•Retranslocation of iron (Fe) from source leaves to sinks requires soluble Fe binding forms. As much of the Fe is protein-bound and associated with the leaf nitrogen (N) status, we investigated the role of N in Fe mobilization and retranslocation under N deficiency- vs dark-induced leaf senescence.
•By excluding Fe retranslocation from the apoplastic root pool, Fe concentrations in source and sink leaves from hydroponically grown barley (Hordeum vulgare) plants were determined in parallel with the concentrations of potential Fe chelators and the expression of genes involved in phytosiderophore biosynthesis.
•N supply showed opposing effects on Fe pools in source leaves, inhibiting Fe export out of source leaves under N sufficiency but stimulating Fe export from source leaves under N deficiency, which partially alleviated Fe deficiency-induced chlorosis. Both triggers of leaf senescence, shading and N deficiency, enhanced NICOTIANAMINE SYNTHASE2 gene expression, soluble Fe pools in source leaves, and phytosiderophore and citrate rather than nicotianamine concentrations.
•These results indicate that Fe mobilization within senescing leaves is independent of a concomitant N sink in young leaves and that phytosiderophores enhance Fe solubility in senescing source leaves, favoring subsequent Fe retranslocation.
The retranslocation of iron (Fe) in crops plays a physiologically and agronomically important role in the biofortification of seeds as well as in enhancing plant tolerance to Fe deficiency-induced chlorosis (Zhang et al., 1995; Welch & Graham, 2005). Although the phloem has been proposed as the primary route for Fe loading of grains in cereal crops (Borg et al., 2009), the quantitative extent of Fe retranslocation via the phloem remains elusive. When Fe retranslocation was calculated from bean (Phaseolus vulgaris) leaves that were pre-incubated with 59Fe, only 20% was found to be exported to sink leaves, but this number increased to 34% when leaves were shaded (Zhang et al., 1995). Studies on zinc (Zn) retranslocation in wheat (Triticum aestivum) yielded similar or even higher retranslocation rates relative to those of Fe (Waters et al., 2009) and showed that Zn retranslocation profits from ample N supply or leaf shading (Erenoglu et al., 2011). Another study showed that Zn retranslocation was higher in Zn-deficient rice (Oryza sativa) than under adequate Zn supply (Hajiboland et al., 2001). Thus, besides the nutritional status of the plant, metal retranslocation is also favored by leaf senescence.
In graminaceous plant species, reported Fe retranslocation rates differ greatly, ranging from 10% or < 30% (Hocking, 1994; Waters & Grusak, 2008) to 77% of shoot Fe being retranslocated to grains in sand-grown wheat (Garnett & Graham, 2005). High Fe retranslocation rates of up to 66% were also observed in hydroponically grown wheat plants when they were starved of Fe from anthesis onwards (Waters et al., 2009). However, a major drawback in the calculations of many retranslocation studies is that the contribution of the root pool and in particular the apoplastic Fe pool in roots was not determined or adequately considered. Even if root Fe contents do not significantly change during the course of the study, the apoplastic pool of Fe in the root may significantly contribute to the Fe translocation to sink leaves. As these pools are usually large, a small, even nonsignificant change in the size of this pool may make a major contribution to Fe translocation to sink organs. This requires particular attention as real-time translocation studies have indicated that root-to-shoot translocation via the xylem makes a substantial contribution to Fe import into sink leaves (Tsukamoto et al., 2009).
Before retranslocation from source to sink tissues, most of the Fe must be solubilized and converted into mobile forms. We refer to this as the mobilization process, which involves a change in Fe binding forms. When Fe is liberated from internal stores or degraded proteins, low-molecular-weight chelators are thought to complex Fe and retain it in a soluble form suitable for long-distance transport (Briat et al., 2007; Curie et al., 2009). Among the potential ligands, attention has particularly focused on nicotianamine (NA), phytosiderophores, organic acids, and small peptides (von Wirén et al., 1999; Krüger et al., 2002; Haydon & Cobbett, 2007; Inoue et al., 2008). NA is a hexadentate Fe chelator synthesized from S-adenosyl-methionine by the activity of nicotianamine synthase (NAS; Mori, 1999). In Fe-deficient grasses, NA is further converted by nicotianamine aminotransferase (NAAT) to an unstable 3′-oxo acid and finally to the phytosiderophore deoxymugineic acid (DMA) by DMA synthase (DMAS; Nagasaka et al., 2009). In barley (Hordeum vulgare), seven NAS genes have been identified, named HvNAS1 to 7 (Herbik et al., 1999; Higuchi et al., 1999), together with two NAAT genes (HvNAAT-A and HvNAAT-B) and one DMAS gene (HvDMAS1; Takahashi et al., 1999; Bashir et al., 2006). Most of these genes were up-regulated in Fe-deficient roots, while their expression levels in leaves were usually low (Higuchi et al., 2001). Upon chelation of Fe, complexes of Fe(III)-DMA but also of Fe(II)-NA and Fe(III)-NA have been shown to be transported by the plasma membrane-localized transporter ZmYS1 (Zea mays YELLOW STRIPE1) (Curie et al., 2001; Schaaf et al., 2004). In addition, based on the growth complementation of an Fe uptake-defective yeast mutant supplied with Fe and NA and from oocyte expression studies, evidence has been obtained that ZmYS1-related yellow-stripe-like (YSL) proteins mediate Fe-NA transport (Koike et al., 2004; Waters et al., 2006; Ishimaru et al., 2010), although none of these studies properly verified which Fe complexes were actually formed and transported (Schaaf et al., 2005). Thus, Fe binding forms transported in planta still await direct characterization.
In particular, NA has been proposed as a major candidate for Fe complex formation for subsequent phloem loading, as: NA efficiently complexes both Fe(II) and Fe(III) at neutral pH and prevents Fe(II) participating in Fenton reactions (von Wirén et al., 1999); loading of Fe into the seed decreased in transgenic plants with lower NA levels (Takahashi et al., 2003); and Fe translocation to sink organs ceased in transgenic lines with repressed expression of YSL genes (Waters et al., 2006). Based on the observation that DMA concentrations increased in Zn-deficient leaves, DMA has been proposed to contribute to Zn retranslocation from Zn-deficient leaves (Suzuki et al., 2006), whereas a role for DMA in the long-distance translocation of Fe has not yet been demonstrated.
Recent reports have highlighted a significant impact of N on Fe or Zn translocation/retranslocation in cereals. Therefore, we designed experiments to test two opposing hypotheses on the possible effect of N on Fe mobilization and retranslocation from source to sink leaves during senescence. On the one hand, a high N nutritional status may promote both processes, as the biosynthesis of relevant Fe chelators or transport peptides such as NA or the Iron Transport Peptide (ITP) requires N (von Wirén et al., 1999; Krüger et al., 2002). This hypothesis is supported by studies showing a positive correlation between N supply and the uptake and translocation of Fe or Zn (Kutman et al., 2010; Shi et al., 2010) or an elevated retranslocation of radiolabeled Zn in N-supplied wheat (Erenoglu et al., 2011). On the other hand, a high N nutritional status may decrease Fe mobilization and retranslocation as the higher protein concentrations in N-fertilized leaves tend to immobilize Fe and delay senescence (Marschner, 1995) as well as the subsequent retranslocation of N, which is closely linked with the retranslocation of Fe. This hypothesis is supported by the parallel depletion of copper (Cu) and N in senescing wheat leaves (Hill et al., 1979) and by the lower retranslocation of N and Fe in transgenic wheat lines with repressed expression of the NAC-type transcription factor NAM-B1 (Uauy et al., 2006; Waters et al., 2009). To test these opposing hypotheses, we decided to first compare Fe mobilization and retranslocation in N-sufficient vs N-deficient barley plants as N deficiency triggers senescence. Presuming that Fe deficiency acts as a driving force for Fe retranslocation, concomitant N and Fe deficiencies should give rise to the highest Fe retranslocation rates. In an alternative approach, source leaves were shaded to induce senescence and changes in Fe pools were compared between Fe-sufficient and Fe-deficient plants. As Fe retranslocation is a complex process requiring Fe solubilization and consisting of phloem loading in the source tissue, phloem transport and phloem unloading in the sink tissue, we focused here on the determination of pools of soluble Fe and potential Fe chelators relevant for Fe mobilization in source leaves and for subsequent Fe retranslocation to sink organs.
Materials and Methods
Plant culture and sampling
Barley seeds (Hordeum vulgare L. cv Henni) were germinated on filter paper. After 5 d, seedlings were transferred to split-root boxes, and two root portions were guided separately via an inverted Y-shaped plastic tube into aerated nutrient solution of the following composition (in mM): NH4NO3 1.0, K2SO4 0.75, MgSO4 0.5, CaCl2 2.0, KH2PO4 0.1; (in μM): H3BO3 1.0, MnSO4 0.5, ZnSO4 0.5, CuSO4 0.2 and (NH4)6Mo7O24 0.01; and 10 mM MES (2-[N-Morpholino]ethanesulfonic acid) (pH 6.0). To one root fraction, 50 μM [correction added after online publication 29 May 2012: in the preceding text, the previously published concentration value of 0.5 μM has been corrected to read as 50 μM] Fe(III)EDTA was supplied while the other remained Fe-free. After another 9 d of preculture, Fe-supplied root parts were cut off and all plants were transferred to 5-l pots containing saturated CaSO4 solution for the first 24 h and then to nutrient solution supplied or not supplied with 50 μM [correction added after online publication 29 May 2012: in the preceding text, the previously published concentration value of 0.5 μM has been corrected to read as 50 μM] Fe(III)EDTA (Supporting Information Fig. S1). After another 4 d of growth, a first harvest was conducted. Fe-sufficient and Fe-deficient plants were then divided into two portions and either subjected to N deficiency or continued to be supplied with 2 mM N for 8 d (Fig. S1a). Then the second harvest was conducted. In the shading experiment (Fig. S1b), preculture was the same but instead of the transfer of plants to N deficiency, aluminum foil was used to cover the oldest leaves of Fe-supplied or Fe-deficient plants (Zhang et al., 1995). Plants were then harvested after 7 d of shading. The aerated nutrient solution was renewed every 2 d. Plants were grown in a growth chamber with a 16 h : 8 h, light : dark and 24°C : 22°C temperature regime at 60% humidity.
As an approximation for chlorophyll levels, spectral plant analysis diagnostic (SPAD) readings were taken just before the final harvest using a SPAD meter 520 (Minolta Co. Ltd, Osaka, Japan). Measurements were taken from ten leaves (middle part of the first expanded young leaf) and mean values were recorded in each treatment. At harvest, most organs were harvested separately and fractionated into the root (R), the oldest leaf (OL), the second-oldest leaf (SOL), the youngest leaf (YL) and the rest of the shoot (RS). An aliquot of the OL fraction was stored in liquid N2, while all other samples were oven-dried at 65°C for 72 h.
Sample fractionation and metal and chelator analysis
The dried samples were ground in an agate mill and powdered samples were wet-digested using nitric acid in a microwave oven. Metal concentrations were determined by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) (Elan 6000; Perkin Elmer, Boston, USA). Deep-frozen OL samples were homogenized using a mortar and pestle in liquid N2 and extracted in double-distilled water. After centrifugation at 3500 g for 30 min at 4°C, the supernatant was collected as the water-soluble fraction. The pellet, containing the water-insoluble fraction, was dried at 65°C for 72 h and wet-digested in the microwave. In both fractions Fe was determined by ICP-MS. The water-soluble fraction was additionally subjected to HPLC analysis to measure citrate (Carvalhais et al., 2011) or NA and DMA. For this purpose, the fluorescing reagent 9-fluorenyl methoxycarboxyl chloride (FMOC) was used for a stable derivatization procedure. One mg of FMOC was dissolved in 1 ml of acetonitrile and incubated for 10 min at 55°C. This solution was kept on ice and used for derivatization of the samples by adding 80 μl of 1 M sodium borate (pH 8.0) to 10 μl of the sample followed by the addition of 10 μl of FMOC reagent. The mixture was incubated at 55°C for 10 min and separated immediately by HPLC using a SynergyHydro C18 column (4 μm, 4.6 × 50 mm) for DMA or a Luna C18 column (5 μm, 4.6 × 250 mm) for NA (Phenomenex, Aschaffenburg, Germany). The HPLC system consisted of a gradient pump, a degasing module, an autosampler and a fluorescence detector (Waters GmbH, Eschborn, Germany). Chromatograms were recorded using the software program Millennium 32 (Waters GmbH, Eschborn, Germany). For DMA determinations, samples were separated for 15 min using an eluent consisting of 63.5% acetonitrile, 36% HPLC water and 0.5% formic acid, whereas for NA a gradient was formed using eluent A containing 0.5% formic acid in HPLC grade water and eluent B consisting of 0.5% formic acid in pure acetonitrile. The HPLC gradient was produced by the following concentration changes: 80% A and 20% B was used at the beginning and the ratio of A to B was set to 60 : 40 within the first 10 min and changed to 40 : 60 for the following 5 min to purify the column from contamination. The A to B ratio was set back to the initial ratio of 80% A to 20% B for another 5 min to recondition the column for the next sample. The column was equilibrated at a flow rate of 1 ml per min and tempered at 30°C. Concentrations of the phytosiderophores mugineic acid and epi-/hydroxy-mugineic acid were below 2% of those of DMA and thus not quantified. Relative concentrations of Fe(III)-citrate complexes were determined by hydrophilic interaction chromatography/Fourier transform ion cyclotron resonance coupled to high-resolution MS (HILIC-FTICR/MS Köster et al., 2011).
RNA extraction and real-time quantitative RT-PCR
Total RNA was extracted from frozen OL samples with Trizol reagent and treated with DNAase (Invitrogen, Germany). One microgram of total RNA was reverse-transcribed using Superscript II reverse transcriptase and random primers (Invitrogen). 18S rRNA (GenBank accession number: AY552749) was used as a control for cDNA synthesis and amplification. Real-time quantitative PCR analysis was performed using a commercial PCR kit containing fluorescent dyes (QuantiTectTM SYBR® Green; Qiagen, Valencia, CA, USA) in the presence of 1 μg of gene-specific primers. After enzyme activation at 95°C for 15 min, amplification was carried out in a two-step PCR procedure. Dissociation curves for each amplicon were analyzed to verify the specificity of each amplification reaction. The dissociation curve was obtained by heating the amplicon from 60 to 95°C. Relative gene expression was determined using the method (Livak & Schmittgen, 2001). The levels of the RNA for each sample were normalized with respect to 18S rRNA. Quantifications were expressed relative to control plants grown in complete nutrient solution.
For statistical analysis, a one-way ANOVA was performed and the significance of differences between treatments was determined using Tukey’s test (P <0.05).
Nitrogen deficiency-induced changes in Fe translocation from old to young leaves in barley plants
To investigate the impact of the N nutritional status on Fe retranslocation between source and sink leaves, hydroponically grown barley plants were precultured in a split-root system that allowed for the removal of Fe-supplied roots and thereby the exclusion of Fe retranslocation from this apoplastic Fe pool (Fig. S1). Plants then continued to grow in the absence or presence of Fe for 12 d and an additional N deficiency treatment was imposed for 8 d in addition to both Fe treatments to induce leaf senescence. Removing part of the roots evidently had no major impact on the Fe uptake capacity, as chlorophyll and leaf Fe concentrations in Fe-supplied plants indicated that the remaining roots still covered the Fe demand of the shoots (Fig. 1). As a consequence of N shortage, leaves of Fe-sufficient plants became slightly chlorotic. As expected, Fe-deficient plants developed severe chlorosis in young leaves, which was partially alleviated under additional N deficiency (Fig. 1a). The induction of senescence by the N deficiency treatment was reflected by a lower dry matter content of OL compared to N-sufficient plants irrespective of the Fe supply (Fig. 1b). By contrast, dry matter increased in the sink organs, either in the YL in the case of Fe-deficient plants, or in the RS fraction encompassing the second and third youngest leaves in the case of Fe-sufficient plants. As all barley plants were still in the vegetative growth phase, even root dry matter production profited from the superimposed N deficiency in Fe-starved plants (Fig. 1b). In OL, Fe concentrations dropped considerably under N deficiency and appeared to be almost completely depleted under concomitant Fe and N deficiencies (Fig. 1c). The Fe concentrations in the SOL were also low under N deficiency. In Fe-sufficient but N-deficient plants, roots accumulated more Fe, whereas under concomitant Fe and N deficiencies lower root Fe concentrations were found and only the YL fraction accumulated more Fe than control plants. Thus, the parallel increase in dry matter and in the Fe concentration in the YL of Fe-deficient plants indicated that an induction of senescence by N deficiency improved Fe retranslocation from the oldest to the youngest leaf.
Concentrations of other metals showed the most marked changes in response to the Fe treatments (Fig. S2). Under Fe deficiency, Zn and manganese (Mn) concentrations strongly increased, particularly in OL, which was probably caused by the up-regulation of the Fe acquisition machinery (von Wirén et al., 1994; Vert et al., 2002). Moreover, Cu accumulated especially in roots, probably favored by Fe deficiency-induced acquisition processes (Chaignon et al., 2002) and the large Cu fixation capacity in the root cell walls (Iwasaki et al., 1990). This excess accumulation of other metals in Fe-deficient plants probably reduced the driving force for their retranslocation. However, in Fe-sufficient plants N deficiency significantly decreased Cu concentrations in all vegetative organs and also tended to decrease Zn and Mn concentrations, supporting the notion that N supply stimulates the acquisition of metal micronutrients, as has recently been shown for Zn (Erenoglu et al., 2011).
We then performed quantitative RT-PCR on OL samples and first determined whether N deficiency really induced premature senescence before investigating genes involved in the biosynthesis of the Fe chelators NA and DMA. Indeed, mRNA levels of the senescence marker HvS40 (Krupinska et al., 2002) were strongly up-regulated under N deficiency in OL of both Fe-sufficient and Fe-deficient plants (Fig. 2a). By contrast, mRNA levels of the NA synthase gene HvNAS1 showed no signs of senescence-dependent regulation and were strongly induced under Fe deficiency in N-sufficient but not in N-deficient plants (Fig. 2b). Interestingly, the NA synthase gene HvNAS2, which has also been described as the senescence-induced gene Senic3 (Ay et al., 2008), exhibited the same expression pattern as HvS40 (Fig. 2c). This up-regulation under senescence was also observed for mRNA levels of the DMA synthase gene HvDMAS1, but not for the NA-aminotransferase gene HvNAAT-B (Fig. 2d,e), which was the only one of the two NAAT homologs expressed in leaves in our experiments. These observations indicated that the onset of senescence in OL induced the biosynthesis of NA as well as of DMA.
Next, we determined the concentrations of three major Fe chelators (NA, DMA, and citrate) in the aqueous extract of OL and observed that NA concentrations showed a significant increase in plants deficient in Fe and sufficient in N (Fig. 3a), coinciding with the highest expression level of HvNAS1 (Fig. 2b). By contrast, DMA concentrations showed a clear increase under N deficiency, in particular under concomitant Fe deficiency, finally reaching a 10-fold higher level than those of NA (Fig. 3b). Interestingly, a strong increase under Fe and N deficiency was also evident for citrate concentrations (Fig. 3c). The formation of corresponding Fe(III)-citrate complexes in a 1 : 2, 2 : 2 and 3 : 4 stoichiometry was then confirmed by HILIC-FTICR/MS analysis (Fig. S3). Relative concentration changes of these complexes in response to Fe and N deficiency were similar to those of total citrate. Thus, citrate and especially DMA strongly responded to the induction of senescence by N deficiency, while under Fe deficiency NA accumulated especially when sufficient N was available.
Iron availability in leaves under varied N supply
To evaluate the influence of N deficiency on the retranslocation of Fe, differences in Fe contents between the intermediate and final harvests were determined (Fig. S1, Table S1). Under Fe-sufficient conditions, sink tissues (roots, RS and YL) experienced a net gain of Fe between the two harvesting dates resulting from the continuous supply and uptake of Fe from the nutrient solution (Table 1). Despite the continuous supply of Fe and a similar total Fe uptake (Table S2), N deficiency led to a weaker Fe accumulation in younger shoot organs, a net loss of Fe from the OL and SOL and elevated Fe accumulation in roots. This suggested that N is required for efficient root-to-shoot translocation of Fe. Under Fe deficiency, OL and SOL lost more Fe when plants were N deficient (Table 1). For instance, Fe depletion from the OL rose from 16 to 94% under N deficiency. As a consequence, N-deficient YL profited from a larger Fe import, indicating that N deficiency supported net Fe retranslocation from source to sink leaves under Fe deficiency.
Table 1. Absolute and relative changes in iron (Fe) contents of different barley (Hordeum vulgare) organs under different nitrogen (N) and Fe treatments
Changes in Fe content (μg Fe per organ)
Before harvest, plants were cultured hydroponically for 13 d with or without Fe and for 8 d with or without N supply (Supporting Information Fig. S1a). +Fe+N, control, complete nutrient solution; +Fe−N, N-free nutrient solution; −Fe+N, Fe-free nutrient solution; −Fe−N, Fe- and N-free nutrient solution. OL, oldest leaf; SOL, second-oldest leaf; RS, rest of shoot; YL, youngest leaf. Values represent absolute differences in Fe contents with per cent changes in brackets (%) between the intermediate and final harvest. Each value represents the average of five replicates. Negative values denote a loss of Fe.
− 0.1 (− 6)
− 0.1 (− 4)
− 1.0 (− 56)
− 1.7 (− 60)
− 0.2 (− 16)
− 0.4 (− 18)
− 0.6 (− 21)
− 1.2 (− 94)
− 1.2 (− 52)
− 0.7 (− 24)
Taking a closer look at changes in the pool sizes of the water-soluble and water-insoluble Fe fractions showed that N deficiency significantly depleted the water-insoluble Fe fraction irrespective of the Fe supply (Table 2). Consequently, the proportion of water-soluble Fe relative to the total Fe content increased by 2–2.5 times under N deficiency. Among the other metal micronutrients, Mn and Zn followed the same trend, with their water-soluble fractions increasing slightly under N deficiency. This indicated that N deficiency increased the soluble Fe pool that is suitable for Fe retranslocation.
Table 2. Influence of nitrogen (N) deficiency on absolute and relative pool sizes of water-soluble (w-s) and water-insoluble (w-in) iron (Fe) in the oldest barley (Hordeum vulgare) leaf
Fe content (μg per leaf)
w-s/total micronutrient (%)
Before harvest, plants were cultured hydroponically for 13 d with or without Fe and for 8 d with or without N supply (Supporting Information Fig. S1a). +Fe+N, control, complete nutrient solution; +Fe−N, N-free nutrient solution; −Fe+N, Fe-free nutrient solution; −Fe−N, Fe- and N-free nutrient solution. For comparison, relative pool sizes in the oldest leaf were also determined for manganese (Mn), zinc (Zn) and copper (Cu). Values represent means and different letters denote significant differences within treatments according to one-way ANOVA (Tukey test) at P <0.05, n =5.
Dark-induced changes in Fe mobilization in source leaves of barley
To investigate the effect of dark-induced senescence on Fe mobilization, barley plants were again precultured in a split-root system under varying supplies of Fe and the OL were shaded using aluminum foil (Fig. S1b). As expected, Fe deficiency led to severe chlorosis in the youngest leaves while shading had no visible impact on the appearance of Fe-deficiency symptoms (Fig. 4a). This was accompanied by a severe biomass reduction of YL under Fe deficiency (Fig. 4b). While the dark treatment had no significant impact on dry matter production in YL, it led to a significant decrease in the dry weight of OL irrespective of the Fe supply. In parallel, Fe concentrations in OL significantly decreased upon shading whereas they tended to increase in the RS and YL fractions (Fig. 4c). This indicated that shading promoted Fe retranslocation from source to sink leaves. A similar effect of shading was also observed for Cu and Mn but not for Zn (Fig. S4). Similar to the N deficiency treatment, shading depleted the water-insoluble Fe pool in OL and thereby significantly increased the water-soluble Fe fraction relative to total Fe when compared with the control plants (Table 3), indicating that shading also solubilized Fe from less available binding forms.
Table 3. Influence of shading on absolute and relative pool sizes of water-soluble (w-s) and water-insoluble (w-in) iron (Fe) in the oldest barley (Hordeum vulgare) leaf
Fe content (μg per leaf)
w-s/total micronutrient (%)
Before harvest, plants were cultured hydroponically for 13 d with or without Fe and for 7 d with the oldest leaf being shaded or not (Supporting Information Fig. S1b). +Fe, control, complete nutrient solution; +Fe+S, complete nutrient solution with the oldest leaf being shaded; −Fe, Fe-free nutrient solution; −Fe+S, Fe-free with the oldest leaf being shaded. For comparison, relative pool sizes in the oldest leaf were also determined for manganese (Mn), zinc (Zn) and copper (Cu). Values represent means and different letters denote significant differences within treatments according to one-way ANOVA (Tukey test) at P <0.05, n =5.
Although visible symptoms were not obvious yet (Fig. 4a), quantitative RT-PCR confirmed the induction of senescence in shaded leaves with high levels of HvS40 mRNA (Fig. 5a). mRNA levels of HvNAS1 showed a comparatively weak response to shading, while Fe deficiency alone led to a slightly more pronounced induction of gene expression (Fig. 5b), which is in agreement with the up-regulation of HvNAS1 observed in Fe-deficient roots (Higuchi et al., 1996). By contrast, HvNAS2/Senic3 was strongly responsive to shading (Fig. 5c) and followed a very similar expression pattern to HvS40. HvNAAT-B and HvDMAS1 did not significantly change expression levels after shading (Fig. 5d,e).
Similar to the N-deficiency treatment, NA concentrations in OL responded to Fe deficiency and almost doubled after the shading of Fe-deficient leaves (Fig. 6a). By contrast, DMA only tended to increase under Fe deficiency but strongly accumulated after shading, reaching 13-fold higher levels than in Fe-deficient nonshaded leaves (Fig. 6b). Regarding citrate, shading led to a slightly lower level in Fe-sufficient leaves, while in Fe-deficient leaves shading increased the concentrations of this potential Fe chelator in the OL (Fig. 6c). HILIC-FTICR/MS analysis identified Fe(III)-citrate complexes in a 1 : 2, 2 : 2 or 3 : 4 stoichiometry (Fig. S5). In particular, the 1 : 2 complex showed similar changes to the total citrate concentrations in response to Fe and shading treatments. With respect to the total concentrations found in OL, citrate concentrations were at least 6-fold higher than DMA concentrations, which in turn, were at least 50 times higher than NA concentrations. Taken together, the results showed that leaf shading, in particular under Fe deficiency, strongly increased the pool of potential Fe chelators.
The N nutritional status during leaf senescence may display opposing effects on Fe retranslocation in plants. On the one hand, N favors Fe acquisition and allocation (Kutman et al., 2011) which, in graminaceous species, heavily rely on the involvement of phytosiderophores and NA which both derive from methionine biosynthesis (Aciksoz et al., 2011). On the other hand, a high N nutritional status tends to fix Fe in proteins (Marschner, 1995), thereby potentially decreasing its availability for retranslocation processes. Using a split-root preculture which avoided Fe retranslocation from apoplastic root pools and employing two independent approaches to trigger leaf senescence in barley allowed us to demonstrate that: either N effect may prevail depending on the Fe nutritional status; Fe mobilization and retranslocation during senescence do not depend on N deficiency and thus an increased N demand in the sink tissue; and NA and DMA synthesis are induced in leaves not only by Fe deficiency but also by senescence. Taken together, the findings of this study emphasize the importance of Fe mobilization processes in senescing source leaves for the subsequent retranslocation of Fe to sink organs.
The influence of N on Fe retranslocation depends on the Fe nutritional status
When barley plants were grown under adequate Fe supply, N supply clearly improved total plant Fe uptake (Table S2) and Fe translocation from roots to shoots (Fig. 1c). This promotive effect of N on Fe is highly reminiscent of the promotive effect of N on Zn uptake, translocation and retranslocation (Shi et al., 2010; Erenoglu et al., 2011) which was less evident for Zn in our study, probably because of the higher Zn supply. The mechanisms behind this stimulative effect of N on Zn uptake remain to be identified, but have mainly been ascribed to the N-dependent biosynthesis of phytosiderophores, as derived from a larger Fe mobilization capacity in root exudates of N-supplied wheat, and to the N-mediated stimulation of Fe-phytosiderophore uptake (Aciksoz et al., 2011). In fact, the present study indicated a link between the N nutritional status and the formation of the Fe chelator NA in leaves. N-supplied leaves had higher NA concentrations (Fig. 3a) and under Fe deficiency this coincided with an enhanced expression of HvNAS1 (Fig. 2b). Taking into account a moderate up-regulation of HvNAS1 under Fe deficiency (Fig. 5b), N supply indeed appeared to be an additional trigger for HvNAS1 expression (Fig. 2b) and the subsequent elevation of NA concentrations (Fig. 3a). Thus, HvNAS1 expression in barley is not restricted to roots, as previously suggested by Higuchi et al. (2001), but occurs also in Fe-deficient leaves, in particular under an adequate N nutritional status.
Interestingly, a promotive effect of N on Fe retranslocation was lost under Fe deficiency, as with N supply there was a lower exchange of Fe between different shoot tissues (Table 1). In Fe-deficient plants, N deficiency enhanced Fe solubility within the plant as indicated by a depleted pool of Fe in the water-insoluble fraction and an increased proportion of soluble Fe relative to total Fe (Table 2). This increased Fe solubilization was accompanied by an enhanced formation of the Fe chelators DMA and citrate in source leaves (Fig. 3). Thus, DMA concentrations far exceeded those of NA and may have profited from the N deficiency-induced up-regulation of HvNAS2/Senic3 and HvDMAS1, which underwent a similar N-dependent regulation to the senescence marker HvS40 (Figs 2, 3; Krupinska et al., 2002). Hence, N-deficient source leaves depleted more Fe in favor of higher Fe accumulation and concomitant biomass increase in sink leaves (Fig. 1, Table 1). In conclusion, N supply shows opposing effects on Fe pools in source leaves, inhibiting Fe export out of source leaves under N sufficiency but stimulating Fe export from source leaves under N deficiency.
Senescence promotes Fe retranslocation independent of the N nutritional status
The approach of inducing leaf senescence by N deficiency suffers from the drawback that this treatment also increases the sink strength for N in young leaves, so that Fe mobilization and retranslocation may just follow N dynamics. We therefore chose a second approach and induced leaf senescence by shading, which presumably does not alter the N demand in sink tissues and thus allows for the investigation of Fe allocation independent of a driving force for N retranslocation. In fact, shading resulted in low Fe concentrations and even lower Fe contents relative to illuminated leaves in favor of a higher Fe accumulation in sink tissues (Fig. 4). Although in this experiment Fe retranslocation was not quantified by undertaking an intermediate harvest, a comparison of Fe concentrations in different treatments indicated that shading promoted Fe export from OL and Fe import into YL. Interestingly, this held true not only for Fe-deficient plants but to a weaker extent also for Fe-sufficient plants (Fig. 4). A comparable observation was made in wheat, where repression of a NAC-type transcription factor delayed senescence in source leaves without evidence for an altered sink strength for N. This inhibited the subsequent retranslocation of N concomitantly with that of Fe and Zn (Uauy et al., 2006; Waters et al., 2009). Thus, even though Fe retranslocation during senescence is mostly accompanied by a parallel retranslocation of N, it is not dependent on an enhanced sink strength for N in YL. These findings are also relevant in the context of agricultural plant production, where shading is a common stress situation encountered by lower leaves in dense plant stands and where N deficiency-induced senescence may occur as a consequence of suboptimal N fertilization or when drought impairs the mass flow of nitrate to roots (Burkey & Wells, 1991; Marschner, 1995).
Most Fe chelators form complexes not only with Fe but also with other metals. In particular, NA and DMA also chelate Zn, Mn and Cu, and NA has been proposed to play a role in their long-distance translocation in plants (Broadley et al., 2007; Puig et al., 2007; Curie et al., 2009). When plants were adequately supplied with Fe, shading and N deficiency tended to decrease Cu and Mn and less consistently also Zn concentrations in OL without increasing concentrations of the corresponding metals in YL when compared with control plants (Figs S2, S4). This may suggest a stimulated export of other metals out of senescing leaves. However, because of the continuous root supply with Zn, Cu and Mn the present experimental conditions were not optimized to verify a dependence of the retranslocation of these metals on the N nutritional status.
A role of DMA for Fe mobilization within senescing source leaves
Shading also decreased the water-insoluble Fe pool and increased the soluble Fe fraction in OL (Table 3), which was accompanied by a dramatic increase in DMA concentrations (Fig. 6b) and the expression of HvNAS2/Senic3 but not of HvDMAS1 (Fig. 5c,e). The lack of up-regulation of HvDMAS1 under dark-induced senescence was unexpected and requires deeper investigation with particular consideration of the low expression levels observed in Fe-deficient shoots (Figs 2, 5; Bashir et al., 2006). Perhaps the lack of expression was compensated for by other, so far unidentified, DMAS genes. HvNAS2/Senic3 turned out to be reliably up-regulated in senescing leaves irrespective of whether senescence was induced by N deficiency or shading (Figs 2, 5). Interestingly, in both cases NA concentrations remained constantly low, varying only within a factor of 2. By contrast, DMA concentrations increased in all four senescence treatments to a larger extent, thereby exceeding NA concentrations by far (Figs 3, 6). These observations suggest that NA concentrations rather reflected a steady-state pool for the downstream synthesis of DMA and that HvNAS2/Senic3 may take on a regulatory or even limiting role for the biosynthesis of DMA under senescence. A limiting role for NAS in metal retranslocation has been inferred from the analysis of an activation-tagged rice line overexpressing OsNAS3 which accumulated more NA and DMA as well as Fe and Zn in seeds (Lee et al., 2009). Likewise, activation tagging of OsNAS2 also enhanced the expression of genes involved in NA and DMA synthesis as well as NA and DMA accumulation, leading to an enhanced uptake and translocation of Fe and Zn to the seeds (Lee et al., 2011). In rice plants, Zn deficiency led to an up-regulation of OsNAS3 and of OsNAAT1, which resulted in elevated DMA concentrations in the shoot while NA appeared to be depleted (Suzuki et al., 2008). All these studies show a common link between the up-regulation of an NAS gene and NA or DMA concentrations in shoots, while it remained unclear to what extent these two chelators contributed directly to Fe or Zn retranslocation. Against this background, the present study now shows that leaf senescence in barley represents an endogenous process that enhances predominantly the leaf concentrations of DMA, probably triggered by induction of the strongly senescence-responsive HvNAS2/Senic3 gene.
It is currently unclear which are the relevant sources for Fe mobilization during leaf senescence. While Fe release from ferritin appears to play a minor role (Ravet et al., 2009), Fe may be recycled via degradation of Fe-containing proteins via the vacuole and then released to the cytosol via NRAMP (Natural Resistance-Associated Macrophage Protein) transporters (Lanquar et al., 2005; Briat et al., 2007). It is further unclear which Fe form may then prevail. Although the affinity constant of ferrous and ferric Fe is one to two orders of magnitude higher for NA than for DMA (Hider et al., 2004), in vitro studies at cytosolic pH have shown that ferrous Fe strongly prefers the formation of Fe(II)-NA complexes, while ferric Fe preferentially forms Fe(III)-DMA or Fe(III)-phytosiderophore complexes (Weber et al., 2006). However, neither complex occurred in substantial amounts in senescing leaves when samples from the shading experiment were subjected to zwitterionic hydrophilic interaction liquid chromatography coupled to electrospray ionization mass spectrometry (Köster et al., 2011; data not shown). In part, this may be attributable to the low chromatographic stability of Fe(II)-NA complexes (Xuan et al., 2006). Instead, Fe-deficient shaded leaves showed higher levels of Fe-citrate complexes (Fig. S5). On the one hand, this may be a result of the Fe deficiency-induced synthesis of organic acids and especially of citrate, as reported for Fe-deficient sugar beet (Beta vulgaris) leaves (Rellan-Alvarez et al., 2011). As recently reported for Arabidopsis, citrate takes on a prominent role in long-distance transport and mobilization of apoplastic Fe throughout plant development and whenever symplastic disconnections between tissues must be overcome (Roschzttardtz et al., 2011). On the other hand, our analysis showed that the accumulation of citrate in barley leaves further increased when Fe-deficient plants were senescing irrespective of whether N deficiency or shading triggered senescence (Figs 3c, 6c). Thus, with regard to the remarkable competitiveness of citrate as an Fe chelator (von Wirén et al., 1999), the participation of the large citrate pool in Fe chelation in senescing leaves cannot be disregarded and requires more attention in further studies.
Taken together, our results show that leaf senescence enhances Fe mobilization in source leaves and subsequent retranslocation of Fe to sink leaves. The senescence-induced increase in soluble Fe pools was clearly associated with elevated DMA and citrate but not with elevated NA concentrations, suggesting that DMA and citrate are major chelators involved in the mobilization of Fe in the senescing source tissue. Whether and to what extent the corresponding DMA- and citrate-Fe chelates contribute to phloem loading remains to be shown.
The study was supported by the Deutsche Forschungsgemeinschaft, Bonn, Germany (Wi1728/6-3 and 14-1), the trilateral PLANT-KBBE Initiative funded by the Bundesministerium für Bildung und Forschung (grant no. 0315458B) and the 973 project (No. 2009CB118605) of the National Natural Science Foundation of China (30871592). We thank Dr Ben Gruber and Dr Ricardo Giehl, IPK Gatersleben, for critically reading the manuscript.