Inhibition of the Iron-induced ZmFer1 Maize Ferritin Gene Expression by Antioxidants and Serine/Threonine Phosphatase Inhibitors*

Two pathways have been implicated in the regulation of maize ferritin synthesis in response to iron. One of them involves the plant hormone abscisic acid (ABA) and controls the expression of ZmFer2 gene(s). Another pathway, ABA-independent, has been characterized in a de-rooted maize plantlet system and involves an oxidative step. The ZmFer1 maize ferritin gene is not regulated by ABA, and it is shown in this paper that the corresponding mRNA accumulates in de-rooted maize plantlets and BMS (Black Mexican Sweet) maize cell suspension cultures in response to iron via the oxidative pathway described previously. To investigate ZmFer1 gene regulation further, the BMS cell system has been used to develop a transient expression assay using aZmFer1-β-glucuronidase fusion. Both iron induction and antioxidant inhibition of ZmFer1 gene expression were observed in this system. Using Northern blot analysis and transient expression experiments, it was shown that both okadaic acid and calyculin A, two serine/ threonine phosphatase inhibitors, specifically inhibit ZmFer1 gene expression. These data indicate that an okadaic acid-sensitive protein phosphatase activity is involved in the regulation of the ZmFer1 ferritin gene in maize cells, and this activity is required for iron-induced expression of this gene.

Two pathways have been implicated in the regulation of maize ferritin synthesis in response to iron. One of them involves the plant hormone abscisic acid (ABA) and controls the expression of ZmFer2 gene(s). Another pathway, ABA-independent, has been characterized in a de-rooted maize plantlet system and involves an oxidative step. The ZmFer1 maize ferritin gene is not regulated by ABA, and it is shown in this paper that the corresponding mRNA accumulates in de-rooted maize plantlets and BMS (Black Mexican Sweet) maize cell suspension cultures in response to iron via the oxidative pathway described previously. To investigate ZmFer1 gene regulation further, the BMS cell system has been used to develop a transient expression assay using a ZmFer1-␤-glucuronidase fusion. Both iron induction and antioxidant inhibition of ZmFer1 gene expression were observed in this system. Using Northern blot analysis and transient expression experiments, it was shown that both okadaic acid and calyculin A, two serine/ threonine phosphatase inhibitors, specifically inhibit ZmFer1 gene expression. These data indicate that an okadaic acid-sensitive protein phosphatase activity is involved in the regulation of the ZmFer1 ferritin gene in maize cells, and this activity is required for iron-induced expression of this gene.
Iron homeostasis needs to be controlled tightly in living cells because this element is both essential for important metabolic processes and potentially toxic (1). Iron deficiency or overload leads to important physiological disorders in plant and animal cells. Iron toxicity is linked to the high reactivity of this transition metal with oxygen species and the hydroxyl radicals produced, which can damage all cell components (2). Therefore, free iron storage is a key step in the prevention of oxidative stress. Among the mechanisms controlling iron homeostasis, ferritins are a class of ubiquitous multimeric proteins able to store iron in a soluble and nontoxic form (3). Injection of H-type ferritins into the cytoplasm of endothelial cells prevents H 2 O 2induced cytolysis (4), proving that ferritin iron storage capacity can confer oxidative stress resistance.
Ferritin synthesis is induced by iron excess, mainly at the translational level in mammalian cells (5), whereas a tran-scriptional control has been demonstrated in soybean cell suspension culture (6). Ferritin cDNAs and genes have been isolated from various plant species (6 -10). In maize, two mRNA subclasses have been identified, and the corresponding genes, named ZmFer1 and ZmFer2, have been cloned (7,11). Amino acid sequences corresponding to the open reading frames encoded by the ZmFer1 and ZmFer2 transcripts are highly similar, suggesting a common function for both subunits (7). In contrast, promoter sequence analysis of the two genes reveals a high level of divergence (11), leading to the hypothesis of differential regulation of these ferritin genes.
Iron overload of maize plantlets has been found to induce ferritin mRNA and subunit accumulation (7). Two transduction pathways involved in this response have been identified. The first one requires the plant hormone abscisic acid (ABA) 1 as a cellular relay (12) and leads to ZmFer2 mRNA accumulation (11). No ZmFer1 transcript accumulation has been observed in response to exogenous ABA or abiotic stress (drought) known to raise cellular ABA concentrations (11). When de-rooted maize plantlets are incubated in iron citrate solution, a rapid accumulation of ferritin mRNA and subunits has been detected (13). This response has even been observed in an ABA-deficient maize mutant, demonstrating that this activated pathway is ABA-independent. Ferritin mRNA accumulation in this derooted maize plantlet system in response to iron overload was inhibited in the presence of antioxidant agents such as Nacetylcysteine (NAC) (13). This result suggests that an oxidative step occurs in this pathway, and this is supported by the fact that hydrogen peroxide treatment induced ferritin transcript accumulation. Whether ZmFer1 is regulated by iron through this ABA-independent pathway remained to be determined.
Eukaryotic cells may elicit adaptative responses to environmental stresses by recruiting intracellular signal cascades, many of which involve reversible protein phosphorylations, controlled by protein kinases and protein phosphatases. Physiological responses to various cellular stresses involve mitogenactivated protein kinases, in mammalian (14 -17) and yeast cells (18 -20). In plants, cellular responses to some abiotic stresses (cold, drought, wounding, osmotic shock) might also be mediated by plant mitogen-activated protein kinase cascades (21)(22)(23). Protein phosphatases have also been shown to regulate some cellular signaling cascades in response to stress. In yeast, a calcineurin (PP2B) has been found to confer tolerance to salt stress (24), and PPZ has been demonstrated to play a role in the maintenance of cell integrity (25) as well as in the control of salt homeostasis (26). Arabidopsis PP2C (ABI1) is required for stomatal closure, seed dormancy, and growth inhibition, indicating its key role in some of the transduction pathways involving ABA (27)(28)(29).
Recent insights into the role of protein phosphorylation in cellular signal transduction have been established by the use of drugs that are permeable to biological membranes and interact specifically with protein kinases and protein phosphatases. Among such pharmacological agents, okadaic acid (OA) and calyculin A (CA) are two potent inhibitors of PP1 and PP2A (30 -32). Using these toxins, initial evidence for the involvement of PP1/2A activities was established for numerous cellular events in plants (33).
In this paper, using Northern blot experiments and a transient expression assay, we demonstrate that ZmFer1 is the maize ferritin gene induced in de-rooted plantlets and cell suspension cultures, through the previously identified ABAindependent oxidative pathway. This iron-induced expression of ZmFer1 is inhibited in the presence of OA and CA, suggesting that a serine/threonine phosphatase is involved in the regulation of this gene.

EXPERIMENTAL PROCEDURES
Plant Materials, Treatment Conditions, and Reagents-Maize plants (Zea mays var. M017) were grown hydroponically in iron-free medium as described previously (7). After 7 days of iron starvation, plantlets were cut at the level of the collar, and stems were dipped into the various treatment solutions. Samples were harvested at various times after treatment, frozen in liquid nitrogen, and stored at Ϫ70°C.
Maize BMS (Black Mexican Sweet) cells were cultivated in the dark and with constant shaking at 70 rpm, in 100 ml of Murashige and Skoog (MS) liquid culture medium as described (34), containing macronutrients and vitamins purchased from Sigma. For 2-week subcultures, 10 ml of 15-day-old cell suspension was added to 90 ml of fresh medium. For iron-starved cell cultures, the subculture was carried out with 20 ml of the 15-day-old cell suspension added to 80 ml of iron-free fresh medium. Experiments were carried out using iron-starved cell suspensions, 1 week after subculture. For H 2 O 2 treatments, cells were grown in iron-containing medium, and cells were resuspended in fresh complete medium before the addition of H 2 O 2 . The iron citrate mixture for the induction of ferritin synthesis, NAC, reduced gluthathione (GSH), and H 2 O 2 (Sigma) were prepared and diluted as described previously (13). OA (Biomol) and CA (Life Technologies, Inc.) were dissolved in ethanol and dimethyl sulfoxide, respectively. To obtain accurate cell treatments when using OA over a concentration range, 500 M OA solution (10% dimethyl sulfoxide) was purchased from Life Technologies, Inc. After treatments, cells were filtered and washed with a solution of 10 mM EDTA and 20 mM KCl, then immediately frozen in liquid nitrogen and stored at Ϫ70°C.
RNA Extraction and Northern Blot Analysis-Total RNA, from frozen maize plantlets and BMS cells, was extracted using the guanidine method (35) with modifications (7). For Northern blot analysis, 10 g of total RNA for each sample was electrophoresed through a 1.2% agarose and formaldehyde gel (36) and blotted onto a nylon filter (Hybond N, Amersham). Blotted RNA was hybridized with the probes labeled with [␣-32 P]dCTP using a random priming kit (Pharmacia Biotech Inc.). The FM probe, which hybridizes to ferritin ZmFer1 and ZmFer2 transcripts, was a 620-bp SacI-StuI fragment from the FM1 cDNA (7). The FM1 probe, which hybridizes specifically to the ZmFer1 transcripts, was a 53-bp StuI-BclI fragment from the 3Ј-untranslated region of the ZmFer1 cDNA (11). The FM2 probe corresponds to a 56-bp fragment of the ZmFer2 cDNA 3Ј-untranslated region (11). Probe H1 corresponded to a maize H1 histone cDNA (37).
Protein Extraction and Western Blot Analysis-Total proteins were extracted from 0.2 g of frozen maize BMS cells as described previously (13). 20 g of total protein for each sample was electrophoresed through an SDS-polyacrylamide (15%) gel and electroblotted onto nitrocellulose membrane (Hybond C). Ferritin subunit immunodetection was per-formed using polyclonal antibodies raised against maize seed ferritin (7) and anti-rabbit IgG sheep immunoglobulins coupled to peroxidase (Biosys), as described previously (39).
Plasmid Constructs-The plasmid pRD109 containing the ␤-glucuronidase reporter gene (GUS) fused to the maize histone H3C4 promoter and rice actin introns was a gift from R. DeRose (Rhône Poulenc Agrochimie, Lyon, France).
The plasmid pAHC18 containing the luciferase reporter gene fused to the promoter, the 5Ј-untranslated region, and the first intron of the the maize ubiquitin 1 gene was a gift of Prof. P. Quail (UCLA, Berkeley).
The 674-bp fragment SacI-ClaI from ZmFer1 gene, spanning exons 1-4, was subcloned into SacI-AccI-digested M13mp19, and singlestranded DNA was produced according to standard procedures (40). Site-directed mutagenesis was performed to introduce an NcoI site into the sequence corresponding to exon 3 using the Sculptor in vitro mutagenesis system (Amersham) and the deoxyoligonucleotide 5Ј-GCT-GAAAAGCCCATGGAGTAC-3Ј. Mutations were selected by sequencing using the T7 Sequencing kit (Pharmacia). The mutated fragment was excised from double-stranded M13mp19 using SacI-NcoI restriction enzymes and introduced into the corresponding sites of the pRD109 to obtain pSL3 and produce a translational fusion between exon 3 of the ZmFer1 gene and the uidA open reading frame encoding GUS. Then, a 1.9-kbp NsiI-SacI fragment from the ZmFer1 gene was subcloned into PstI-SacI sites of pSL3 to produce the pSL5 construct (see Fig. 6A).
Transient Expression Analysis-Plasmid DNA was prepared using the polyethylene glycol method (41). Gold particles (Bio-Rad) were washed in 100% ethanol and resuspended in 50% glycerol at 130 g/l. For DNA coating of microcarrier, 6 g of construct (pSL5 or pRD109) and 6 g of the internal standard plasmid (pAHC18) were added to 25 l of gold particles. Then, 32 l of 2.5 M CaCl 2 and 12.5 l of 100 mM spermidine were added, mixed, and incubated for 5 min on ice. The DNA-carrier mix was centrifuged briefly, and then 64 l of supernatant was removed. 2 l of DNA-coated gold particles was used for each transformation.
8 days after subculture, 0.5 g of BMS cells cultured in the presence of 100 M Fe-EDTA was filtered, and the cells on the filter paper were transferred to iron-free MS solid medium. After a 3-h incubation at 28°C in the dark, cells were transformed with a helium biolistic device using 2 l of DNA-microcarrier preparation. For bombardments, the particle gun was set to 25 mbar of vacuum, 4 bars of helium pressure, and the distance between the microcarrier and target cells was 21 cm. Transformed cells were transferred to 3 ml of iron-free liquid MS medium in six-well microplates and incubated overnight at 28°C on a rotary shaker. If required, a 6-h incubation was performed before iron treatment in the presence of the appropriate inhibitor. Then when indicated, iron citrate was added. Cells were incubated subsequently under the same conditions for 24 h.
For reporter gene activity detection, cells were washed in 0.1 M phosphate buffer, pH 7.8, and 400 l of extraction buffer (0.1 M phosphate buffer, 0.1% Triton X-100, 1 mM dithiothreitol) was added. After a 10-s sonication, lysates were centrifuged for 10 min at 13,000 ϫ g, 4°C, and supernatants were recovered. 20 l of extract was used to measure the reporter gene activities using the GUS light kit (Tropix) and the luciferase assay system (Promega), with a luminometer (1203 Bio-Orbit, Turku, Finland). Results are expressed as relative light units corresponding to the ratio between the GUS (assay) and the luciferase (internal standard) activity.

An Increased Abundance of ZmFer1 Transcripts via an Oxidative Pathway Is Induced by Iron in De-rooted Maize Plantlets-It
has been previously demonstrated that iron overload of de-rooted maize plantlets leads to an increase of ferritin transcript abundance (13). This response was shown to be mediated by an ABA-independent pathway because ferritin mRNA still accumulates in response to iron in the ABA-deficient maize mutant vp2. This pathway was defined as requiring an oxidative step because antioxidant agents inhibit the iron-induced accumulation of ferritin mRNA in this system. In this study, Northern blot experiments were performed using a probe called FM, which hybridized to all of the different subclasses of maize ferritin mRNA identified so far. However, using specific probes hybridizing specifically the ZmFer1 and ZmFer2 subclasses of maize ferritin transcripts, it has been shown that in irontreated hydroponic cultures of maize plantlets ZmFer1 mRNA accumulates independently of ABA (11). To determine if in de-rooted plantlets the increased abundance of ZmFer1 mRNA is stimulated in response to iron through the previously identified oxidative and ABA-independent pathway, Northern blot experiments were performed using the FM1-specific probe in this system (Fig. 1A). As described previously (13), a significant increase in ferritin mRNA level was detected using the FM probe when de-rooted plantlets were dipped into a 500 M iron citrate solution (Fig. 1A, FM probe). A transient accumulation was observed, reaching a maximum 3 h after the addition of iron and similar kinetics detected using the FM1-specific probe (Fig. 1A, FM1 probe). In contrast, hybridizations using the FM2 probe revealed that no ZmFer2 mRNA accumulated in response to the iron excess during the time course of the experiment (Fig. 1A, FM2 probe). The iron treatment has no effect on the maize H1 histone mRNA (Fig. 1A, H1 probe). Therefore, iron overloading of de-rooted maize plantlets induces specifically the accumulation of ZmFer1 ferritin mRNA, and the level of ZmFer2 mRNA remains undetectable. To determine if this increased abundance of ZmFer1 mRNA is sensitive to antioxidant agents, plantlets were pretreated for 3 h in 10 mM NAC or 1 mM GSH before a 3-h iron treatment (Fig. 1B). Hybridizations using FM or FM1 probes, respectively, revealed that the iron-induced accumulation of ferritin mRNA was inhibited completely in the presence of NAC and partially diminished when plantlets were cotreated with GSH. These data are in agreement with the results obtained previously using the FM probe (13). As mentioned above, no ZmFer2 mRNA accumulation was detected in response to an iron overload (Fig. 1B, FM2 probe). Hybridizations using the H1 probe show that the H1 mRNA level is not affected in the presence of NAC or GSH (Fig. 1B, H1 probe). Furthermore, it has been shown recently in Brassica napus seedlings that ferritin and acorbate peroxidase mRNA abundances were increased in response to iron with the same kinetics and dose dependence (42). Iron-induced accumulation of ascorbate peroxidase mRNA was independent of an ABA pathway and not affected by NAC treatment, whereas ferritin mRNA accumulation induced by an iron excess was blocked by NAC treatment in this system. Together these observations demonstrate that NAC treatment does not lead to a general inhibition of gene expression but rather to a specific inhibition of ZmFer1 mRNA accumulation in our system. Therefore, the ZmFer1 transcripts are the ferritin transcripts that accumulate in response to iron overload of de-rooted plantlets through the activation of the oxidative and ABA-independent pathway.
Iron Overload Induces Ferritin Synthesis and Increased Abundance of ZmFer1 mRNA in Maize BMS Cells-To study further the iron-induced ZmFer1 mRNA accumulation, maize cell suspension cultures were chosen as model. Indeed, it has already been reported that plant cell suspension cultures provide a suitable cell system for the study of ferritin synthesis (6). The cell system used in this study consists of nonphotosynthetic BMS maize cells. Cells were subcultured for 1 week in iron-free medium before iron treatment during the exponential growth phase. ZmFer1 transcripts were not detected in these cells before the iron treatment ( Fig. 2A). When BMS cells were incubated in the presence of 500 M iron citrate, an increased abundance of ZmFer1 transcripts was detected, reaching its maximum 24 h after iron treatment and decreasing at 48 and 72 h (Fig. 2A). Similar kinetics of ferritin mRNA accumulation have been observed in iron-overloaded maize plantlet leaves (7). ZmFer1 transcript accumulation was proportional to increasing iron citrate concentrations from 0 to 500 M, supplied in the culture medium for 3 h (Fig. 2B). To determine if iron overload of maize cell suspension cultures leads to ferritin accumulation, total protein extracts from iron-induced BMS cells were separated by electrophoresis and analyzed by West- . After the various treatments, RNA extracted from de-rooted maize plantlets was analyzed by Northern blots hybridized to FM1, FM2, FM, and H1 probes. FM1 and FM2 probes hybridize specifically with ZmFer1 and ZmFer2 gene transcripts, respectively, whereas the FM probe recognizes both classes of transcripts (11). The maize histone H1 cDNA probe was used as a non-iron-inducible control. Alternatively, the cells were treated with increasing iron citrate concentrations for 3 h (panel B). Afterward, the extracted RNA was analyzed on Northern blots hybridized to the FM1 probe (panels A and B) or by Western blot using polyclonal antibodies raised against the maize seed ferritin subunit (panel C).
ern blot, using an antibody raised against maize seed ferritin subunits. A very low level of ferritin subunits was detected in iron-starved cells (Fig. 2C), consistent with previous observations using plants or soybean cell suspension cultures (6). The addition of iron citrate led to an increase in the amount of ferritin subunits up until 72 h after the beginning of the treatment (Fig. 2C), whereas the ZmFer1 mRNA level reached its maximum 24 h after the addition of iron ( Fig. 2A). This discrepancy could be the result of the almost constitutive expression of the ZmFer2 gene in the BMS cell suspension cultures (data not shown). Such a deregulation of ABA responsive genes in cell suspension cultures has already been observed. 2 Also, a post-translational effect of iron on ferritin stability has been evidenced in animal and plant cells (12,43,44), the presence of the iron core inside the ferritin shell stabilizing the protein structure in vivo. Then, in iron-overloaded cells, ferritins translated from ZmFer2 mRNA would be stabilized by iron. The antibody used in our experiment does not allow discrimination of the two maize ferritin subunits (such an antibody is not available). Taken together, our results demonstrate that iron induces an increased abundance of ZmFer1 transcripts and ferritins in maize BMS cells.

Increased Abundance of ZmFer1 Transcripts in Maize BMS Cells, in Response to Iron Overload, Depends on an Oxidative
Step-The results described above demonstrated the involvement of an oxidative pathway in ZmFer1 transcript accumulation in iron-treated de-rooted maize plantlets. To determine if the same pathway is activated by iron in maize cell suspension cultures, the effect of antioxidant and oxidant agents on ZmFer1 mRNA accumulation in BMS cells was investigated. When maize cells were pretreated for 3 h by an antioxidant such as NAC before a 3-h iron treatment, the increase in ZmFer1 transcript abundance was inhibited (Fig. 3A, compare  lanes 2 and 3). This inhibition is specific to ZmFer1 transcripts as maize histone H1 mRNA abundance, which is not increased on iron treatment (Fig. 3A, lanes 1 and 2), was not affected significantly by NAC treatment (Fig. 3A, lanes 3 and 4). Furthermore, cellular evidence was obtained to demonstrate that BMS cell integrity was not affected by such an antioxidant agent. After NAC treatments, BMS cells were incubated with fluorescein diacetate and then examined by fluorescence microscopy. Using this method, it was shown that after a 6-h incubation with 10 mM NAC, 95% of BMS cells were still viable (data not shown). In addition, iron uptake in iron-treated BMS cells in the presence or without NAC was compared by using an iron radioactive tracer ( 55 Fe). The NAC treatment did not sig-nificantly affect the iron uptake into BMS cells during the time course of the experiment (data not shown). Thus, the increased abundance of ZmFer1 transcripts in iron-treated BMS cells, as in iron-treated de-rooted maize plantlets, is dependent on an oxidative step.
A pro-oxidant treatment with hydrogen peroxide has been shown to induce an increase in ferritin transcript abundance in de-rooted maize plantlets (13). When BMS cells were treated with 5 mM hydrogen peroxide for 6 h, the abundance of ZmFer1 transcripts was increased markedly (Fig. 3B, FM1 probe), whereas the abundance of the histone H1 gene transcripts was not affected (Fig. 3B, H1 probe). Using fluorescein diacetate as mentioned above, it was observed that BMS cell viability was also not affected by H 2 O 2 treatment (data not shown). Therefore, it was concluded that maize BMS cells constitute a suitable cellular model for further study of the oxidative pathway regulating ZmFer1 gene expression in response to iron.

OA Specifically Inhibits ZmFer1 Transcript Accumulation in Iron and H 2 O 2 -treated BMS Cells-To investigate how the
ZmFer1 gene is regulated in response to iron overload, pharmacological studies were performed. Molecular agents that interfere with signaling cascade pathways were tested against BMS cells. Initial experiments used OA, a toxin that specifically inhibits Ser/Thr protein phosphatases (30 -32). This inhibitor has been shown to inhibit plant PP1/2A (33). No effect on ZmFer1 mRNA abundance was observed when cells were treated with 500 nM OA for 6 h with no iron treatment (Fig. 4A,  compare lanes 1 and 4). When cells were incubated for 3 h in 500 nM OA followed by treatment by iron for 3 h, the increase 2 M. F. Niogret and M. Pages, personal communication.  1 and 2) with 500 nM OA for 3 h and then treated (lanes 2 and 3) or not (lanes 1 and 4)  in ZmFer1 mRNA abundance in response to the iron treatment was not observed (Fig. 4A, compare lanes 2 and 3). In contrast, an increased transcript abundance was observed on OA treatment when the same RNA preparation was hybridized with H1 probe irrespective of whether the cells had or had not been iron treated (Fig. 4A, H1 probe). This result is in agreement with a previous report showing that OA treatment does not lead to a general inhibition of gene expression (45) and with our results (see below) showing that both transcription and translation are active in the presence of OA. In addition, as indicated by fluorescein diacetate labeling, OA-treated cell viability was not altered after a 6-h incubation with 500 nM OA (data not shown). Thus, OA specifically inhibits the increased abundance of ZmFer1 transcripts in iron-treated cells. Furthermore, iron uptake measurements using an iron radioactive tracer ( 55 Fe) demonstrated that OA did not affect iron uptake into BMS cells during the time course of OA treatment (data not shown).

FIG. 4. Specific inhibition of iron and H 2 O 2 -induced accumulation of ZmFer1 gene transcripts in maize BMS cells by OA. Panel A, maize BMS cells were pretreated (lanes 3 and 4) or not (lanes
To determine whether or not OA is able to inhibit ZmFer1 gene expression in response to H 2 O 2 treatment, BMS cells were pretreated for 3 h with 500 nM OA, and then 5 mM H 2 O 2 was added in the culture medium. As can be seen in Fig. 4, OA also inhibited the increased in ZmFer1 transcript abundance usually observed in H 2 O 2 -treated cells (Fig. 4B; compare lanes 2  and 3), whereas the abundance of H1 transcripts was again increased by OA treatment (H1 probe).
Therefore, it can be concluded that both inducers of ZmFer1 gene expression, iron and H 2 O 2 , are antagonized by OA treatment.

OA-and CA-sensitive Protein Phosphatase Activities Are Required for the Accumulation of ZmFer1 Gene Transcripts in
Response to Iron Overloading-To confirm that the inhibitory effect of OA on ZmFer1 mRNA accumulation results from the inhibition of protein phosphatase activities, the effect of CA, another Ser/Thr inhibitor structurally unrelated to OA, was investigated (30 -32). CA has been shown to inhibit PP1 and 2A activities in plant and animal cells (33). To compare the effects of both inhibitors, BMS cells were treated for 15 h with increasing concentrations of OA, from 0 to 250 nM, and CA, from 0 to 100 nM, followed by a 3-h treatment with 500 M iron citrate. Potent inhibition of ZmFer1 mRNA accumulation was obtained with a 100 nM concentration of either inhibitor (Fig. 5, panel A,  lane 4; and panel B, lane 6). Preincubation of BMS cells for 6 h instead of 15 h gave the same results for both inhibitors (data not shown). OA and CA specifically inhibited the increase in ZmFer1 transcript abundance; histone H1 mRNA accumulation was unaffected by these treatments (Fig. 5, A and B, H1  probe). Thus, both OA and CA strongly inhibit iron-induced ZmFer1 mRNA accumulation in BMS cells, indicating that a phosphatase activity is required for maximal accumulation of this transcript.
Activation of a ZmFer1-GUS Fusion by Iron in BMS Cells Using a Transient Expression Assay-ZmFer1 transcripts are encoded by a unique nuclear gene ZmFer1, located on chromosome 4 of the maize genome (UMC genetic map of maize; Ref. 46). This gene has been cloned and sequenced recently (11). To investigate the regulation of the ZmFer1 gene by transient expression experiments, a ZmFer1-GUS fusion gene was constructed. A DNA fragment from ZmFer1 containing 1.6 kbp of promoter and the region spanning from exon 1 to 3 was fused in-frame with GUS to obtain the pSL5 plasmid (Fig. 6A). Introns were included in this construct because it has been documented extensively that introns are required for optimal expression of genes in monocotyledonous plant cells (47). The pSL5 DNA was introduced into maize BMS cells by biolistic transformation. In each experiment, the pAHC18 vector was cotransformed as an internal transformation control for stand-ardization of the results. The pAHC18 construct harbors the luciferase reporter gene fused to the ubiquitin promoter and first intron (48).
In nontreated BMS cells, a low level of expression of the ZmFer1-GUS construct (pSL5) was detected, whereas iron treatment led to an 8-fold increase in the reporter gene expression (Fig. 6B). To assess the specificity of this response, the same experiment was performed using the pRD109 construct instead of pSL5. This plasmid contains the GUS reporter gene under the control of a histone H3C4 maize promoter. Iron overload had no effect on the expression of pRD109. Therefore, the transient expression assay described above is suitable for the study of the regulation of ZmFer1-GUS fusion gene by iron in BMS cells.

An Antioxidant Agent and Phosphatase Inhibitors Inhibit Iron-induced Expression of the ZmFer1-GUS Fusion
Gene-When BMS cells or de-rooted plantlets were treated with the antioxidant agent NAC before iron treatment, no ZmFer1 mRNA accumulation was measured (Figs. 1B and 3A). The effect of such a treatment on the ZmFer1-GUS construct was investigated in transient expression experiments. Cells transformed with pSL5 were treated for 3 h with 10 mM NAC before the addition of 1 mM iron citrate to the culture medium. Reporter gene activity measurement indicated that the antioxidant agent diminished the expression of the construct in ironoverloaded cells as well as in control cells maintained on ironfree medium (Fig. 6B). In iron-overloaded cells, an 80% inhibition of pSL5 expression by NAC was observed when compared with iron-treated control cells. Furthermore, NAC treatment had no significant effect on the control construct pRD109 in the transient expression assay (Fig. 6B). This result is consistent with the data obtained from Northern blot experiments demonstrating that antioxidant agents have a strong inhibitory effect on ZmFer1 gene expression (Fig. 3A).
Northern blot experiments had also shown a specific inhibition of ZmFer1 mRNA accumulation in response to iron by Ser/Thr phosphatase inhibitors (Figs. 4 and 5). To analyze the expression of the pSL5 construct in the presence of these inhibitors, transformed cells were incubated for 6 h in the presence of various concentrations of the appropriate inhibitor fol-lowed by incubation for 24 h with 500 M iron citrate added to the media. Determination of reporter gene activities showed a dose-dependent inhibition of pSL5 expression by OA for both iron-induced cells and untreated control cells (Fig. 7). Incubation of transformed cells in the presence of 50 nM or 200 nM OA before iron overload led to 38 and 71% inhibition, respectively, of ZmFer1-GUS fusion expression. In contrast, no significant effect of OA was detected on the expression of the control construct pRD109 (Fig. 7). Similar results were obtained using the phosphatase inhibitor CA (data not shown). It can be concluded from these transient expression assays therefore that iron-induced ZmFer1-GUS expression can be inhibited by an antioxidant agent and Ser/Thr phosphatase inhibitor treatments. DISCUSSION Two pathways have been implicated in the regulation of maize ferritin synthesis in response to iron. One of them requires the plant hormone ABA as a cellular relay and controls the expression of the maize ZmFer2 gene(s) (11). The other pathway is ABA-independent; and from studies using a derooted maize plantlet system, it has been shown to involve an oxidative step (13). In this paper, using a ZmFer2-specific probe, we show that this ABA-independent pathway does not lead to the accumulation of ZmFer2 transcripts in response to a 6-h iron treatment (Fig. 1). Because an exogenous ABA treatment has no effect on ZmFer1 transcript levels (11), we investigated whether the ZmFer1 maize ferritin could be regulated by this ABA-independent oxidative pathway in this de-rooted maize plantlet system.
Using a ZmFer1-specific probe, it was demonstrated that ZmFer1 mRNA accumulation is induced by iron overload in both de-rooted maize plantlets (Fig. 1) and BMS maize cell suspension cultures (Fig. 2). This increased abundance of ZmFer1 transcripts was inhibited by the antioxidant agent NAC and induced by H 2 O 2 treatment (Figs. 1 and 3). These results proved that ZmFer1 mRNA accumulates in response to iron via the oxidative pathway described previously in derooted plantlets (13). We show here that a similar pathway is activated in BMS cell suspension cultures, although the kinetic of iron induction is slightly different between the two systems. Note also that in BMS cells the expression of the ZmFer2 gene is almost constitutive (data not shown), avoiding the use of this Panel B, expression study of pSL5 in transiently transformed maize BMS cells. 8-day-old maize cells were transformed by particle bombardment using the two plasmids, pSL5 or pRD109, schematized in panel A. pRD109 served as a non-iron-inducible control. Each of the constructs was cotransformed with the pAHC18 plasmid. This construct contains a fusion between the promoter and intron 1 of the maize ubiquitin 1 gene and the luciferase reporter gene and allows standardization of the results. After transformation, cells were transferred to liquid medium without iron for 16 h. Cells were pretreated for 3 h in 10 mM NAC if required, and a 24-h treatment with 1 mM iron citrate was performed. Protein extracts were prepared from the transformed cells, and reporter gene activities were measured using the GUS light kit (Tropix) and the luciferase assay kit (Promega). Values are given in relative light units corresponding to a GUS:luciferase activity ratio and are the mean of at least four independent transformations. Standard deviations are given. cell system to study the iron regulation of this gene. Therefore, maize cell suspension cultures have been chosen as a suitable model for further investigation of ZmFer1 gene regulation in response to iron. As a similar pathway has been identified in Arabidopsis thaliana plantlets (9), it is not unique to monocotyledonous plants. In both de-rooted maize plantlets and A. thaliana, ferritin mRNA accumulation in response to iron is inhibited by antioxidant agents, suggesting that an oxidative step is involved in this pathway. Iron is a strong biological pro-oxidant (2), and iron excess in tobacco causes significant physiological disorders, potentially caused by oxidative damage (49).
When maize cells are pretreated by antioxidant agents or Ser/Thr phosphatase inhibitors before the addition of iron, ZmFer1 mRNA accumulation is reduced strongly, but iron uptake by the cells is unaffected. In these conditions, the iron overload of the cells could stabilize ferritins resulting from the constitutive expression of the ZmFer2 gene (see above), to store this element. Also, in response to iron overload, only 5% of the total iron is found in ferritins in soybean cell suspension cultures (50). With plant ferritin compartmentalized in plastids, it has been suggested that iron excess could be handled by other compartments, such as the vacuole (50).
Using a ZmFer1-specific probe has enabled the corresponding ZmFer1 gene to be mapped to a single locus on chromosome 4 (bin number 4.08) (UMC genetic map of maize; Ref. 46). To study further ZmFer1 gene regulation, a transient assay was developed. A 2.2-kbp ZmFer1 gene fragment containing 1.6 kbp of promoter, and the region from exon 1 to 3, was fused inframe with the GUS reporter gene (ZmFer1-GUS). In monocotyledonous plants, introns are required for the expression of various genes. Introns have been shown to increase reporter gene expression up to 1,000-fold in transient and stable transformation (47,51). Similar results were obtained for ZmFer1 gene expression; no significant reporter gene activity was detected in the absence of introns in our constructs (data not shown). Transient expression of the ZmFer1-GUS fusion in BMS cells revealed an 8-fold increase in reporter gene activity in response to iron overload and a strong inhibition of the expression of this construct in the presence of NAC. Although to a lesser extent than in iron-treated cells, expression of the ZmFer1-GUS fusion was also affected by NAC treatment in iron-untreated cells (Fig. 6B). Whether the antioxidant effect of NAC acts exclusively on the iron activation of the ZmFer1 gene or also has an effect on the basal level of expression of this gene remains, therefore, to be clarified. Nevertheless, from these experiments, we can conclude that the ZmFer1 2.2-kbp fragment used in the pSL5 construct contains the regulatory sequences required for increased expression of the ZmFer1 gene on iron induction and that this expression can be inhibited by an NAC treatment. In soybean cells, ferritin mRNA accumulation in response to iron has been shown, using nuclear run-on experiments, to be mediated by a 40-fold increase in transcription (6). So far, in eukaryotes, the iron-responsive element (IRE), which control ferritin synthesis according to the iron status of the cell, has been characterized only in animal cells. In this animal system, almost no variation of ferritin mRNA levels has been detected in response to iron (52). The IRE is a stem-loop structure localized in the 5Ј-untranslated region of both H and L ferritin subunit mRNA and is involved in a translational control, increasing ferritin mRNA translation when iron concentrations are high. Two animal regulatory proteins, IRP1 and IRP2, are involved in this translational control (5). Binding of the IRP1 protein to the IRE is regulated by the iron status within the cell. In the absence of iron this protein binds to the IRE and blocks ferritin mRNA translation. When iron levels are high, IRP1 contains an iron-sulfur cluster, it cannot bind to IREs and has an aconitase activity (5). This transition of IRP1 from RNA binding to aconitase activity is linked, therefore, to the lability of the iron-sulfur cluster of this regulatory protein. (52). The IRE regulatory sequence is not present in any of the plant ferritin mRNA and genes characterized to date (7,9,11). Furthermore, no plant protein was able to bind animal IRE in vitro (53), and fusion of a frog IRE within the 5Ј-untranslated region of a plant ferritin cDNA failed to confer iron-dependent translational control to this chimeric construct (54,55). Thus, in the future, transient expression analysis of ZmFer1 gene expression in response to iron should allow the localization and characterization of new regulatory elements involved in iron-dependent ferritin gene regulation in eukaryotic cells.
The use of protein phosphatase inhibitors such as OA and CA (30) allowed the initial demonstration of the role of Ser/Thr phosphatase activities in various cellular processes in plants, such as photomorphogenesis (45), hormonal control (56,57), cell cycle regulation (58,59), cellular metabolism (60 -63), ion channel control (64 -66), and pollination (67) (for review, see Ref. 33). Using Northern blot analysis and transient expression analysis, we have shown in this paper that these two structurally unrelated toxins inhibited the iron-induced expression of the ZmFer1 gene. A strong inhibition was observed using 100 nM OA or CA. Such an approach has also been used to demonstrate the involvement of protein phosphatase activity in lightinduced chlorophyll accumulation in maize plantlets (45); these results were obtained with concentrations of phosphatase inhibitors similar to those used in here. Our data indicate therefore that an OA-sensitive protein phosphatase activity is involved in the regulation of the ZmFer1 gene in maize cells, and this activity is required for iron-induced expression of this ferritin gene. However, using the BMS/ZmFer1-GUS transient assay that we have developed, it can be seen that an OA treatment also decreased the basal level of expression of the transgene in iron-untreated cells (Fig. 7). The exact role that the phosphatase activity revealed in this work plays in the regulation of the ZmFer1 gene expression remains therefore to be worked out. Because PP2B is almost insensitive to OA and PP2C are insensitive (68,69), these types of protein phosphatases are not likely to be involved in the inhibition of the ironinduced ZmFer1 gene expression. PP1 and PP2A have been shown to be inhibited by OA and CA in plants (45,70), and they are involved in regulatory circuits. However, other Ser/Thr phosphatases, such as PPX, which are also OA-sensitive, have now been characterized from both plants and animals (33,71). Further experiments will be needed to determine the exact nature of the Ser/Thr protein phosphatase involved in the regulation of ZmFer1 expression.
A link has been suggested recently between a phosphatasedependent pathway and the control of ferritin mRNA translation in animal cells through the IRE/IRP1 interaction mentioned above. It has been shown that H 2 O 2 is also able to activate the RNA binding activity of IRP1 onto an IRE (72). This result suggests a role of iron and oxidant sensor for this protein. The H 2 O 2 response might be mediated by the direct effect of this agent on the iron-sulfur cluster of IRP1, as described for other labile clusters (73,74). Alternatively, it has recently been shown that a transduction pathway could also be involved in the H 2 O 2 activation of IRP1 (75). Indeed, IRP1 RNA binding enhancement by H 2 O 2 is inhibited in the presence of OA, suggesting the involvement of a stress-induced phosphatase/kinase pathway. In maize, iron-and H 2 O 2 -induced ZmFer1 gene expression is inhibited by OA treatments, as demonstrated by Northern blots and transient expression analysis in this paper. OA-sensitive phosphatase activities are probably, therefore, involved in both plants and animal ferritin gene regulation, although the ultimate target (transcription versus translation) is different. These data suggest some common aspects in iron metabolism regulation in plants and animal cells. However, in B6 cells, OA only inhibits H 2 O 2 activation of IRP1 and has no effect on the iron-dependent regulation. In contrast, in maize, induction of ZmFer1 gene expression by both iron and H 2 O 2 treatments is inhibited by Ser/Thr phosphatase inhibitors, suggesting that their effect occurs at the integration point of the two signals or downstream of this step.