Identification and Characterization of the Iron Regulatory Element in the Ferritin Gene of a Plant (Soybean)*

Iron increases ferritin synthesis, targeting plant DNA and animal mRNA. The ferritin promoter in plants has not been identified, in contrast to the ferritin promoter and mRNA iron-responsive element (IRE) in animals. The soybean leaf, a natural tissue for ferritin expression, and DNA, with promoter deletions and luciferase or glucuronidase reporters, delivered with particle bombardment, were used to show that an 86-base pair fragment (iron regulatory element (FRE)) controlled iron-mediated derepression of the ferritin gene. Mutagenesis with linkers of random sequence detected two subdomains separated by 21 base pairs. FRE has no detectable homology to the animal IRE or to known promoters in DNA and bound atrans-acting factor in leaf cell extracts. FRE/factor binding was abrogated by increased tissue iron, in analogy to mRNA (IRE)/iron regulatory protein in animals. Maximum ferritin derepression was obtained with 50 μm iron citrate (1:10) or 500 μm iron citrate (1:1) but Fe-EDTA was ineffective, although the leaf iron concentration was increased; manganese, zinc, and copper had no effect. The basis for different responses in ferritin expression to different iron complexes, as well as the significance of using DNA but not mRNA as an iron regulatory target in plants, remain unknown.

Ferritin, the highly conserved protein that concentrates and mineralizes iron, occurs in animals, plants and microorganisms (1). Although iron is the most abundant element on earth by weight (2) and is required for life, the low solubility of ferric ion (10 Ϫ18 M) (2) and the reaction of ferrous ion with molecular oxygen to generate free radical species (reactive oxygen species) in living cells have led to a number of different strategies to transport iron and to concentrate iron. Ferritin is the only known mechanism to concentrate iron to the level required by cells, to store iron in a soluble and biologically available form, to release iron when needed, and to protect the cells against the toxic effects of excess iron (1,3,4). The higher order structure of ferritin is conserved in plants, animals and microorganisms (1,3,4).
In animals, effects of iron and oxidative stress on ferritin expression have been extensively studied but not in plants.
In plants, the molecular basis for effects of iron and oxidative stress on ferritin expression has been much less studied. Many end effects of ferritin regulation, such as induction by iron and changes during development, are the same in plants and animals, even though mechanisms differ. For example, the ferritin gene in plants contrasts with animals because it enocodes a target peptide to deliver the protein to an organelle, the plastid (27,28). In addition, although iron can change transcription of ferritin genes in both plants and animals, the effect on transcription in animals is amplified by large changes in mRNA translation. In soybean and maize by contrast, increases in transcription and mRNA accumulation of as much as 50 -60fold occur with no detectable, differential effects on translation of mRNA (29 -33). In addition, neither IRE sequences nor IRPs have been detected in plants (28,33,34). 2 Despite the dominance of DNA regulation of ferritin in plants, compared with the ferritin mRNA regulation in animals, the DNA target(s) in plants that are responsible for changes in ferritin expression caused by development, nodule maturation (32,35,36), abscisic acid (38,39), excess iron (31)(32)(33), or reactive oxygen species (32,33) has (have) not been identified.
We describe in this study the identification and characterization of the upstream sequence in the soybean ferritin gene, which responds to iron (FRE), using soybean leaves to analyze soybean gene expression. There is no sequence similarity detectable between FRE and any known promoter or with the IRE, but in analogy to the IRE/IRP interaction in animal mRNA, we observe derepression of the soybean ferritin gene by iron through an iron-sensitive trans-factor. In addition, we show that the form of iron in the medium influences FRE activity in the leaves, with the highest derepression observed with iron fully chelated by citrate (iron:citrate ϭ 1:10), com-pared with iron partially chelated by citrate (iron;citrate ϭ 1;1); Fe-EDTA did not derepress ferritin, even though the leaf iron content increased.

EXPERIMENTAL PROCEDURES
Plant Growth and Medium-Soybean seed (Glycine max var. Williams) was kindly provided by Dr. Joe Burton (Department of Crop Science, North Carolina State University, Raleigh, NC). Seeds were planted in oyster shells and Perlite; after planting, the seedlings were watered twice a day with 250 ml of water for 4 days, followed by watering with 200 ml of water once a day and 200 ml of nutrient solution once a day (37); Ca(NO 3 ) 2 (5.0 mM) was added as the nitrogen source. The non-nodulated soybean plants were grown with ambient light and harvested after 2-3 weeks of growth. For bombardment experiments, the de-rooted soybean plants were pre-incubated in water or 500 M ferric citrate (1:1) for 4 -5 h with exposure to ambient light.
The effects of different physiological signals on ferritin gene expression were tested with soybean plants from which the roots were removed by cutting while holding the roots and stem under water. Incubation was at room temperature in the light. Preliminary experiments showed that the process of removing the roots did not change ferritin expression. Negative and positive controls were water (Ϫiron) and 500 M ferric citrate (1:1), which had been shown previously to give a large induction of ferritin mRNA in cultured tobacco cells (29,32) and in excised soybean leaves (33). Plant hormones, heavy metals, and antioxidants were tested at the following concentrations: 200 M abscisic acid, 100 M giberellic acid, 500 M ZnCl 2 , 500 M MgCl 2 , 50 -500 M ferric citrate (1:1), 50 -500 M ferric citrate (1:10), 500 M ferric nitrate. Iron:citrate (1:10) was made by adding a 10-fold molar excess of citric acid to a solution of iron nitrate dissolved in 1 mM HCl, and the solution was neutralized with NaOH before adding to the plants. Iron:citrate (1:1) is a commercial salt, which, when dissolved in water, becomes a mixture of "free" iron and partly chelated iron. Such "free iron" is "reactive," in contrast to Fe(III) in iron:citrate (1:10), which is fully chelated and unavailable for radical (reactive oxygen species) chemistry. Salt stress was achieved by treating the de-rooted soybean plants with 300 mM NaCl. Drought stress was created by placing soybean leaves on dry filter paper, in a sealed Petri dish, for up to 6 h. Soybean leaves were removed from the de-rooted plants after 6 h of incubation under the test conditions and the tissue frozen immediately in liquid nitrogen.
Construction of Reporter Vectors for Gene Delivery-The source plasmid GmFRE1 (28) encodes the entire soybean ferritin gene plus a 1900-bp promoter region. Reporter vectors were PBI221, in which the glucuronidase (GUS) gene is under the control of the 35 S cauliflower mosaic virus promoter (38) and the TC14 vector (C. Taylor, DuPont, Wilmington, DE), which has the luciferase gene, under the control of the Fed A promoter (39). For all GUS constructs, two primers were designed, which corresponded to the ends of the region to be examined and contained either a BamHI site or a PstI site. PCR amplification of the soybean ferritin promoter fragment and BamHI/PstI sites was followed by restriction enzyme digestion with BamHI and PstI. Digested PCR products were ligated to BamHI/PstI-digested PBI221, in which the 35 S promoter had been removed. Luciferase constructs were obtained by digesting the GUS constructs with HindIII and BamHI and ligating the promoter fragments to TC14 vectors, digested with the same enzymes (Scheme I).
Linker scan mutants LS1, LS2, LS3, and LS4 were made by inserting random sequences, 25 bp in length, into different sites of the 86-bp promoter regions, using PCR. The sites were 22, 43, and 64 in the 86-bp promoter. Those mutants were then ligated to the luciferase reporter vector in TC14.
Delivery of Plasmid DNA into Soybean Leaf Cells by Particle Bombardment-DNA-coated particles were delivered to soybean leaves using procedures described previously (40,41). Briefly, 5 l of plasmid DNA (1 g/l) were added to 200 l of gold (1.6 M) suspension, followed by the addition of 50 l of 2.5 M CaCl 2 and 20 l of 0.1 M spermidine.
After mixing (using a vortex mixer) for 3 min, DNA-coated gold particles were sedimented in a microcentrifuge, washed with 70% ethanol, resuspended in 400 l of 100% ethanol, and the suspension subjected to sonication (Fisher sonicator water bath) for several seconds before spreading 5 l of suspension on a macrocarrier disc. The gold particles on the carrier were then delivered to the soybean leaf fragments using the PDS-1000/helium system; the helium pressure was 1,100 p.s.i., chamber pressure 27 inches of mercury, and the Petri dish was placed on the second shelf from the bottom of the chamber. After bombardment, the Petri dishes were sealed with parafilm and incubated at room temperature for 36 -48 h with light from a fluorescent lamp during the night and ambient light during the day.
Analysis of GUS Activity-GUS activity was analyzed by staining the soybean leaves with 3 ml of a mixture containing 0.1 M Na 2 HPO 4 / NaH 2 PO 4 , pH 7.0, 10 mM EDTA, 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide, 0.1% Triton X-100, and 1 mg/ml X-GLUC for 24 h at 37°C (38). The leaves were then rinsed with 70% ethanol for several hours to remove the chlorophyll. Cells expressing the GUS gene turned blue and could be analyzed with a dissecting microscope. The total number of the blue spots in all leaf fragments from one plate was used to represent the activity of the GUS gene. The data are presented as the average of three to six experiments, and statistics were obtained from a general linear model, using a square root transform in the SAS-GLM program.
Analysis of Luciferase Activity-Luciferase was measured (42) following bombardment and incubation, by grinding soybean leaf fragments from each plate in a chilled mortar with 1.5 ml of buffer (0.1 M potassium phosphate buffer, pH 7.8, 1% Triton X-100, 1 mM dithiothreitol, and 2 mM EDTA). After grinding, the suspension was transferred to a 2-ml microcentrifuge tube and subjected to centrifugation for 5 min at 12,000 rpm (SA600 rotor) at 4°C to remove cell debris, and 100 l of the supernatant solution were mixed with reagents from the enhanced luciferase assay kit (Analytical Luminescence Laboratory, Ann Arbor, MI), as directed by the manufacturer. Another 100 l of the supernatant solution was used to measure the protein concentration with the Bradford assay (Bio-Rad). The data are presented as the average of three to six experiments, and the data were analyzed with a general linear model, using a square root transform in the SAS-GLM program.
Identification of trans-Acting Factors from Cells in Soybean Leaves-Fresh, young soybean leaves (20 g) were collected 2-3 weeks after germination and ground in a chilled mortar with 80 ml of whole cell extraction buffer (40 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 0.5 M sucrose, 10 mM ␤-mercaptoethanol, and 0.5 mM phenylmethylsulfonyl fluoride). The suspension was filtered through four layers of cheesecloth and then two layers of Miracloth before transfer of the filtrate to a beaker chilled on ice. 0.1 volume of 5 M cold NaCl solution was added dropwise, the mixture was incubated on ice for 1 h with slow stirring, and the suspension was centrifuged at 40,000 rpm for 1 h using a Ti-70 rotor. The supernatant solution was recovered and transferred to a second chilled beaker, to which cold saturated (NH 4 ) 2 SO 4 was added to a final concentration of 40%, followed by incubation on ice for 1 h. The protein pellet was collected by sedimentation at 40,000 rpm for 30 min and then resuspended in 1 ml of a buffer containing 10 mM Tris-HCl, pH 7.5, 5 mM KOAc, 10% glycerol, and 0.5 mM phenylmethylsulfonyl fluoride, followed by dialysis against 500 ml of the same buffer for 5 h. The insoluble pellet was removed by centrifugation for 30 min at 15,000 rpm. Aliquots (100 l) of the supernatant solution were frozen in liquid nitrogen and stored at Ϫ80°C until use (1-2 months).
A soybean ferritin promoter probe of 400 bp was prepared from SoF-P304-GUS by the fill-in method with 32 P after restriction enzyme digestion with BamHI and PstI overnight at 37°C (8 l of DNA, 2 l of 10ϫ buffer, 1 l of BamHI, 1 l of PstI, 1 l of bovine serum albumin, 7 l of H 2 O) (43). Wild type and "linker scan mutant" iron regulatory element probes (86 bp) were labeled the same way. Labeled probes were purified by gel electrophoresis (1% agarose) and Qiagen gel extraction kit. Various amounts of cell extract were added to 20 l of binding mix that contained 20 mM Hepes-KOH, pH 7.6, 50 mM KCl, 5 mM MgCl 2 , 5 mM dithiothreitol, and 1.5 g of poly(dI/dC) binding mix. Incubation of the mixture at room temperature for 5 min was followed by the addition of 1 l of labeled DNA (around 10 ng). Further incubation for 20 min was followed by quickly loading the mixture onto a 4% polyacrylamide gel. Free and bound fractions of DNA were separated by electrophoresis in 0.5ϫ TBE buffer at 140 V for 2 h. The experiments were repeated with two to four independently prepared sets of cell extracts.
Northern Hybridization-Total RNA was purified from frozen leaf samples with a Qiagen kit (RNAeasy total RNA). 10 g of the RNA from soybean leaves were used in hybridization reactions. Denatured RNA (heated at 65°C for 10 min in 60% formamide, 5% formaldehyde, pH ϭ SCHEME 1 7) was fractionated by electrophoresis in formamide, 1% agarose gels and transferred to Duralose-UV membranes. Prehybridization (5ϫ SSC, 5ϫ Denhardt's, 0.5% SDS, 100 g/ml denatured sheared sperm DNA) was at 62°C for 2-3 h (36). DNA probes were prepared by digesting SoF-IC1 (28) with EcoRI to release the soybean ferritin insert, followed by electrophoresis (1% agarose), purification of the insert (Qiagen gel extraction kit), and labeling with 32 P (using the oligolabeling kit from Amersham Pharmacia Biotech); probes were purified by gel filtration (Sephadex G-50). Hybridization, 18 h at 62°C, was followed by washing in 1ϫ SSC, 0.1% SDS at room temperature and then at 62°C. For autoradiography, we used a PhosphorImager (Molecular Dynamics), and the images were analyzed with Imagequant (Molecular Dynamics). The data were averaged from three to six experiments. For image treatment, we used Photoshop 3.0 (Adobe).
Soluble Iron Content Soybean Leaf Cell Extracts-Fresh leaves (0.2 g) were ground on ice in a chilled mortar with 2 ml of cold extraction buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM MgCl 2 , 10 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 1% ␤-mercaptoethanol); cell debris was removed by centrifugation at 12,000 rpm (SA600 rotor) for 30 min at 4°C. The supernatant solution (500 l) was used to measure the iron concentration, after boiling for 30 min in 0.1 M HCl, by forming the Fe(II)-1,10-phenanthroline complex and measuring the absorbance at 510 nm, after neutralization (36). Protein concentrations were determined using the Bradford method (Bio-Rad). The data are presented as the average of two to four experiments.

Identification of the Iron-responsive Element in the Soybean
Ferritin Gene-Maximum induction of ferritin expression (50 -60-fold increases in ferritin mRNA accumulation) occurred in soybean leaves within 6 h of adding 500 M ferric citrate to the plants. The high ferritin mRNA concentration persisted for at least 24 h. 3 No RNA was isolated from leaves after 24 h because the tissues displayed symptoms of iron toxicity (brown spots, drying). Ferric citrate (1:1) was used as the iron source in the initial studies, since historically it has been the most common agent for inducing ferritin synthesis in plants. Iron overload had been shown to increase ferritin mRNA synthesis 3-5-fold in intact soybean and maize plants (44). De-rooted soybean plants used here have been used by many other investigators to eliminate the root barrier and to accelerate iron uptake (45,46). The wounding of the plantlets due to the cutting of the roots did not itself induce ferritin mRNA accumulation (data not shown).
The location of the iron regulatory element in the soybean ferritin gene used a set of promoter fragments linked to a luciferase reporter in plasmid TC14. DNA fragments, representing different sequences upstream from the transcription start site of the soybean ferritin gene determined by primer extension analysis and S1 nuclease mapping, were delivered to soybean leaf sections by particle bombardment, followed by incubation of the leaf sections with or without iron. The results are shown in Fig. 1A. The 35 S Luc construct, which showed very high luciferase activity (data not shown), also demonstrated that the particle bombardment was effective in delivering the foreign gene into soybean leaf cells. Luciferase activity was observed in SoF-P84, SoF-P156, and SoF-P218 transformed soybean leaves with and without iron, while luciferase activity encoded in SoF-P304, SoF-P600, and SoF-P1900 was easily detectable only after incubation with iron. The effect of iron on expression of luciferase SoF-P304, SoF-P600, and SoF-P1900 was highly significant (confidence level ϭ 99%) when the data were analyzed with a general linear model, using a square root transform in the SAS-GLM program. The sequence of the region studied is shown in Fig. 1D (28).
Attempts to use co-transfection of the tissue with an 35 S GUS construct (PBI221) to quantitate DNA delivery by bombardment failed because the background luminescence in the leaf tissue obscured the signal. Instead, a series of constructs was made with the set of promoter fragments used in the 3 J. Wei and E. C. Theil, unpublished observations.

FIG. 1. Identification of sequences in the soybean ferritin gene responsible for the iron response.
The effect of iron on the transient expression of different soybean ferritin promoter reporter gene constructs was explored using particle bombardment of soybean leaf fragments; bombardment was followed by incubation with and without iron (500 M ferric citrate (1:1)), as described under "Experimental Procedures." A, luciferase reporter; B, GUS reporter; C, linker scan of the soybean ferritin FRE (random DNA sequences were substituted for promoter subdomains as follows: LS1 ϭ Ϫ218 to Ϫ239, LS2 ϭ Ϫ240 -261; LS3 ϭ Ϫ262-283; LS4 ϭ Ϫ284 -304), using the luciferase reporter constructs); D, ferritin promoter sequence (numbers correspond to those used in A-C). The FRE is underlined. FRE subdomains LS1 and LS3 (required for the iron response illustrated in C), are labeled, and two hexanucleotide sequences, which flank the 5Ј ends of LS1 and LS3 and which can potentially base pair, are shown in italics and large font. CAAT sequences occur at Ϫ115 and Ϫ91, and a TATA sequence occurs at Ϫ25. The data are presented as the average of three to six experiments, and the data were analyzed with a general linear model, using a square root transform in the SAS-GLM program. The sequence is from GenBank™ (accession no. U31648; see Ref. 28). luciferase assay (Fig. 1B) and the GUS reporter in the PBI221. Expression of GUS was measured histologically. The results for cells in the histological analysis paralleled those obtained using the luciferase reporter on tissue extracts. For example, constitutive expression of the GUS was observed in SoF-P84, SoF-P156, and SoF-P218 with or without iron. Further, GUS expression was greatly decreased in SoF-P304, SoF-P600, and SoF-P1900 in the absence of iron, suggesting that a negative control element exists between 218 and 304 bp upstream from transcription start site. Finally, increased GUS activity with SoF-P304, SoF-P600, and SoF-P1900 in iron-treated leaf fragments suggests that the iron signal derepresses, rather than induces, ferritin gene activity.
The ratio of GUS or luciferase activities in iron-treated soybean leaf sections compared with control leaf sections was approximately 1.0 for the 35 S and 84-, 156-, and 218-bp promoter sequences. In contrast, for SoF-P304, the ratio of irontreatment/control was 13.2 for GUS activity and 6.2 for luciferase activity, emphasizing the importance of the region from bp 218 to 304 in the iron response.
The response of the soybean ferritin gene to environmental stimuli appears to be very specific. For example, among the metals tested, at 500 M (CuCl 2 , ZnCl 2 , MnCl 2 , and iron citrate) only iron had any effect on soybean leaf ferritin expression (47). Salt stress (300 mM NaCl) and drought also had no effect on ferritin expression. An oxidative pathway of ferritin induction has been suggested for maize and Arabidopsis thaliana (30,45,46). Hydrogen peroxide had no effect on soybean leaf ferritin expression. However, when the soybean plants were exposed to a mixture of ascorbate and non-inducing levels of partly ferric chelated citrate (iron:citrate ϭ 1:1) (48 -50), ferritin induction occurred, but whether the effect was due to reduction of the Fe(III) or to the complex chemistry associated with the mixture in the presence of respiring plant tissue (51-57) cannot be determined from such observations alone. Among the iron complexes tested, Fe(III) fully complexed to citrate (iron:citrate ϭ 1:10) gave maximum derepression at one-tenth the concentration required for ferric citrate (iron: citrate ϭ 1:1). Fe(III)-EDTA was ineffective, even though the leaf iron content was equal to that obtained with ferric citrate (data not shown).
Two Subdomains Are Detected in the Soybean Ferritin FRE Using Promoter Linker Scan Mutants-To determine if the entire 86 bp region of the soybean ferritin promoter (FRE) is required for the iron response, the region between Ϫ218 bp and Ϫ304 bp in the soybean gene (Fig. 1B) was subdivided into four regions. Four constructs were made in which a different one of the four subdivisions in the iron regulatory element was replaced by a linker of random sequence in SoF-P304 with the luciferase reporter ("linker scan" mutations). The effect of iron on luciferase expression was measured after introduction of the plasmid DNA into soybean leaf sections by particle bombardment (Fig. 1C). Two of the four plasmids, LS2 and LS4, showed responses to iron that were not significantly changed by the mutations, compared with the 86-bp sequence in SoF-P304. In contrast, mutation of LS1 and LS3 sequences abrogated iron derepression, indicating that the LS1 and LS3 each contained sequences for the iron response.
The results indicate that two DNA sequences of 21 base pairs within the 86-bp iron response element of the soybean FRE, and separated from each other by 21 base pairs, are each sufficient to confer derepression of the ferritin gene in the presence of iron (Fig. 1C). Flanking the 5Ј ends of LS1 and LS3 sequence are sequences that have symmetry: CACAGA (Ϫ288 bp to Ϫ282 bp) and AGACAC (Ϫ245 bp to Ϫ239 bp). The significance of the separated iron-responsive sequences in the promoter and the symmetry of bases flanking the two ironresponsive sequence is not clear. Possible explanations are redundancy of trans-acting factor binding sites, sites for multiple trans-acting factors or the formation of DNA loops.
Detection of Iron-sensing trans-Acting Factor(s) in Soybean Leaf Extracts-The data from the soybean ferritin gene promoter (FRE) deletion experiment indicated that the promoter region, Ϫ218 bp to Ϫ304 bp, is required for the iron response of the soybean ferritin gene (Fig. 1). In order to detect (an) ironsensitive trans-acting factor(s), extracts of soybean leaves from plants grown with and without iron (500 M ferric citrate) were incubated with one of two 32 P-labeled DNA probes. One DNA probe was a 400-bp fragment that included 304 bp of the promoter and 96 bp of encoding the 5Ј-untranslated region of soybean ferritin mRNA. The other DNA probe was the 86-bp fragment that corresponded to the frequency between Ϫ304 bp and Ϫ218 bp in the soybean ferritin gene sequence, which was shown to be required for the iron response (Fig. 1). Electrophoresis of DNA with and without cell extracts, displayed in Fig. 2, showed that a complex formed only in extracts from plants incubated without iron. Unlabeled DNA of the same sequence prevented the radioactive DNA from forming the complex but heterologous DNA (plasmid DNA without promoter insert) had no effect.
The DNA complexes formed with leaf extracts of plants grown without iron likely represent a ferritin promoter-repressor complex, suggested by the derepression of the gene by iron in the leaf bombardment experiments (Fig. 1). Experiments with ferritin mRNA from animals also showed the presence of a ferritin mRNA repressor in cell extracts when iron concentrations were low (17). The soybean ferritin DNA repressor was unable to bind the FRE when the cytoplasmic level of iron increased (Fig. 2). Since both the 400-and 86-bp fragments display similar factor binding, the 86-bp sequence is sufficient for both trans-acting factor binding (Fig. 2) and the iron response (Fig. 1). In an experiment to test binding to the ironresponsive sequences within the FRE, with a single set of leaf extracts, only SoF-LS1 and SoF-LS3 DNA formed a complex and then, only with extracts of leaves grown without iron, in analogy to experiments with the full FRE.
Boiling or proteinase K digestion prevented DNA binding but DNase and RNase had no effect. Thus, the iron-sensitive transacting factor that recognizes FRE in the soybean ferritin gene is likely a protein. DISCUSSION Functional similarity is shared by the DNA FRE in the soybean ferritin gene and the RNA IRE, which is found in animal ferritin mRNAs and several other animal mRNAs, with no sequence similarity detectable. In addition, the plant FRE has no detectable sequence similarity to any DNA regulatory sequence currently known. In bacteria, a cis-acting element, Fur, that regulates groups of genes involved in iron/oxygen metabolism also has no detectable structural similarity to either the IRE in animal or the FRE in plants. However, an IRE-like sequence, based on protein binding, has recently been observed in Bacillus subtilis. The bacterial IRE-like sequence occurred in mRNAs encoding proteins of an iron uptake system and of the major cytochrome oxidase (58). In plants, the genetic mechanisms of iron regulation with a DNA target have diverged from the animal mRNA target, even though the environmental signal (iron) and the gene product (the highly conserved ferritin proteins) have been conserved (see also Refs. 9, 28, and 29).
The mechanism of iron-dependent regulation of soybean DNA is gene derepression, in analogy to mRNA derepression in animal mRNA (19,59). Constitutive reporter expression controlled by the FRE was low compared with the 35 S promoter (47). Two domains, separated by 21 bp, were observed in the 86-bp FRE, based on "linker scan mutations" (Fig. 1C). The two FRE subdomains could reflect multiple binding sites for the same repressor or could indicate the occurrence of multiple repressors, just as animal IREs are recognized by two distinct proteins (IRPs). The presence of symmetrical sequences at the end of each of the FRE subdomains (CACAGA (Ϫ288 bp to Ϫ282 bp) in LS1, and AGACAC (Ϫ245 bp to Ϫ239 bp) in LS3) indicates potential for some sort of DNA loops. Whether the novel cis-acting element identified in the soybean ferritin gene is representative of a sequence in the promoters of other plant genes, such as those for iron uptake, to allow coordinated regulation by iron as for mRNA of animals, is not known at this time.
Derepression of the soybean ferritin gene (FRE) by growth in the presence of high concentrations of iron ( Fig. 1) suggested that a trans-acting factor existed that cannot recognize the regulatory element in the presence of iron. Such an hypothesis is supported by the observation that both a 400-bp fragment with the regulatory element and the 86-bp element itself are recognized by a macromolecule in the extract of soybean leaves only when environmental iron is present in low concentrations (Fig. 2). Based on sensitivity to proteinase K or heat and insensitivity to nucleases, the iron-sensing FRE recognition factor appears to be a protein.
Only iron had any detectable effect on FRE regulation, among the possible metal and environmental signals tested (iron, manganese, copper, zinc, abcisic acid, drought, high salt). The iron ligands also influenced the response, affecting either vascular transport or tissue distribution or both, since ferric citrate derepressed ferritin expression, but Fe(III)-EDTA did not. An explanation for the sharing of an iron signals by ferritin regulation in plants and animals but the divergence of the genetic targets (DNA or RNA) can be a shared signal transduction pathway, which a terminal factor specific for DNA recognition in plants and mRNA in animals. The definitive explanation remains unknown.