Iron- and Ferritin-Dependent Reactive Oxygen Species Distribution: Impact on Arabidopsis Root System Architecture

Transcription

1 Research Article Iron- and Ferritin-Dependent Reactive Oxygen Species Distribution: Impact on Arabidopsis Root System Architecture Guilhem Reyt 1,2, Soukaina Boudouf 1, Jossia Boucherez 1, Frédéric Gaymard 1 and Jean-Francois Briat 1, * 1 Biochimie et Physiologie Moléculaire des Plantes, Centre National de la Recherche Scientifique, Institut National de la Recherche Agronomique, Université Montpellier 2, SupAgro. Bat 7, 2 place Viala, Montpellier Cedex 1, France 2 Present address: Institute of Biological and Environmental Sciences, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen, Scotland AB24 3UU, UK *Correspondence: Jean-Francois Briat (briat@supagro.inra.fr) ABSTRACT Iron (Fe) homeostasis is integrated with the production of reactive oxygen species (ROS), and distribution at the root tip participates in the control of root growth. Excess Fe increases ferritin abundance, enabling the storage of Fe, which contributes to protection of plants against Fe-induced oxidative stress. AtFer1 and AtFer3 are the two ferritin genes expressed in the meristematic zone, pericycle and endodermis of the Arabidopsis thaliana root, and it is in these regions that we observe Fe stained dots. This staining disappears in the triple fer1-3-4 ferritin mutant. Fe excess decreases primary root length in the same way in wild-type and in fer1-3-4 mutant. In contrast, the Fe-mediated decrease of lateral root (LR) length and density is enhanced in fer1-3-4 plants due to a defect in LR emergence. We observe that this interaction between excess Fe, ferritin, and root system architecture (RSA) is in part mediated by the H 2 O 2 /O 2 $ balance between the root cell proliferation and differentiation zones regulated by the UPB1 transcription factor. Meristem size is also decreased in response to Fe excess in ferritin mutant plants, implicating cell cycle arrest mediated by the ROS-activated SMR5/SMR7 cyclin-dependent kinase inhibitors pathway in the interaction between Fe and RSA. Key words: roots, development, oxidative stress, reactive oxygen species balance, iron, ferritins Reyt G., Boudouf S., Boucherez J., Gaymard F., and Briat J.-F. (2015). Iron- and Ferritin-Dependent Reactive Oxygen Species Distribution: Impact on Arabidopsis Root System Architecture. Mol. Plant. 8, INTRODUCTION Plant roots play a major role in iron (Fe) homeostasis. Fe deficiency activates chelation or reduction mechanisms coupled to activity of high-affinity root transporters (Curie et al., 2009; Morrissey and Guerinot, 2009; Conte and Walker, 2011), enabling plants to acquire sufficient Fe even when Fe is scarce in the soil. In contrast, Fe excess increases the abundance of ferritin protein. These 24-mer plastid proteins store thousands of Fe atoms in their central cavity. Fe buffering by ferritins avoids Fe reacting with oxygen, which would otherwise lead to an oxidative stress (Briat et al., 2010a, 2010b). Arabidopsis has four ferritin genes: AtFer1 4 (Petit et al., 2001a). AtFer1 is expressed at a higher level than AtFer3 or AtFer2 and its expression is seed specific (Petit et al., 2001a; Ravet et al., 2009a, 2009b). AtFer4 is very weakly expressed and the protein is targeted to both the plastids and mitochondria (Zancani et al., 2004; Galatro and Puntarulo, 2007). Expression of AtFer1 is regulated by Fe, phosphate, oxidative stress, light, and the circadian clock. Various transduction pathways (Lobréaux et al., 1992; Lobréaux et al., 1993; Lobréaux et al., 1995; Savino et al., 1997; Murgia et al., 2002; Murgia et al., 2004; Arnaud et al., 2006; Touraine et al., 2012) and some cis-elements and trans-acting factors regulating AtFer1 expression have been characterized (Petit et al., 2001b; Duc et al., 2009; Bournier et al., 2013). In addition, Fe-dependent post-transcriptional regulation involving the AtFer1 transcript 3 0 UTR is required to return expression of AtFer1 back to basal levels after induction by Fe excess (Ravet et al., 2012). A common point of connection between these regulatory mechanisms is their link with oxidative stress responses (Ravet and Pilon, 2013). Indeed, the major role of ferritin in leaves and Published by the Molecular Plant Shanghai Editorial Office in association with Cell Press, an imprint of Elsevier Inc., on behalf of CSPB and IPPE, SIBS, CAS. Molecular Plant 8, , March 2015 ª The Author

2 Fe Homeostasis and Root Architecture Figure 1. Ferritin Gene Expression in Arabidopsis Roots and Leaves in Response to Fe Excess. Two-week-old plants are harvested before (0 h), and after (6 h, 9 h) adding 500 mm Fe(III)-EDTA in the medium. (A D) Ferritin transcripts abundance. RNA is extracted from leaves (A, B) or roots (C, D). Expression of AtFer1 (A, C), AtFer2, AtFer3, and AtFer4 (B, D) is analyzed by qrt-pcr. Mean values ± standard error are shown (n = 3). Asterisks indicate significant differences after 6 h of Fe excess (* if P < 0.05 and ** if P < 0.005; Student s t-test) (E) Ferritin protein abundance. Protein extracts (10 mg) of leaves or roots are separated by PAGE, blotted, and probed with ferritin antibodies. Loading is evaluated by Coomassie staining. transcription factor, UPB1, modulates the balance between cell proliferation and differentiation in the PR by directly regulating the expression of a particular set of peroxidases acting to alter the balance between H 2 O 2 and O 2 $ (Tsukagoshi et al., 2010). More recently, it was also demonstrated that UPB1 is required to control lateral root (LR) emergence through reactive oxygen species (ROS) signaling (Manzano et al., 2014). ROS homeostasis is therefore emerging as a major parameter controlling root system architecture (RSA). Moreover, Fe homeostasis and ferritins are also known to be involved with ROS production and oxidative stress (Ravet et al., 2009a; Briat et al., 2010b). It therefore seems timely to investigate the potential role of ROS in connecting ferritin function and RSA remodeling in response to Fe excess. flowers is to protect against oxidative stress, as shown by the characterization of multiple ferritin mutants (Ravet et al., 2009a; Sudre et al., 2013). However, very little is known about the synthesis and function of ferritin in roots or their involvement in oxidative stress responses. Differences in superoxide (O 2 $ ) and hydrogen peroxide (H 2 O 2 ) accumulation in the root tip significantly affect primary root (PR) growth and differentiation (Dunand et al., 2007). A bhlh 440 Molecular Plant 8, , March 2015 ª The Author RESULTS Fe-mediated Increase of Ferritin Abundance in Arabidopsis Roots and Leaves When in excess, Fe led to a rapid and significant increase in abundance of the AtFer1 transcript in leaves (Figure 1A) and represented the most highly expressed of the four AtFer genes (Figure 1A and 1B). In roots, AtFer1 was also the most highly expressed ferritin gene in response to Fe excess, although to a lesser magnitude than that observed in leaves (Figure 1A 1D). The AtFer3 gene was the next most highly expressed after AtFer1, both in roots and leaves. AtFer4 mrna abundance was less than either AtFer1 or AtFer3, in particular in roots, and its expression was activated in response to Fe excess in both leaves and roots (Figure 1B and 1D). AtFer2 was the most weakly expressed of the ferritin genes, in both leaves and roots (Figure 1B and 1D). These

3 Fe Homeostasis and Root Architecture Molecular Plant Figure 2. Root Localization of GUS Expression Driven by AtFer1, AtFer3, or AtFer4 Promoters. patfer::gus lines are treated with Fe excess (500 mm of Fe(III)-EDTA) or not (50 mm of Fe(III)- EDTA) during 24 h before harvesting. (A) GUS staining of whole root patfer::gus lines. (B) GUS staining of PRs and LRs, and on transversal thin cross-sections of mature roots or of emerging LR. Tissue-Specific Expression of Ferritin Genes in Roots Tissue-specific expression of ferritin genes in roots in response to Fe excess was investigated in transgenic plants expressing the b-glucuronidase (GUS) gene driven by AtFer1, AtFer3, or AtFer4 promoter sequences. AtFer4 promoter activity was undetectable in roots, even after Fe excess treatment (Figure 2A), consistent with the low level of AtFer4 transcript detected by qrt-pcr (Figure 1). patfer1::gus and patfer3::gus shared a similar pattern of expression. AtFer1 and AtFer3 promoter activity was detected at 50 mm Fe(III)-EDTA, both in PR and LR (Figure 2A and2b). The AtFer1 and AtFer3 promoter::gus fusions reveal a strong staining in the meristematic and elongation zones in a similar way between PR and LR. Cross-sections revealed GUS staining in the central vasculature and in the endodermis and pericycle. Treatment with 500 mm Fe(III)-EDTA led to the enhancement of GUS staining in the same cell layers as observed with a 50 mm Fe(III)-EDTA treatment (Figure 2B, right panel). results are consistent with previous reports (Petit et al., 2001a; Ravet et al., 2009a, 2009b). At the protein level, ferritins are assembled from two polypeptides of 26.5 and 28 kda, which increased in abundance in response to Fe excess in both leaves and roots (Figure 1E). Each polypeptide was attributed to the expression of the various ferritin genes in roots (Supplemental Figure 1), as previously reported with leaves (Briat et al., 2010a). Estimated qualitatively from immunoblots, the most abundant ferritin subunit in roots in response to Fe excess was AtFer1, present in two polypeptides of 28 and 26.5 kda, respectively. AtFer3 and AtFer4 subunits were less abundant and were present only in the form of a 26.5 kda polypeptide (Supplemental Figure 1). However, these data are not a direct measurement of the amount of each subunit, since they are derived from signal differences on Western blots performed on mutants and mutant combinations. Role of Ferritins in Fe Content and Distribution in Roots After treatment with 500 mm Fe(III)-EDTA, 90% of Fe accumulated in the root was trapped in the apoplast (Lobréaux et al., 1992). This extracellular Fe was removed according to Bienfait et al. (1985) on plants treated with 50 or 500 mm Fe(III)-EDTA during 48 h in order to estimate the amount of Fe accumulated within the cells of the root. Under such conditions, the total root Fe concentration decreased by 23% in the fer1-3-4 ferritin mutant compared with wild-type after exposure to 500 mm Fe(III)-EDTA (Figure 3A). However, after exposure to 50 mm Fe(III)-EDTA, no statistically significant difference was observed in cellular root Fe concentrations between wild-type and fer1-3-4 plants. In contrast, the Fe concentration in leaves was unaffected either by 48 h of Fe excess treatment and/or by ferritin mutations (Figure 3B). These data suggest that a 48 h Fe excess treatment is no longer enough to observe a change in total leaf Fe concentration and/or that a checkpoint operates at the root to shoot translocation level. Since loss of function of fer1, 3, and 4 in the fer1-3-4 triple mutant caused a significant decrease in Fe accumulation in roots after exposure to 500 mm Fe(III)-EDTA, we investigated the role of the Fer genes Molecular Plant 8, , March 2015 ª The Author

4 Fe Homeostasis and Root Architecture Figure 3. Fe Content and Localization. Fe content in roots (A) and leaves (B) from 2-week-old Col0 and fer1-3-4 plants, treated with 50 or 500 mm Fe(III)-EDTA during 48 h. Mean values ± standard error are shown (n R 3). Asterisks indicate significant differences between the two genotypes (P < 0.05; Student s t-test). Fe localization in roots by Perls-DAB staining the tip of PR and in young emerged LR (C) from Col0 plants grown at 50 mm Fe(III)-EDTA. Longitudinal (top) and transversal (bottom) cross-sections in the meristematic apical zone of roots from Col0 and fer1-3-4 plants (D) grown at 50 mm Fe(III)-EDTA. Transversal cross-sections of mature roots at different magnifications (E), from 2-week-old Col0 and fer1-3-4 plants, treated with 50 or 500 mm of Fe(III)-EDTA during 48 h. Arrows indicate Fe-rich structures. c, cortex; end, endodermis; ep, epidermis; lrc, LR cap; p, pericycle. in tissue and cellular Fe distribution using Perls-DAB staining (Figure 3C 3E; Roschzttardtz et al., 2009). Strong staining with Perls-DAB was observed in the apical and LR meristems after exposure to either 50 mm or 500 mm Fe(III)-EDTA (Figure 3C). Longitudinal and transverse cross-sections in these zones from wild-type plants grown at 50 mm Fe(III)-EDTA (Figure 3D, left 442 Molecular Plant 8, , March 2015 ª The Author panel) revealed strong Fe staining in the apoplast of the internal cell layers, as well as strong and extensive dot-like staining within cells. In contrast, no such intracellular dot-like staining was observed in sections from fer1-3-4 roots of plants grown in 50 mm Fe(III)-EDTA, strongly suggesting that this dot-like staining corresponded to ferritin Fe (Figure 1B, right panel). Perls-DAB

5 Fe Homeostasis and Root Architecture Molecular Plant Figure 4. Fe-Mediated Oxidative Stress in Roots of Col0 and fer1-3-4 Plants. Catalase activity (A) and malondialdehyde (MDA) content (B) of roots from Col0 and fer1-3-4 plants grown on MS/2 medium containing 50 mm Fe(III)- EDTA during 7 days, and then transferred on the same medium containing 50 or 250 mm Fe(III)- EDTA during 7 days. Mean values ± standard error are shown (n R 5). Different letters (a, b, or c) indicate statistical differences (P < 0.05; Student s t-test). this, we conclude that root ferritins contribute to protecting this organ from oxidative stress caused by Fe excess. staining of cross-sections of differentiated older roots from plants grown without Fe excess treatment revealed apoplastic Fe staining of internal cell layers (pericycle, endodermis, and xylem vessels, Figure 3E), which was more or less intense independent of the genotype considered. After Fe excess treatment, the apoplast of the epidermal and cortical cells remains heavily stained both in wild-type and fer1-3-4 plants. In addition, dotlike staining was observed mainly in the endodermis, pericycle, and stele, and to a lesser extent in the cortex of wild-type roots. This dot-like staining is absent from the same cellular locations in roots of the fer1-3-4 plants (Figure 3E, lower panels), suggesting again that this dot-like staining corresponds to ferritin Fe. Consistent with this observation, Perls-DAB was shown previously to stain ferritin Fe in leaves of Arabidopsis plants treated with Fe excess (Roschzttardtz et al., 2013). It is noteworthy that this dot-like staining with Perls-DAB occurred within the same cell layers in which we observe GUS staining in the patfer1::gus and patfer3::gus transgenic plants (Figure 2), both in mature roots of wild-type plants treated with Fe excess (Figure 3E) and in the apical root meristem in plants grown at 50 mm Fe(III)- EDTA (Figure 3D). Ferritin-Dependence of Fe-Mediated Oxidative Stress in Arabidopsis Roots Previously, it was established that ferritins participate in protection against oxidative stress in leaves (Ravet et al., 2009a; Briat et al., 2010b; Sudre et al., 2013). Here, we show that the distribution of Fe is disturbed in the root of the fer1-3-4 triple mutant (Figure 3). We therefore tested whether or not this modification in Fe allocation correlated with an enhanced oxidative stress. Catalase activity was not statistically different in roots of fer1-3-4 plants untreated with Fe compared with wild-type roots. However, after treatment with Fe excess, catalase activity in roots increased in wild-type, but this increase was 40% higher in roots from the fer1-3-4 triple mutant plants (Figure 4A). In addition, malonyldialdehyde (MDA), a product of membrane lipid peroxidation, was observed to be significantly higher in roots of fer1-3-4 plants compared with wild-type after treatment with either sufficient or excess Fe (Figure 4B). From Influence of Altered Fe Homeostasis on Arabidopsis Root Architecture ROS signaling is known to influence PR (Dunand et al., 2007; Tsukagoshi et al., 2010) and LR (Manzano et al., 2014) development in Arabidopsis. Furthermore, Fe readily reacts with oxygen to generate ROS (Halliwell and Gutteridge, 1992), and root ferritins contribute both to managing root Fe content and distribution and to protecting this organ against oxidative stress (Figures 3 and 4). Fe homeostasis in roots could therefore be an important determinant of their development. To address this hypothesis, we measured PR and LR length and LR density in both wildtype and the fer1-3-4 triple mutant grown at four different Fe concentrations in the agar solidified nutrient medium (Figure 5). An example of the RSA of both wild-type and fer1-3-4 grown at 50 or 500 mm Fe(III)-EDTA is presented in Figure 5A. Quantification of RSA traits in plants growing in the presence of increased concentrations of Fe establishes that PR length decreases with increasing Fe concentration in the nutrient medium in an identical manner in both wild-type and the fer1-3-4 triple mutant (Figure 5B). LR length of both fer1-3-4 and wild-type plants decreased by 30% at 150 mm Fe-EDTA (Figure 5C). However, at 300 and 500 mm Fe-EDTA, LR length in the fer1-3-4 triple mutant was significantly reduced compared with wild-type. The LR density of wild-type plants, measured according to Dubrovsky and Forde (2012), was observed to increase as a function of the Fe concentration in the nutrient medium. Such an increase in LR density was also observed in fer1-3-4 plants but to a lesser extent, and only up to a 300 mm Fe-EDTA concentration (Figure 5D). At 500 mm Fe-EDTA, the LR density in fer1-3-4 decreased and was 35% lower than the LR density of wild-type plants at the same Fe concentration (Figure 5D). This difference in LR density between wild-type and fer1-3-4 triple mutant plants at 500 mm Fe-EDTA could be due to either a lower number of initiation events or a defect in LR primordia (LRP) emergence. In order to discriminate between these two possibilities, the number of LRP as a percentage of the total LR initiation events was determined within the LR branching zone, according to Dubrovsky and Forde (2012). The number of LR initiation events was not statistically different between wild-type and the fer1-3-4 triple mutant at the two Fe concentrations tested (Supplemental Table 1). In contrast, the number of non-emerging Molecular Plant 8, , March 2015 ª The Author

6 Fe Homeostasis and Root Architecture Figure 5. Root Architecture Analysis. (A) Root architecture from representative wild-type Col0 and fer1-3-4 plants grown on MS/2 medium containing 50 mm Fe(III)-EDTA during 7 days and then transferred on the same medium containing 50 or 500 mm Fe(III)-EDTA during 7 days. (B) PR length of plants grown on MS/2 medium containing 50 mm Fe(III)-EDTA during 7 days and then transferred on the same medium containing 50, 150, 300, or 500 mm Fe(III)-EDTA during 7 days. (C) Average length of LR on the same plants as in (B). (D) LR density (number of emerged LR per centimeter of root branching zone) on the same plants as in (B). (E) Percentage of LRP of the total LR initiation events within the LR branching zone on the same plants as in (B) but only transferred on a medium containing 50 or 500 mm Fe(III)-EDTA during 3 days. LR length (C), density (D), and proportion of aborted primordia (E) were measured only on LR formed below the location of the PR tip at the moment of the transfer. Mean values ± standard error are shown (n R 20 plants). Asterisks indicate significant differences between the two genotypes (P < 0.05; Student s t-test). LRP as a percentage of the total LR initiation events was significantly higher in fer1-3-4 than in wild-type plants at 500 mm Fe-EDTA, whereas it was not significantly different at 50 mm Fe- EDTA (Figure 5E). This observation strongly suggested that the lower LR density observed in the ferritin fer1-3-4 triple mutant at 500 mm Fe-EDTA is due to a defect in LR emergence rather than to lower LR initiation events. In all the experiments we have described, Fe was provided chelated with EDTA. In order to confirm that the observed phenotypes were specifically due to high Fe concentration and not to the organic chelate, similar experiments were performed using Fe-citrate instead of Fe-EDTA (Supplemental Figure 2). Root architecture parameters of wild-type and fer1-3-4 plants were similarly affected by Fe in excess, whatever the chelate considered (Figure 5B 5D and Supplemental Figure 2). 444 Molecular Plant 8, , March 2015 ª The Author It can therefore be concluded that ferritins are involved in controlling LR emergence and growth under Fe excess. However, ferritins are not involved in maintaining PR growth under these same conditions. Modification of O 2 $ and H 2 O 2 Distribution in Arabidopsis Roots in Response to Alteration of Fe Homeostasis PR growth depends upon the localization of the transition between the proliferation and elongation zones, which is correlated with the relative distribution of O 2 $ and H 2 O 2 in the root tip (Tsukagoshi et al., 2010). Furthermore, LR emergence requires ROS signaling (Manzano et al., 2014). We observe that Fe in excess promotes oxidative stress in roots and ferritins provide protection against this stress (Figure 4). It was also observed

7 Fe Homeostasis and Root Architecture Molecular Plant Figure 6. Meristem Size of PR and LR of Col0 and fer1-3-4 Plants Grown at 50 or 500 mm Fe-EDTA. The meristem size of PR (A) or LR (B) was measured on plants grown on MS/2 medium containing 50 mm Fe(III)-EDTA during 7 days and then transferred on the same medium containing 50 or 500 mm Fe(III)-EDTA. For LR, the sizes were determined on the six youngest LR per plant longer than 1.5 mm. The size is obtained by measuring the distance between above the root cap and the transition zone, after propidium iodine staining. Mean values ± standard error are shown (n R 18 plants). Different letters (a, b, or c) indicate statistical differences (P < 0.05; Student s t-test). that Fe homeostasis influences RSA (Figure 5). These observations suggest the possibility that Fe and ferritins are involved in remodeling of RSA by affecting the production and distribution of O 2 $ and H 2 O 2 in the meristematic and elongation zones, and/or the size of the root meristems. In order to investigate these possibilities, the distributions of H 2 O 2 and O 2 $, were visualized by co-staining the same root with the fluorescent probes 3 0 -(p-hydroxyphenyl) fluorescein (HPF) and dihydroethidium (DHE), specific for H 2 O 2 and O 2 $, respectively. ROS distribution using this co-staining method and meristem size were determined in both PR and LR of wild-type and the fer1-3-4 triple mutant after exposure to control Fe conditions (50 mm Fe- EDTA) or Fe excess (500 mm Fe-EDTA). Loss of function of ferritins in the fer1-3-4 triple mutant had no effect on the size of the meristematic zone (Figure 6A and the position of the arrows in Figure 7) or on the distribution of H 2 O 2 or O 2 $ between the proliferation and elongation zones (Figure 7) in the PR. In contrast, plants treated with Fe excess showed a 15% decrease in the average meristem size, regardless of the genotype (wild-type or fer1-3-4) (Figure 6A). This decrease in meristem size correlated with an increase in H 2 O 2 fluorescence in the transition zone (Figure 7A and 7D). Concomitant with this increase in H 2 O 2 fluorescence, we observed a decrease in O 2 $ fluorescence in both the meristematic and proliferation zones (Figure 7B). Merging the H 2 O 2 and O 2 $ fluorescence images (Figure 7C) clearly shows that Fe excess increased the area in which H 2 O 2 is abundant and decreased the area in which O 2 $ is abundant. Concerning LR, their meristem size was decreased both after treatment with Fe excess and in the ferritin triple mutant plants (Figure 6B and position of arrows in Figure 8). ROS distribution in LR was affected both by Fe excess and by loss of function of the ferritins in the triple mutant (Figure 8A 8C). Image analysis of the H 2 O 2 fluorescence (Figure 8D) indicated that H 2 O 2 distribution at 50 mm Fe-EDTA was very similar between wildtype and fer1-3-4 plants, except for a higher H 2 O 2 content in the tips of fer1-3-4 LR compared with wild-type plants. The high H 2 O 2 content decreased away from the tip through the meristematic zone until reaching the transition zone, where it increased into the elongation zone. Fe excess strongly decreased H 2 O 2 content at the root tip and the elongation zone of both wild-type and the fer1-3-4 triple mutant, whereas Fe excess increased H 2 O 2 content in the transition zone compared with the 50 mm Fe- EDTA Fe control treatment. This increase was more pronounced in the fer1-3-4 LR compared with the wild-type (Figure 8D). Both in wild-type and fer1-3-4 plants, the highest O 2 $ fluorescence was observed in the meristematic zone after growth in the presence of 50 mm Fe-EDTA (Figure 8E), and it strongly decreased from the transition zone throughout the elongation zone. Exposure to Fe excess in the growth media decreased O 2 $ concentration in the meristematic and transition zones compared with Fe control conditions, and this decrease was exacerbated in the LR of the fer1-3-4 triple mutant. Taken together, these results suggested that Fe in excess modified ROS distribution both in PR and LR. These modifications were correlated with effects on PR and LR length (Figure 5B and 5D). Furthermore, the loss of function of ferritin in the fer1-3-4 triple mutant specifically altered H 2 O 2 and O 2 $ distribution in the transition and meristematic zones of LR, particularly after growth in the presence of Fe excess. In order to establish whether there is a direct link among Fe homeostasis, ROS distribution, and RSA, wild-type and fer1-3-4 plants were grown under control or Fe excess conditions and treated with either KI, a known H 2 O 2 scavenger (Tsukagoshi et al., 2010), or SHAM, a peroxidase inhibitor (Spreen Brouwer et al., 1986), prior to measuring root length. KI treatment did not affect the size of the PR under Fe control conditions and partly restored its growth under Fe excess conditions for both wild-type and fer1-3-4 plants (Figure 9A). In contrast, SHAM treatment reduced PR length in both wild-type and fer1-3-4 plants, irrespective of the Fe concentration used in the growth medium (Figure 9B). These results indicate that the decrease in the size of the PR in response to Fe treatment was correlated with ROS production and distribution, in agreement with the model of Tsukagoshi et al. (2010). Concerning LR, the only significant effect of KI was to reduce their average length Molecular Plant 8, , March 2015 ª The Author

8 Fe Homeostasis and Root Architecture Figure 7. ROS Distribution in the PR Tips. (A) 3 0 -(p-hydroxyphenyl) fluorescein (HPF) and (B) DHE fluorescence photographs in the PR from wild-type Col0 and fer1-3-4 mutant grown on MS/2 medium containing 50 mm Fe(III)-EDTA during 7 days and then transferred on the same medium containing 50 or 500 mm Fe(III)-EDTA during 4 days. (C) Overlay of the photographs presented in (A) and (B). (D) Fluorescence intensity of HPF along PR tip. (E) Staining intensity of DHE along PR tip. Mean values ± standard error are shown (n R 18 plants). Arrows indicate the transition zone. in wild-type plants at 500 mm Fe-EDTA (Figure 9C). SHAM treatment decreases the average LR length for both wild-type and fer1-3-4 plants, regardless of the Fe concentration used, and this decrease was more pronounced in the fer1-3-4 triple mutant at the highest Fe concentration (Figure 9D). The impact of Fe homeostasis on LR growth appeared therefore to be connected to ROS homeostasis. The observed effect of SHAM treatment on LR growth supports the model of Tsukagoshi et al. (2010), whereas the KI effect that we observed does not. Influence of Fe Homeostasis on Root Meristem Activity In Arabidopsis, an increase in the rate of root growth is correlated with an increase in meristem size as a result of increased rates of cell division (Ubeda-Tomás et al., 2009). Since Fe homeostasis affects root length (Figure 5), it was important to determine the 446 Molecular Plant 8, , March 2015 ª The Author effect of Fe excess and of loss of function of ferritins on the size of the PR and LR meristems. We observed that Fe excess treatment significantly decreased the size of the PR and LR meristems, both in wild-type and the fer1-3-4 triple mutant (Figure 6A and 6B). Under control Fe conditions (50 mm Fe-EDTA), the size of the LR meristems was already decreased in ferritin triple mutant fer1-3-4 compared with wild-type plants (Figure 6B). An additive effect between the loss of function of the Fer1, 3, and 4 genes in the triple mutant and the decrease in size of LR meristems after growth in excess Fe was also observed (Figure 6B). The size of the PR meristems was reported to be controlled by a ROS pathway involving the UPB1 transcription factor, which regulates the expression of a specific set of peroxidases (Tsukagoshi et al., 2010). In order to investigate the function of UPB1 in control of RSA, the same parameters shown in Figure 5B 5D were

9 Fe Homeostasis and Root Architecture Molecular Plant Figure 8. ROS Distribution in the LR Tips. (A) 3 0 -(p-hydroxyphenyl) fluorescein (HPF) and (B) DHE fluorescence photographs in the LR from Col0 and fer1-3-4 mutant grown as indicated in Figure 7. (C) Overlay of the photographs presented in (A) and (B). (D) Fluorescence intensity of HPF along the LR tip. (E) Staining intensity of DHE along the LR tip. Fluorescence intensity is measured on the six youngest LRs per plant longer than 1.5 mm. Mean values ± standard error are shown (n R 18 plants). Arrows indicate the transition zone. measured in the upb1 mutant in response to Fe excess (Supplemental Figure 3). The upb1 mutant exhibited a longer PR (Supplemental Figure 3A) and LR (Supplemental Figure 3B) than wild-type plants, whatever the Fe concentration used. These data support the conclusion that UPB1-dependent control of root growth is still functional under Fe excess, and this transcription factor also regulates LR growth. The upb1 mutant also exhibits a higher LR density under Fe excess. Meristem sizes were increased in the upb1 mutant and decreased in UPB1 over expressor lines, correlating the quantity of UPB1 to the size of the root meristems. Furthermore, it is known that UPB1 expression is positively regulated by oxidative stress (Tsukagoshi et al., 2010). This knowledge prompted us to measure the transcript abundance of UPB1 in roots in response to Fe excess and/or in a ferritin-less mutant genetic background (Figure 10A). An increase in abundance of UPB1 mrna was observed in roots of the fer1-3-4 triple mutant compared with wild-type plants, and this was enhanced by Fe excess (Figure 10A), consistent with LR meristem decrease in size measured under these various conditions (Figure 6). This result suggested that Fe excess and/or ferritin loss of function could be negative regulators of the cell cycle through the ROS pathway, leading ultimately to a decrease in root meristem size. It was recently documented that two members of the SIAMESE-RELATED (SMR) cyclin-dependent kinase inhibitors (CKI) gene family, namely SMR5 and SMR7, were transcriptionally activated by ROS, leading to activation of cell division arrest by cell cycle checkpoint (Yi et al., 2014). We therefore measured SMR5 and SMR7 transcript abundance in roots of wild-type and the fer1-3-4 triple mutant after growth in Fe control conditions or after a 9-h Fe excess treatment (Figure 10B and 10C). SMR5 and SMR7 expression appeared to be significantly downregulated in wild-type after Fe excess treatment, whereas expression of Molecular Plant 8, , March 2015 ª The Author

10 Fe Homeostasis and Root Architecture Figure 9. Effects of ROS Level Changes on PR and LR Length. PR length (A and B) and average LR length (C and D) of plants grown on MS/2 medium containing 50 mm Fe(III)-EDTA during 7 days and then transferred on the same medium containing 50 or 500 mm Fe(III)-EDTA supplemented with 100 mm potassium iodide (KI [A and C]) or 100 mm salicylhydroxamic acid (SHAM [B and D]) during 7 days. LR length (C, D) was measured only on LRs formed below the location of the PR tip at the moment of the transfer. Mean values ± standard error are shown (n % 10 plants). Different letters (a, b, c, or d) indicate statistical differences (P < 0.05; Student s t-test). SMR5 and SMR7 was upregulated in roots of the fer1-3-4 triple mutant. SMR1 expression, known to be unaffected by oxidative stress (Yi et al., 2014), was unaffected under our conditions (Figure 10D). These observations suggest that altered Fe homeostasis, and consequently, altered ROS production in the fer1-3-4 triple mutant, affect the expression of CKIs genes known to be activated by oxidative stress. DISCUSSION Varying initiation of LRP, frequency of LR emergence, and growth rate of both PR and LR give a huge combinatorial potential to modifying RSA (Petricka et al., 2012). RSA plasticity constitutes a major determinant of plant adaptation to soil heterogeneity, contributing to optimization of foraging for minerals and water acquisition from soil (Giehl et al., 2014). RSA is genetically determined and auxin plays a critical role in the control of root development, both at the embryonic root formation level and for LR formation (Petricka et al., 2012; Lavenus et al., 2013). More recently, ROS homeostasis has emerged as a major regulator of RSA by modulating PR growth (Dunand et al., 2007; Tsukagoshi et al., 2010) and LR emergence (Manzano et al., 2014). Some of the steps in the genetic developmental program of root are highly responsive to environmental cues, with nutrient availability being one of the more critical (Lopez- Bucio et al., 2003; Gruber et al., 2013; Kellermeier et al., 2014). 448 Molecular Plant 8, , March 2015 ª The Author Most of the information available on the impact of nutrient deficiency on RSA has been obtained for N, P, K, and S, singly or in combination (Gruber et al., 2013; Kellermeier et al., 2014), but much less is known about how other elements reprogram RSA. Among them, Fe is known to reshape root architecture by modifying root hair patterning (Schmidt et al., 2000), by acting on LR elongation through modification of auxin distribution (Giehl et al., 2012), or by interacting with phosphorus to control PR growth (Ward et al., 2008). It is also well established that Fe and ROS homeostasis are intimately linked, with ferritins as major actors in this interaction (Briat et al., 2010b; Sudre et al., 2013). We observe that AtFer1 and AFer3 ferritin genes are expressed in the meristems and in the endodermis and pericycle cell layers, which are essential tissues in determining RSA. Part of the root intracellular Fe appears to be contained within the ferritin protein, since in fer1-3-4 roots (1) Fe concentration decreased by 20% compared with Col0 plants (Figure 3A) and (2) Fe allocation was modified (Figure 3D and 3E). As a consequence of their ability to store excess Fe, ferritins participated in protecting roots against Fe-mediated oxidative stress (Figure 4), as previously reported for other organs (Ravet et al., 2009a; Briat et al. 2010b). Consequently, Fe homeostasis (Fe excess and/or ferritins) interfered with ROS distribution in PR and/or LR (Figures 6 and 7) and could therefore be involved in RSA. Indeed, Fe excess led to a significant decrease of PR length (Figure 5), in agreement with the report by Gruber et al. (2013), showing a significant decrease in PR length at 75 mm Fe, compared with 5 or 10 mm. This correlated with a decrease in the size of the meristem (Figure 6A) and with a decrease in the O 2 $ content of the proliferation zone, whereas H 2 O 2 fluorescence was slightly increased in the transition zone (Figure 7). These observations were in agreement with the model proposed by Tsukagoshi et al. (2010), establishing the role of ROS signaling in PR growth. It was also consistent with a partial restoration of PR growth by KI treatment and with enhancement of the decrease in PR length in response to SHAM treatment under Fe excess conditions (Figure 9A and

11 Fe Homeostasis and Root Architecture Molecular Plant Figure 10. Fe Regulation of Genes Controlling Meristem Activity. Relative transcript abundance of UPB1 (A), SMR1 (B), SMR5 (C), and SMR7 (D) from roots of Col-0 and fer1-3-4 plants grown on MS/2 medium containing 50 mm Fe(III)-EDTA during 7 days and then transferred on the same medium containing 50 or 500 mm Fe(III)-EDTA for 9 h. Mean values ± standard error are shown (n = 4). Different letters (a, b, or c) indicate statistical differences (P < 0.05; Student s t-test). response to Fe, but in a contrasted manner compared with PR (Figure 9). SHAM treatment gave expected results predicted by the model, whereas KI treatment gave unexplained opposite results from those predicted by this model. 9B). However, ferritin mutations had no effect on PR growth (Figures 5 and 8) or on PR meristem size (Figure 6A). AtFer1 and AtFer3 ferritin genes were nevertheless expressed in the PR meristems (Figure 2A). In conclusion, PR shortening under Fe excess could be attributed to the ROS signaling pathway independently of the presence of ferritins. The effect of Fe excess on LR growth appears more complex than for PR. After a rapid decrease of LR length between 50 and 150 mm Fe-EDTA, both in wild-type and ferritin-less mutant plants, the length of the root was less, or even not, affected at higher Fe concentrations (Figure 5C). LR length of Col0 plants was identical at 50 or 500 mm Fe-EDTA, despite the fact that Fe excess decreased meristem size and altered O 2 $ /H 2 O 2 balance (Figures 6B and 7). This suggests that at least two pathways with opposite effects on LR growth were at work. In fer1-3-4 mutant plants, O 2 $ /H 2 O 2 balance was more altered in response to Fe compared with wild-type plants, and LR length was 20% shorter than in wild-type plants at 500 mm Fe-EDTA, indicating that ferritins were an element of these opposite pathways. Therefore, the model proposed by Tsukagoshi et al. (2010) for PR development does not fully explain LR growth. This statement is reinforced by the observation that KI and SHAM treatments modify LR length in The most important LR phenotype in response to increasing Fe concentration in the culture medium concerned the increase of LR density in wild-type plants, which was alleviated in fer1-3-4 plants (Figure 5D). This indicated that increasing Fe concentration had a positive effect on the number of productive primordia, and that ferritins were required for this phenomenon to take place. It was explained by a higher LR emergence frequency rather than an effect on primordia initiation events (Figure 5E and Supplemental Table 1). It was consistent with the report showing that ROS signaling involving UPB1-regulated peroxidases controlled LR emergence (Manzano et al., 2014), and with the tissue-specific expression of ferritin in pericyle cells and LRP (Figure 2), where LR initiate (Malamy and Benfey, 1997). Although PR and LR growth is controlled by common mechanisms such as auxin transport through the PIN transporters (for review, see Tian et al., 2014), ferritin mutations do not affect PR and LR growth similarly (Figure 5). This can be related to the fact that different PR or LR pathways are known to participate in the growth control of these two types of roots, in particular in response to environmental constraints (Tian et al., 2014). For example, a phosphate deficiency is known to stimulate LR growth and repress PR growth. These opposite responses are due to better auxin transport and sensitivity, specifically in LR (Nacry et al., 2005; Perez-Torres et al., 2008), and PR growth is specifically repressed by the multicopper oxidases LPR1 and 2 (Svistoonoff et al., 2007). Ferritins could therefore interact differentially with these distinct regulatory mechanisms occurring in LR or PR, explaining their different impact on the growth of these two organs. Fe homeostasis had an impact on PR and LR length, and LR density (Figure 5), consistent with a decrease in the size of meristems (Figure 6) and with a decrease in LR emergence efficiency Molecular Plant 8, , March 2015 ª The Author

12 (Figure 5E and Supplemental Table 1), suggesting an interaction of Fe homeostasis with control of the cell cycle. The ROS signaling pathway involved in the control of RSA required UPB1 with an abundance that correlated with the size and length of the PR meristem (Tsukagoshi et al., 2010; Supplemental Figure 3A), the efficiency of LR emergence affecting LR density (Manzano et al., 2014; Supplemental Figure 3C), and LR root length (Supplemental Figure 3B). UPB1-dependent control of root growth is still functional under Fe excess, and this transcription factor also regulates LR growth and density under Fe excess (Supplemental Figure 3). Increase in the abundance of UPB1 transcript in response to Fe excess and in ferritin-less plants (Figure 10A) was therefore consistent with a role of Fe homeostasis in RSA via a ROS pathway. A decrease in meristem size indicated a lower rate of cell divisions (Ubeda- Tomás et al., 2009). In yeast, Fe overload induced translational repression of Cln1 and Cln2 cyclins, arresting cell division at the G1 regulatory start point (Philpott et al., 1998). In Arabidopsis, hydroxyurea treatment led to ROS-dependent transcriptional activation of SMR5/SMR7 genes whose products belong to a class of CKIs (Yi et al., 2014). We found that these genes upregulated in roots of fer1-3-4 plants under Fe excess conditions (Figure 10B and 10C), where intensity of oxidative stress (Figure 4), decrease in meristem size (Figure 6), and decrease in LR density were more pronounced compared with other conditions tested. This suggested that the control of Fe homeostasis was necessary to allow the cell cycle to proceed. SMR5/SMR7 gene activation was triggered by ROS rather than replication problems, linking these genes with cell cycle checkpoint activation upon the occurrence of ROS-mediated DNA damage (Yi et al., 2014). This suggested that Fe-mediated DNA damage could lead to cell division arrest in root meristems. In contrast, cell division arrest in yeast by Fe overload did not require activation of the DNA damage checkpoint governed by RAD9 (Philpott et al., 1998). Such a difference between yeast and plants could be related to the unexpected high Fe concentration found in plant nuclei (Roschzttardtz et al., 2011). Fe homeostasis is an important parameter for RSA by affecting both PR and LR growth and development. Independent of ferritins, Fe excess decreases PR length through modification of O 2 $ /H 2 O 2 balance, according to the model proposed by Tsukagoshi et al. (2010). Fe excess affects the growth of LR to a lesser extent than that of PR. Ferritins are required to optimize LR emergence under high Fe concentration, and Fe excess and/or ferritin mutations decrease PR and LR meristem sizes. This is consistent with the upregulation of ROS-induced genes at high Fe concentration, which inactivate cell division (Figure 10; Yi et al., 2014). LR elongation triggered by localized Fe supply is mediated by an alteration of the AUX1-mediated auxin distribution (Giehl et al., 2012). How this hormonal pathway and the ROS pathway interact to adapt the RSA to changes in Fe availability remains to be elucidated. METHODS Plant Materials Arabidopsis thaliana (Columbia ecotype, Col0) plants were used for all experiments. Atfer1-3-4 T-DNA insertion mutant (Ravet et al., 2009a, 2009b) and patfer1::gus transgenic plants (Tarantino et al., 2003) have already been reported. A 1.8 kb and a 1.4 kb DNA fragment from the AtFer3 and AtFer4 promoter regions, respectively, were amplified from genomic DNA using primers (Supplemental Table 2) and cloned in the NcoI and BamHI sites of pcambia These plasmids were used to produce transgenic plants, as reported for the patfer1::gus construct (Tarantino et al., 2003). Plants were grown under hydroponic conditions for AtFer1 4 expression analysis, GUS and Perls-DAB staining, and Fe content measurement. The liquid nutrient solution, renewed every 5 days, contained 1.5 mm Ca(NO 3 ), 1.5 mm KNO 3, 0.75 mm KH 2 PO 4, 1.5 mm KCl, 0.75 mm MgSO 4,50mM KCl, 10 mm MnCl 2, 1.5 mm CuCl 2, 2 mm ZnCl 2,50mM H 3 BO 3, mm (NH 4 ) 6 Mo 7 O 24,50mM Fe(III)-EDTA, 0.5 g/l 2-(N-morpholino)ethanesulfonic acid (Mes), ph adjusted to 5.7 with KOH. For RSA analysis, catalase activity, MDA content, ROS staining, meristem size measurement, and gene expression analysis, plants were grown on MS/2 with 6 g/l agar, 0.5 g/l Mes, ph adjusted to 5.7 with KOH, 50 mm Fe(III)-EDTA for 7 days, and then transferred on the same medium containing different Fe(III)-EDTA concentrations as indicated. Plants were grown under 16 h light (150 mmol/s/m light intensity)/8 h dark at 21 C and 18 C, respectively, and at 70% hygrometry. Root and leaf samples were collected and ground in liquid nitrogen prior to storage at 60 C, before protein or RNA extractions for qrt-pcr, Western blot, or enzyme activity analysis. RNA Extraction and Analysis Total RNA was isolated from rosettes or roots with the Tri-reagent according to the manufacturer s instructions (Molecular Research Center). One microgram of total RNA treated with RQ1 DNase (Promega) was used for reverse transcription (Goscript reverse transcriptase; Promega) with oligo(dt) 18 and 0.4 mm dntp. cdna was diluted twice with water, and 1 ml of each sample was assayed by qrt-pcr in a LightCycler 480 (Roche) using LC480-SYBR-Green master I (Roche). Expression levels were calculated relatively to the PP2 gene (At1g13320) using the comparative threshold cycle method. Primers are presented in Supplemental Table 2. Protein Extraction and Western Blot Analysis Proteins were extracted, resolved by 12% SDS-PAGE, and electroblotted according to Ravet et al. (2009a). Ferritin immunodetection was performed using a rabbit polyclonal antiserum recognizing the four ferritins subunits (Ravet et al., 2009a) and the Immun-Star AP Substrate (Bio-Rad). GUS Staining GUS root staining with 1 mm 5-bromo-4-chloro-3-indolyl-b-D-glucuronide and root cross-sections were performed according to Tarantino et al. (2003). Determination of Fe Content and Localization Apoplastic Fe was washed out according to Bienfait et al. (1985). Root and shoot samples (10 30 mg dry weight) were mineralized with a Speed Wave Two (Berghof) in 1 ml of 48% HNO 3 and 7.5% H 2 O 2, 3 min at 100 C, 20 min at 180 C, and 10 min at 140 C. Fe concentration was determined as previously described (Lobréaux et al., 1992). Histochemical Fe staining was achieved by the Perls-DAB method reported by Roschzttardtz et al. (2009). Catalase and MDA Measurements Fe Homeostasis and Root Architecture Plants were grown on plates containing 50 mm Fe(III)-EDTA for 7 days and then transferred on plates containing 50 or 250 mm Fe(III)-EDTA for 7 more days. Catalase activity and MDA content were determined in crude root extract at 20 C, as previously described (Cakmak and Marschner, 1992; Hodges et al., 1999). Protein concentration was determined using Bio- Rad Protein Assay Dye Reagent using bovine serum albumin as a standard. 450 Molecular Plant 8, , March 2015 ª The Author 2015.

13 Fe Homeostasis and Root Architecture Determination of Root Architecture Parameters Plants were grown on MS/2 plates containing 50 mm Fe(III)-EDTA for 7 days and then transferred on the same medium containing 50, 150, 300, or 500 mm Fe(III)-EDTA for 7 additional days. Plants were scanned with a V750 scanner (Epson). Root length was measured with ImageJ using the NeuronJ plug-in (Meijering et al., 2004). PR elongation, LR density, and LR length were measured only on the part of root that was newly formed after the plate transfer. No measurement was made on LR formed above the location of the PR tip at the moment of the transfer. LR density was defined as the number of emerged LR per unit length of the root branching zone, according to Dubrovsky and Forde (2012). Measurements were performed on at least 20 different plants and were repeated in at least two independent experiments. For the determination of the number of LRP, roots were fixed and cleared according to Truernit et al. (2008). The roots were then transferred onto microscope slides, covered with Hoyer s solution (3 g gum arabic, 20 g chloral hydrate, 2 g glycerol, and 5 ml water) and observed by Nomarski optics on an Olympus BX61 microscope. ROS Staining DHE was used for O 2 $ staining and HPF for H 2 O 2 determination (Dunand et al., 2007; Tsukagoshi et al., 2010). Each root was double stained with both probes. Plants were incubated in darkness in liquid MS/2 medium containing 10 mm DHE for 25 min, then 5 mm HPF for 3 min, before washing with MS/2 medium for 3 min. Fe(III)-EDTA concentrations (50 or 500 mm) were maintained in this liquid MS/2 medium. Microscope Observations and Image Analysis GUS and Fe staining images were collected with an Olympus SZX16 stereomicroscope, and images of cross-sections and ROS staining with an Olympus BX61 microscope. For HPF fluorescence, excitation was in the range of nm and detection in the range of nm. For DHE fluorescence, excitation was in the range of nm and detection with a longpass filter of 590 nm. The ImageJ (Schneider et al., 2012) Plot profile function was used to assess HPF and DHE intensity along the root tips, with a line width one-third of the width of the root. Image analysis was performed by quantifying the pixel intensity through three virtual lines chosen arbitrarily on both edges and in the central part of the root. The mean value of these three measurements was calculated and assumed to be representative of the overall ROS staining of a given root tip. The position of the transition zone in the root tip was determined after propidium iodine staining. The meristem size was obtained by measuring the distance between above the root cap and the transition zone. HPF and DHE fluorescence and the transition zone position in LR were measured on the six youngest LR longer than 1.5 mm. SUPPLEMENTAL INFORMATION Supplemental Information is available at Molecular Plant Online. FUNDING G.R. was supported by a fellowship from the French Ministère de l Enseignement Supérieur et de la Recherche. ACKNOWLEDGMENTS We acknowledge Professor David Salt (University of Aberdeen, UK) for critical reading of this manuscript, Professor Philip N. Benfey (University of Duke, USA) for the gift of upb1 mutant seeds, and Mrs Chay (B&PMP, Montpellier, France) for her help with the Fe measurements. No conflict of interest declared. Received: September 3, 2014 Revised: October 20, 2014 Accepted: November 2, 2014 Published: November 9, 2014 Molecular Plant REFERENCES Arnaud, N., Murgia, I., Boucherez, J., Briat, J.F., Cellier, F., and Gaymard, F. (2006). An iron-induced nitric oxide burst precedes ubiquitin-dependent protein degradation for Arabidopsis AtFer1 ferritin gene expression. J. Biol. Chem. 281: Bienfait, H.F., van den Briel, W., and Mesland-Mul, N.T. (1985). Free space iron pools in roots: generation and mobilization. Plant Physiol. 78: Bournier, M., Tissot, N., Mari, S., Boucherez, J., Briat, J.F., and Gaymard, F. (2013). Arabidopsis ferritin 1 (AtFer1) gene regulation by the phosphate starvation response 1 (AtPHR1) transcription factor reveals a direct molecular link between iron and phosphate homeostasis. J. Biol. Chem. 288: Briat, J.F., Duc, C., Ravet, K., and Gaymard, F. (2010a). Ferritins and iron storage in plants. Biochim. Biophys. Acta 1800: Briat, J.F., Ravet, K., Arnaud, N., Duc, C., Boucherez, J., Touraine, B., Cellier, F., and Gaymard, F. (2010b). New insights into ferritin synthesis and function highlight a link between iron homeostasis and oxidative stress in plants. Ann. Bot. 105: Cakmak, I., and Marschner, H. (1992). A rapid, sensitive, and specific method for the determination of protein in dilute solution. Plant Physiol. 98: Conte, S.S., and Walker, E.L. (2011). Transporters contributing to iron trafficking in plants. Mol. Plant 4: Curie, C., Cassin, G., Couch, D., Divol, F., Higuchi, K., Le Jean, M., Misson, J., Schikora, A., Czernic, P., and Mari, S. (2009). Metal movement within the plant: contribution of nicotianamine and yellow stripe 1-like transporters. Ann. Bot. 103:1 11. Dubrovsky, J.G., and Forde, B.G. (2012). Quantitative analysis of lateral root development: pitfalls and how to avoid them. Plant Cell 24:4 14. Duc, C., Cellier, F., Lobréaux, S., Briat, J.F., and Gaymard, F. (2009). Regulation of iron homeostasis in Arabidopsis thaliana by the clock regulator time for coffee. J. Biol. Chem. 284: Dunand, C., Crèvecoeur, M., and Penel, C. (2007). Distribution of superoxide and hydrogen peroxide in Arabidopsis root and their influence on root development: possible interaction with peroxidases. New Phytol. 174: Galatro, A., and Puntarulo, S. (2007). Mitochondrial ferritin in animals and plants. Front. Biosci. 12: Giehl, R.F., Lima, J.E., and von Wirén, N. (2012). Localized iron supply triggers lateral root elongation in Arabidopsis by altering the AUX1- mediated auxin distribution. Plant Cell 24: Giehl, R.F., Gruber, B.D., and von Wiren, N. (2014). It s time to make changes: modulation of root system architecture by nutrient signals. J. Exp. Bot. 65: Gruber, B.D., Giehl, R.F., Friedel, S., and von Wirén, N. (2013). Plasticity of the Arabidopsis root system under nutrient deficiencies. Plant Physiol. 163: Halliwell, B., and Gutteridge, J.M. (1992). Biologically relevant metal iondependent hydroxyl radical generation. An update. FEBS Lett. 307: Hodges, D.M., DeLong, J.M., Forney, C.F., and Prange, R.K. (1999). Improving the thiobarbituric acid-reactive-substances assay for estimating lipid peroxidation in plant tissue containing anthocyanin and other interfering compounds. Planta 207: Kellermeier, F., Armengaud, P., Seditas, T.J., Danku, J., Salt, D.E., and Amtmann, A. (2014). Analysis of the root system architecture of Arabidopsis provides a quantitative readout of crosstalk between nutritional signals. Plant Cell 26: Molecular Plant 8, , March 2015 ª The Author