AtNRAMP3, a multispecific vacuolar metal transporter involved in plant responses to iron deficiency

Authors

  • Sébastien Thomine,

    Corresponding author
    1. Institut des Sciences du Végétal, CNRS UPR2355, Avenue de la Terrasse, 91198 Gif-sur-Yvette Cedex, France, and
    2. Cell and Developmental Biology Section, Division of Biology and Center for Molecular Genetics, University of California, San Diego, La Jolla, CA 92093-0116, USA
    Search for more papers by this author
  • Françoise Lelièvre,

    1. Institut des Sciences du Végétal, CNRS UPR2355, Avenue de la Terrasse, 91198 Gif-sur-Yvette Cedex, France, and
    Search for more papers by this author
  • Elise Debarbieux,

    1. Institut des Sciences du Végétal, CNRS UPR2355, Avenue de la Terrasse, 91198 Gif-sur-Yvette Cedex, France, and
    Search for more papers by this author
  • Julian I. Schroeder,

    1. Cell and Developmental Biology Section, Division of Biology and Center for Molecular Genetics, University of California, San Diego, La Jolla, CA 92093-0116, USA
    Search for more papers by this author
  • Hélène Barbier-Brygoo

    1. Institut des Sciences du Végétal, CNRS UPR2355, Avenue de la Terrasse, 91198 Gif-sur-Yvette Cedex, France, and
    Search for more papers by this author

For correspondence (fax +33 1 69 82 37 68; e-mail thomine@isv.cnrs-gif.fr).

Summary

Metal homeostasis is critical for the survival of living organisms, and metal transporters play central roles in maintaining metal homeostasis in the living cells. We have investigated the function of a metal transporter of the NRAMP family, AtNRAMP3, in Arabidopsis thaliana. A previous study showed that AtNRAMP3 expression is upregulated by iron (Fe) starvation and that AtNRAMP3 protein can transport Fe. In the present study, we used AtNRAMP3 promoter β-glucoronidase (GUS) fusions to show that AtNRAMP3 is expressed in the vascular bundles of roots, stems, and leaves under Fe-sufficient conditions. This suggests a function in long-distance metal transport within the plant. Under Fe-starvation conditions, the GUS activity driven by the AtNRAMP3 promoter is upregulated without any change in the expression pattern. We analyze the impact of AtNRAMP3 disruption and overexpression on metal accumulation in plants. Under Fe-sufficient conditions, AtNRAMP3 overexpression or disruption does not lead to any change in the plant metal content. Upon Fe starvation, AtNRAMP3 disruption leads to increased accumulation of manganese (Mn) and zinc (Zn) in the roots, whereas AtNRAMP3 overexpression downregulates Mn accumulation. In addition, overexpression of AtNRAMP3 downregulates the expression of the primary Fe uptake transporter IRT1 and of the root ferric chelate reductase FRO2. Expression of AtNRAMP3::GFP fusion protein in onion cells or Arabidopsis protoplasts shows that AtNRAMP3 protein localizes to the vacuolar membrane. To account for the results presented, we propose that AtNRAMP3 influences metal accumulation and IRT1 and FRO2 gene expression by mobilizing vacuolar metal pools to the cytosol.

Introduction

Transition metals fulfill crucial functions in the living cells. They act as co-factors for many enzymes such as proteases or superoxide dismutases, and catalyze electron transfer in mitochondria and chloroplasts. However, excessive accumulation of metals is potentially toxic. Therefore, the living organisms have to regulate their intracellular metal concentrations tightly. Recent research points out to critical roles of metal cation transporters in maintaining metal homeostasis (Nelson, 1999). For example, mutations in homologous copper pumps cause severe pathologies in humans (Vulpe et al., 1993) and ethylene signaling phenotype in Arabidopsis (Hirayama et al., 1999; Woeste and Kieber, 2000).

With the sequencing of the whole genomes, many putative metal-transporter genes have been identified, and their functions can now be investigated. NRAMP genes encode a family of integral membrane proteins in bacteria, fungi, plants, and animals (Cellier et al., 1995, 2001; Williams et al., 2000). Several NRAMP proteins have been shown to function as metal transporters that transport multiple classes of cations including the essential metals manganese (Mn) and iron (Fe), and the toxic metal cadmium (Cd; Agranoff et al., 1999; Chen et al., 1999; Goswami et al., 2001; Gunshin et al., 1997; Makui et al., 2000; Picard et al., 2000; Thomine et al., 2000). NRAMP proteins play central roles in metal absorption and metal homeostasis in yeast and mammalian cells (Fleming and Andrews, 1998; Nelson, 1999). For example, in mammalian cells, NRAMP1 modulates sensitivity to intracellular pathogenic bacteria by transporting Mn and Fe across the phagosomal membranes in the macrophages (Goswami et al., 2001; Gruenheid et al., 1997; Vidal et al., 1996), whereas the homologous transporter NRAMP2 mediates the dietary Fe absorption in the duodenum and Fe recycling in the red blood cells (Canonne-Hergaux et al., 1999; Fleming et al., 1998; Gruenheid et al., 1999; Gunshin et al., 1997). Defects in NRAMP1 lead to increased sensitivity to intracellular pathogens (Skamene et al., 1998; Vidal et al., 1993), and defects in NRAMP2 lead to microcytic anemia (Fleming et al., 1998).

NRAMP genes have been identified in many plant species (Belouchi et al., 1995, 1997; Mäser et al., 2001; Williams et al., 2000). In Arabidopsis, genomic analysis has revealed the existence of six sequences with strong homology to NRAMP in addition to the bifunctional regulator of ethylene responses, EIN2 (Alonso et al., 1999; Mäser et al., 2001). Functional studies have shown that AtNRAMP transporters modulate Cd and Fe toxicity in plants (Curie et al., 2000; Thomine et al., 2000). The fact that several AtNRAMP proteins can transport Fe when expressed in yeast, and that NRAMP gene expression is increased upon Fe starvation suggests that these proteins are involved in plant Fe nutrition. Genetic analyses have identified, at the molecular level, IRT1 as the transporter mediating primary uptake of Fe from the soil to the roots of strategy I plants (dicots and non-graminaceous monocots; Vert et al., 2002), and YS1 as the transporter mediating primary uptake of siderophore–Fe complexes in graminaceous species (strategy II) (Curie et al., 2001). In contrast, the presence of highly homologous NRAMP genes in both the dicots and the graminaceous species (Belouchi et al., 1997; Curie et al., 2000; Mäser et al., 2001; Thomine et al., 2000) suggests common basic functions for NRAMP proteins in cellular metal homeostasis, independent of the strategy used for Fe acquisition.

In this study, we investigated the function of AtNRAMP3 in metal homeostasis under Fe starvation. AtNRAMP3 was previously shown to mediate Fe transport when expressed in yeast, and to be upregulated in roots upon Fe starvation (Thomine et al., 2000). Here, we show that AtNRAMP3 is mainly expressed in the root stele and the vascular bundles of leaves and cotyledons. In a previous study, we showed that AtNRAMP3 modulates Cd sensitivity in planta (Thomine et al., 2000). AtNRAMP3 overexpression leads to Cd hypersensitivity, whereas AtNRAMP3 disruption slightly decreases Cd sensitivity of root growth. Here, we analyze the function of AtNRAMP3 in Fe nutrition by comparing several responses to Fe starvation between control, AtNRAMP3-overexpressing plants, and atnramp3-1 mutant, carrying a T-DNA disrupted allele of AtNRAMP3. We find that the accumulation of Mn and zinc (Zn) in response to Fe starvation is increased in the atnramp3-1 mutant, whereas it is reduced in the AtNRAMP3-overexpressing plants. Furthermore, the expression levels of the plasma membrane Fe uptake transporter, IRT1, and the root ferric chelate reductase, FRO2, as well as the root ferric chelate reductase activity are downregulated in the AtNRAMP3-overexpressing plants. The study of the subcellular localization of AtNRAMP3 protein reveals that it resides in the vacuolar membrane. A model is described in which possible functions of AtNRAMP3 in vacuolar metal release during Fe starvation are discussed.

Results

Localization and regulation of AtNRAMP3 expression

Previous data indicated that AtNRAMP3 is expressed both in roots and shoots of the Arabidopsis seedlings. To determine in which tissues AtNRAMP3 is expressed, we have analyzed the activity of the AtNRAMP3 promoter in Arabidopsis using β-glucoronidase (GUS) as a reporter gene (Jefferson et al., 1987). Arabidopsis (ecotype Columbia 0 and WS) were transformed either with a vector containing a promoter-less uidA gene or uidA gene downstream of 1889 base pairs of genomic sequence located upstream of the AtNRAMP3 initiation codon. Whereas none of the control lines showed any staining, the AtNRAMP3 promoter drove the transcription of GUS in the stele of primary and secondary roots and in the vascular bundles of leaves and stems (Figure 1). The staining initiated in the early differentiation zone of the root and became stronger in older parts of the roots (Figure 1b,c). In some cases, GUS staining was also present in the columella at the root tips (Figure 1b). Eight independent transgenic lines were analyzed, and all showed a similar pattern of expression for AtNRAMP3 with different intensities of staining. Thus, the expression of AtNRAMP3 is restricted to the vascular tissues of both roots and shoots. This localization suggests a possible function in the control of long-distance metal transport in plants. We analyzed the effect of Fe starvation on GUS activity and GUS staining localization in Arabidopsis seedlings carrying the GUS gene under the control of the AtNRAMP3 promoter in five independent transformed lines. Figure 2 shows that the GUS activity was upregulated in both the roots and the shoots under Fe starvation. The induction of AtNRAMP3 promoter activity under Fe starvation in roots was consistent with our previous Northern blot analyses. In the roots, the GUS activity increased by fourfold upon Fe starvation, whereas the induction was only 2.5-fold in the shoots. However, no change in the localization of the staining could be observed when control and Fe-starved Arabidopsis seedlings were compared (data not shown).

Figure 1.

AtNRAMP3 expression is localized in vascular tissues.

β-glucoronidase (GUS) staining of Arabidopsis seedlings expressing the uidA gene under the control of the AtNRAMP3 promoter (1.8 kb upstream of AtNRAMP3 coding sequence). Photographs by C. Belin.

(a) Expression in the vascular bundles of leaves, cotyledons, and hypocotyl.

(b) Expression in the central cylinder at the tip of the primary root.

(c) Expression in the central cylinder in older and secondary roots.

(d) AtNRAMP3 is expressed in the central cylinder, but not in cortical or epidermal cells or root hairs.

Figure 2.

The AtNRAMP3 promoter activity is upregulated upon iron (Fe) starvation in roots and shoots.

Arabidopsis seedlings were grown for 7 days on plates containing Fe-replete medium. Then, the plants were transferred either on the same medium (+Fe) or on a medium in which Fe was omitted and 100 µm Ferrozine was added (–Fe), and grown for another 7 days. β-glucoronidase (GUS) enzymatic activity was measured independently on protein extracts from roots or shoots. The results represent the mean GUS activity for five independent lines ± SD.

AtNRAMP3 regulates metal accumulation upon iron starvation

AtNRAMP3 expression is regulated by Fe deficiency, and the AtNRAMP3 protein transports Fe when expressed in yeast (Thomine et al., 2000). To investigate the role of AtNRAMP3, we analyzed the accumulation of metals in the atnramp3-1 mutant and in the AtNRAMP3-overexpressing plants under Fe starvation. As previously reported (Thomine et al., 2000), there was no significant difference in Fe, Mn, Zn, and copper (Cu) content between the control and the atnramp3-1 mutant Arabidopsis seedlings under Fe-replete conditions (Figure 3). Recently, Vert et al. (2002) have demonstrated that the IRT1 metal transporter is required for the accumulation of Mn and Zn in Arabidopsis roots upon Fe starvation. The analysis of Mn and Zn content in the atnramp3-1 mutant seedlings revealed that the accumulation of Zn and Mn under Fe starvation was significantly higher in the mutant (Figure 3b,c; n = 3; P < 0.05). Whereas control seedlings accumulated 3.3× more Mn and 3.4× more Zn upon Fe starvation, mutant seedlings accumulated 5.1× more Mn and 8.4× more Zn when compared to Fe-replete conditions. In contrast, Cu accumulation was not significantly different between the control and the atnramp3-1 mutant plants grown under Fe deficiency (Figure 3d). Fe contents were not significantly different in the control and the atnramp3-1 mutant Arabidopsis seedlings (Figure 3a).

Figure 3.

AtNRAMP3 disruption leads to enhanced manganese (Mn) and zinc (Zn) accumulation in Arabidopsis seedlings under iron (Fe) deficiency.

Fe (a), Mn (b), Zn (c), and Cu (d) contents of control (open bars) or atnramp3-1 mutant (black bars) Arabidopsis ecotype WS under Fe-replete (+Fe) or Fe-starvation (–Fe) conditions. Results are shown as mean ± SE of the three independent experiments.

Independent analysis of metal levels in the roots and the shoots confirmed the mutant phenotype, and showed that under our experimental conditions, the accumulation of Mn and Zn is restricted to the roots both in the mutant and in the wild-type control. Table 1 shows the representative results obtained from one experiment out of the two that gave qualitatively similar results.

Table 1.  Metal accumulation in shoots and roots under Fe-sufficient and Fe-deficient conditions in the control and the atnramp3-1 mutant
MetalOrgan+Fe–Fe
Controlatnramp3-1Controlatnramp3-1
  1. The numbers represent the metal content in p.p.m. (mg kg−1 dry weight (D.W.)) except bold numbers in brackets, which represent the accumulation ratio between Fe-sufficient and Fe-deficient conditions.

CuShoots11.48.812.5 (1.1)11.1 (0.9)
Roots15.210.314.8 (1)13.3 (1.3)
FeShoots424441 (1)35 (0.8)
Roots16018969 (0.4)53 (0.3)
MnShoots177156169 (1)157 (1)
Roots3082121142 (3.7)2483 (11.7)
ZnShoots10091103 (1)132 (1.5)
Roots3442391348 (3.9)2264 (9.5)

To further analyze the role of AtNRAMP3 in metal accumulation upon Fe starvation, we measured the metal content of the AtNRAMP3-overexpressing Arabidopsis seedlings grown under Fe-replete or Fe-starvation conditions. Two independent 35S-AtNRAMP3 Arabidopsis lines were analyzed to ascertain that the differences in metal accumulation observed were correlated with AtNRAMP3 overexpression. As observed with the atnramp3-1 mutant, no significant difference in Fe, Mn, and Zn content was observed between the control and the AtNRAMP3-overexpressing Arabidopsis seedlings under Fe-replete conditions (Figure 4). However, under Fe starvation, the accumulation of Mn was significantly reduced in the AtNRAMP3-overexpressing Arabidopsis seedlings in comparison with the control Arabidopsis transformed with the empty vector (Figure 4b; n = 3 for each line; P < 0.05). The control seedlings accumulated 4.4× more Mn upon Fe starvation. In contrast, the AtNRAMP3-overexpressing seedlings accumulated only 2.3× (line 1) and 1.9× (line 2) more Mn when compared to Fe-replete conditions. The average accumulation of Zn was also lower in the 35S-AtNRAMP3 lines, although the difference did not reach statistical significance (P = 0.11 and 0.12 for lines 1 and 2, respectively; Figure 4c) as a result of an overall high variability of the Zn content in the Columbia ecotype. A similar high variability in the Cu levels prevented the analysis of Cu content in these plants. Fe contents were not significantly different in the control Arabidopsis and the 35S-AtNRAMP3 lines (Figure 4a).

Figure 4.

AtNRAMP3 overexpression leads to decreased manganese (Mn) and zinc (Zn) accumulation in Arabidopsis seedlings under iron (Fe) deficiency.

Fe (a), Mn (b), and Zn (c) contents of control (open bars) or AtNRAMP3-overexpressing (hatched and black bars) Arabidopsis ecotype Columbia under Fe-replete (+Fe) or Fe-starvation (–Fe) conditions. Results are shown as mean ± SE of the three independent experiments.

Thus, the atnramp3-1 mutant and the AtNRAMP3-overexpressing plants display opposite conditional metal accumulation phenotypes upon Fe starvation. Together, these data indicate that AtNRAMP3 downregulates Mn and Zn accumulation in roots under Fe deficiency.

Constitutive ectopic expression of AtNRAMP3 downregulates the expression of the iron uptake transporter gene, IRT1, and of the ferric chelate reductase gene, FRO2

In strategy I plant species such as Arabidopsis, the uptake of Fe in the root epidermal cells includes two steps: chelated Fe(III) is first reduced to soluble Fe(II) at the root surface; Fe(II) is then taken up into the root cells by Fe transporters (Marschner and Romheld, 1994). The genes encoding the membrane-bound ferric chelate reductase enzyme, FRO2, and the root Fe uptake transporter, IRT1, have been identified (Eide et al., 1996; Robinson et al., 1999; Vert et al., 2002). Similar to AtNRAMP3, FRO2 and IRT1 are upregulated upon Fe starvation (Connolly et al., 2002; Eide et al., 1996; Robinson et al., 1999; Vert et al., 2002). We used RNA blots to quantify the expression of IRT1 and FRO2 in the roots of 35S-AtNRAMP3 plants and atnramp3-1 mutant. Under our experimental conditions, IRT1 was expressed at measurable levels, even under Fe-sufficient conditions (Figure 5). As previously described, IRT1 expression was increased upon Fe starvation (Connolly et al., 2002; Eide et al., 1996; Vert et al., 2002). Under Fe-sufficient conditions, the level of IRT1 expression was consistently lower in the 35S-AtNRAMP3 (Figure 5). IRT1 expression was reduced by 55 ± 11% (n = 3) in the 35S-AtNRAMP3 lines. Under Fe starvation, a decrease of 36 ± 9% (n = 3) was observed between the control and the 35S-AtNRAMP3 plants (Figure 5). In contrast, no reproducible change in IRT1 expression could be observed in the atnramp3-1 mutant in comparison to the control under Fe-sufficient or Fe-deficient conditions (data not shown). The expression of FRO2 followed the same trend as IRT1 (Figure 5). Under Fe-sufficient conditions, FRO2 was repressed by 40% in the 35S-AtNRAMP3 plants. As previously described, FRO2 expression was stimulated under Fe deficiency (Robinson et al., 1999), but there was no significant difference between the control and the overexpressing lines. To determine whether the decrease in FRO2 expression in the 35S-AtNRAMP3 lines could be correlated with changes in root ferric chelate reductase activity, we measured root ferric chelate reductase activity in AtNRAMP3-overexpressing plants (Yi and Guerinot, 1996). Under Fe-replete conditions, root ferric chelate reductase activity was significantly lower in AtNRAMP3-overexpressing plants when compared with the control plants transformed with the empty vector (n = 5; P < 0.05). Root ferric chelate reductase activity was reduced by 25 and 32% in the overexpressing lines 1 and 2, respectively. This result is in agreement with the downregulation of FRO2 expression observed in the 35S-AtNRAMP3 lines. In contrast, when the root ferric reductase activity of the control and the atnramp3-1 mutant Arabidopsis was compared, no significant difference was found, whether plants were grown under Fe-replete or Fe-starvation conditions.

Figure 5.

FRO2 and IRT1 expression levels are reduced in the roots of the AtNRAMP3-overexpressing plants.

Northern blot of 10 µg total RNA from roots of control or 35S-NRAMP3-1Arabidopsis ecotype Columbia grown under Fe-replete (+Fe) or Fe-starvation (–Fe) conditions were hybridized either with FRO2 (top) or IRT1 probe (middle). Hybridization with a probe for the constitutively expressed gene, tubulin 1 (bottom) was used to normalize the data.

Thus, under Fe-replete conditions, strong ectopic expression of AtNRAMP3 downregulates IRT1 and FRO2 expression and the root ferric chelate reductase activity that are activated by Fe starvation. In contrast, AtNRAMP3 disruption does not alter the activity of the root ferric chelate reductase or the IRT1 and the FRO2 expression.

NRAMP3 protein resides on the vacuolar membrane

To obtain further insight into the mechanism of AtNRAMP3 function, we investigated the subcellular localization of the AtNRAMP3 protein. For this purpose, a plant optimized green fluorescent protein (GFP; smRS-GFP) was fused to the carboxy terminus of AtNRAMP3, and placed under the control of the CaMV 35S promoter (Davis and Vierstra, 1998; Haseloff et al., 1997). The construct was transiently expressed in the onion epidermal cells by particle bombardment. Confocal imaging showed that, in contrast to free GFP or free DsRed2, the NRAMP3::GFP fusion was excluded from the nucleus (Figure 6d,g). Fluorescence intensity was sharper and lined only the side of the nucleus facing the vacuole and not the side facing the plasma membrane (>40 cells from four independent experiments, Figure 6g). This pattern was reminiscent of the immunolocalization of γ-TIP protein in barley root tip cells, suggesting vacuolar localization (Jauh et al., 1999). Closer examination at higher magnification (>25 cells from four independent experiments) and double-labeling with free DsRed2 (seven cells from one experiment) revealed that the fluorescent signal corresponding to NRAMP3::GFP lined the transvacuolar strands going through the main vacuole (Figure 6i) and was also concentrated at the periphery of the smaller vacuoles (Figure 6h). In contrast, the signal from free DsRed2 filled the cytoplasmic strands and was homogeneous in the cytosol (Figure 6e,f). To confirm the localization of the AtNRAMP3 protein in the vacuolar membrane observed in the onion epidermal cells, we expressed the AtNRAMP3::GFP fusion in protoplasts prepared from Arabidopsis cell suspensions. In this homologous expression system, the AtNRAMP3::GFP protein was also localized to the vacuolar membrane (Figure 6m–o). Altogether, our data indicate that the NRAMP3 protein is targeted to the vacuolar membrane.

Figure 6.

AtNRAMP3::GFP fusion protein is targeted to the vacuolar membrane.

(a–i) Confocal cross-sections of epidermal onion cells co-bombarded with DsRed2 and NRAMP3::GFP.

(a–c) Transmission images of one cell and two details of different cells.

(d–f) DsRed2 fluorescence (excitation 543 nm, emission measured between 555 and 620 nm) localized in the cytosol and nucleus.

(g–i) Green fluoroscent protein (GFP) fluorescence (excitation 488 nm, emission measured between 495 and 530 nm) is excluded from the nucleus and lines the vacuole indicating a vacuolar membrane localization.

(j–l) Overlay between GFP and DsRed2 fluorescence signals.

(m) Confocal cross-section through an Arabidopsis protoplast transformed with NRAMP3::GFP.

(n) Overlay: the fluorescence lines the vacuoles.

(o) Transmission image of the same protoplast.

Discussion

In this study, we investigated the localization and function of AtNRAMP3 in Arabidopsis. The AtNRAMP3 promoter–reporter gene fusion shows that AtNRAMP3 is expressed in the vascular bundles of leaves and in the central cylinder of roots, irrespective of the Fe-nutritional conditions. 1.8 kb of the AtNRAMP3 promoter is sufficient to confer upregulation of gene expression upon Fe starvation. By analyzing the metal content of an AtNRAMP3 disruption mutant and the AtNRAMP3-overexpressing plants, we demonstrated that AtNRAMP3 downregulates Mn and Zn accumulation upon Fe deficiency. Furthermore, strong ectopic expression of AtNRAMP3 downregulates the expression of the primary Fe uptake transporter, IRT1, and the root ferric chelate reductase, FRO2, indicating complex regulation among these mechanisms. Finally, using transient expression of the GFP-tagged AtNRAMP3 protein, we show that AtNRAMP3 resides on the vacuolar membrane. Taken together, our results support the hypothesis that AtNRAMP3 regulates the Fe-starvation-dependent accumulation of Mn-, Zn-, and Fe-acquisition mechanisms by modulating metal transport across the vacuolar membrane.

AtNRAMP3 metal transporter is expressed in conductive tissues

The analysis of the tissue-specific expression of the AtNRAMP3 gene shows highest levels of expression in the root stele and in the vascular bundles of leaves and stems. This result is in agreement with the previous Northern analyses showing expression of AtNRAMP3 both in the roots and in the shoots of the Arabidopsis seedlings (Thomine et al., 2000). This localization is distinct from the expression domain of the primary Fe uptake transporter, IRT1 (Vert et al., 2002), and its close homolog, IRT2 (Vert et al., 2001), which are expressed specifically in root epidermal and cortical cells, respectively. Together with the vacuolar membrane intracellular localization of the AtNRAMP3 protein, the tissue-specific expression pattern of the AtNRAMP3 gene does not support the hypothesis of a role of AtNRAMP3 in the primary uptake of metals from the soil. This localization around the vascular tissues would rather suggest a possible involvement of AtNRAMP3 in long-distance transport of metals between the root tissues, where they are stored, and the aerial tissues. However, the demonstration of this hypothesis will require further investigation.

In agreement with the data obtained from the Northern blot experiments, the GUS activity driven by the AtNRAMP3 promoter is upregulated under Fe-deficient conditions in roots (Thomine et al., 2000). The AtNRAMP3 promoter is sufficient to confer upregulation of GUS activity under Fe starvation. This is consistent with the previous findings that other Fe-regulated genes in plants, IRT1, IRT2, and AtFER, are also regulated at the transcriptional level (Petit et al., 2001; Vert et al., 2001, 2002). In contrast, in mammalian cells, the regulation by Fe is post-transcriptional: iron regulatory elements (IRE), located on the untranslated regions of the Fe-regulated mRNAs, confer increased stability to these mRNAs upon Fe deficiency (Casey et al., 1988; Klausner et al., 1993).

In shoots, the GUS activity driven by the AtNRAMP3 promoter is higher than in the roots, and is upregulated under Fe starvation. These observations do not agree with the results from Northern blot experiments showing lower levels of AtNRAMP3 in shoots than in roots and no upregulation of AtNRAMP3 mRNA levels in shoots (Thomine et al., 2000). It is possible that, in shoots, a post-transcriptional regulation downregulating AtNRAMP3 levels is superimposed on the transcriptional control by the AtNRAMP3 promoter.

Even though the AtNRAMP3 promoter GUS fusions retain induciblity under Fe starvation, the pattern of AtNRAMP3 expression is not altered under Fe starvation. Thus, the upregulation of AtNRAMP3 mRNA levels in roots is not achieved by extension of the AtNRAMP3 expression domain, but by higher expression levels in the same cell types.

AtNRAMP3 downregulates iron-starvation-induced metal accumulation and Fe acquisition mechanisms

We found that AtNRAMP3 disruption leads to an increase in Mn and Zn accumulation, specifically upon Fe starvation. In agreement with this observation, AtNRAMP3 overexpression leads to a decreased accumulation of Mn and Zn after Fe starvation. This coherent set of results excludes that AtNRAMP3 can mediate Mn and Zn accumulation upon Fe starvation. It has been demonstrated recently that IRT1 is required for the accumulation of Mn and Zn under Fe deficiency: the irt1-1 knockout mutant fails to accumulate Mn and Zn upon Fe starvation (Vert et al., 2002). It is possible that Mn or Zn can substitute for Fe in some Fe-dependent processes; however, Fe is specifically required for essential functions such as the formation of hemes and FeS clusters. Therefore, the accumulation of Mn and Zn under Fe starvation is more likely a mere consequence of the broad-metal selectivity of IRT1, rather than a compensatory process (Korshunova et al., 1999). In contrast to frd3 (man1), which displays constitutive overaccumulation of Fe, Mn, and Zn when grown under Fe-replete conditions (Delhaize, 1996; Rogers and Guerinot, 2001; Yi and Guerinot, 1996), the atnramp3-1 metal-accumulation phenotype is observed exclusively under Fe-starvation conditions. The accumulation of Fe is not altered in the AtNRAMP3 mutant or in the overexpressing lines. Likely, this is because of the fact that under Fe starvation, Fe accumulation is limited by Fe availability in the medium, and, on the other hand, our results indicate that AtNRAMP3 does not modulate metal accumulation in Fe-sufficient conditions.

Overexpression of AtNRAMP3 downregulates the level of IRT1 and FRO2 expression and root ferric chelate reductase activity. The downregulation of IRT1 under Fe starvation could account partly for the decreased accumulation of Mn and Zn in the AtNRAMP3-overexpressing lines (Vert et al., 2002). In contrast, ferric chelate reductase activity and the level of IRT1 and FRO2 expression were not significantly altered in the atnramp3-1 mutant. Two hypotheses could explain this result. First, AtNRAMP3 is expressed only in the stele, whereas IRT1 expression and Fe reduction take place in the epidermal cells. It is possible that the expression of AtNRAMP3 in the epidermal cells is necessary to alter IRT1 expression and Fe reduction. In this case, the downregulation of IRT1 and FRO2 in 35S-AtNRAMP3 lines would not reflect a physiological role of AtNRAMP3, but would result from its ectopic expression, which may lead to enhanced cellular Fe availability via vacuolar Fe release (Figure 7). Secondly, it is possible that a redundant function mediated by the close homologs of AtNRAMP3 (AtNRAMP2, -4, and -5) could mask the effect of the T-DNA insertion mutation.

Figure 7.

Model for AtNRAMP3 functions in plant cells.

This model illustrates the hypothesis that AtNRAMP3 could export metals from the vacuole. In 35S-AtNRAMP3 lines, increased Fe remobilization could account for IRT1 and FRO2 repression, and increased cadmium (Cd) re-mobilization could account for higher Cd sensitivity.

In the case of the atnramp3-1 mutant, Mn and Zn levels are increased without a significant reproducible upregulation of IRT1. It is possible that the IRT1 protein level is upregulated in the atnramp3-1 mutant without any change in the IRT1 mRNA level. Fe independently regulates the IRT1 mRNA levels and the IRT1 protein stability (Connolly et al., 2002). It is also possible that a defect in Mn and Zn transport across the vacuolar membrane links AtNRAMP3 disruption to metal accumulation via increased sequestration inside the vacuole.

NRAMP3 is a vacuolar metal transporter

The investigation of the intracellular localization of the AtNRAMP3::GFP fusion protein reveals that AtNRAMP3 resides in the vacuolar membrane. The fluorescence localized exclusively to the vacuolar membrane suggests a strong control of AtNRAMP3 vacuolar localization. In this respect, it is surprising that AtNRAMP3 was able to complement yeast mutants impaired in the plasma membrane Mn and Fe uptake (Thomine et al., 2000). However, yeast complementation required high levels of expression of AtNRAMP3, and overexpression of the NRAMP3::GFP fusion protein in yeast under the same promoter showed that the protein was targeted mainly to the vacuolar membrane and, also to a minor extent, to the plasma membrane (S. Thomine and E. Debarbieux, unpublished data). Therefore, functional heterologous expression in yeast does not necessarily provide information about the intracellular targeting of the protein in its native environment. However, we cannot exclude that AtNRAMP3 might localize to different membranes according to the physiological conditions, as reported for the NRAMP protein, SMF1, in yeast (Liu and Culotta, 1999a,b). Other non-plant members of the NRAMP family are targeted to the vacuolar membrane or to the analogous cell compartments. This is the case for yeast SMF3p (Portnoy et al., 2000) and mammalian NRAMP1, which reside in the phagosomal membrane (Gruenheid et al., 1997). In plants, previous studies have identified metal transport activities (Salt and Wagner, 1993) as well as metal transporter proteins on the vacuolar membrane. For example, CAX2, a member of the calcium proton exchanger family, transports Ca and Mn into the vacuole (Hirschi et al., 2000) and AtMHX, a magnesium (Mg) proton exchanger transports Mg and Zn into the vacuoles (Shaul et al., 1999). In addition, it has been proposed that ZAT1, a protein belonging to the cation diffusion facilitator (CDF) family, sequesters Zn into the vacuole (van der Zaal et al., 1999).

AtNRAMP3 may control metal accumulation and IRT1 expression by mobilizing vacuolar metal pools

To account for the data presented, we propose the hypothesis that AtNRAMP3 functions in the export of metals from the vacuolar compartment to the cytosol. Indeed, increased AtNRAMP3-mediated Fe transport in 35S-AtNRAMP3 lines would result in increased Fe mobilization from the vacuoles (Figure 7). The increase in available Fe in the cytoplasm could, in turn, downregulate IRT1 and FRO2 expression (Figure 7), and therefore, reduce metal accumulation upon Fe starvation. The hypothesis that AtNRAMP3 mediates intracellular Fe transport is compatible with the observation that AtNRAMP3 overexpression leads to alteration of IRT1 and FRO2 expression without changes in the Fe content of the whole plant (Figure 3). In the atnramp3-1 mutants, a defect in metal export from the vacuole could lead to increased vacuolar sequestration and accumulation of Mn and Zn under Fe starvation. Furthermore, mobilization of vacuolar Cd by AtNRAMP3 could account for the previously reported increase in Cd sensitivity in the AtNRAMP3-overexpressing lines (Figure 7) without change in the Cd content of the whole plant (Thomine et al., 2000).

A similar function in Zn mobilization from the vacuolar compartment has been demonstrated for the yeast Zn transporter, ZRT3 (MacDiarmid et al., 2000). A function in vacuolar Fe mobilization has also been proposed for the yeast NRAMP homolog, SMF3 (Portnoy et al., 2000, 2002) and the yeast high-affinity Fe-transport system, Fth1/Fet5 (Urbanowski and Piper, 1999).

Finally, we propose that AtNRAMP3 controls Fe-acquisition genes, Mn and Zn accumulation upon Fe starvation, and Cd sensitivity by mobilizing Fe, Cd, and other metals from the vacuolar compartment. Further research will be needed to directly test the hypothesis that AtNRAMP3 drives metal efflux from the vacuole as well as to determine the physiological relevance of mobilization of the vacuolar pools of metals.

Experimental procedures

Plant material

The isolation of the insertion mutant in AtNRAMP3 and the construction of the overexpressing lines have been described previously (Thomine et al., 2000). Briefly, atnramp3-1 was obtained by PCR screening on genomic DNA from 14 000 lines of Arabidopsis thaliana ecotype WS, containing random T-DNA insertions. The T-DNA insertion that we identified in AtNRAMP3 gene is located in the last exon between GLU467 and VAL468. The overexpressing lines have been constructed by transformation with the pMON530 vector containing AtNRAMP3 cDNA under the control of the CaMV 35S promoter.

Promoter β-glucuronidase fusions, β-glucuronidase activity measurements, and histologic analysis

The genomic sequence upstream of the AtNRAMP3 initiation codon was amplified by PCR from the BAC T20D16 using Advantage cDNA Taq polymerase (BD Bioscience Clontech, Palo Alto, CA, USA) and the primers ProNR1S (5′-CTT ATT TAC AAT GGT TGC CAC TGG-3′) and ProNR1R (5′-GGA TTC TCT GTT TCA GTT TCG GG-3′). The PCR fragment was cloned in pGEMTeasy (Promega, Madison, WI, USA), and was subcloned at the EcoR1 site in pCAMBIA 1381Z. The constructs in which the AtNRAMP3 promoter was driving the expression of the uidA gene were used in parallel with empty pCAMBIA 1381Z for Agrobacterium tumefaciens-mediated Arabidopsis transformation. Transgenic lines were obtained on the Columbia and WS backgrounds, and eight independent lines were analyzed.

For histologic analysis, the Arabidopsis seedlings were grown on plates containing 2.5 mm H3PO4, 5 mm KNO3, 2 mm MgSO4, 1 mm Ca(NO3)2, 50 µm FeEDTA, Murashige and Skoog microelements, 1% sucrose, 1% BactoAgar, and 1 mm MES adjusted with KOH to pH 6.1 (Fe-replete medium). After 12 days, the seedlings were prefixed in 90% acetone for 20 min, washed briefly in GUS buffer (50 mm NaPO4, 2 mm ferrocyanide, 2 mm ferricianide, 0.2% Triton X-100, pH 7.2) and then incubated for 10 h in GUS buffer with 2 mm GUS substrate X-Gluc (5-bromo-4-chloro-3-indolyl-β-O-glucopyranoside) at 37°C. The seedlings were then incubated in ethanol for de-staining, counter-stained with 95% ethanol containing eosine, and rinsed again with ethanol 100% before observation. For analysis of the expression pattern under Fe starvation, the plants were grown vertically on plates containing Fe-replete medium for 10 days. Then, the plants were transferred to plates containing either the same medium (+Fe) or a medium in which Fe was omitted and 100 µm Ferrozine was added (–Fe), and grown for another 4 days before GUS staining.

For GUS activity assays, five independent hygromycine-resistant T2 lines were grown for 7 days on plates containing Fe-replete medium, and then either on the same medium (+Fe) or on a medium in which Fe was omitted and 100 µm Ferrozine was added (–Fe), and grown for another 7 days. Roots and shoots were harvested separately and ground in 0.3 and 0.6 ml of GUS extraction buffer for roots and shoots, respectively (Jefferson et al., 1987). GUS activity was measured using 1 mm 4-methylumbeliferyl-β-d-glucuronide as a substrate, and the formation of the fluorescent product methyl-umbeliferone was quantified by fluorimetry (Jefferson et al., 1987). The GUS activity was normalized to the protein concentration in the extract determined according to Bradford (1976).

Metal ion content measurements

Two hundred seeds were germinated in a flask containing 50 ml of liquid Murashige and Skoog medium supplemented (–Fe) or not (+Fe) with 1 mm Ferrozine, and grown for 10 days on a shaker at 100 r.p.m. The plants were harvested by vacuum filtration and washed with 10 ml CaCl2 (0.1 m) and 10 ml EDTA (10 mm). For experiments in which the roots and the shoots were analyzed independently, the seedlings were grown vertically for 12 days on plates containing the same media with 1% agar. The roots and the shoots were excised and washed with 10 ml CaCl2 (0.1 m) and 10 ml EDTA (10 mm). The dry weight (D.W.) of the samples was measured after drying overnight at 60°C. Subsequently, the samples were digested in 0.2–0.4 ml of 70% HNO3 (trace metal grade, Fisher Scientific, Hampton, NH, USA) for 3 days, and complete digestion was achieved by heating the samples to 100°C for 30 min. After dilution, the metal content of the samples were determined by Inductively Coupled Plasma Optical Emission Spectroscopy (Perkin Elmer Optima 3000LX; Applied Biosystem, Foster City, CA, USA). The results are given in p.p.m. (mg metal kg−1 D.W.). A Student's t-test was performed to evaluate the significance of the differences observed.

Root ferric reductase activity

The root ferric reductase activity was measured as described by Yi and Guerinot (1996). Briefly, the Arabidopsis seedlings were grown vertically on plates containing 2.5 mm H3PO4, 5 mm KNO3, 2 mm MgSO4, 1 mm Ca(NO3)2, 50 µm FeEDTA, Murashige and Skoog microelements, 1% sucrose, 1% BactoAgar, and 1 mm MES adjusted with KOH to pH 6.1. After 10 days, the plants were transferred either onto the same medium (+Fe) or onto a medium in which Fe was omitted and 100 µm Ferrozine was added (–Fe). After another 4 days, the roots of the 10–20 seedlings were washed in distilled water and incubated for 90 min in a reaction medium containing 100 µm FeEDTA and 300 µm Ferrozine. The amount of reduced Fe was calculated by the measurement of the absorbance of the reaction mixture at 562 nm using an extinction coefficient of 28.6 mm−1cm−1 for the Fe(II)–ferrozine complex.

Gene expression

The Arabidopsis seedlings were vertically grown for 10 days on plates containing Murashige and Skoog medium with 1% sucrose and 1% Agar A (Sigma, St Louis, MI, USA; Figure 5) or in the conditions described above in the ‘root ferric reductase activity’ section (Figure 6). Total RNA was extracted from about 60 seedlings using the Trizol reagent (Invitrogen, Life Technologies, Carlsbad, CA, USA) or the Plant RNA Mini kit (Qiagen, Hilden, Germany). For Northern blots, 10–25 µg total RNA was separated on denaturing 1.2% formaldehyde agarose gel, transferred to a Hybond-N nylon membrane (Amersham Biosciences, Piscataway, NJ, USA) or Positive TM Membrane (Q-BIOgene, Montreal, Canada), and hybridized with probes synthetized from full-length NRAMP3, IRT1, or FRO2 cDNAs, and actin2 or tubulin1 cDNA for loading controls. Relative abundances of the transcripts were calculated as the ratio between the hybridization intensity for the gene of interest versus the intensity for the loading control gene.

GFP fusion and intracellular localization

To construct a cDNA encoding a translational fusion between AtNRAMP3 and the GFP, RSsm-GFP (Davis and Vierstra, 1998; Haseloff et al., 1997), AtNRAMP3 cDNA was amplified with the oligonucleotides AtNRAMP3 BamH1S (5′-CGG GAT CCA TGC CAC AAC TCG AGA AC-3′) and AtNRAMP3 BamH1R (5′-CGG GAT CCA ATG ACT AGA CTC CGC-3′), which remove the stop codon and introduce BamH1 restriction sites. AtNRAMP3 was subsequently cloned at the BamH1 site on the plant transient expression vector CD3-327 (http://weedsworld.arabidopsis.org.uk/vol3ii/sol-modGFP.html) to generate a fusion with the GFP as an additional carboxy terminal domain under the control of the CaMV 35S promoter. The AtNRAMP3::GFP fusion construct or the empty vector CD3-327, or the pOL-DsRed2 vector were transiently expressed in the onion epidermal cells by biolistic bombardment or in the protoplast from Arabidopsis cell suspensions by PEG-mediated transformation. For onion bombardment, gold microcarriers (1.06 µm) coated with plasmid DNA (1.7 µg mg−1 microcarrier) were bombarded on onion scales at a distance of 6 cm using the PSD-1000/He instrument (Bio-Rad, Hercules, CA, USA) with 650 psi He pressure under 28 in. Hg vacuum. The onions were then incubated under high humidity in the dark at 21°C. After 24–48 h, epidermal peels were detached from the onion and mounted in water for microscopic observation. For protoplast transformation, Arabidopsis suspension cells were digested in Gamborg's B5 medium supplemented with 0.17 m glucose, 0.17 m mannitol, 1% cellulase, and 0.2% macerozyme. The protoplasts were purified by floatation in Gamborg's B5 medium supplemented with 0.28 m sucrose. For transformation, 0.2 million cells were mixed with 5 µg of plasmid DNA in a solution containing PEG 6000 25%, mannitol 0.45 m, Ca(NO3)2 0.1 m (pH 9), and incubated in the dark for 20 min. Then, the PEG was washed twice with 0.275 m Ca(NO3)2, and the protoplasts were re-suspended in Gamborg's B5 medium supplemented with 0.17 m glucose, 0.17 m mannitol in which they were maintained until microscopic observation. Fluorescent cells were imaged by confocal microscopy (Leica TCS SP2, Leica Microsystem, Wetzlar, Germany) with excitation at 488 nm, and the fluorescence emission signal was recovered between 495 and 530 nm for GFP and GFP fusion, and excitation at 543 nm and emission signal recovered between 555 and 620 nm for DsRed2.

Acknowledgements

We thank E. Rogers and M.L. Guerinot for the gift of IRT1 and FRO2 cDNAs, Ian Small and Claire Lurin for the gift of the pOL-DsRed2 plasmid, and Carmela Giglione for the gift of the CD3-327 GFP plasmid. We wish to thank Celine Charon for help with the biolistic particle gun, Carine Alcon for help with the Arabidopsis protoplast transformation, and Spencer Brown for his invaluable help with the confocal microscopy. We thank the Scripps Institution of Oceanography elemental analysis facility for use of the ICP spectrometer and the Cell Biology facility of the IFR87 for use of the confocal microscope. We thank C. Curie and G. Vert for critical reading of the manuscript. This work was supported by a Human Frontier Science Program Fellowship to S.T. by N.I.E.H.S. grant 1P42-ESI0337 to J.I.S., and by the Centre National de la Recherche Scientifique.

Ancillary