fep1 241 fusion proteins (Fig. 4D). To test the DNA binding activity of the three Cys mutant Fep1 proteins, an in vitro gel shift assay was conducted. As shown in Fig. 4E, purified MBP alone did not shift the target DNA, whereas the wild type MBP-2 Fep1 241 shifted the 5Ј-(A/T)-GATAA-3Ј-containing probe. Quantitative in vitro DNA binding assays of these mutants with the fio1 ϩ promoter GATAs revealed a much lower affinity that correlates with failure to repress fio1 ϩ gene expression. Indeed, we found that Fep1 C70A/C76A, Fep1 C85A/C88A, and Fep1 C70A/C76A/C85A/ C88A bind to the GATA sequences with a K d(app) of 4.0 ϫ 10 Ϫ7 , 6.7 ϫ 10 Ϫ7 , and 5.6 ϫ 10 Ϫ7 M, respectively. This represents a reduction in binding affinity of approximately 1 order of magnitude (6.4-, 10.6-, and 8.8-fold) compared with the wild type protein. Together, these results revealed that, whereas each of the Cys mutant Fep1 proteins tested are weakly capable of DNA binding in vitro, they failed to repress fio1 ϩ gene expression during high iron growth conditions, exhibiting the same level of constitutive expression as found with a complete lossof-function fep1⌬ null mutant.
Effective Repression of fio1 ϩ Requires the Presence of a Putative Dimerization Domain within the C Terminus of Fep1-Analysis of the amino acid sequence of Fep1 using different computer modeling programs such as COILS (51), GOR4 (52), and MULTICOIL (53) revealed a putative C-terminal amphipathic ␣-helix that may be involved in protein-protein in-  teractions. Interestingly, this potential ␣-helix was also observed in four other iron transcriptional repressors in other fungi. 2 Further analysis suggested that the Fep1 ␣-helix (residues 522-536) is formed of two heptad repeats of Leu and Ile residues. Based upon these observations, we hypothesized that this ␣-helix motif serves to specifically interact with the corresponding ␣-helix in a second Fep1 molecule. Two such ␣-helices would stably dock along the non-polar face that includes Leu 522 (position a), Leu 525 (position d), Ile 529 (position a), Ile 532 (position d), and Leu 536 (position a) residues (Fig. 6D) to form a coiled-coil structure, thereby generating an intermolecular Fep1-Fep1 interaction. To test this hypothesis, a series of truncations were created from the C-terminal end of a functional TAP-1 Fep1 564 fusion protein (Fig. 5A). The mutant alleles were returned to S. pombe fep1⌬ cells and assayed for iron-dependent regulation of the fio1 ϩ iron-responsive gene to determine whether the C-terminal amino acids encompassing the ␣-helix motif are important for Fep1 function in vivo. Importantly, the mutant proteins, like the wild type TAP-Fep1 fusion protein, were targeted to the nucleus (data not shown). The first construct (TAP-Fep1-M541) in which the last 23 amino acids of Fep1 were removed showed iron-dependent regulation of the fio1 ϩ gene. The ability of this mutant to regulate fio1 ϩ gene expression was similar to that observed in cells expressing the full-length TAP-fep1 ϩ allele, except for the magnitude of the response, which was more pronounced with ϳ12-fold induction in response to BPS, and ϳ3-fold repression under iron-replete conditions (Fig. 5, B and C). Further deletion to amino acid residue 491 (mutant TAP-Fep1-M491) gave higher basal levels of fio1 ϩ gene expression with failure to fully repress gene expression in response to elevated iron concentrations. On the other hand, increased gene expression was detected under conditions of iron deprivation (Fig. 5, B and C). When the TAP-Fep1-M491 fusion protein was further truncated to amino acid residues 359, 318, and 241, the fio1 ϩ gene was highly expressed and unresponsive to the presence of iron or BPS (Fig. 5, B and  C). Using an iron-activated antibiotic phleomycin plate assay (19,26) for detection of perturbation of the homeostatic control of iron acquisition, we found that fep1⌬ cells expressing the TAP-fep1-M491 allele displayed a higher sensitivity to phleomycin as compared with the wild type TAP-fep1 ϩ allele expressed under the same circumstances (Fig. 5D). The increased basal level expression of fio1 ϩ as a result of this truncation rendered the cells more sensitive to phleomycin, an antibiotic that cleaves nucleic acids in the presence of excess iron when cells are grown aerobically (54,55). Moreover, we consistently observed that fep1⌬ cells expressing TAP-fep1-M359, TAP-fep1-M318, and TAP-fep1-M241 alleles were hypersensitive and unable to grow in the presence of phleomycin, reflecting the sustained expression of the iron transport fio1 ϩ gene detected in these mutants (Fig. 5D). Given the results from the analysis of the truncations generated from the C-terminal end of Fep1, we next tested whether specific residues within the putative ␣-helix are important for Fep1 function. Site-directed mutagenesis was used to convert Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 codons to those encoding Ala (Fig. 6A). These Leu and Ile residues, which have the potential to form a coiled-coil structure, were also changed to Asp in the same fashion (data not shown). To assess the effects of these mutations on Fep1 function, plasmids expressing the mutant proteins were integrated into a fep1⌬ strain. The ability of Fep1 to repress fio1 ϩ gene expression in the presence of iron was tested by RNase protection assay. Fig. 6B shows that mutations of the Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 amino acid residues to Ala, denoted Mut LI 3 A, resulted in a partially functional Fep1 protein that failed to fully repress the fio1 ϩ gene expression under low and high iron conditions. In contrast, a strain expressing the wild type Fep1 protein showed complete repression of fio1 ϩ expression in the presence of both low and elevated iron concentrations (Fig. 6, B and C). Taken together, these data indicate that the five Leu and Ile residues of the two heptad repeats may constitute a dimer-forming region of transcription factor Fep1, which is required for effective iron transport gene regulation as function of changes in iron levels in fission yeast.
Substitutions of the Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 Amino Acid Residues within the C-terminal Region of Fep1 Do Not Block the Interaction with Tup11-Because fio1 ϩ gene repression by iron is controlled by Fep1 through its association with the corepressor Tup11 (21), we sought to examine whether mutations of Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 residues to Ala in Fep1 affected its ability to interact with Tup11 in vivo.
We coexpressed GST alone or GST-Tup11 (21) with the TAP tag fused to the C-terminal residues 319 -564 of Fep1 or its mutant derivative. Each combination was coexpressed in S. pombe and GST pull-down experiments were performed (Fig.  7). Examination of the proteins bound to the beads by immunoblotting with purified rabbit anti-mouse IgG revealed that both wild type TAP-319 Fep1 564 and TAP-319 Fep1 564 Mut LI 3 A were present in the bound fraction (Fig. 7). Neither TAP-319 Fep1 564 nor TAP-319 Fep1 564 Mut LI 3 A were found in the bound fraction of control cells expressing only TAP-319 Fep1 564 or TAP-319 Fep1 564 Mut LI 3 A with GST alone (Fig. 7). Singularly, we note that the TAP-319 Fep1 564 fusion protein displays a slightly faster electrophoretic mobility to that observed for the TAP-319 Fep1 564 Mut LI 3 A, despite an identical predicted molecular mass. Perhaps, the formation of an amphipathic ␣-helix structure affects its mobility. To ascertain the specificity of the pull-down experiments, the total cell lysates and bound fractions were analyzed by immunoblotting using an antibody directed against PCNA, a soluble protein like the GST-Tup11, TAP-319 Fep1 564 , and TAP-319 Fep1 564 Mut LI 3 A fusion proteins. As shown in Fig. 7, PCNA is present in the total cell extracts but not in the immunoprecipitates. This suggests that the GST-Tup11 and TAP-319 Fep1 564 or TAP-319 Fep1 564 Mut LI 3 A fusion proteins specifically interact with each other to form a stable heteroprotein complex that can be pulled down from whole cell extracts. To assess GST or GST-Tup11 steady state levels, immunoblot analysis of the protein preparations and bound fractions was conducted using anti-GST antibody (Fig. 7). Taken together, the data show that the higher level of fio1 ϩ expression observed in a fep1⌬ strain expressing the fep1-GFP Mut LI 3A mutant allele is not because of a lack of physical interaction between the Tup11 corepressor and the Fep1 transcription factor.
Fep1 Self-association in S. pombe-Our data suggest that a putative coiled-coil motif within the C-terminal region of Fep1 is required to fully repress iron-responsive gene expression. Based on this observation, we examined if an intermolecular Fep1-Fep1 interaction was detectable in vivo. The C-terminal region of Fep1 from residues 360 to 541 was fused to the His 6 tag. Likewise, the TAP tag was added in-frame to a second Fep1 molecule from residues 360 to 541. We coexpressed His 6 -360 Fep1 541 with TAP-360 Fep1 541 in a fep1⌬ mutant strain. A metal chelation affinity matrix was used to selectively isolate His 6 -360 Fep1 541 and associated proteins. As shown in Fig. 8, TAP-360 Fep1 541 co-purified with His 6 -360 Fep1 541 . Selective enrichment of Fep1-associated proteins was demonstrated by the lack of binding of the soluble PCNA protein. Furthermore, the association between TAP-360 Fep1 541 and His 6 -360 Fep1 541 in vivo was extinguished when the Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 amino acid residues were mutated to Ala (data not shown). The specific interaction of TAP-360 Fep1 541 with His 6 -360 Fep1 541 in native cell extracts strongly suggested that Fep1 dimerizes in vivo.
The C-terminal Region of Fep1 Dimerizes in Bakers' Yeast Cells in a Two-hybrid Assay-Given the fact that a Fep1-Fep1 intermolecular interaction was detected and found to be critical for maximal repression of the fio1 ϩ gene expression, we further investigated the hydrophobic amino acid residues that are required for the apparent Fep1 dimerization. We carried out two-hybrid analyses using the chimeric protein that contains the N-terminal 211 amino acids of LexA DNA-binding domain and the C-terminal 205 amino acids of Fep1 (residues 360 -564) as a bait. As second hybrids, different coding regions of the fep1 ϩ gene fused to the coding region of the VP16 acidic activation domain were used (Fig. 9A). When coexpressed with LexA-360 Fep1 564 , the VP16-360 Fep1 564 and VP16-360 Fep1 541 fu-sion proteins exhibited significant levels of ␤-galactosidase activity. When the C-terminal region of 360 Fep1 541 was truncated to Pro 491 , the ␤-galactosidase activity was completely abolished (Fig. 9B). We then tested if the Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 amino acid residues located in the predicted coiled-coil motif were involved in the intermolecular Fep1-Fep1 interaction. When the Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 residues were changed to Ala, the mutant protein failed to interact with LexA-360 Fep1 564 , as no ␤-galactosidase activity was detected (Fig. 9, A and B). As shown in Fig. 9C, the loss of protein-protein interactions was not because of lack of protein expression because all the fusion proteins tested for two-hybrid interactions were produced as confirmed by immunoblot analyses (Fig. 9C). Taken together, we conclude that the Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 amino acid residues that compose the predicted coiled-coil domain in the C-terminal segment of Fep1 are required for a Fep1-Fep1 interaction by two-hybrid assay.
Dimerization of the Coiled-coil Region of Fep1-To determine the stoichiometry of Fep1 molecules present in the predicted coiled-coil structure, cross-linking experiments were carried out on a bacterially expressed form of the Fep1 protein. The C-terminal region of Fep1 from residues 360 to 564 was fused to the His 6 tag. The fusion protein was purified to homogeneity by affinity chromatography using adsorption on a Ni 2ϩ -agarose column and incubated with increasing concentrations of EGS. This cross-linker reacts predominantly with the ⑀-amine group of Lys residues, which are predicted to be accessible for such reaction in the C-terminal region of Fep1 if the molecules associate. As the EGS concentration is increased, His 6 -360 Fep1 564 forms a ϳ44-kDa complex consistent with the size expected for a homodimer (Fig. 10). As the EGS concentration was increased to 1.0 mM or above, no formation of additional higher molecular weight complexes at this or higher EGS concentrations was detected (Fig. 10, and data not shown). To determine precise residues responsible for Fep1 dimerization, we mutated the five hydrophobic residues Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 , which form the two heptad repeats within the C-terminal region of Fep1. Once expressed in and purified from E. coli cells, the His 6 -360 Fep1 564 Mut LI 3 A mutant protein failed to dimerize (Fig. 10). Taken together, these results strongly suggest that the Fep1 protein could form a homodimer through two amphipathic ␣-helices that would adhere to one another via a stripe of hydrophobic residues, Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 , forming a coiled-coil structure.

DISCUSSION
Because iron is both an essential cofactor and a toxic metal, it is critical that organisms maintain homeostatic mechanisms to acquire sufficient iron, yet prevent its accumulation to detrimental levels (1, 2). In S. pombe, several genes encoding FIG. 7. Mutations of the Fep1 Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 amino acid residues do not disrupt the in vivo association between Tup11 and the C terminus of the Fep1 protein. Total cell extracts were prepared from fep1⌬ cells containing GST alone or GST-tagged-1 Tup11 614 and the indicated TAP-tagged-319 Fep1 564 fusion proteins. Lysates (Total) were incubated with a glutathione affinity resin. Following washing, bound fractions were analyzed by immunoblotting using the anti-mouse IgG antibody. A portion of the total protein lysates (ϳ2%) was also analyzed to verify the presence of the immunoblotted proteins prior to chromatography. TAP, TAP-319 Fep1 564 , and TAP-319 Fep1 564 Mut LI 3 A are indicated with arrows. As specific controls, aliquots of cell lysates and bound fractions were probed with anti-GST antibody and anti-PCNA antibody. WB, Western blot. components of the reductive and non-reductive iron transport systems are repressed by the iron-dependent Fep1 transcription factor in response to high extracellular iron concentrations (18,19). Fep1 is a member of the GATA factor family that binds to DNA sequences containing 5Ј-(A/T)GATAA-3Ј (18,19). Interestingly, regulators with similar sequences and functions to Fep1 have been found in other fungi, e.g. U. maydis, N. crassa, A. nidulans, and C. albicans, but not in S. cerevisiae (30). Interestingly, these proteins are distinct from other known yeast GATA transcription factors by the presence of two consensus Cys 2 /Cys 2 -type zinc fingers and a highly conserved amino acid sequence with four Cys located between the two zinc fingers (30,56). Additionally, the sequence homology in these proteins also extends into their C-terminal region in which a predicted coiled-coil domain is present that may promote dimerization (25,26). Because the iron-responsive GATA repressors are structurally distinct from other known classes of GATA DNA-binding proteins, determination of the domains by which these iron-dependent regulators function is important in formulating a comprehensive understanding of iron-responsive cell signaling and iron homeostasis.
In this report we show that the stability of Fep1 is not affected by iron status. We also demonstrate that the loss of in vivo repression activity of Fep1 is not because of iron starvation-mediated export of Fep1 from the nucleus, as Fep1 is localized within the nucleus in both iron-deficient and ironreplete cells. Interestingly, when Fep1-GFP-(5-261⌬) was expressed, most of the mutant protein is localized in the cytosol, and is largely excluded from the nucleus (Fig. 2A). 2 This suggests that the Fep1-GFP-(5-261⌬) truncated form lost a nuclear localization sequence that prevented Fep1 from localizing to the nucleus. Consistently, we have identified a 5-amino acid stretch KRRKR (residues 217-221) that may be responsible to confer Fep1 nuclear localization. However, whether this sequence is sufficient for nuclear localization must await a fine dissection of this region to determine the contribution of the above mentioned basic residues to Fep1 nuclear localization.
Fep1 has two Cys 2 /Cys 2 -type zinc fingers that could potentially interact with 5Ј-(A/T)GATAA-3Ј motifs in DNA. As a first step toward understanding the role(s) of these fingers, we have mutated or deleted individually, and in combination, the se-quences encoding the GATA-type zinc fingers. Although the mutant proteins were properly localized in the nucleus (except for Fep1-GFP-(5-261⌬)), mutating or deleting either one or both zinc fingers gave rise to higher basal levels of fio1 ϩ and lack of repression by iron. In parallel, the wild type and corresponding mutant forms of the N-terminal region of Fep1 were systematically expressed as fusion proteins in E. coli and tested for their ability to interact with the 5Ј-(A/T)GATAA-3Ј sequence in vitro. The binding experiments revealed that Fep1 can bind 5Ј-(A/T)GATAA-3Ј motifs using only the C-terminal finger (ZF2) and not the N-terminal finger (ZF1). The requirement of only the ZF2 for DNA binding is reminiscent to the situation in U. maydis where only the second finger is sufficient for specific DNA binding (23). However, as opposed to Urbs1, Fep1 requires both zinc fingers for in vivo function. Given the fact that the second zinc finger is necessary for the recognition of 5Ј-(A/T)GATAA-3Ј motifs yet insufficient for iron repression of target gene expression, what role could ZF1 play in Fep1 function? One possibility is that ZF1 may form an essential part of a protein-protein interaction domain to facilitate intermolecular repression. Whereas the ZF1 of Fep1 cannot bind the 5Ј-(A/T)GATAA-3Ј sequence in isolation, it may play a critical role in increasing the affinity of the Fep1-DNA complexes. In support with this latter possibility, we determined that the MBP-Fep1-(5-59⌬) and MBP-Fep1-(5-129⌬) mutant proteins, in which ZF1 was removed, bind to the GATA motifs, but at a K d(app) of 3.3 ϫ 10 Ϫ7 and 4.0 ϫ 10 Ϫ7 M, respectively (Fig. 3B). The results show that the mutant proteins appear to have a much lower binding affinity (5.2-and 6.3-fold less) than the wild type protein. This might explain why without ZF1, Fep1-(5-59⌬) and Fep1-(5-129⌬) were found to be insufficient to mount a normal transcriptional repression of the wild type fio1 ϩ promoter in vivo.
Between the two zinc fingers of Fep1, Urbs1, SRE, SREP, SREA, and Sfu1, a conserved N-terminal 27-residue segment with four highly conserved Cys residues is present (25,26,30). We examined the importance of these Cys residues by mutating either the first two (C70A and C76A), the last two (C85A and C88A), or all four. Cells expressing these mutants exhibited sustained expression of the fio1 ϩ gene, even in the presence of 100 M iron (Fig. 4C), and a concomitant increased activity of FIG. 8. Self-association of the C terminus of Fep1 in native S. pombe extracts. Cell extracts were prepared from fep1⌬ mutant cells coexpressing the TAP-360 fep1 ϩ541 and His6-360 fep1 ϩ541 alleles. The cell extracts were incubated with a Ni 2ϩ affinity resin and washed, and the bound fraction was eluted with 150 mM imidazole (nickel-nitrilotriacetic acid pull-down). A portion (ϳ2%) of the total cell extract was also included to monitor the presence of the proteins prior to chromatography (Total). All samples were subjected to immunoblotting with the indicated antibodies. WB, Western blot. the cell surface reductase Frp1. 2 To gain additional insight into the reason why the Cys mutants have lost the ability to turn off fio1 ϩ transcription, we determined for each of the three Cys mutants the apparent dissociation constant of the protein-DNA complex in an electrophoretic mobility shift analysis (Fig. 4E). We found that Fep1 C70A/C76A, Fep1 C85A/C88A, and Fep1 FIG. 9. C terminus segments of Fep1 containing the heptad repeats specifically interact with each other. A, schematic representation of the LexA DNA-binding domain (DBD) fused downstream of and in-frame to the Fep1 coding region that corresponds to codons 360 through 564. The bait molecule was coexpressed with the VP16 activation domain (AD) or different VP16-Fep1 fusion derivatives. The amino acid sequences of the Tup11 and Fep1 proteins are numbered relative to their first initiator codons, respectively. The VP16-360 Fep1 564 fusion protein marked with five "X" means that the five Leu and Ile amino acids found in the heptad repeats were mutated to alanines. LexA-360 Fep1 564 and VP16-Tup11 fusion proteins served as controls for the assay (21). B, positive interactions between the proteins were detected by liquid ␤-galactosidase assays. The values are the averages of triplicate determinations Ϯ S.D. C, whole cell extracts were prepared from aliquots of cultures used in B and analyzed by immunoblotting using either anti-LexA or anti-VP16 antibody. As a control, total extract preparations were probed with anti-PGK antibody. Fep1 564 or His6-360 Fep1 564 Mut LI 3 A was incubated with 0, 0.1, 0.5, and 1.0 mM EGS for 30 min at room temperature. The EGS-cross-linked complexes were analyzed by SDS-polyacrylamide gel electrophoresis and subjected to Western blotting using anti-His monoclonal antibody. Monomeric (ϳ22-kDa, 1 oval) and dimeric (ϳ44-kDa, 2 ovals) forms were detected. M, reference marker. C70A/C76A/C85A/C88A exhibit much lower binding affinities for GATA sequences in vitro, with a K d(app) of 4.0 ϫ 10 Ϫ7 , 6.7 ϫ 10 Ϫ7 , and 5.6 ϫ 10 Ϫ7 M, respectively. This represents an approximately 1 order of magnitude (6.4-, 10.6-, and 8.8-fold) reduction in binding affinity compared with the K d(app) observed for the wild type protein. Clearly, the K d(app) of the Cys mutant proteins reflects a much lower affinity that correlates with failure to repress fio1 ϩ gene expression. As opposed to our data, it should be noted that mutation of conserved Cys residues between the two zinc fingers of SRE failed to derepress target gene expression (25). We suggest the following explanation. In N. crassa, inactivation by deletion of the SRE locus does not completely abolish iron-repression and iron starvation-mediated activation of gene expression, suggesting that additional factor(s) may be involved in regulation of iron homeostasis (24,30). In contrast, in S. pombe, Fep1 appears to be the sole regulatory factor for iron-mediated repression of iron transport genes. Perhaps, in N. crassa, compensatory mechanisms may mask any potential derepression because of the Cys mutant SRE proteins.
In this study, we have identified and demonstrated a role for the C-terminal amphipathic ␣-helix domain of the S. pombe Fep1 protein. This domain consists of two heptad repeats. The model predicts that these repeats of two Fep1 molecules clasp together along a non-polar face, forming a Fep1-Fep1 intermolecular interaction. Deletion mapping analyses that impaired the ability of Fep1 to self-associate strongly reduced its ability to repress transcription, suggesting that proper assembly of a number of Fep1 molecules at a promoter is required for full repression. Analysis of the coiled-coil structure using computer algorithms predicted that the Fep1 Leu 522 , Leu 525 , Ile 529 , Ile 532 , and Leu 536 residues in the repeats were critical to stabilize helix dimerization between the two Fep1 molecules through hydrophobic and van der Waals interactions. 2 We tested the roles of these residues in Fep1-Fep1 interaction by replacing Leu 525 , Ile 529 , Ile 532 , and Leu 536 with alanine (or aspartic acid) and found that these mutations abolished Fep1 self-association. Whereas we have not ascertained the contributions of each of the 14 residues that comprise the putative coiled-coil domain, we observed that residues in positions e in one helix, and g in the other are charged (e.g. Asp and His), thus possibly forming interhelical electrostatic interactions, therefore stabilizing and promoting the assemblage. The dimerization region (residues 522-536) at the C terminus of Fep1 overlaps the C-terminal amino acids from residues 405 to 541 that are required for physical interaction of Fep1 with the corepressor Tup11 (21). Using pull-down experiments on S. pombe extracts, we showed that mutations of Leu 525 , Ile 529 , Ile 532 , and Leu 536 result in a Fep1 protein that can still interact with Tup11. This suggests that the Fep1 coiled-coil domain is not required for association between Fep1-Tup11. The exact biological role of Fep1 self-association is not certain, but there are a number of potential advantages. The most simple effect of self-association is to increase the local concentration of Fep1 and presumably therefore to increase its potency as a transcriptional repressor. It is notable that many iron-responsive promoters contain multiple GATA sites, and it may be that Fep1 self-association is involved in the ordered assembly of higher order complexes at these promoters. Furthermore, Fep1 dimerization may play a specific role in regulating this subset of genes. It is interesting to note that our data concerning Fep1 self-association is reminiscent to that observed for the S. cerevisiae Dal80 protein (57), which is a member of the GATA superfamily of transcription factors (56). Dal80, which has a single zinc finger, is a negative regulator of genes involved in nitrogen metabolism (58). Using a two-hybrid analysis, it has been shown that Dal80 could form a homodimer via a leucine zipper domain near its C terminus (57). Furthermore, it has been shown that the Dal80 leucine zipper was able to interact with the leucine zipper containing domain of Deh1, which is another GATA factor involved in utilization of the nitrogen source in S. cerevisiae (57). Considering this information, one can envision the possibility that Fep1 may be involved in interactions with one or more other proteins that harbor leucine zipper motifs.