Differential Regulation of nramp and irt Metal Transporter Genes in Wild Type and Iron Uptake Mutants of Tomato*

Metal transporters regulated by iron can transport a variety of divalent metals, suggesting that iron regulation is important for specificity of iron transport. In plants, the iron-regulated broad-range metal transporter IRT1 is required for uptake of iron into the root epidermis. Functions of other iron-regulated plant metal transporters are not yet established. To deduce novel plant iron transport functions we studied the regulation of four tomato metal transporter genes belonging to the nramp and irt families with respect to environmental and genetic factors influencing iron uptake. We isolated Lenramp1 and Lenramp3 from tomato and demonstrate that these genes encode functional NRAMP metal transporters in yeast, where they were iron-regu-lated and localized mainly to intracellular vesicles. Lenramp1 and Leirt1 revealed both root-specific expression and up-regulation by iron deficiency, respectively, in contrast to Leirt2 and Lenramp3 . Lenramp1 and Leirt1 , but not Lenramp3 and Leirt2 , were down-regu-lated in the roots of fer mutant plants deficient in a bHLH gene regulating iron uptake. In chloronerva mutant plants lacking the functional enzyme for synthesis of the plant-specific metal chelator nicotianamine Leirt1 and Lenramp1 were up-regulated despite sufficient iron supply independent of a functional fer gene. Lenramp1 was expressed in the vascular root parenchyma in a similar cellular pattern as the fer gene. However, the fer gene was not (mutant chloronerva Homozygous fer or chloronerva mutant seeds were obtained after grafting mutant shoots onto wild type root stocks, which rescued the phenotypes. Double homozygous mutants fer and chloronerva ( fer / fer ; chln / chln were obtained in the following way. Among the 1:8 segregating progeny of a double heterozygous plant ( ; chln / (cid:1) chlorotic F2 were and grafted onto wild type root stocks to obtain seeds genotypes / ; chln / chln , fer / (cid:1) ; chln / chln , (cid:1) / (cid:1) ; chln / chln , fer / fer (cid:1) (cid:1) and fer / fer ; chln / (cid:1) We identified F3 that segregated 1:3 for double mutants (yellow very small / chln / in a chloronerva (yellow green cotyledons ; ; chln or (yellow Non-segregating double mutant were also mutants from a segregating line in the chloronerva background were rescued Genomic DNA was from individuals and checked for the presence of the mutant fer allele by restriction fragment length polymorphism to

The transition metal iron is essential for many cellular electron transfer reactions. Because of its redox potential free iron may become toxic and produce radicals. In addition, uptake of various divalent metals can interfere with that of iron and vice versa, as iron transport mechanisms often do not discriminate metals. These characteristics may explain the presence of multiple transport and regulatory mechanisms for storage and mobilization of iron in all cells and organisms.
Molecular characterization of iron mobilization in plants ex-posed to iron deficiency has been a recent subject of investigation. In dicotyledon plants, iron is mobilized by reduction and taken up across the plasmalemma into the root epidermis via IRT1 1 (1)(2)(3). IRT1 is a member of the ZIP family of broad range metal transport proteins identified from eukaryotes and, recently, also bacteria (4 -6). Atirt1 transcription is predominant in roots and induced upon iron deficiency (2). Loss of function of irt1 leads to reduced viability in Arabidopsis unless excess iron is supplied, suggesting that IRT1 is a major structural component of iron uptake in plants (3,7). Grasses may take up iron differently because they mobilize iron by secreting potent phytosiderophores from the root and subsequently take up the iron-siderophore complexes (1). Internal transport and mobilization of iron inside the plant root is still unclear. Iron is transported via the xylem to the shoot essentially as iron citrate (for an example, see Ref. 8). Furthermore, iron can be translocated in the phloem, for example as complex with nicotianamine, a diffusible plant-specific small non-proteinogenic tripeptide derived from methionine (for review, see Ref. 9), or iron-binding proteins (10). Because plants play a major role for human nutrition and because iron deficiency is a worldwide nutritional problem, one challenge is to produce iron-rich crops (for examples, see Refs. 11 and 12). In this context, it is of great interest to study internal iron mobilization and transport in plants. The specific roles of iron transporters other than IRT1 are not known yet. Atirt2, although in sequence and expression pattern similar to Atirt1, is not capable of functionally complementing irt1 mutants, and loss of function causes no obvious phenotype upon low iron supply (7,13). To date six members of the Arabidopsis nramp gene family have been assigned and partially characterized. NRAMP proteins are broad range membrane-bound metal transporters found in all eukaryotes (for review, see Ref. 14). Their name was derived from phagosomal NRAMP1 (natural resistance-associated macrophage protein 1), which functions as an efflux pump in the membrane and in this way enhances resistance against intracellular bacteria by reducing metal availability (for review, see Ref. 15). Mammalian NRAMP2 is an important factor for iron transport from the duodenum lumen into epithelial cells (16,17). The three yeast NRAMP proteins, SMF1, SMF2, and SMF3, are differentially regulated by iron and manganese and may serve to release these metals from intracellular stores (18,19). It was found that Atnramp1, Atnramp3, and Atnnramp4 encode functional plant metal transporters and are expressed at higher levels upon low iron supply in plants (20,21). Specificity of plant transporters may be achieved essentially by differential regulation at transcriptional and post-transcriptional levels as shown for IRT1 and metal transporters in other eukaryotes (2,(22)(23)(24). In plants, iron acquisition is controlled by yet unknown iron signals. A promising clue about signaling components involved in iron acquisition was recently derived from the molecular identification of the tomato fer gene encoding a root-specific bHLH protein (25). The tomato fer mutant is unable to develop physiological and morphological iron-deficiency responses such as iron reduction and induced Leirt1 transporter gene expression in roots and accumulates less iron than wild type, leading to severe chlorosis (25)(26)(27). The fer gene seems to be required for sensing iron availability in the root tip and subsequently regulating the appropriate physiological and morphological responses (25). The regulatory mechanism and concrete molecular responses affected by the fer mutation are not known.
An opposite effect on strategy I responses is observed in the chloronerva mutant from tomato, which lacks nicotianamine due to a mutation in nicotianamine synthase (28). chloronerva mutant plants display increased iron reductase activity at sufficient iron supply and over-accumulate metals compared with the wild type, suggesting that nicotianamine synthase is required for adequate uptake and distribution of iron in plants (for review, see Ref. 9).
Here, we analyzed the possible role and regulation of two nramp genes and two irt genes from tomato in response to environmental and genetic factors related to iron nutrition. We could establish a model of gene interactions and regulation involved in iron uptake in plants. From our results we propose a novel role for nramp1 in plants.

EXPERIMENTAL PROCEDURES
Plant Growth-10-Day-old tomato seedlings were transferred into moderately aerated hydroponic culture with Hoagland solution containing 10 M FeNaEDTA (normal iron supply) (according to Stephan and Prochazka (29)). After 2 days, the experiment was started by transferring the seedlings into Hoagland solution with varying concentrations of iron for up to 8 days. The limiting supply was 0.1 M FeNaEDTA; excess supply was 100 M FeNaEDTA.
Plant Material-Seeds were derived from the lines Lycopersicon esculentum cv. Moneymaker (wild type), T3238fer (mutant fer), and chloronerva (mutant chloronerva). Homozygous fer or chloronerva mutant seeds were obtained after grafting mutant shoots onto wild type root stocks, which rescued the phenotypes. Double homozygous mutants for fer and chloronerva (fer/fer; chln/chln) were obtained in the following way. Among the 1:8 segregating progeny of a double heterozygous plant (fer/ϩ; chln/ϩ), chlorotic F2 individuals were selected and grafted onto wild type root stocks to obtain seeds (F2 genotypes fer/fer; chln/chln, fer/ϩ; chln/chln, ϩ/ϩ; chln/chln, fer/fer; ϩ/ϩ and fer/fer; chln/ϩ). We identified F3 lines that segregated 1:3 for double mutants (yellow very small leaves, yellow veins, green-yellow cotyledons ϭ fer/fer; chln/chln) in a chloronerva background (yellow leaves, green veins, green cotyledons ϭ fer/ϩ; chln/chln or ϩ/ϩ; chln/chln) or fer background (yellow leaves, yellow veins, green cotyledons ϭ fer/fer; chln/ϩ or fer/fer; ϩ/ϩ). Non-segregating double mutant lines were also identified. Individual mutants derived from a segregating line in the chloronerva background were rescued by grafting. Genomic DNA was prepared from individuals and checked for the presence of the mutant fer allele by restriction fragment length polymorphism according to Ling et al. (25). This way it was confirmed that double mutants with severe shoot phenotype were homozygous for the mutant fer allele, whereas mutants with chloronerva phenotype were either heterozygous for the mutant fer allele or contained the wild type allele only. Transgenic tomato plants expressing a fer cDNA behind the constitutive cauliflower mosaic virus 35S promoter in a fer mutant background have been described by (25).
Gene Isolation and Characterization-Lenramp1c DNA was obtained as follows. Degenerate primers 5Ј-aytcyccagcataagtdcc-3Ј and 5Ј-aatstmttcytscaytcdgc-3Ј, corresponding to conserved motifs of Atnramp1 and Osnramp1 genes (20,30), were used to amplify DNA fragments from cDNA prepared from iron-starved tomato root RNA. Amplification was performed for 35 cycles using ExTaq polymerase (Taqara) in the presence of 2 M primers, 92°C denaturing for 1 min, 45°C annealing for 1 min, and extension at 72°C for 1 min. A main amplification product of 334 bp was obtained that was subcloned into pCRII (Invitrogen) and sequenced. The labeled purified 334-bp fragment was used to screen a tomato BAC library (LE_Hba, L. esculentum cv. Heinz 1706, Clemson University). Lenramp1 contained on BAC clone 24I11 was sequenced to obtain full exon and intron sequences. Specific primers were designed (5Ј-atggagaatcaacagcaaaatcaa-3Ј and 5Ј-ttattgtggcaatggaatatcagc-3Ј) to amplify full-length Lenramp1-coding sequence from root cDNA. The Lenramp1 cDNA fragment was subcloned into pCRII and fully sequenced (accession number AY196091).
A Lenramp3 cDNA clone was identified by a BLAST search of Arabidopsis and rice NRAMP amino acid sequences (20,21,30) against the tomato EST data base (tigrblast.tigr.org/tgi). Several EST clones corresponding to four assembled transcription units were obtained from Clemson University and sequenced. EST clone cLET21E13 fully contained the Lenramp3-coding sequence (accession number AY196092).
Yeast Complementation Assay-Lenramp1-and Lenramp3-coding sequences subcloned as PCR fragments into pCRII (Invitrogen) were excised using EcoRI and cloned in correct orientation into the EcoRI site from pYES2 (Invitrogen) behind the galactose-inducible GAL1 promoter, yielding pYES2-LeNR1 and pYES2-LeNR3, respectively. Saccharomyces cerevisiae strain smf1 (smf1⌬, SLY8, (31) was verified for its genotype (MAT␣ his3 ade2 leu2 trp1 ura3 smf1::HIS3) by plating onto selective media and transformed according to Dohmen et al. (32) with pYES2-LeNR1, pYES2-LeNR3, and as a control, with empty pYES2. Transformed cells were selected on SD medium containing amino acid supplements without uracil (SDϪura) and containing 2% glucose. Liquid overnight cultures of transformed cells and, as further control, wild type Y190 (Clontech) were each diluted to optical densities of 0.5, 0.05, and 0.005 and spotted onto SDϪura plates supplemented with 2% galactose and containing 5, 10, 20, and 50 mM EGTA, respectively, in the presence of 50 mM MES at pH 6.0 for complementation assays. Photos of yeast colonies were taken after 2 days of growth at 30°C.
GFP Fusion Protein Localization in Yeast-Full-length coding sequences for Lenramp1 and Lenramp3 were amplified using primers that contained the Gateway-compatible attachment sites. Following the Gateway protocol (Invitrogen), these PCR fragments were cloned into destination vector pDESTFY (provided by E. Walker, University of Massachusetts, Amherst, MA), a Gateway-compatible derivative of pFL61 (33), allowing the expression of a protein in-frame with a Cterminal GFP behind the strong PGK (phosphoglycerate kinase) promoter. The resulting plasmids were named pEXFY-NR1 and pEXFY-NR3 and transformed into yeast strain L40coat according to Dohmen et al. (32). Positive colonies selected on SDϪura medium were each inoculated into SDϪura medium (normal) or SDϪura medium containing either 50 M BPDS, 20 mM EGTA, or both. After 2 days of growth cells were observed using a Axiophot microscope (Zeiss) integrated into a digital optical three-dimensional microscope system (Digital Optics Inc.). Only cells devoid of strong autofluorescence as visible in red (filter set XF 102-2; Omega Optical Inc.) have been considered for analysis. Image stacks were taken in differential interference contrast (DIC) and GFP fluorescence (filter set XF 100 -2) to generate extended focus images and to combine them for three-dimensional-three-dimensional co-localization. To identify nuclei, cells were embedded in 0.2 g/ml 4Ј,6-diamidino-2-phenylindole, dihydrochloride (DAPI) in Vectashield mounting medium (Vector Laboratories, Inc.) and observed using the 4Ј,6-diamidino-2-phenylindole, dihydrochloride filter set F31-013 (AHF Analysentechnik AG).
In Situ Hybridization-Roots tips from plants grown for 8 days in Hoagland medium with 0.1 M FeNaEDTA were fixed and embedded in paraplast according to Kyozuka et al. (37). 10-m transverse root sections were hybridized using digoxigenin-labeled RNA probes according to Jackson (38). RNA probes were prepared from 500 -600 bp of amplified cDNA fragments subcloned into pCRII (Invitrogen) using either T7 or SP6 RNA polymerase for in vitro transcription. Hybridization signals were detected using primary anti-Dig antibodies coupled to alkaline phosphatase (violet staining, according to Jackson (38). Probe cDNA fragments were obtained using the following primer pairs: 5Ј-ggccaatttatcatgcaaggatttc-3Ј and 5Ј-ttgtggcaatggaatatcagcaagat-3Ј for Lenra-mp1; 5Ј-atggagagtggtaatgcatcaatgga-3Ј and tgattgctggataataggttgtgaaat-3Ј for fer (25); 5Ј-tcaagcacatcgccatcatctttattctca-3Ј and 5Ј-acccatttcttgatcagcaccaccagttt-3Ј (Leirt2a fragment) and 5Ј-ggagttgaaggcactaaattactacg-3Ј and 5Ј-gagggaaatacatagcagttcaaat-3Ј (Leirt2b fragment) for Leirt2 (35). Leirt2a and Leirt2b probe fragments covered Leirt2 cDNA and were mixed to yield the Leirt2 probe. The Leirt2 probe was specific for Leirt2. Probe fragments covering Leirt1 did not result in any hybridization signals, indicating that there was no cross-hybridization of Leirt1 and Leirt2 RNA probes with the respective mRNAs (not shown). In control reactions labeled sense probes were used that resulted in no signals (Fig. 5, G-I).

Identification of nramp Metal
Transporter Genes from Tomato-NRAMP proteins are encoded by evolutionarily conserved genes. By performing a BLAST search of plant NRAMP peptide sequences against the tomato EST data base we could identify several EST sequences potentially encoding NRAMP transporters. Several EST clones covering four non-overlapping nramp sequences with similarity to Atnramp3 have been fully sequenced. This way we could determine that these EST nramp sequences were derived from different regions of the same gene. Because of its similarity with Atnramp3 (21) we have termed this gene Lenramp3 (accession number AY196092). The Lenramp3 coding sequence was fully contained in EST clone cLET21E13.
In addition degenerate primers were designed recognizing two conserved regions between Atnramp1 and Osnramp1, one of them representing the conserved transport motif (20,39). Using the degenerate primers, a 334-bp nramp cDNA fragment was amplified from iron-starved root cDNA. Sequencing revealed best sequence similarity with Atnramp1 so that this gene was termed Lenramp1. Full-length Lenramp1 sequence was obtained from a tomato BAC clone containing the 334-bp DNA fragment, namely 24I11. Full-length Lenramp1 cDNA was amplified from root cDNA, subcloned, and sequenced (accession number AY196091). Lenramp1 cDNA sequence was identical to the exon genomic sequence obtained from the BAC clone.
The deduced protein sequences of LeNRAMP1 and LeNRAMP3 were compared with those of the six Arabidopsis and three rice NRAMP protein sequences (20,21,30). Sequence comparison showed that plant NRAMP sequences can be divided into two subgroups and that LeNRAMP1 and LeNRAMP3 fall into distinct subgroups (Fig. 1). Comparing Arabidopsis and tomato NRAMP sequences showed that LeNRAMP1 is most similar with AtNRAMP1 and AtNRAMP6. LeNRAMP3 shows the highest sequence similarity with At-NRAMP3 and AtNRAMP4. This finding indicates that distinct NRAMP functions might be conserved in plants.
Functional Analysis of Lenramp1 and Lenramp3 in Yeast-To determine whether Lenramp1 and Lenramp3 might encode functional metal NRAMP transporters, full-length coding sequences have been cloned into the yeast expression vector pYES2 behind the galactose-inducible GAL1 yeast promoter. These constructs were expressed in the yeast smf1 mutant, deficient in a broad range metal transporter nramp homolog (40,41). To test for functional complementation, smf1 mutants transformed with pYES2 containing either Lenramp1, Lenramp3, or empty control vector and, as positive control, wild type yeast were spotted in different dilutions on minimal medium in the presence of 5, 10, 20, and 50 mM divalent metal chelator EGTA. Under these conditions, Lenramp1-and Lenramp3-transformed smf1 strains grew significantly better than control mutant smf1 strain transformed with empty vector ( Fig. 2A). Growth of LeNR1-and LeNR3-transformed yeast cells was comparable with that of wild type if the medium was supplemented with up to 20 mM EGTA ( Fig. 2A). At 50 mM EGTA, growth of LeNR1-and LeNR3-transformed cells was retarded compared with that of wild type but significantly better than that of control pYES2-smf1 mutant cells ( Fig. 2A). Growth of wild type was also reduced at 50 mM EGTA compared with lower EGTA concentrations ( Fig. 2A). Therefore, Lenramp1 and Lenramp3 both encode functional metal transporters able to restore growth of a yeast metal uptake mutant.
To determine in which compartments LeNRAMP1 and LeNRAMP3 could be mediating metal transport, we localized the proteins in yeast cells using GFP fusion constructs. We observed different types of staining patterns when GFP was fused to a NRAMP protein (Fig. 2, C-H), whereas fluorescence was fairly uniform when GFP was expressed in its free form without fusion protein (Fig. 2B). Generally, only few cells expressed LeNRAMP-GFP fusion proteins. However, their number increased up to 40-fold when 50 M BPDS or 50 M BPDS and 20 mM EGTA was added to the growth medium (data not shown). The number of stained cells increased up to 5-fold in the presence of only 20 mM EGTA (data not shown). BPDS is a synthetic chelator for iron-causing iron deficiency. EGTA chelates manganese, causing manganese deficiency. LeNRAMP1 was found mainly in single or multiple vesicles upon normal growth conditions (70 -80% of GFP-expressing cells) (examples for these staining patterns are given in Fig. 2, C-E). In the presence of BPDS, LeNRAMP1 was found to an even greater extent in single and multiple vesicles (80 -90% of GFP expressing cells). In the remaining 10 -30% of cells LeNRAMP1-GFP was localized to the vacuolar membrane (not shown), to intracellular structures (an example is shown in Fig. 2E), or partially to the plasma membrane (not shown). LeNRAMP3 was localized mainly in the vacuole upon normal growth conditions (20 -50% of GFP expressing cells) and/or in single vesicles (30 -60% of GFP stained cells) and only occasionally in multiple vesicles (Fig. 2, F-H). Upon inducing iron deficiency the staining was less frequently detected in the vacuole (less than 20% of GFP expressing cells) but more frequently in single or multiple vesicles (around 60 -70% of GFP expressing cells) (Fig. 2, G and H). Occasionally we observed patterns where the vacuolar membrane was stained (Fig. 2F), or staining was found in the cytoplasm (not shown) or, partially, in the plasma membrane (Fig. 2H). Vesicles showing LeNRAMP1 or LeNRAMP3-GFP staining were presumably of similar origin. These vesicles were in distinct location from the nucleus (Fig.  2D, shown for LeNRAMP1-GFP) and located underneath the plasma membrane outside of the vacuole (Fig. 2, C and H). In summary, LeNRAMP1 and LeNRAMP3 protein expression and localization patterns are regulated by iron and to a lower extent manganese supply in yeast. LeNRAMP1 was predomi-nantly localized to intracellular vesicles, which might indicate that LeNRAMP proteins could mediate release of metals from such vesicles.
Expression Analysis of Lenramp and Leirt Genes-Specificity of metal transporters for certain metals may be essentially achieved by transcriptional regulation (2,22,23). Therefore, expression of Lenramp1, Lenramp3, and two previously identified zip genes, Leirt1 and Leirt2 (35) was analyzed in different parts of iron-starved tomato plants. We found that all four genes were expressed in roots. Lenramp1, Leirt1, and Leirt2 transcripts were only found in roots but not leaves and cotyledons (Fig. 3A). Lenramp3 was highest expressed in roots, but transcripts were also found in green parts of the shoot (Fig. 3A).
To study whether the genes were induced by iron deficiency in roots, tomato plants were grown in a hydroponic system. Root samples were harvested for analysis 2, 4, and 8 days after supply of iron-limiting (0.1 M FeNaEDTA) or iron-sufficient (10 M FeNaEDTA) growth medium. We observed that expression of Leirt1 and Lenramp1 was enhanced at low iron supply in roots compared with sufficient iron supply conditions (Fig.  3B). Expression of Lenramp3 was slightly enhanced upon iron limitation (Fig. 3B). Expression of Leirt2 was not significantly affected by iron supply (Fig. 3B). Thus, Leirt1, Leirt2, Lenramp1, and Lenramp3 show differential expression with respect to tissue specificity and regulation by iron supply.
Expression Analysis of Metal Transporter Genes in fer Mutant Plants-We tested whether the fer defect affected the expression of metal transporter genes in fer plants grown in the presence of limiting (0.1 M iron), sufficient (10 M iron), or excess amounts of iron (100 M iron). We found that Lenramp1 showed a significant decrease in transcription level in roots of fer mutant plants compared with wild type in all three iron supply conditions analyzed (Fig. 4). Leirt1 expression was reduced in fer roots compared with wild type, however, to a lower extent than that of Lenramp1 (Fig. 4, also see Ref. 25). In contrast, expression of Lenramp3 and Leirt2 were not significantly altered in the fer mutant (Fig. 4). Lenramp3 was found to be induced in roots after iron starvation even in the absence of an intact fer gene (Fig. 4). Therefore, expression of Leirt1 and Lenramp1 but not Leirt2 and Lenramp3 is dependent on fer.
To investigate in which tissues LeNRAMP1 may be active, the expression pattern of the fer-dependent Lenramp1 gene was studied using in situ hybridization on transverse sections of iron-starved roots. We detected hybridization signals in the root epidermis and cortex behind the root tip (not shown) as well as in the vascular parenchyma between xylem and phloem in the root hair zone of wild type roots (Fig. 5A). In fer mutant roots, Lenramp1 hybridization signals were not visible in the vascular parenchyma (Fig. 5D) but were still apparent in the outer cell layers close to the root tip (not shown). As a control we utilized Leirt2 as an in situ hybridization probe. We found that Leirt2 was expressed in the epidermis and the vascular parenchyma (Fig. 5B). Leirt2 expression in the vascular parenchyma cells of the root hair zone was apparent in wild type as well as in mutant fer roots (Fig. 5E). fer expression was localized to the vascular cylinder of the root hair zone and in the outer cells close to the root tip (Fig. 5, C and F; see also Ref. 25). Thus, Lenramp1 and Leirt2 expression co-localize with that of fer in the vascular parenchyma. Only the cellular expression pattern of Lenramp1 but not that of Leirt2 was affected by the fer mutation and only within the vascular cylinder.
To analyze whether fer may be able to directly activate metal transporter genes, we examined Lenramp1 and Leirt1 gene expression in roots and leaves of transgenic tomato plants expressing an intact fer cDNA behind the constitutive cauliflower mosaic virus 35S promoter in a fer mutant background. We found that in the roots of these transgenic plants, the level of expression of Lenramp1 and Leirt1 was comparable with that in wild type plants at low and sufficient iron supply (Fig.  6, Roots ϪFe, Roots ϩFe, compare 35s1 and 35s2 with wild type (WT)). In leaves where fer was ectopically expressed in the transgenic plants, Leirt1 and Lenramp1 were not expressed irrespective of iron supply (Fig. 6, Leaves ϪFe and ϩFe, 35s1  and 35s2). Therefore, we concluded that fer is required but not sufficient to direct expression of Leirt1 and Lenramp1.
Expression Analysis of Metal Transporter Genes in chloronerva and fer/chloronerva Double Mutants-The mutant chloronerva, to date the only nicotianamine-free plant mutant, over-accumulates metals and displays up-regulation of iron reduction upon sufficient iron supply (9). We examined whether in chloronerva mutant plants iron transporter gene regulation was affected. We found that in chloronerva roots low iron supply did not significantly induce expression of Leirt1 and Lenramp1 as was the case in wild type plants (Fig. 7). However, upon sufficient and excess iron supply, the expression level of these two genes was significantly increased in chloronerva plants compared with wild type (Fig. 7). Expression of Leirt2 appeared unaltered in chloronerva mutant versus wild type plants regardless of available iron in the medium. Lenramp3 appeared to be expressed at comparable or slightly increased level in chloronerva mutant versus wild type plants (Fig. 7). Thus, nicotianamine synthase is required for proper regulation of Lenramp1 and Leirt1.
We tested whether a functional fer gene was needed for increased expression of Leirt1 and Lenramp1 in chloronerva mutants. Therefore, we constructed non-segregating double homozygous mutant lines (fer/fer; chloronerva/chloronerva) as well as lines segregating 1:3 for the double mutant in an otherwise fer or chloronerva mutant background, respectively. We found that double homozygous mutants could be recognized morphologically by their shoot appearance. fer/chloronerva double mutant plants had yellow-green cotyledons and very small yellow leaves with yellow veins at both low and sufficient iron supply (Fig. 8A, only ϪFe shown). Shoots of double mutant plants were significantly smaller at iron-limiting and sufficient iron supply conditions than shoots of either fer or chloronerva single mutants (Fig. 8B). These observations indicate that double mutants were more severely affected than the single mutants and could not be partially rescued by increasing iron concentrations in the growth medium as was the case for single mutants. Double mutants like the single fer mutant plants did not show iron reductase activity (data not shown). CHLOR-ONERVA transcripts were detectable in fer mutant roots, and FER transcripts were found in chloronerva mutant roots (Fig.  8, C and D), indicating that fer and chloronerva genes are not involved in regulating gene expression of each other.
We found that in the double mutant roots, expression of Leirt1 and Lenramp1 was significantly decreased at the three iron supply conditions tested, as was the case for fer mutant roots (Fig. 7). Expression of Lenramp3 and Leirt2 was not significantly affected in the double mutants (Fig. 7). Therefore, double fer/chloronerva mutants had a similar expression pattern of metal transporter genes as the single fer mutant plants showing that the fer gene was required for increased Leirt1 and Lenramp1 expression observed in single chloronerva mutant plants. DISCUSSION Regarding plants, only for IRT1 a well determined role as membrane-bound transporter mediating iron uptake into roots was proposed to date (2,3,7). In the present study we characterized two novel nramp genes from tomato, and we investigated the regulation of four different metal transporter genes belonging to the irt and nramp gene families. Based on our results we suggest a novel function for nramp1 in mobilizing iron upon iron deficiency in the vascular parenchyma in plants.
We could establish that a fer/nicotianamine synthase-dependent signal transduction pathway exists that controls expression of iron-regulated genes Leirt1 and Lenramp1. Another regulatory mechanism not dependent on fer and nicotianamine synthase involves Leirt2 and Lenramp3.
Differential Functions for Metal Transporter Genes-Sequence comparisons between plant NRAMP amino acid sequences indicate that NRAMP1-type proteins form a conserved subfamily of NRAMP metal transporters in the plant kingdom.
Lenramp1, which encodes a functional metal transporter, as shown in yeast complementation experiments, is induced by iron limitation in tomato roots. Lenramp1 expression was altered in mutants of the bHLH gene fer and nicotianamine synthase, which are both required for proper regulation of iron uptake. Furthermore, LeNRAMP1 mRNA coding sequence or protein gene products are iron-regulated at the post-transcriptional or post-translational level. Therefore, Lenramp1 is likely to play a role in acquisition and/or redistribution of iron. We showed that Lenramp1 expression took place in the vascular parenchyma in the root hair zone and that this expression pattern was dependent on a functional fer gene. It is interesting to note that Lenramp1 was expressed in the same vascular parenchyma cells as the fer gene. Our finding thus indicates that the vascular parenchyma cells may play an important role in regulating iron nutrition of the plant. Because the vascular parenchyma cells lie between phloem and xylem in tomato roots, it is tempting to speculate that these cells receive and/or release iron or iron signals.
Leirt1 can be a crucial factor in iron uptake, since the Arabidopsis homolog Atirt1 has been recently reported as the ma- FIG. 7. Expression analysis of metal transporter genes in chloronerva (chln) and fer/chloronerva double mutants (fer/chln). Expression was monitored before the start of the experiment (0) and after 8 days of different iron supply treatments. fer mutant plants and wild type tomato serve as controls. Elongation factor 1a (Leef-1a), which occurs in all tissues, growth conditions, and tomato lines, is the control. WT, wild type. jor factor in iron uptake from the soil (3). Three observations support that this could be also the case for Leirt1. First, among tomato EST sequences representing putative zip genes, Leirt1 showed the highest sequence similarity with Atirt1. 2 Second, Leirt1 was induced in response to iron starvation. Finally, Leirt1 expression was altered in mutants with well defined defects in iron uptake.
Leirt1 and Lenramp1 were down-regulated in the fer mutant irrespective of iron supply conditions. It was demonstrated by Ling et al. (25) that the fer gene acts upon iron limitation and sufficient iron supply. Only excessive amounts of iron (100 M) can rescue fer mutant plants. Therefore, Leirt1 and Lenramp1 are presumably not only involved in transporting iron upon iron limitation but also function perhaps to a lower extent in iron transport upon sufficient iron supply. Interestingly, Arabidopsis irt1 mutant plants also require high iron concentrations for rescue (more than 100 M iron), indicating that Atirt1 functions at sufficient iron supply conditions (10 -50 M iron) where wild type plants would show no obvious symptoms of iron deficiency (3,7). Different expression behaviors were observed for Leirt2 and Lenramp3. Leirt2 was found constitutively expressed in roots irrespective of iron supply. Lenramp3, which also encodes a functional metal transporter, was found induced by iron deficiency in roots and is expressed in roots and leaves. Lenramp3 and Leirt2 do not require functional fer and nicotianamine synthase genes for regulation of their expression. Apparently, Leirt2 and Lenramp3 cannot functionally compensate the reduced Leirt1 and Lenramp1 expression in fer mutant plants. Therefore, their function remains rather speculative. It is possible that Leirt2 and Lenramp3 mediate iron uptake only in the absence of Leirt1 and Lenramp1 provided that excessive iron is available. This may also explain why in Arabidopsis neither Atnramp3 nor Atirt2 loss of function leads to an obvious chlorotic phenotype (7,21). Alternatively, LeIRT2 or LeNRAMP3 may supply iron to particular compartments after iron has been mobilized, function upstream of the fer gene, or transport predominantly other metals than iron.
Distinct functions for nramp genes have also been found in yeast. SMF1 was reported to be present in portions of the plasma membrane, where it may serve metal uptake (42). SMF2 was found in intracellular vesicles of an unknown nature, where it may mobilize metals from intracellular stores (18). Upon metal replete conditions, SMF1 and SMF2 become targeted to the vacuole for degradation (18,42). SMF3 was only detected in the vacuolar membrane, where it may mobilize iron from vacuolar stores (18). LeNRAMP1 and LeNRAMP3 could complement the smf1 defect. Because LeNRAMP1 and LeNRAMP3 were mainly localized to vesicles upon manganese and iron deficiency in yeast, smf1 complementation may work essentially by release of internal metals, as was proposed for SMF2 function (18). Both NRAMP gene products may be subject to post-transcriptional or post-translational control by iron and, to a lower degree, manganese. GFP fusion protein signals were detected in more cells and localized less frequently to the vacuole upon iron-deficient growth conditions than upon sufficient iron supply. This can be explained by increased protein degradation in the vacuole or lower mRNA stability upon iron supply. Therefore, LeNRAMP1 and LeNRAMP3 may contain signals, allowing mRNA and/or protein regulation in response to iron in yeast. We are currently generating transgenic plants expressing LeNRAMP1-and LeNRAMP3-GFP fusion proteins to test ironmediated post-transcriptional or post-translational regulation in the plant system. Iron-regulated and ubiquitin-mediated protein degradation was suggested for IRT1 in plants (24). As a working model we propose the following new function for NRAMP1 in plants. Upon iron deficiency FER may upregulate Lenramp1 transcription. In addition, LeNRAMP1 may become stabilized at the protein level and mobilize iron in the vascular parenchyma cells. This iron may be utilized for transport toward the shoot to meet rapid iron requirements. LeNRAMP1 may also be involved in further signaling of the iron status. Upon sufficient iron supply, Lenramp1 expression is down-regulated to a low level. This system may allow the plant to rapidly respond to even small changes in iron requirements.
Regulation of Iron Uptake by Nicotianamine and fer-Our expression analysis of metal transporter genes in fer and chloronerva mutant plants suggests that the mutant phenotypes can be explained by inadequate iron transporter regulation. Lenramp1 and fer genes were expressed in similar tissues, opening the possibility that FER might directly activate Lenramp1. However, we failed to demonstrate Leirt1 or Lenramp1 induction when ectopically expressing the fer gene. We have been unsuccessful so far in obtaining binding of FER to E-box sequences contained in the Lenramp1 promoter. 3 bHLH factors usually bind E-box target sequences (CANNTG) as homoor heterodimers. Several bHLH proteins bind their targets when activated by another protein partner (for example, see Refs. [43][44][45]. If there is direct interaction between FER and the Lenramp1 promoter we predict that FER requires for transcriptional activation the interaction with a root-specific binding partner. chloronerva (nicotianamine synthase) mutant plants showed decreased expression of Leirt1 and Lenramp1 upon iron starvation and increased expression upon sufficient iron supply compared with wild type. Iron over-accumulation in chloronerva mutant plants, thus, can be explained by increased metal transporter activity despite sufficient iron supply. Altered regulation of metal transporter genes due to the nicotianamine synthase mutation in the chloronerva mutant plants was dependent on a functional fer gene. Regarding iron transporter gene expression and iron reduction in the double mutant, the fer mutant allele acted in an epistatic way to the chloronerva mutant allele. This finding suggests that the fer gene would act before the chloronerva nicotianamine synthase gene in a same pathway. Regarding shoot chlorosis, the two mutations clearly behaved in a synergistic way as the double mutant shoot phenotypes were more severe than those of either single mutant. It can be interpreted that fer and chloronerva genes act together in the same pathway. We could exclude that the fer gene acted upstream of nicotianamine synthase by regulating nicotianamine synthase gene expression and vice versa. To explain our results, we propose the following model. At low iron supply nicotianamine produced by nicotianamine synthase may be required for the fer gene-mediated induction of iron uptake responses. At sufficient iron supply, nicotianamine may repress iron uptake responses via FER. We suggest that nicotianamine acts in the fer gene signaling pathway as the sensor for iron availability. Modulation of nicotianamine action could be achieved by differential binding of metals to nicotianamine. For example, upon sufficient iron nutrition, iron could be bound to nicotianamine, resulting in suppression of these responses, whereas upon iron deficiency other metals may be bound, causing induction of iron deficiency responses. Future studies will show whether nicotianamine may bind directly to a regulator protein or whether nicotianamine acts indirectly by delivering metals to final targets, which then regulate the fer gene pathway.