NRAMP genes function in Arabidopsis thaliana resistance to Erwinia chrysanthemi infection


(fax +33 1 44 08 16 31; e-mail


AtNRAMP3 and AtNRAMP4 are two Arabidopsis metal transporters sharing about 50% sequence identity with mouse NRAMP1. The NRAMP1/Slc11A1 metal ion transporter plays a crucial role in the innate immunity of animal macrophages targeted by intracellular bacterial pathogens. AtNRAMP3 and AtNRAMP4 localize to the vacuolar membrane. We found that AtNRAMP3 is upregulated in leaves challenged with the bacterial pathogens Pseudomonas syringae and Erwinia chrysanthemi, whereas AtNRAMP4 expression is not modified. Using single and double nramp3 and nramp4 mutants, as well as lines ectopically expressing either of these genes, we show that AtNRAMP3 and, to a lesser extent, AtNRAMP4 are involved in Arabidopsis thaliana resistance against the bacterial pathogen E. chrysanthemi. The susceptibility of the double nramp3 nramp4 mutant is associated with the reduced accumulation of reactive oxygen species and ferritin (AtFER1), an iron storage protein known to participate in A. thaliana defense. Interestingly, roots from infected plants accumulated transcripts of AtNRAMP3 as well as the iron-deficiency markers IRT1 and FRO2. This finding suggests the existence of a shoot-to-root signal reminiscent of an iron-deficiency signal activated by pathogen infection. Our data indicate that the functions of NRAMP proteins in innate immunity have been conserved between animals and plants.


During microbial infection, there is aggressive competition between the host and the microorganism for essential nutritional resources. As an essential metal for most forms of life, iron can play a critical role in such competitive relationships. Owing to its ability to engage in one-electron oxidation reactions between the ferric (Fe3+) and the ferrous (Fe2+) states, iron is a privileged co-factor for a variety of proteins mediating redox and electron transfer reactions. In the presence of oxygen, Fe3+ and Fe2+ can catalyze the formation of highly reactive oxygen radicals through the Fenton/Haber-Weiss reactions (Pierre and Fontecave, 1999). These radicals can trigger chain reactions, resulting in protein denaturation, DNA damage and lipid peroxidation. In addition to its high reactivity, iron is poorly soluble under aerobic, aqueous and neutral pH conditions, which means that iron is rarely present as a free element in biological tissues. Therefore, pathogens use specialized iron uptake systems allowing them to acquire this vital element from host tissues. Reciprocally, hosts express innate immune responses aimed at depriving the invader of nutritional iron. The host can also use iron as a modulator of the oxidative burst (Zwilling et al., 1999; Schaible and Kaufmann, 2004; Ong et al., 2006).

The natural resistance-associated macrophage proteins (NRAMPs) form a highly conserved family of integral membrane proteins that are involved in divalent metal transport. NRAMP genes have been identified in a wide range of organisms, including bacteria, fungi, plants and animals (Cellier et al., 1996). The first member of this family, NRAMP1 (SLC11A1), has been characterized in mice as a resistance gene to intracellular pathogens such as Leishmania, Mycobacterium and Salmonella (Vidal et al., 1993). Numerous studies have established that NRAMP1 is a key component of the host resistance in response to intracellular bacterial infection in mammals (Forbes and Gros, 2001). In infected macrophages, NRAMP1 is rapidly recruited to the membrane of maturing phagosomes (Gruenheid et al., 1997), and mutations in this transporter lead to increased sensitivity to bacterial infection (Vidal et al., 1995). NRAMP1 regulates the concentrations of essential divalent metals such as Fe2+ and Mn2+ in the macrophage compartments, but the exact function of NRAMP1 in host resistance is still a matter of debate. DCT1/DMT1/NRAMP2 (SLC11A2), the second NRAMP gene identified in mammals, is an H+/divalent metal symporter involved in intestinal iron absorption, and in iron release from endosomes, during the transferrin cycle (Gunshin et al., 1997; Canonne-Hergaux et al., 1999; Gruenheid et al., 1999).

Over the past 10 years, NRAMP genes have been identified in various plant species (Belouchi et al., 1997; Curie et al., 2000; Thomine et al., 2000; Bereczky et al., 2003; Kaiser et al., 2003; Mizuno et al., 2005). The AtNRAMP family in Arabidopsis thaliana is represented by six members, in addition to EIN2 (Alonso et al., 1999; Maser et al., 2001), sharing between 40 and 50% amino acid sequence identity with mouse NRAMP1. Most AtNRAMPs complement yeast mutants deficient for iron or manganese uptake, revealing their conserved function as metal transporters among both the plant and animal kingdoms (Curie et al., 2000; Thomine et al., 2000). The close homologs AtNRAMP3 and AtNRAMP4 accumulate in response to iron deficiency in both roots and aerial parts of Arabidopsis plants, and localize to the vacuolar membrane, indicating a function in intracellular metal homeostasis (Thomine et al., 2003; Lanquar et al., 2004, 2005). Using a double knock-out mutant, AtNRAMP3 and AtNRAMP4 were shown to act redundantly to mobilize iron from vacuolar globoids during early seedling development (Lanquar et al., 2005). To date, studies on NRAMP functions in plants mainly focused on their roles in nutrition. Unlike for animals, there have been no reports on their putative functions in plant innate immunity.

In addition to specialized responses following pathogen attack, plants and animals have developed common innate defenses, such as the rapid accumulation of reactive oxygen species (ROS) (Wojtaszek, 1997; Gloire et al., 2006) and the production of antimicrobial peptides (Broekaert et al., 1995; Medzhitov and Janeway, 1998). As iron plays key roles in animal resistance to bacterial pathogens, it is worthwhile investigating whether the role of iron in host resistance is conserved in plants. Consistent with this hypothesis, recent studies have emphasized the importance of changes in plant iron trafficking during the establishment of plant innate immunity (Dellagi et al., 2005; Liu et al., 2007). Control of iron homeostasis is a critical factor for the phytopathogenic enterobacterium Erwinia chrysanthemi (syn. Dickeya dadantii) that causes soft-rot disease in A. thaliana. During the infection process E. chrysanthemi multiplies within intercellular spaces, and produces several plant cell wall degrading enzymes, causing systemic rotting symptoms on a wide range of plants (Franza et al., 2005; Fagard et al., 2007). Survival of E. chrysanthemi in planta requires the production of two high-affinity iron acquisition systems mediated by two siderophores. Interestingly, this bacterium is able to modulate its virulence in relation to its iron requirements. Expression of several genes necessary for successful infection, including the genes involved in siderophore-mediated iron uptake, and those encoding pectin-degrading enzymes, are controlled by the availability of iron (Expert, 1999; Franza et al., 2002, 2005). In response to E. chrysanthemi infection, A. thaliana expresses several defense responses, including the production of an oxidative burst, and the activation of the salicylic acid (SA), jasmonic acid (JA) and ethylene (ET) pathways, which are the three major signaling pathways involved in the plant immune network (Fagard et al., 2007). In addition to these defense responses, A. thaliana synthesizes the ferritin AtFER1, an iron storage protein required for Arabidopsis resistance to E. chrysanthemi infection. Interestingly, the upregulation of AtFER1 gene expression during infection is mediated by the bacterial siderophores. It was thus suggested that these iron sequestering molecules could cause severe iron depletion in Arabidopsis leaf tissues. This local iron deficiency could lead to a redistribution of intracellular iron, resulting in AtFER1 upregulation (Dellagi et al., 2005; Boughammoura et al., 2007).

The data obtained in animal models, and the recent evidence supporting the role of iron in plant resistance, prompted us to address the possible involvement of Arabidopsis NRAMP genes in basal resistance to bacterial pathogens. To focus our work on the relevant AtNRAMP gene members, we examined the microarray data available. Among the six AtNRAMP genes, only AtNRAMP3 was found to be strongly upregulated in response to several biotic stresses in the microarray data (; Zimmermann et al., 2004). Because Lanquar et al. (2005) found a functional redundancy between AtNRAMP3 and AtNRAMP4 in seed germination, we analyzed the role of both genes in plant resistance to pathogens. In this work, we unravel the role of AtNRAMP3 and AtNRAMP4 in Arabidopsis interaction with E. chrysanthemi. We show that the AtNRAMP3 and AtNRAMP4 genes are involved in a basal level of resistance of Arabidopsis to E. chrysanthemi, and that they contribute to the amplification of several plant defense reactions. Furthermore, we provide evidence that this bacterium is able to trigger local and long-distance signaling, leading to the activation of several iron-deficiency responses in the plant.


AtNRAMP3 is induced in response to bacterial infection

To test whether AtNRAMP3 and AtNRAMP4 are involved in plant response to bacterial attack, we analyzed AtNRAMP3 and AtNRAMP4 gene expression levels in Arabidopsis after infection. According to the microarray data, both the necrotrophic fungal pathogen Botrytis cinerea and the phytopathogenic bacterium Pseudomonas syringae pv. tomato (P.s.t.) DC3000 appeared to upregulate AtNRAMP3 expression in Arabidopsis leaves, whereas the expression of AtNRAMP4 was unaltered (;Zimmermann et al., 2004). In agreement with the microarray data, we found no modification in AtNRAMP4 gene expression after P.s.t. DC3000 treatment, whereas AtNRAMP3 was upregulated in infected leaves at 3, 7 and 24 h post-infection (p.i.) (Figure 1a). These data indicate that AtNRAMP3 could be involved in the Arabidopsis interaction with P.s.t. DC3000. We wondered whether E. chrysanthemi infection modifies AtNRAMP3 or AtNRAMP4 expression. Leaves from the wild-type ecotypes Wassilewskija (Ws) and Columbia (Col-0) were harvested at 8, 16 and 24 h p.i. Following bacterial challenge, AtNRAMP3 expression in Ws was strongly upregulated at 16 and 24 h p.i. (Figure 1b), whereas AtNRAMP4 transcripts either constitutively accumulated at a steady-state level or were undetected, depending on the experiment. Similar expression profiles were observed in Col-0 (data not shown). Thus, both P.s.t. DC3000 and E. chrysanthemi infections trigger a signal leading to the upregulation of AtNRAMP3 expression, but have no effect on AtNRAMP4 expression.

Figure 1.

 Expression of AtNRAMP3 and AtNRAMP4 genes in response to infection by Erwinia chrysanthemi and Pseudomonas syringae (P.s.t. DC3000).
(a and b) Northern blot analysis of AtNRAMP3 and AtNRAMP4 expression in leaves infiltrated with 10 mm MgSO4 (mock), P.s.t. DC3000 or with E. chrysanthemi 3937 ( at the indicated times.
(c) GUS staining of Arabidopsis leaves expressing the uidA gene under the control of the AtNRAMP3 or AtNRAMP4 promoter. Leaves were harvested at the indicated times after inoculation. Experiments in a–c were performed three times with similar results.

To further document AtNRAMP3 and AtNRAMP4 gene expression under E. chrysanthemi infection conditions, we analyzed the expression of AtNRAMP3 and AtNRAMP4 promoter fusions. We used two transformed lines of the Ws ecotype containing the uidA gene fused downstream of the AtNRAMP3 promoter (proAtNRAMP3::GUS) (Thomine et al., 2003), and two transformed Ws lines harbouring the uidA gene fused downstream of the AtNRAMP4 promoter (proAtNRAMP4::GUS) (Lanquar et al., 2005). GUS staining was monitored at 16 and 24 h p.i. In agreement with the accumulation of AtNRAMP3 transcripts observed by northern blot hybridization, the AtNRAMP3 promoter drove the transcription of the gene fusion in the zone infiltrated with bacteria at 16 h p.i. (Figure 1c). At 24 h p.i., the GUS fusion was also upregulated in the non-infiltrated part of the leaf. Similar results were obtained with both proAtNRAMP3-GUS lines tested. These data indicate that the AtNRAMP3 promoter is sufficient to confer upregulation of GUS expression in response to pathogen attack. They also show that the signal leading to AtNRAMP3 upregulation spreads from the infected zone to healthy parts of the leaf. In contrast, the expression of the uidA gene fused to the AtNRAMP4 promoter was the same in mock- and bacteria-treated leaves (Figure 1c) in both proAtNRAMP4-GUS lines tested. These results confirm that AtNRAMP4 transcription is not induced by the bacteria. In order to determine whether the level of AtNRAMP4 protein was modified upon infection, we performed western blots with leaf proteins harvested 10 and 24 h p.i. (Figure 2a) using an anti-AtNRAMP4 polyclonal antibody (Lanquar et al., 2004). No difference in the protein level was observed between control and infected leaves, whereas, as expected, iron starvation resulted in an accumulation of AtNRAMP4 in leaves and roots (Figure 2).

Figure 2.

 AtNRAMP4 protein level in infected Arabidopsis plants.
AtNRAMP4 accumulation was monitored by western blot with an anti-AtNRAMP4 polyclonal antibody. A total of 15 μg of proteins extracted from leaves or roots harvested at the indicated times after treatment with 10 mm MgSO4 (mock), or with Erwinia chrysanthemi ( were used in each lane: nr3nr4, nramp3 nramp4 double knock-out. Controls are untreated iron-sufficient (+Fe) or iron-starved (−Fe) plants. Data are representative of three biological replicates.

AtNRAMP3 and AtNRAMP4 genes are involved in a basal level of resistance of Arabidopsis to E. chrysanthemi

In order to investigate the possible involvement of AtNRAMP3 and AtNRAMP4 genes in plant resistance or susceptibility to E. chrysanthemi infection, we used A. thaliana T-DNA insertion lines harboring single mutations in AtNRAMP3 (Thomine et al., 2000) and AtNRAMP4 (Lanquar et al., 2004) genes. We compared the susceptibility of these nramp mutants with that of the wild-type ecotype Ws. The nramp3 mutant showed a significantly increased susceptibility to E. chrysanthemi (< 0.04; Figure 3b). The nramp4 single mutant did not show significantly different susceptibility.

Figure 3.

 Involvement of AtNRAMP3 and AtNRAMP4 genes in Arabidopsis resistance to Erwinia chrysanthemi.
(a) Left panel: healthy leaf; right panels: typical symptoms reported.
(b and c) Disease severity in Arabidopsis thaliana plants, expressed as the number of plants harboring the symptoms shown in (a) 3 days after infection. The data represent the mean values from three independent biological replicates, performed with 24 plants each (error bars: SD).
nr3, single nramp3 mutant; nr4, single nramp4 mutant; nr3nr4, nramp3 nramp4 double knock-out; 35S-NR3-1 and 35S-NR3-2, two independent AtNRAMP3 overexpressing lines; 35S-NR4-1 and 35S-NR4-2, two AtNRAMP4 overexpressing lines; control, empty vector transformed plants; nr3nr4::NR3 and nr3nr4::NR4, double nramp mutants complemented with AtNRAMP3 or AtNRAMP4, respectively, with their own promoter.

To investigate the possible compensatory mechanisms of AtNRAMP3 and AtNRAMP4 genes, we analyzed their expression profiles in nramp3 or nramp4 single mutants, and in AtNRAMP3 or AtNRAMP4 overexpressing lines. At 24 h p.i., the AtNRAMP3 mRNA accumulation was similar in the wild-type, in the nramp4 mutant and in the 35S-AtNRAMP4 plants (Figure 4a). Conversely, disruption or overexpression of AtNRAMP3 did not alter the expression profile of AtNRAMP4 at 24 h p.i. (Figure 4b). Thus, AtNRAMP3 and AtNRAMP4 are differentially and independently regulated in response to pathogen attack. This result indicates that there is no compensatory mechanism between these two genes.

Figure 4.

AtNRAMP3 and AtNRAMP4 gene expression in AtNRAMP mutants and overexpressing lines
Northern blot analysis of AtNRAMP3 and AtNRAMP4 expression in leaves inoculated with 10 mm MgSO4 (mock) or with Erwinia chrysanthemi ( nr3, single nramp3 mutant; nr4, single nramp4 mutant; 35S-NR3 and 35S-NR4, AtNRAMP3 and AtNRAMP4 overexpressing lines, respectively. Experiments were repeated three times with similar results.

As AtNRAMP3 and AtNRAMP4 have been shown to function redundantly during seed germination (Lanquar et al., 2005), we tested the susceptibility of the nramp3 nramp4 double knock-out mutant to bacterial infection. The spreading of the symptoms was faster and the number of systemic infections was significantly higher in the double mutant line than in the wild type (< 0.01; Figure 3b). These results show that functional AtNRAMP3 and AtNRAMP4 genes are required to delay the initiation of the symptomatic phase of the disease. The more pronounced phenotype of the double mutant compared with that of the single mutants suggests that the functions of these genes in plant resistance are partly redundant or additive. To confirm that the observed phenotype was caused by the disruption of both transporters, we tested whether the double mutant susceptibility could be rescued by transformation with AtNRAMP3 or AtNRAMP4 genes harboring their own promoter. The nramp3 nramp4::NR3 complemented lines displayed a significantly reduced number of symptoms compared with the double mutant (< 0.001; Figure 3b), indicating that AtNRAMP3 is involved in Arabidopsis defense against E. chrysanthemi. The higher level of resistance of the nramp3 nramp4::NR3 complemented line compared with the nramp4 mutant may be explained by the insertion of several copies of the AtNRAMP3 transgene, and a positional effect of the T-DNA insertions in the vicinity of enhancers. The nramp3 nramp4::NR4 complemented lines did not display a significantly reduced number of symptoms compared with the double mutant.

To further document the role of AtNRAMP3 and AtNRAMP4 in pathogen susceptibility, we investigated whether AtNRAMP3 or AtNRAMP4 overexpressing lines were more resistant to E. chrysanthemi infection. We used the wild-type ecotype Col-0 transformed with the empty vector as a control, and two independent lines overexpressing either AtNRAMP3 (Thomine et al., 2000) or AtNRAMP4 (Lanquar et al., 2004). The number of macerated plants in both 35S::AtNRAMP3 lines tested was strongly and significantly reduced compared with the control line (< 0.001; Figure 3c). Three Ws lines overexpressing AtNRAMP3 displayed similar high levels of resistance (data not shown). No significant difference was observed between the number of infected plants in the control and the 35S::AtNRAMP4-1 line plants, whereas a significantly lower susceptibility was observed for the 35S::AtNRAMP4-2 line (P < 0.05). Taken together, these results indicate that AtNRAMP3 and, to a lesser extent, AtNRAMP4 are involved in resistance to E. chrysanthemi.

To further analyze the increased susceptibility of the double mutant, we tested whether it correlated with an enhanced bacterial growth in plant tissues. Thus, we analyzed the growth of E. chrysanthemi in inoculated leaves of Ws or nramp3 nramp4 plants (Figures S1 and 5a). No difference between the two genotypes was observed at 8 h p.i. After 24 h, an increase in bacterial populations occurred in both Ws and the nramp3 nramp4 double mutant plants, but this increase was more important in the double mutant. Finally, at 48 h p.i., we observed a decrease in bacterial counts in the wild-type ecotype, whereas the bacteria were still multiplying in the nramp3 nramp4 mutant. Thus, the enhanced susceptibility of the double mutant is associated with an increased bacterial growth in leaves. Together, our data indicate that AtNRAMP3 and AtNRAMP4 contribute to the limitation of bacterial growth in the first steps of infection.

We analyzed the susceptibility of the nramp3 nramp4 mutant to two pathogens that induce AtNRAMP3 expression, P.s.t. DC3000 and B. cinerea. We observed no difference in the susceptibility of the double mutant in response to these fungal and bacterial challenges (data not shown). Furthermore, bacterial growth of P.s.t. DC3000 was similar in the wild-type and in the double mutant leaves (Figure 5b). These results indicate that the role of AtNRAMP3 and AtNRAMP4 in Arabidopsis resistance is restricted to a limited group of pathogens.

Figure 5.

 Bacterial growth in nramp3 nramp4 double mutants.
(a and b) Bacterial growth of Erwinia chrysanthemi 3937 and Pseudomonas syringae pv. tomato (P.s.t.) DC3000, respectively, monitored in Ws and nramp3 nramp4 mutant (nr3nr4) plants. Values represent the average of three replicates ± SD (see Experimental procedures). Experiments were repeated three times with similar results.

The upregulation of AtNRAMP3 gene expression is independent of the plant defense signals triggered by E. chrysanthemi infection

We then sought to determine whether AtNRAMP3 induction is controlled by signaling pathways that are known to be triggered during pathogen attack. For this purpose, we monitored AtNRAMP3 expression in response to E. chrysanthemi in A. thaliana mutants defective in SA-, JA- or ET-dependent responses. We used the sid2 mutant that is impaired in the ICS1 gene required for the major part of SA signaling in the pathosystem used (Nawrath and Metraux, 1999; Fagard et al., 2007), the ein2 mutant that no longer expresses a transmembrane protein required for ET perception (Guzman and Ecker, 1990; Alonso et al., 1999), and the jar1 mutant that fails to produce an active form of JA (Staswick et al., 1992, 2002). In these different mutants, we observed that AtNRAMP3 expression was still upregulated 24 h p.i. with E. chrysanthemi (Figure 6). Therefore, functional SA, JA and ET signaling pathways are not required for the upregulation of AtNRAMP3 gene expression in the leaves, in response to bacterial infection.

Figure 6.

AtNRAMP3 expression in Arabidopsis mutants affected in the salicylic acid (SA), jasmonic acid (JA) and ethylene (ET) signaling pathways.
Expression profiles of AtNRAMP3 in Col-0 and in defense signaling mutants, as indicated, analyzed by northern blot hybridization. RNAs were isolated from leaves harvested 24 h after the indicated treatments. Representative data of three independent RNA extractions are shown.

Defense responses mediated by SA, JA or ET signaling pathways are not attenuated in the nramp3 nramp4 mutant

To test whether the defense responses mediated by SA, JA or ET signaling pathways were modified in the nramp3nramp4 mutant, we studied the expression patterns of AtERF1 and CHI-B, two genes synergistically induced in JA and ET signaling pathways (Lorenzo et al., 2003), and PR1, as a representative gene of the SA-mediated signaling pathway (Glazebrook et al., 2003). The expression of the different marker genes was very weak at 10 h p.i. in both the nramp3 nramp4 mutant and in wild-type Ws. After 24 h, we found that CHI-B, AtERF1 and PR1 gene expression was upregulated in Ws and in the nramp3 nramp4 double mutant after bacterial infection. We noted that CHI-B and AtERF1 induction upon infection was stronger in the double mutant compared with the wild type. Furthermore, we also detected AtERF1 and PR1 transcripts in mock-treated leaves of the nramp3 nramp4 mutant (Figure 7). These data indicate that the defense responses mediated by SA, JA or ET are constitutively activated, or that their activation is enhanced in nramp3 nramp4 mutants compared with wild-type plants.

Figure 7.

 Expression of defense-related marker genes of the salicylic acid (SA), jasmonic acid (JA) and ethylene (ET) signaling pathways in the nramp3 nramp4 mutant.
RT-PCRs were performed with RNAs extracted at the indicated times from Ws or nramp3 nramp4 double mutant (nr3nr4) leaves inoculated with 10 mm MgSO4 (mock) or with Erwinia chrysanthemi ( The constitutive EF1α gene is used as a control. Representative data of three independent RNA extractions are shown.

Infection with E. chrysanthemi triggers the expression of several iron-deficiency genes in roots

Under iron starvation, AtNRAMP3 and AtNRAMP4 are upregulated both in shoots and roots (Thomine et al., 2003; Lanquar et al., 2005). As E. chrysanthemi infection leads to the spreading of AtNRAMP3 upregulation from the infected zone to healthy parts of the leaf (Figure 1c), we tested whether AtNRAMP expression was also upregulated in roots. Hydroponically grown Col-0 plants were challenged by leaf infiltration of E. chrysanthemi, and then AtNRAMP3 and AtNRAMP4 gene expression was monitored in roots by RT-PCR and northern blotting. AtNRAMP4 transcript levels remained unaltered (Figure 8a,b). Similar results were obtained for the AtNRAMP4 protein, whereas as expected, iron starvation resulted in an accumulation of AtNRAMP4 (Figure 2b). Interestingly, AtNRAMP3 transcripts accumulated in the roots of infected plants (Figure 8a,b). These data suggest that bacterial infection of A. thaliana leaves generates a long-distance signal that triggers AtNRAMP3 upregulation in roots. These results prompted us to check for the possible activation of other iron deficiency-induced responses in roots upon infection with E. chrysanthemi. A. thaliana, like other non-grasses, activates a reductive process to acquire iron from the soil under iron starvation conditions. Ferric iron is first reduced into ferrous iron by the ferric chelate reductase FRO2 (Robinson et al., 1999). Ferrous iron is then taken up into the roots by the iron transporter IRT1 (Eide et al., 1996). Using RT-PCR and northern blotting to monitor the expression of IRT1 and FRO2 genes in roots, we found that IRT1 expression was upregulated 24 and 48 h after E. chrysanthemi infection, whereas FRO2 expression was activated 24 h p.i. and then declined (Figure 8a,b). As Fe is available in the growth medium, the decrease in FRO2 expression may reflect a feedback downregulation of FRO2 under Fe-sufficient conditions. Upregulation of IRT1 and FRO2 upon iron deficiency requires the presence of local apoplastic iron (Vert et al., 2003). In order to check whether the same process was required to upregulate the iron-deficiency markers after infection, we washed apoplastic iron with the strong iron chelator bathophenantroline, and then starved the plants of iron for 5 days. Under these conditions, the upregulation of IRT1 and FRO2 by bacteria was abolished (Figure 8c,d). AtNRAMP3 upregulation was also compromised, but was not completely abolished. Collectively, these data suggest the existence of a signaling pathway that is reminiscent of the iron-deficiency signal, triggered by infection with E. chrysanthemi, which leads to the induction of iron-acquisition genes in the roots upon infection of the leaves.

Figure 8.

 Expression of iron deficiency responsive genes in roots of Arabidopsis plants infected by Erwinia chrysanthemi.
Expression patterns of iron-deficiency genes in roots of Arabidopsis thaliana ecotype Col-0, monitored by RT-PCR (a, c) and northern blot (b, d). Roots were harvested at the indicated times after leaf inoculation with 10 mm MgSO4 (mock) or with E. chrysanthemi ( Plants were either grown in iron-sufficient (+Fe, or if not specified) or iron-deficient (−Fe) conditions.
In (c) and (d), roots were harvested 24 h after the treatments. The constitutive EF1α gene is used as a control. Data are representative of three biological replicates.

AtNRAMP3 and AtNRAMP4 contribute to the upregulation of AtFER1 expression in response to E. chrysanthemi

We have previously shown that AtFER1 induction is required for resistance to E. chrysanthemi (Dellagi et al., 2005). To investigate the role of AtNRAMP3 and AtNRAMP4 in AtFER1 gene upregulation and host resistance, we compared the accumulation of AtFER1 transcripts and protein in the nramp3 nramp4 double mutant with that of the wild-type ecotype Ws during infection. Northern blot analysis showed that the upregulation of AtFER1 expression was reduced in the nramp3 nramp4 mutant (Figure 9a). At the protein level, western blot experiments confirmed an important increase in the ferritin content in Ws, whereas the ferritin content in the nramp3 nramp4 mutant was only slightly increased (Figure 9b). This result indicates that AtFER1 accumulation following infection with E. chrysanthemi requires functional AtNRAMP3 or AtNRAMP4 genes.

Figure 9.

 Ferritin expression in the nramp3 nramp4 double mutant.
RNAs and proteins were extracted at the indicated times from Ws or nramp3 nramp4 double mutant (nr3nr4) leaves inoculated with 10 mm MgSO4 (mock) or with Erwinia chrysanthemi (
(a) Expression patterns of AtFER1 analyzed by northern blot.
(b) Western blot detection of AtFER1. AtFER1 purified from Escherichia coli extract (Dellagi et al., 2005) was used as a positive control. Representative data of three independent RNA and protein extractions are shown.

The oxidative burst induced by E. chrysanthemi infection is reduced in the nramp3 nramp4 double knock-out mutant

The production of ROS is another important line of defense deployed by the host in response to pathogen attack. In A. thaliana the oxidative burst observed during E. chrysanthemi infection starts at 9–10 h p.i. (Fagard et al., 2007). We compared the level of hydrogen peroxide (H2O2) accumulated in leaves of the nramp3 nramp4 mutant with that in the wild-type ecotype at 10 h p.i., using 3,3′-diaminobenzidine (DAB) staining (Figure 10a). The quantification of DAB precipitate indicates that the accumulation of H2O2 after E. chrysanthemi infection is significantly weaker in the nramp3 nramp4 mutant compared with the wild type (P < 0.0001; Figure 10b). As AtNRAMP3 gene expression is upregulated upon E. chrysanthemi infection (Figure 1b), and ectopic expression of AtNRAMP3 confers increased resistance to E. chrysanthemi (Figure 3b,c), we performed and then quantified DAB staining in E. chrysanthemi-infected, AtNRAMP3-overexpressing plants. The intensity of brown precipitates was significantly increased in the 35S::AtNRAMP3 plants compared with the control line (< 0.0001; Figure 10b). Mock-treated leaves of all lines did not display any brown precipitate, indicating that DAB staining was strictly related to infection.

Figure 10.

 H2O2 production and AtRBOH-D expression in the nramp3 nramp4 double mutant after Erwinia chrysanthemi infection.
(a) Plants showing representative 3,3′-diaminobenzidine (DAB) staining after E. chrysanthemi treatment or mock treatment.
(b) Quantitative analysis of DAB staining on leaves 10 h p.i. Error bars represent the SDs calculated from three independent experiments with 20 inoculated leaves each; control, empty vector transformed line.
(c) Expression profiles of the AtRBOH-D gene in Ws or nramp3 nramp4 (nr3nr4) leaves monitored by RT-PCR. Representative data of three independent RNA extractions are shown.

Altogether, these data show that the lack of functional AtNRAMP3 and AtNRAMP4 genes results in the reduced production of H2O2, suggesting a role of these genes in the modulation of the oxidative burst.

In the A. thalianaE. chrysanthemi interaction, the oxidative burst is mainly generated by the NADPH oxidase AtRBOH-D (Fagard et al., 2007). To determine whether the lack of functional AtNRAMP3 and AtNRAMP4 modified the transcriptional activation of the AtRBOH-D gene, we monitored the expression of AtRBOH-D by RT-PCR in E. chrysanthemi or mock-treated leaves. AtRBOH-D gene expression was upregulated to the same extent in infected leaves of both Ws and nramp3 nramp4 double mutant plants (Figure 10c). Thus, the transcriptional activation of the AtRBOH-D gene after E. chrysanthemi infection is not altered in the nramp3 nramp4 mutant.


In this work, we have identified AtNRAMP3 and AtNRAMP4 as novel immunity genes in A. thaliana. We found that the lack of functional AtNRAMP3 and AtNRAMP4 genes confers an enhanced susceptibility of Arabidopsis plants to E. chrysanthemi, whereas the ectopic expression of AtNRAMP3 reduces the plant susceptibility. These results show that these two genes contribute to plant basal resistance to bacterial infection. They indicate that the importance of NRAMPs in host resistance to pathogen attack is not restricted to animals.

AtNRAMP gene upregulation triggered by E. chrysanthemi is part of a global change in plant iron homeostasis gene expression

AtNRAMP3 gene expression is upregulated in leaves infected with either E. chrysanthemi or P.s.t. DC3000, whereas no change was observed in AtNRAMP4 expression. However, this does not exclude the possibility that a basal expression level of AtNRAMP4 is sufficient to confer enhanced resistance.

AtNRAMP3 upregulation after E. chrysanthemi infection is not restricted to the infiltrated zone of the leaves. After 24 h, upregulation occurs in healthy parts of the infected leaf, and, furthermore, is also found in roots. We never observed the presence of bacteria in healthy parts of the plant, including the roots. These results raise the question of the nature of the signaling mechanisms triggered by the bacteria during infection, leading to the induction of AtNRAMP3 in non-infected areas of leaves and in roots. Because AtNRAMP3 gene expression confers some resistance to the plant, we checked whether the major signals controlling defenses in A. thaliana, SA, ET and JA, could mediate the regulation of this gene. None of these signals seems to be involved in the upregulation of AtNRAMP3 in leaves. The major signal known to induce AtNRAMP3 expression is iron deficiency (Thomine et al., 2003). Interestingly, we found that the expression of AtNRAMP3 in roots is induced together with that of IRT1 and FRO2 genes, which are relevant markers of the iron-deficiency response in dicots. It is thus possible that an iron deficiency caused by the bacteria during infection triggers the responses observed locally, and at distance. In a previous study, Dellagi et al. (2005) showed that the presence of siderophores in leaf tissues causes an upregulation of the ferritin gene AtFER1, suggesting the existence of a local remobilization of iron, consecutive to iron depletion mediated by siderophores. Transcriptional activation of AtNRAMP3 in the infected leaf may be a part of the response triggered by the siderophores produced by the bacteria. The systemic response implies the existence of a signaling pathway that could involve a shoot-borne signal, similar to that regulating the root response to iron starvation (Vert et al., 2003). This hypothesis is supported by the observation that the removal of apoplastic Fe abolishes IRT1 and FRO2 induction by Fe deficiency in leaves (Vert et al., 2003), as well as by E. chrysanthemi infection. However, AtNRAMP3 upregulation in roots is not totally abolished in iron-starved plants, indicating that it does not only depend on local iron availability. Thus, it cannot be excluded that another pathway contributes to the activation of this gene. How the iron-deficiency signal is transmitted from the shoot to the root still remains to be discovered. Interestingly, nitric oxide is a critical molecule implicated in iron availability in maize (Graziano et al., 2002), and recent studies show the importance of this compound in tomato root responses to iron deficiency (Graziano and Lamattina, 2007), and in AtFER1 upregulation in Arabidopsis (Murgia et al., 2002; Arnaud et al., 2006). Whether NO takes part in the signaling pathway triggered by E. chrysanthemi during infection deserves attention, as it is well established that NO is a signaling molecule involved in plant defense (Wendehenne et al., 2004).

Role of AtNRAMP3 and AtNRAMP4 genes in Arabidopsis innate immunity

The use of insertional single and double nramp3 and nramp4 mutants, and double nramp3 nramp4 mutants complemented by either AtNRAMP3 or AtNRAMP4, with their own promoter or driven by the 35S promoter, enabled us to show that AtNRAMP3 is involved in resistance to E. chrysanthemi. We observed that the loss of AtNRAMP4 function reduces resistance. However, the ectopic expression of AtNRAMP4 did not enhance the plant resistance compared with controls in all the genetic combinations tested. This observation, in addition to the absence of upregulation of AtNRAMP4 by infection, indicates that this gene and its paralogue, AtNRAMP3, play distinct roles. The AtNRAMP4 basal expression level is necessary, but not limiting, for resistance to E. chrysanthemi. In contrast, increased expression of AtNRAMP3 contributes to the resistance to infection by these bacteria. Analysis of the defense responses in the nramp3 nramp4 double mutant revealed a higher level of expression of defenses mediated by SA, JA and ET. However, the nramp3 nramp4 double mutant does not display any constitutive morphological difference or growth defect relative to the wild-type ecotype. In addition, no oxidative stress was observed in healthy nramp3 nramp4 leaves.

In contrast, consecutively to infection, two lines of defenses, AtFER1 ferritin accumulation and the oxidative burst, are reduced in the double mutant, and these two defects may account for the increased susceptibility of the nramp3 nramp4 mutant to E. chrysanthemi infection. What is the significance of these results in light of the function assigned to the cognate proteins AtNRAMP3 and AtNRAMP4? These proteins are metal transporters contributing to the increase in iron availability to the cytosol under iron-deficient conditions (Thomine et al., 2003; Lanquar et al., 2005). Thus, we may expect that an efflux of iron mediated by these transporters from the vacuole to the cytosol during infection results in ferritin accumulation, and enhancement of the oxidative burst (Figure 11). Ferritin is known to accumulate in response to iron overload (Gaymard et al., 1996) or to an excess of ROS (Petit et al., 2001; Murgia et al., 2002). The oxidative burst mainly results from AtRBOH-D activity, and it is conceivable that an iron efflux mediated by AtNRAMP3 and AtNRAMP4 could positively regulate the activity of this enzyme. AtRBOH-D is similar to the gp91phox subunit of the respiratory burst oxidase (Torres et al., 2002), which requires a heme prosthetic group to reduce oxygen into superoxide (Vignais, 2002). It is also possible that elevated iron levels in the cytosol participate in iron/oxygen reactions in the cell, and exacerbate ROS production (Pierre and Fontecave, 1999). The involvement of AtNRAMP3 and AtNRAMP4 in the modulation of the oxidative burst is reminiscent of the role proposed for NRAMP1 as a mediator of the oxidative burst described in macrophages challenged with Mycobacterium avium (Kuhn et al., 1999; Zwilling et al., 1999). As Arabidopsis NRAMPs can also transport various divalent metals (Curie et al., 2000; Thomine et al., 2000; Lanquar et al., 2004), iron might not be the only metal involved in these reactions.

Figure 11.

 Working model of changes in plant iron trafficking upon Erwinia chrysanthemi infection.
This model illustrates the hypothesis that E. chrysanthemi invasion triggers iron depletion in leaves, leading to a mobilization of vacuolar iron mediated by AtNRAMP3 and AtNRAMP4. The reactive iron released in the cytosol first contributes to amplify the production of reactive oxygen species (ROS), resulting in the inhibition of the bacterial growth at 10 h p.i., and then induces AtFER1 synthesis at 48 h p.i., which then deprives the bacteria of iron. Infection also results in iron mobilization in roots from both the vacuole and the soil. C, chloroplast; V, vacuole.

The contribution of AtNRAMP genes to host resistance seems to be efficient against a restricted number of bacterial pathogens. Although we observed, in agreement with microarray data, that AtNRAMP3 gene expression is upregulated after infection with P.s.t. DC3000, the growth of this pathogen is not affected in the nramp3 nramp4 mutant. Determining the protein level of AtNRAMP3 in plants infected with P.s.t. DC3000 could help us to understand why the corresponding gene does not seem to participate in resistance against this bacterium. The role of iron in P.s.t. DC3000 pathogenicity is under investigation (Jones et al., 2007), and it appears that the importance of this metal in pathogenesis depends on the mechanisms of bacterial attack and invasion. Zaharik et al. (2002) showed that bacterial growth of the mammalian pathogen Salmonella typhimurium increases in mice lacking functional NRAMP1. Interestingly, both E. chrysanthemi and Salmonella species belong to the Enterobacteriaceae family. Altogether, our work indicates that the involvement of NRAMPs in host resistance to pathogenic bacteria, well documented in mammals, also occurs in plants.

Experimental procedures

Plant material and growth conditions

Arabidopsis thaliana seeds from the Col-0 and Ws ecotypes were obtained from the INRA Versailles collection. The nramp3 knock-out mutant, the AtNRAMP3 overexpressing lines, and the transgenic control line transformed with the empty vector used for overexpressing lines (pMON530), were all previously described in Thomine et al. (2000). The transgenic lines expressing the AtNRAMP3 promoter fused to the uidA gene (GUS) were described in Thomine et al. (2003). The nramp4 knock-out, the nramp3 nramp4 double mutant, the transgenic lines expressing the AtNRAMP4 promoter fused to the uidA gene, and the nramp3 nramp4-complemented lines with AtNRAMP3 or AtNRAMP4 genes under the control of their own promoter, were described in Lanquar et al. (2005). The AtNRAMP4 overexpressing lines were described in Lanquar et al. (2004). The sid2-1 mutant was kindly donated by J.-P. Métraux (Switzerland). Seeds of the ein2-1 (Guzman and Ecker, 1990) and jar1-1 (Staswick et al., 1992) mutants were provided by the NASC (Scholl et al., 2000). Plants were grown as described in Fagard et al. (2007). For hydroponic cultures, seeds were first stratified for 2 days at 4°C in a solution containing 0.1% agar and 50% of the nutrient solution (described below). Seeds were then individually sown in Eppendorf tubes cut at the bottom and filled with 0.75% agar. They were placed in PVC holders floating on the nutrient solution. Plants were allowed to grow for 5–6 weeks. The nutrient solution contains 0.25 mm Ca(NO3)2.4H2O, 1 mm KH2PO4, 0.5 mm KNO3, 1 mm MgSO4.7H2O, 50 μm H3BO3, 19 μm MnCl2.4H2O, 10 μm ZnCl2, 1 μm CuSO4.5H2O, 0.02 μm Na2MoO4.2H2O and 50 μm FeNa-EDTA. Plants were subjected to an 8-h light/16-h dark cycle, at 19°C, with 70% relative humidity. Plants were first grown under the above described conditions for 5 weeks, and were then transferred to iron-deficient medium after washing the roots for 5 min with medium containing the reductant sodium dithionite (5.7 mm) and the chelator bathophenanthrolinedisulfonic acid (0.3 mm), both from Sigma-Aldrich (

Bacterial strains and culture conditions

The wild-type strain E. chrysanthemi 3937 (our collection) was isolated from Saintpaulia ionantha H. Wendl. (African violet). P.s.t. DC3000 was kindly provided by A. Bent (Wisconsin, USA) (Kunkel et al., 1993). Growth conditions were as described in Dellagi et al. (2005).

Plant inoculations and determination of bacterial growth

Plant inoculation with E. chrysanthemi was performed as described in Fagard et al. (2007). To determine P.s.t. DC3000 growth in planta, entire leaves were infiltrated using a syringe, without a needle, containing a bacterial suspension at a density of 5 × 106 CFU ml−1, in sterile water.

For bacterial counts, three leaves per plant were inoculated. Three pools of six leaves from two inoculated plants were harvested in 0.9% NaCl solution, and were ground using a pestle and sterile sand. The resulting suspensions were used for serial dilutions, followed by plating on an appropriate medium. The means and standard deviations were calculated from the three values. For the scoring of symptoms, only one leaf per plant was inoculated. For RNA and protein extractions, GUS fusions, and DAB staining, we used a syringe without a needle to infiltrate the entire leaf, or a portion of the leaf (half a leaf for DAB staining and a quarter of the leaf for GUS staining), with bacterial suspensions at a density of 5 × 107 CFU ml−1 in 10 mm MgSO4.

RNA extraction, northern blotting and RT-PCR

Northern blot hybridization was carried out as described in Dellagi et al. (2005). AtFER1, AtNRAMP3 and AtNRAMP4 probes were synthesized from cloned cDNAs (Gaymard et al., 1996; Thomine et al., 2000). IRT1 and FRO2 probes were synthesized from purified PCR fragments obtained with the primers described below.

For RT-PCR analysis, reverse transcription was performed as described in Fagard et al. (2007). PCR runs were of 94°C for 4 min, 26–30 cycles, and each cycle consisted of 94°C for 30 s, 54–58°C for 30 s, and 72°C for 1 min, with a final step of 72°C for 10 min to complete polymerization. Primers for EF1α, CHI-B, AtERF1, PR1 and AtRBOH-D were described in Fagard et al. (2007). The other gene-specific primers were: AtNRAMP3-F (At2g23150), 5′-CGAGCCACTTCTAATCAACGAGG-3′; AtNRAMP3-R, 5′-CAAAACCGGTATAGACTATGCCAC-3′; AtNRAMP4-F (At5g67330), 5′-AGATAGCGGACACCATCGGTCTTGC-3′; AtNRAMP4-R, 5′-TCCCTCTGTGGTTCTTTATGTGAAG-3′; FRO2-F (At1g01580), 5′-CTCGAACCAGAGAAGCTAGTATTG-3′; FRO2-R, 5′-ATTTTGATGTTAATGTCGGAGGATA-3′; IRT1-F (At4g19690), 5′-AAAGCTTTGATCACGGTTGG-3′; IRT1-R, 5′-TATGAATCGTGGGGCCTATC-3′.

Protein extraction and western blotting

For western blots with anti-AtFER1 antibody, AtFER1 protein extract, total protein extractions from leaves and blotting, were performed as described in Dellagi et al. (2005). Protein extraction from hydroponically grown plants and western blotting with the anti-AtNRAMP4 antibody (Lanquar et al., 2004) were performed as described in Lanquar et al. (2005). Controls were 2-week-old plants grown under sufficient iron conditions (+Fe) or plants grown under deficient iron conditions, by omitting the iron in the medium (−Fe) 1 week before harvesting.

Detection of reactive oxygen species

To detect H2O2 in leaves, we used the DAB-uptake (Sigma-Aldrich) coloration method described by Torres et al. (2002). Quantification of the staining was performed with ImageJ (, using 20 leaves from 20 plants in each experiment. The index staining was calculated for each leaf as the difference between the index of brown pixels in a representative area of the injected half of the leaf, and the index of brown pixels in a representative area of the non-injected half of the leaf.

In plantaGUS expression detection

The leaves were harvested at the indicated time points and incubated for 4 h at 37°C in a GUS buffer containing: 50 mm Na2HPO4, 10 mm EDTA and 0.05% 5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid cyclohexylammonium salt (X-gluc; Fermentas Life Sciences, They were then incubated overnight in ethanol to dissolve the chlorophyll.

Statistical analysis

Statistical analyses were performed with the χ2 test for severity of the symptoms, and the Fisher’s test for the DAB-staining data.


We thank Thierry Franza for helpful discussion and Frederic Gaymard for the anti-AtFER1 antibody. We thank Jean-Pierre Métraux for kindly providing the seeds of sid2 mutants, and Andrew Bent for kindly providing us with the Pseudomonas strain. This work was supported by grants from the Institut National de la Recherche Agronomique (INRA). DE is a researcher at the Centre National de la Recherche Scientifique (CNRS). DS and VL’s PhDs were funded by the Ministère de l’Enseignement Supérieur et de la Recherche.