Fitting into the Harsh Reality: Regulation of Irondeficiency Responses in Dicotyledonous Plants

Transcription

1 Molecular Plant Volume 5 Number 1 Pages January 2012 REVIEW ARTICLE Fitting into the Harsh Reality: Regulation of Irondeficiency Responses in Dicotyledonous Plants Rumen Ivanov, Tzvetina Brumbarova and Petra Bauer 1 Department of Biosciences Plant Biology, Saarland University, Campus A2.4, D Saarbrücken, Germany ABSTRACT Iron is an essential element for life on Earth and its shortage, or excess, in the living organism may lead to severe health disorders. Plants serve as the primary source of dietary iron and improving crop iron content is an important step towards a better public health. Our review focuses on the control of iron acquisition in dicotyledonous plants and monocots that apply a reduction-based strategy in order to mobilize and import iron from the rhizosphere. Achieving a balance between shortage and excess of iron requires a tight regulation of the activity of the iron uptake system. A number of studies, ranging from single gene characterization to systems biology analyses, have led to the rapid expansion of our knowledge on iron uptake in recent years. Here, we summarize the novel insights into the regulation of iron acquisition and internal mobilization from intracellular stores. We present a detailed view of the main known regulatory networks defined by the Arabidopsis regulators FIT and POPEYE (PYE). Additionally, we analyze the root and leaf ironresponsive regulatory networks, revealing novel potential gene interactions and reliable iron-deficiency marker genes. We discuss perspectives and open questions with regard to iron sensing and post-translational regulation. Key words: Iron uptake; gene expression; transcription factors; post-transcriptional regulation. IRON ACQUISITION IN DICOTYLEDONOUS MODEL PLANTS As the primary producer in land ecosystems, plants are responsible for the main supply of bioavailable nutrients. The efficiency at which these become accessible can be crucial to the fitness of the species and the stability of the ecosystem as a whole. Iron (Fe) is one of the most abundant elements on the planet and is also essential for life on Earth, as both plants and animals require it for a high number of enzymatic reactions. However, due to its low solubility in aerobic and neutral ph environments, iron availability to the living organisms is very low. This is a problem concerning directly human nutrition, as the lack of iron results in anemia, which can manifest in fatigue and illness and, in infants, may result in developmental or mental retardation. Once germinated, sessile plants have to cope with their local environment and therefore they have developed a number of adaptations in order to accommodate themselves to the surrounding conditions. The need for efficient acquisition of iron from the soil has resulted in the evolution of two phylogenetically distinct uptake strategies (Marschner and Römheld, 1986; Römheld and Marschner, 1986). Dicotyledonous plants, such as fruit trees, pea (Pisum sativum), tomato (Solanum lycopersicum), potato (Solanum tuberosum), rapeseed (Brassica napus), as well as non-graminaceous monocots, employ a reduction-based strategy, named Strategy I (Figure 1A), where the rhizosphere is acidified by proton extrusion, thus rendering the Fe 3+ complexes soluble, followed by the reduction of Fe 3+ to Fe 2+ and the subsequent uptake of the latter (Marschner, 1995). Studies on tomato and the model plant Arabidopsis (Arabidopsis thaliana) have helped define the role of several key components of this system. On the other hand, graminaceous monocots, such as barley (Hordeum vulgare), maize (Zea mays), and rice (Oryza sativa), apply a chelation-based strategy, named Strategy II. These plants extrude phytosiderophores that chelate iron and the resulting complexes are imported by the roots. The present review concentrates primarily on Strategy I. Several overviews on the major advances on the regulation of Strategy II from recent years have been published (Walker and Connolly, 2008; Morrissey and Guerinot, 2009; Kobayashi et al., 2010). Members of the Arabidopsis H + -ATPase (AHA) protein family have emerged as the most probable factors responsible for soil acidification. The role of AHA2 in proton extrusion under low iron availability has been suggested in the early nineties 1 To whom correspondence should be addressed. p.bauer@mx.unisaarland.de, tel. (+49) , fax (+49) ª The Author Published by the Molecular Plant Shanghai Editorial Office in association with Oxford University Press on behalf of CSPB and IPPE, SIBS, CAS. doi: /mp/ssr065, Advance Access publication 26 August 2011 Received 28 May 2011; accepted 04 July 2011

2 28 Ivanov et al. d Regulation of Iron-Deficiency Response Figure 1. Reduction-Based Strategy for Iron Uptake in the Model Plant Arabidopsis thaliana. (A) Iron is first solubilized by rhizosphere acidification through the action of the H + -ATPase AHA2, and is then reduced from ferric (Fe 3+ )to ferrous (Fe 2+ ) iron by the reductase FRO2. Bivalent iron is then imported into the root cell by the metal transporter IRT1. The activity of this uptake system is dependent on the action of the transcription factor FIT. Under iron deficiency, up-regulation of the iron-responsive genes is achieved through a complex including FITand at least one of the two bhlh proteins (bhlh038 and bhlh039), and presumably also bhlh100 and bhlh101. The FIT gene is induced by this system and thus it undergoes a feed-forward regulation, where the gene product positively regulates the source gene. Post-transcriptional regulation of protein abundance is not depicted here. Induction of FRO2 and IRT1 activity is co-regulated in response to iron deficiency, while that of AHA2 seems to be regulated in an independent manner (yellow arrows). (B) Expression of the AHA2 gene is dependent on FIT. Expression was analyzed by quantitative RT PCR in wild-type Arabidopsis (Col-0 ecotype), a FIT overexpressing line (FIT Ox), and the fit-3 mutant line under iron-sufficient (50 lm Fe) or iron-deficient (0 lm Fe) conditions. AHA2 gene was induced by iron deficiency in both Col-0 and the FIT Ox line but not in the fit-3 mutant. Therefore, FIT is necessary but not sufficient for AHA2 expression. Expression of FIT and IRT1 was monitored as control. t-tests showed all inductions to be statistically significant (p-value, 0.005). (Sussman, 1994) and was further supported by the recent analysis of three available aha2 loss-of-function lines (Santi and Schmidt, 2009). Reduction of iron occurs through the action of the factors from the FERRIC REDUCTASE-OXIDASE (FRO) family. In Arabidopsis, the frd1 mutants, carrying mutations in the coding sequence of the FRO2 gene, are deficient in iron reductase activity (Yi and Guerinot, 1996; Robinson et al., 1999). The IRON-REGULATED TRANSPORTER 1 (IRT1) is the primary iron importer in the roots, as seen by the phenotypes of IRT1-defective plants (Eide et al., 1996; Henriques et al., 2002; Varotto et al., 2002; Vert et al., 2002). IRT1 is a predicted plasma membrane protein with eight transmembrane domains (Eide et al., 1996). The major transcriptional regulator of iron-deficiency responses in Arabidopsis is the bhlh transcription factor

3 Ivanov et al. d Regulation of Iron-Deficiency Response 29 FIT. It is homologous to the previously identified FER transcription regulator in tomato (Ling et al., 2002; Bauer et al., 2004; Colangelo and Guerinot, 2004; Jakoby et al., 2004; Yuan et al., 2005; Bauer et al.,2007). Arabidopsis plants containing loss-of-function mutations in the FIT gene do not manifest iron reductase activity and cannot mobilize iron (Colangelo and Guerinot, 2004; Jakoby et al., 2004; Yuan et al., 2005; Zhang et al., 2006). Redistribution of iron, both between the plant organs and at subcellular level, is a crucial step for its proper storage and utilization. It can be chelated and transported as iron nicotianamine and iron citrate complexes by transporters, such as the YELLOW STRIPE-LIKE (YSL) family proteins (Curie et al., 2001, 2009; Conte and Walker, 2011) and FERRIC REDUCTASE DE- FICIENT3 (FRD3) (Rogers and Guerinot, 2002; Green and Rogers, 2004; Durrett et al., 2007), respectively, or as free iron by various divalent metal transporters (Curie et al., 2009; Jeong and Guerinot, 2009; Conte and Walker, 2011). Here, we summarize the latest developments in studying the regulation of iron-deficiency responses in the root on transcriptional and post-transcriptional levels. We present new information in the context of the already established models. Detailed aspects of iron transport throughout the plant and on a subcellular level have been extensively discussed recently (Curie et al., 2009; Giehl et al., 2009; Jeong and Guerinot, 2009; Conte and Walker, 2011; Nouet et al., 2011) and therefore are not specifically described here. TRANSCRIPTOME CHANGES IN RESPONSE TO IRON DEFICIENCY Several recent studies have shown that most of the changes in the root transcriptome in response to iron deprivation occur early, up to 24 h after exposure to iron deficiency (Dinneny et al., 2008; Buckhout et al., 2009) and seem to be predominantly general stress responses (Dinneny et al., 2008; Schuler et al., 2011) or anticipated responses, like the changes of suggested copper and zinc redistribution genes in response to iron deficiency (Yang et al., 2010), because of the unspecific uptake of other bivalent metals together with iron (Korshunova et al., 1999; Vert et al., 2002). The study by Dinneny et al. (2008) identified that the highest transcriptional response to iron deficiency occurs in the tissues of the central cylinder. A high portion of these genes comprises the plant general stress response. In comparison, the epidermis is far less responsive at the level of gene expression and these changes are more specific to iron deficiency. While the different sets of data seem to be in good agreement, building a model based on them is difficult due to the fact that changes in gene expression are rarely mirrored on the level of protein abundance (Brumbarova et al., 2008; Donnini et al., 2010; Rellan-Alvarez et al., 2010; Lan et al., 2011) and posttranscriptional regulation plays an important role in irondeficiency responses in both tomato and Arabidopsis (Connolly et al., 2002; 2003; Colangelo and Guerinot, 2004; Jakoby et al., 2004; Brumbarova and Bauer, 2005; Kerkeb et al., 2008; Ravet et al., 2009; Lingam et al., 2011; Sivitz et al., 2011). TOPOLOGY OF THE IRON-DEFICIENCY RESPONSE IN THE ROOT In Strategy I plants, roots respond to iron deficiency through morphological changes that result in an increased root surface area for the reduction and transport of iron. These include the formation and branching of root hairs, swelling of root tips, enhanced lateral root development, reduced lateral root growth (Schmidt, 1999; Muller and Schmidt, 2004), and certain plants develop specialized transfer cells (Kramer et al., 1980; Landsberg, 1986). Additionally, riboflavin is synthesized in the roots (Susin et al., 1993; Welkie, 2000). Expression patterns of the Arabidopsis FIT and IRT1 as observed by using promoter reporter fusions suggest that the main activity of the iron uptake system is along the root hair zone including the part that precedes the root hair zone (Jakoby et al., 2004; Seguela et al., 2008). This is in good correlation with the results from early experiments in pea demonstrating that the activity of the iron reductase is present mainly in the root hair zone of the primary and the lateral roots. Almost no iron reduction could be observed at the root tip (Grusak et al., 1990). However, similar experiments in tomato revealed that the majority of the iron reductase activity occurs in the sub-apical region (Brown and Ambler, 1974; Bell et al., 1988). Once iron is solubilized, it can freely enter the apoplastic space of the root. At this point, its passive distribution towards the conductive tissues is limited by the cellular membranes and the Casparian strip of the endodermis (Marschner, 1995). In order to be made available for the plant and further delivered to the aboveground organs, iron needs to be actively transported into the cells. Theoretically, any cell in contact with apoplastic iron may be able to import iron, which would result in a large total surface and high uptake efficiency. Expression data of iron uptake genes from different species, however, indicate that iron uptake may happen predominantly in the epidermal layer. This has been shown for the pea FRO1 (Waters et al., 2002), tomato IRT2 (Bereczky et al., 2003), and Arabidopsis IRT1 (Vert et al., 2002), FRO2 (Connolly et al., 2003), and FIT (Colangelo and Guerinot, 2004; Jakoby et al., 2004). Additionally, immunolocalization of the tomato FER protein showed a strong localization in the epidermis and in the tissues of the central cylinder. This pattern of protein localization is preserved even when FER mrna expression is controlled by a ubiquitous viral promoter, suggesting that such an expression pattern is important to the plant and is ensured at two regulatory levels (Brumbarova and Bauer, 2005). A second regulatory system has recently been identified that seems to be responsible for the internal mobilization of iron in the root and its transport to the shoot. It is controlled by another bhlh transcription factor, POPEYE (PYE), which targets directly several genes involved in metal ion homeostasis, including a nicotianamine synthase (NAS4), a FRO family reductase (FRO6), and the zinc transport facilitator ZIF1 (Long et al., 2010).

4 30 Ivanov et al. d Regulation of Iron-Deficiency Response REGULATION OF IRON UPTAKE COMPONENTS OF THE FIT REGULATORY NETWORK The first success in identifying a gene encoding a transcriptional regulator of the iron-deficiency response in plants was the identification of the FER gene in tomato (Ling et al., 2002). FER is induced by iron deficiency and encodes a root-specific bhlh protein that localizes in the nucleus, activates transcription, and up-regulates several iron-responsive genes, such as IRT1 and NRAMP1 (Ling et al., 2002; Bereczky et al., 2003; Brumbarova and Bauer, 2005). A search for the homolog of FER in Arabidopsis identified the gene BHLH029, which was simultaneously investigated by three different groups under three different names: FIT1, FRU, and BHLH029 (Bauer et al., 2004; Colangelo and Guerinot, 2004; Jakoby et al., 2004; Yuan et al., 2005). To avoid further confusion, the name FIT was agreed upon and is currently in use (Bauer et al., 2007). Expression of this gene in tomato was sufficient to complement the fer mutation, demonstrating experimentally that it is the functional ortholog of FER in Arabidopsis (Yuan et al., 2005). FIT, similarly to FER, is located in the nucleus. Mutagenesis of four arginine residues in the basic region resulted in cytoplasmic localization, suggesting that this region contains the nuclear localization signal (Zhang et al., 2006). FIT regulates the expression of the iron reductase FRO2 and the transporter IRT1 (Colangelo and Guerinot, 2004; Jakoby et al., 2004). However, only in certain weak fit alleles, fit-1, and some FIT antisense lines, IRT1 expression can still be observed (Colangelo and Guerinot, 2004; Zhang et al., 2006). Interestingly, it was discovered that overexpression of FIT does not result in the expected strong induction of FRO2 and IRT1 in roots (Colangelo and Guerinot, 2004; Jakoby et al., 2004). In tomato, immunoblot experiments and immunolocalization on lines with constitutive FER expression showed that the protein availability was controlled independently of the FER mrna abundance, both on whole root and tissue level (Brumbarova and Bauer, 2005). Ectopic expression of FIT in Arabidopsis leaves led to the up-regulation of FRO2 and IRT1 only under iron deficiency (Jakoby et al., 2004). Therefore, it can be speculated that the activity of FIT is regulated at the post-transcriptional level. In two recent reports, Sivitz et al. (2011) and Lingam et al. (2011) demonstrated that, similarly to the FIT mrna, FIT protein abundance is enhanced upon iron deprivation. Chemical treatments of Arabidopsis lines expressing FIT fusions point at turnover of FIT, which is dependent on the 26S proteasome. Treatment with the protein synthesis inhibitor cycloheximide resulted in low FIT abundance, while the application of the proteasome inhibitor MG132 had the opposite effect. The authors proposed that, under iron deficiency, FIT binds to its target promoters and is rapidly inactivated and degraded. New protein is freshly synthesized in order to ensure the presence of active FIT and this leads to a very high FIT turnover rate (Sivitz et al., 2011). In addition, Lingam et al. (2011) reported that the abundance of FIT is dependent on ethylene signaling via the ETYLENE INSENSITIVE3 (EIN3) and ETYLENE INSENSI- TIVE3-LIKE 1 (EIL1) transcription factors involved in response to the phytohormone ethylene. Ethylene is known to promote the iron-deficiency responses (Romera and Alcantara, 1994; Lucena et al., 2006; Waters et al., 2007) and this effect seems to be closely related to the increased gene expression of FIT, FRO2, and IRT1 and for the post-translational regulation of FIT protein. Stability of FIT is enhanced in the presence of EIN3/EIL1. FIT levels are strongly reduced in the ein3 eil1 double mutant and in the wild-type upon treatment with the ethylene inhibitor AVG. These effects are reversed by application of MG132, suggesting that the ethylene-mediated stabilization of FIT also involves the 26S proteasome. FIT can interact with EIN3/EIL1 and the authors speculated that this interaction may prevent the premature degradation of FIT. Thus, the ethylene-dependent stabilization of FIT may be a prerequisite for successful response (Lingam et al., 2011). A lysine-63-linked ubiquitin conjugase, UBC13, was identified by its accumulation in response to iron deficiency in cucumber (Cucumis sativus). The double ubc13a ubc13b mutant in which the two homologous Arabidopsis genes UBC13A and UBC13B have been mutated shows abnormal iron-deficiency responses, which include the enhanced regulation of IRT1 and AHA2, together with increased iron reductase activity (Li and Schmidt, 2010a). At this moment, though, there is no evidence for a molecular link between UBC13A/B and FIT. Further studies are needed to fully explain the role of FIT stability in regulation of iron uptake. A potential way of regulating the activity of FIT has been suggested, involving four additional members within the Arabidopsis BHLH gene family: BHLH038, BHLH039, BHLH100, and BHLH101 (Heim et al., 2003; Toledo-Ortiz et al., 2003). An early indication that these are involved in the regulation of iron uptake was that all four of them are up-regulated under iron starvation (Vorwieger et al., 2007; Wang et al., 2007). By applying the yeast two-hybrid approach and bimolecular fluorescence complementation in leaf protoplasts, Yuan et al. (2008) were able to demonstrate that FIT can interact with itself and with bhlh038 and bhlh039. Therefore, it was proposed that, in order to be active as transcriptional regulator, FIT forms a complex with one of these bhlh proteins. In the study of Yuan et al. (2008), overexpression of any of these BHLH genes alone did not result in misregulation of FRO2 or IRT1, similarly to the FIT overexpression. On the contrary, when FIT was overexpressed together with either BHLH038 or BHLH039, a strong up-regulation of the iron uptake machinery was observed even at sufficient iron supply (Yuan et al., 2008). While neither bhlh100 nor bhlh101 has been reported to interact with FIT so far, based on their induction in the root by iron starvation and sequence similarity, it can be speculated that overexpression and interaction experiments could lead to a similar outcome. One remaining problem is the ultimate proof that the four bhlh proteins are involved in iron-deficiency responses. The single bhlh38-1, bhlh039-1, bhlh100-1, and bhlh101-1 loss-of-function mutants show no iron uptake-related

5 Ivanov et al. d Regulation of Iron-Deficiency Response 31 phenotype perhaps due to functional redundancy (Wang et al., 2007), so the generation of higher-order mutants is required in this case. While all of these five BHLH genes are up-regulated upon iron starvation, the expression pattern of the four homologs differs from that of FIT. Promoter GUS fusions, supported by RT PCR experiments, showed expression for all four of them in the root hair zone but not at the root tip, and very strong expression in leaves, unlike FIT, which is only weakly expressed outside the root. Additionally, split root experiments and mutant analysis indicated that the expression of FIT and the other four BHLH genes is regulated by different pathways, possibly by different signals in response to iron deficiency (Wang et al., 2007). From the analysis of FIT expression in the fit-4 mutant, which cannot produce FIT protein due to the presence of an early stop codon, it was suggested that the FIT protein acts as a positive regulator of its own gene (Jakoby et al., 2004). This feed-forward regulation could later be confirmed using a FITpromoter GUS fusion in wild-type and fit mutant backgrounds. At the same time, expression of BHLH038, BHLH039, BHLH100, and BHLH101 was not dependent on FIT (Wang et al., 2007). The H + -ATPase AHA2 is responsible for the main acidification activity in response to iron deprivation and the AHA2 gene is up-regulated upon iron deficiency (Santi and Schmidt, 2009). In order to find out whether this induction is dependent on FIT, similarly to the case of FRO2 and IRT1, we analyzed the AHA2 expression in wild-type, FIT gain, and loss-of-function lines by quantitative RT PCR (Figure 1B). Lack of iron resulted in the up-regulation of AHA2, together with the control genes IRT1 and FIT in the wild-type. In the fit-3 mutant, no induction was observed for AHA2 and only a very weak one for IRT1. In the FIT Ox gain-of-function line, AHA2 and IRT1 remained iron-responsive, despite the FIT overexpression, supporting the idea that FIT activity is subject to post-transcriptional control as discussed above. Therefore, FIT is necessary but not sufficient for the up-regulation of the AHA2 gene in response to the iron status of the plant. A recent study on the post-transcriptional regulation of AHA2 showed that the phosphorylation of a threonine residue (Thr 947) might lead to the increase of the ATPase activity (Haruta et al., 2010). A similar regulatory mechanism has been observed for yeast plasma membrane ATPases (Lecchi et al., 2007). Interestingly, AHA2 was shown to be redundant to AHA1 and inactivation of one of them leads to enhanced phosphorylation of the other (Haruta et al., 2010). AHA1, however, does not seem to be directly involved in the response to iron deficiency (Santi and Schmidt, 2009). The soil acidification activity of AHA2, required for the initial solubilization of iron in the soil, seems to be regulated independently of the iron reduction and transport. Despite the altered soil acidification in the aha2 mutants, the reductase activity remained unchanged (Santi and Schmidt, 2009). In a similar manner, in the frd1-1 mutant, defective for the FRO2 gene, the soil acidification activity corresponds to that of the wild-type (Yi and Guerinot, 1996). This shows that the two functions are not coupled and may be used independently of each other in processes not related to iron uptake. Another potentially important AHA isoform is AHA7. The AHA7 gene is also induced under iron deficiency and its up-regulation is dependent on FIT (Colangelo and Guerinot, 2004). Mutant plants lacking a functional AHA7 gene could induce acidification activity under iron deprivation similarly to the wild-type; however, aha7 mutant roots showed compromised root hair density, more pronounced under iron deficiency (Santi and Schmidt, 2009). This suggests that AHA7 might be involved in iron deficiency not as an acidification factor, but as a regulator of root developmental responses. Expression of FRO2 and IRT1 genes was shown to be tightly co-regulated both in response to the iron supply and diurnally (Vert et al., 2003), suggesting that they might be physically controlled by the same set of regulators. However, while FIT seems to control the up-regulation of FRO2 under iron deficiency, the expression pattern of IRT1 in some weak fit mutant alleles (Colangelo and Guerinot, 2004; Yuan et al., 2008) suggests that another factor might be able to activate this gene in the absence of FIT. As is the case for the transcriptional regulators, the gene expression of FRO2 and IRT1 is not a direct measurement of the activity of their products. Overexpression of FRO2 did not result in misregulation of the iron reductase activity (Connolly et al., 2003). A FIT bhlh couple seems to be necessary and sufficient for the activation to happen, even at sufficient iron (Yuan et al., 2008). It is possible that an active FIT bhlh complex controls in addition the expression of a post-translational regulator of FRO2. Additionally, Durrett et al. (2007) described two frd4 mutants lacking a thylacoid protein importer, cpftsy, which was unable to induce FRO2 activity in the root under iron starvation, despite the up-regulation of the FRO2 gene. Therefore, lack of cpftsy seems to target a post-transcriptional regulator of FRO2. Localization of a functional FRO2 HA fusion was suggested to be at the plasma membrane based on immunolocalization and was not changed in the frd4 mutant, suggesting that the lack of FRO2 activity is not due to mislocalization (Durrett et al., 2006). A similar scenario might be applicable in the case of IRT1. Transgenic plants expressing IRT1 mrna under the control of a ubiquitous viral promoter are unable, under iron-deficient conditions, to produce IRT1 protein in any plant tissue except the root (Connolly et al., 2002). Additionally, in the weak fit-1 mutant, despite the relatively high IRT1 expression under low iron availability, IRT1 protein is not detectable. It has been proposed that FIT regulates a factor that is required for the stability of IRT1 (Colangelo and Guerinot, 2004). Additional evidence for the post-translational control of IRT1 is its rapid disappearance upon resupply of iron (Connolly et al., 2002). Vert et al. (2002, 2009) suggested that the IRT1 protein fused to GFP in protoplasts is targeted to the plasma membrane, which is consistent with its proposed function of a metal importer. In contrast, IRT2 GFP fusion is localized in intracellular compartments (Vert et al., 2009), suggesting a sequential role of the two

6 32 Ivanov et al. d Regulation of Iron-Deficiency Response in the transport of iron in the cell. In a recent study, Barberon et al. (2011) used an immunolocalization approach with an antibody against IRT1 on Arabidopsis roots and obtained results that strongly contradict earlier data. IRT1 signal was found in the trans-golgi network/early endosomes (TGN) and almost no signal could be detected at the plasma membrane. Using chemical treatments, the authors demonstrated that IRT1 functions for a short time at the plasma membrane and is then internalized to the TGN. This finding corresponds well to earlier reports of proteins accumulating at locations different from the place of their action, as is the case with other transmembrane proteins (Ivanov and Gaude, 2009a, 2009b), or even transcriptional regulators (Ryu et al., 2007; Ivanov et al., 2008; Lindermayr et al., 2010). A GFP fusion of an IRT1 homolog from apple (Malus xiaojinensis), MxIRT1 GFP, was expressed in yeast and localized transiently to the plasma membrane under iron deficiency, after which it re-localized to internal compartments. This effect was more strongly pronounced when iron deficiency was enhanced by the addition of the metal chelator ferrozine (Li et al., 2006). No such dynamics have been demonstrated in planta, but it suggests that the cellular trafficking machinery might be a regulator of the iron-deficiency response. ZRT1, a member of the ZIP family transporters to which IRT1 belongs (Guerinot, 2000), was shown to be ubiquitinated at a specific lysine residue within the loop region, which results in its internalization (Gitan and Eide, 2000). Two lysines are present in the large cytoplasmic loop of IRT1 and their mutation into arginine resulted in an enhanced stability of the protein (Kerkeb et al., 2008). As lysine is the target amino acid for protein ubiquitination and on the basis of the ZRT1 data, it could be speculated that the mutations disturb ubiquitin attachment sites and the protein cannot be endocytosed and targeted for degradation. Using immunoprecipitation of non-mutated and mutated forms of IRT1, and a combination of anti-ubiquitin antibodies, it was recently shown that IRT1 is monoubiquitinated before being sent for vacuolar degradation (Barberon et al., 2011). This process was first demonstrated for the animal EGFR receptor kinase (Haglund et al., 2003) and seems to be common for both plant and animal cells (Gohre et al., 2008; Gimenez-Ibanez et al., 2009). SIGNALS MODULATING THE EFFICIENCY OF THE IRON UPTAKE MACHINERY The iron uptake system is known to respond to local and longdistance shoot-to-root signals, shown by mutant to wild-type grafting, leaf excision, and split-root experiments (Grusak et al., 1990; Welch and Larue, 1990; Grusak and Pezeshgi, 1996; Vert et al., 2003; Wang et al., 2007; Enomoto and Goto, 2008). Therefore, extensive cross-talk and fine balancing of uptake in the context of the homeostasis of the whole plants have been inferred. Within the root, several processes are known to occur as a prerequisite for a successful induction of the iron-deficiency response by FIT. The induction of the uptake machinery in the root is preceded by rapid nitric oxide (NO) synthesis. Indeed, this was demonstrated to be an essential step in both tomato and Arabidopsis,as chemical treatments that deplete internal NO or NO biosynthesis mutants are not able to induce the iron uptake machinery, while, on the contrary, treatment with NO donors results in its stimulation (Graziano et al., 2002; Graziano and Lamattina, 2007; Besson-Bard et al., 2009; Chen et al., 2010). Additionally, both carbon dioxide (CO 2 ) and carbon oxide (CO) can stimulate the iron-deficiency-responsive genes in tomato an effect that, at least in the case of CO 2,still requiresnoproduction(jinetal.,2009;kongetal.,2010). A recent study by Chen et al. (2010) demonstrated that plants that are defective in auxin transport are unable to induce the expression of either FIT, or any of its target genes. Crosstalk between auxin biosynthesis and signaling, and iron uptake has already been suggested earlier (Schmidt, 1999; Schmidt et al., 2000; Rampey et al., 2006), and this finding offers great opportunities for further research. At the moment, it is not clear whether auxin acts locally or it can serve as a long-distance signal as well. The phytohormone ethylene has been known to positively regulate iron-deficiency responses (Romera and Alcantara, 1994; Li and Li, 2004; Lucena et al., 2006; Waters et al., 2007; Zuchi et al., 2009; Garcia et al., 2010; Lingam et al., 2011; Romera et al., 2011). As already mentioned, one of the mechanisms by which this is achieved is the regulation of the FIT protein levels through interaction with the transcription factors EIN3 and EIL1 (Lingam et al., 2011). Effects of other hormones on iron acquisition have also been investigated. Cytokinin can act as a negative regulator of iron uptake. Strong down-regulation of FRO2 and IRT1 expression in the fit-1 mutant in the presence of cytokinins suggests that this negative regulation is independent of FIT (Seguela et al., 2008). Additionally, ABA suppresses iron-deficiency responses (Seguela et al., 2008) and jasmonic acid was also shown to participate in fine tuning the responses to low iron as a negative regulator (Maurer et al., 2011). The iron-deficiency sensing mechanism has not yet been identified in plants. A recent study of the Arabidopsis iron homeostasis mutant nas4x-1 by Klatte et al. (2009) showed that the lack of proper distribution of iron within the leaf might affect the induction of the root uptake system. This mutant has strongly reduced levels of the metal chelator nicotianamine and shows phenotypes similar to the NA-free tomato mutant chloronerva (Becker et al., 1995; Ling et al., 1999; Bereczky et al., 2003) and the transgenic tobacco plants with reduced NA content (Herbik et al., 1999; Takahashi et al., 2003). In vegetative nas4x-1 leaves, the expression of the iron-responsive marker gene BHLH100 was increased, despite the presence of sufficient iron amounts (Klatte et al., 2009), suggesting that iron fails to reach its sensor due to the lack of the chelator NA. Similar conclusions can be made from the analysis of two other mutants (frd3 and opt3-2) defective in long-distance iron transport. The frd3 mutant lacks a MATE family transporter and also displays constitutive activity of the iron uptake system (Delhaize, 1996; Yi and Guerinot, 1996; Rogers and Guerinot, 2002). FRD3 is involved in loading citrate Fe complexes into

7 Ivanov et al. d Regulation of Iron-Deficiency Response 33 the xylem of the root prior to their delivery to the leaf mesophyll cells (Green and Rogers, 2004; Durrett et al., 2007). The opt3-2 mutant has a strongly suppressed expression of an oligonucleotide transporter and fails to export iron from the leaf vasculature, resulting in constitutive iron-deficiency response (Stacey et al., 2008). Additionally, Durrett et al. (2006), using a chloroplast protein translocation mutant, were able to show that the chloroplast integrity is essential for the induction of the iron reductase activity in the root under iron deficiency. The involvement of the chloroplast Fe S clusters in iron-deficiency sensing has been speculated; however, at the moment, there is no experimental evidence that would support this (Jeong and Guerinot, 2009). As with the sensor, the mechanism through which shootto-root signaling is achieved remains a mystery. Several hints have been obtained in recent years. As already mentioned, a role of auxin as a mediator in iron uptake induction has been uncovered and further investigations might demonstrate whether it may be used as a long-distance iron-deficiency signal. Another possible signal has emerged from the data of Buhtz et al. (2010). In this study, the authors identified five small RNAs, whose abundance in the phloem sap specifically decreased upon iron deficiency. Interestingly, these were upregulated upon copper starvation and one of them, mir399, had already been suggested as a long-distance signal for phosphate deficiency responses (Aung et al., 2006; Chiou et al., 2006; Pant et al., 2008). Interestingly, mir158b was up-regulated in the phloem sap under iron deficiency. This micro RNA was also identified in the study of Kong and Yang (2010) among 60 small RNAs up-regulated in Arabidopsis seedlings by insufficient iron. Currently, the role and the mechanism of action of these small RNAs remain unknown; however, it is possible that their differential regulation upon iron deficiency is a result of secondary adjustments of the metabolism. REGULATION OF IRON MOBILIZATION BY PYE Fine tissue-specific analysis of gene expression showed that the majority of iron deprivation-induced changes occur in the tissues of the central cylinder. Interestingly, the largest functional group among the differentially regulated genes was that of the transcriptional regulators (Dinneny et al., 2008; Long et al., 2010). At present, two genes, encoding the bhlh protein POPEYE (PYE) and the putative DNA-binding E3 ubiquitin protein ligase BRUTUS (BTS), have been investigated in more detail (Long et al., 2010). PYE expression could be detected in all root tissues but it was found strongly up-regulated in the central cylinder under iron deficiency. However, the nuclear localized PYE GFP fusion seems equally present in all root layers, suggesting that it functions in multiple cell layers and it may be subject to post-transcriptional regulation. Under iron deficiency, the pye-1 mutant is severely chlorotic, shows retarded growth and defective root hair formation. Several genes, among them BHLH039 and BHLH101, were strongly up-regulated in the mutant compared to the wild-type, which, together with its early up-regulation under iron starvation, suggests a potential role of PYE in the regulation of iron mobilization. Additionally, several genes, which show specific endodermis/central cylinder expression, are highly upregulated in the mutant. These include the two key genes OPT3 and FRD3, involved in the long-distance iron transport (Green and Rogers, 2004; Stacey et al., 2008) and the gene encoding the NRAMP4 transporter (Lanquar et al., 2005). Three other stele expressed genes ZIF1 (Haydon and Cobbett, 2007), NAS4, and FRO3 (Dinneny et al., 2008) were up-regulated in the pye-1 mutant and were identified as direct targets of PYE. The significance of PYE for iron homeostasis is underlined by the strong retention of iron in the pye-1 mutant roots (Long et al., 2010). Therefore, in the root, PYE seems to regulate the mobilization of iron and its further translocation to the green organs. Yeast two-hybrid experiments suggested that PYE may interact with other members of the bhlh family, bhlh115 and ILR3 (bhlh105) possibly forming a functional heterodimer. Interestingly, ILR3 has previously been proposed as a link between auxin availability and metal homeostasis (Rampey et al., 2006). Both bhlh115 and ILR3 interact in yeast two-hybrid assay with the protein BRUTUS (BTS). The BTS gene is also expressed strongly in the stele under iron deficiency. It is also up-regulated in the pye-1 mutant, which indicates that it may be a target of PYE. Due to the fact that BTS and PYE do not interact and due to the opposite phenotypes of bts-1 and pye-1 mutants, it was suggested that their simultaneous presence in the cell ensures the balanced expression of iron homeostasis genes (Long et al., 2010, Figure 2). This hypothesis still needs to be confirmed by an in planta demonstration of these interactions. IDENTIFICATION OF RELIABLE MARKER GENES FOR ARABIDOPSIS IRON DEFICIENCY Investigating iron deficiency requires standardizing conditions for plant growth and using reliable marker genes as controls. In order to obtain optimal responses to phytohormones under different iron supply conditions, seedlings may be preferred over older plants. If large amounts of root material are required, the hydroponic system might be more suitable than the growth on agarose plates. Additionally, the contents of the growth medium may vary, especially with the use or not of chelators for the removal of iron ion traces. As a result, plants may respond to the presence of factors in their environment that are not necessarily linked to iron deficiency. Here, we thoroughly reviewed published microarray data in order to identify genes that are differentially regulated by iron irrespective of growth conditions, or the age of plants. Such genes are more likely to respond specifically to iron, rather than represent a secondary effect from the treatment, and could serve as reliable markers for iron deficiency. For this

8 34 Ivanov et al. d Regulation of Iron-Deficiency Response Figure 2. Hypothetical Model of the Regulation of Iron Mobilization by PYE. PYE and BTS have opposite effects on plant mobilization judged by the phenotypes of the corresponding mutants. Both proteins were shown to interact with ILR3 and bhlh115. It was suggested that the PYE ILR3/bHLH115 interaction might negatively regulate the known PYE target genes ZIF1, FRO3, and NAS4. On the other hand, it can be speculated that the BTS ILR3/bHLH115 interaction might have the opposite effect. Additionally, PYE represses BTS expression, thus probably influencing BTS protein abundance. Thus, a fine balance in the regulation of iron mobilization is achieved. This model is based on the assumption that these protein protein interactions occur in plant cells and that the BTS ILR3/bHLH115 complex influences directly or indirectly the three target genes. purpose, we have compared two new independent datasets from microarray analyses on 6-day-old wild-type seedlings (Lingam et al., 2011) and 6-week-old plants (Schuler et al., 2011) in order to select genes regulated by iron supply in both conditions and growth stages. In 6-week-old plants, roots and leaves were analyzed separately. The entire lists of iron-deficiency-regulated genes in wild-type from both sets of experiments are shown in Supplemental Table 1. The genes selected from the comparison of the two microarray datasets as being commonly regulated in both experiments are listed in Table 1. Among the differentially expressed genes, we were able to identify genes from three distinct groups based on the pattern and topology of expression. These were up-regulated in roots under iron deficiency (five genes), up-regulated in leaves under iron deficiency (13 genes), and up-regulated in leaves in ironsufficient conditions (one gene). The potential group up-regulated in roots under sufficient iron yielded no genes. We have compared this list of genes to several other available genome-wide expression datasets on Arabidopsis roots from plants grown under deficient versus sufficient iron supply (Colangelo and Guerinot, 2004; Dinneny et al., 2008; Buckhout et al., 2009; Yang et al., 2010). The genes selected by our comparison had been identified as iron deficiency up-regulated in these studies as well, with two exceptions. The gene At3g56980 encoding for bhlh039 was not identified by Colangelo and Guerinot (2004). This is surprising, as it has since been shown to be stably up-regulated by iron deficiency and is a possible regulator of the iron uptake (Wang et al., 2007; Yuan et al., 2008). Additionally, the gene At3g50740 (encoding for UDP glucosyl transferase 72E1) was not identified among the early up-regulated genes targeted in the study of Buckhout et al. (2009). As high amounts of zinc were shown to cause the up-regulation of the iron-deficiency responses, we have also taken into account the transcriptomic data obtained by Becher et al. (2004) and van de Mortel et al. (2006), describing the responses to zinc excess in shoots and roots, respectively. In roots, all the five genes up-regulated stably at low iron could be identified as up-regulated by zinc excess in the van de Mortel et al. (2006) dataset; however, only two of the genes, which we have identified as reliably up-regulated in leaves, matched the dataset of Becher et al. (2004). This observation suggests that a strong cross-talk exists between the regulation of irondeficiency responses and zinc homeostasis in the root, while, in leaves, the responses to iron and zinc differ. To gain further insight into the networks of genes that comprise the iron-deficiency response, we performed an analysis of the co-expression networks using the online tool ATTED-II and the Cytoscape software (Obayashi et al., 2009) (Figures 3 and Figure 4). The genes identified in our comparisons (see Table 1) were used as a query. As a result, the generated clusters correspond to the three above-described expression groups. Inroots,thefivegenesup-regulatedunderirondeficiencywere grouped in a single cluster with several sub-clusters of tightly coregulatedgenes.fourofthequerygenes(at3g07720,at3g50740, At3g58810, At4g19680) with predicted function in iron transport and metal tolerance formed a large cluster, together with other known or suggested iron-deficiency-related genes, such as those encoding the transporter IRT1 and the iron-induced transcription factormyb10(buckhoutetal.,2009)(figure3).unfortunately,the FITgene(At2g28160) wasnotavailableforanalysisinthetool. The second sub-cluster, containing BHLH039, could clearly be identified as the PYE BTS regulon, also observed in the study by Long et al. (2010). It contains PYE and BTS as well as their direct targets FRO3, NAS4,and ZIF1. Additionally, this cluster contains the irondeficiency up-regulated BHLH101 and BHLH039. The cluster of genes up-regulated by sufficient iron in leaves was generated on the basis of only one gene, FER1, and predictably included another ferritin gene, FER4. Analysis of the 13 genes up-regulated by iron deficiency in leaves resulted in three separate clusters, with two of them based on the single genes At1g12030 (Figure 4B) and At2g35190 (Figure 4C). The cluster around At1g12030 (unknown function) contains genes with functions predominantly linked to sulfur assimilation (e.g. the thioredoxins APR1, APR2, and APR3), sulfate transport (e.g. SULTR4;1 and SULTR4;2), and genes induced by sulfur deficiency (e.g. SDI1, LSU1, and LSU2). Therefore, this cluster implies the necessity of rebalancing the sulfur metabolism in response to iron deficiency. An additional connection between iron and sulfur homeostasis comes by the fact that many of the genes up-regulated by sulfur deficiency responses are regulated by the FIT interactor and ethylene response

9 Ivanov et al. d Regulation of Iron-Deficiency Response 35 Table 1. List of Genes Differentially Regulated between Iron-Sufficient and Deficient Conditions in Wild-Type 6-Day-Old Seedlings and 6- Week-Old Plants and Comparison with Related Transcriptomic Datasets. ATI gene code Description Regulation according to other genome-wide gene expression studies Genes up-regulated at + Fe versus Fe in leaves Becher et al. (2004) At5g01600 FER1, ferretin 1 Genes up-regulated at Fe versus + Fe in leaves Becher et al. (2004) At1g12030 Protein of unknown function (DUF506) At1g23020 FRO3, ferric reductase oxidase 3 + At1g47400 Unknown protein At2g04050 MATE efflux family protein At2g35190 NPSN11, novel plant snare 11 At3g07800 Thymidine kinase At3g oxoglutarate and Fe(II)-dependent oxygenase superfamily protein At3g18290 BTS, EMB2454 At3g27060 TSO2, ferritin/ribonucleotide reductase-like family protein At3g61630 CRF6, cytokinin response factor 6 At4g16370 OPT3, oligopeptide transporter + At4g17240 Unknown protein At5g05180 Unknown protein Genes up-regulated at + Fe versus Fe in roots No genes Genes up-regulated at Fe versus + Fe in roots Colangelo and Guerinot (2004) Dinneny et al. (2008) At3g07720 Galactose oxidase, kelch repeat superfamily protein Buckhout et al. (2009) Yang et al. (2010) At3g50740 UGT72E1, UDP-glucosyl transferase 72E At3g56980 BHLH039, ORG At3g58810 MTP3, MTPA2, metal tolerance protein A At4g19680 IRT2, iron regulated transporter van de Mortel et al. (2006) Nineteen genes were identified as being significantly differentially regulated in wild-type Col-0 plants between iron-sufficient (+Fe) and deficient ( Fe) conditions in two different growth setups 6-day-old seedlings and 6-week-old plants (Lingam et al., 2011; Schuler et al., 2011). The list of differentially expressed genes in leaves was compared to the results on gene expression in leaves under zinc excess presented in Becher et al. (2004). The list of differentially expressed genes in roots was compared to the results of five other transcriptome analyses in roots under iron starvation (Colangelo and Guerinot, 2004; Dinneny et al., 2008; Buckhout et al., 2009; Yang et al., 2010) and zinc excess (van de Mortel et al., 2006). Zinc excess causes up-regulation of iron-deficiency-induced genes in roots. A plus sign indicates that the gene appeared regulated in the same manner in the corresponding analysis. A minus sign indicates that the gene was not identified as differentially regulated in the corresponding analysis. Cases of opposite regulation of the same gene in different microarray datasets were not observed. transcription factor EIN3 (Maruyama-Nakashita et al., 2006). Ethylene response and iron transporter-encoding genes, such as OPT3, seem to be also affected upon compromised sulfur response (Lewandowska et al., 2010; Watanabe et al., 2010). The At2g35190 cluster (Figure 4C) is enriched in cytokinesis-related genes, like two tubulins and NPSN11, a SNARE protein potentially required for the formation of the cell plate during cytokinesis together with KNOLLE (Zheng et al., 2002). The third and largest cluster contains several subdivisions (Figure 4D). One oftheseleafsub-clusterscoincideswiththepye BTSregulonthat was observed also in roots (Figure 3), suggesting that this is not a root-specific network, supporting the expression data for BHLH039 and BHLH101 (Wang et al., 2007). This might indicate thatpyeand thesetwobhlhsmayalso be theregulatorsof irondeficiency responses in shoots. Interestingly, another sub-cluster could be identified that contained numerous genes encoding proteins related to cell cycle control, like the cyclin CYCB1;1, the replication protein A, RPA70C, a DNA polymerase (At1g49980), and the telomerase TRFL10. The same cluster contains also DNA damage repair-related genes, such as RAD17, RAD51, BRCA1, and PARP1. While it is known that activation of cell cycle causes enhanced activity of the DNA repair machinery (Gospodinov et al., 2003; Kunz et al., 2005), this finding was surprising, as there are no reports of enhanced DNA replication

10 36 Ivanov et al. d Regulation of Iron-Deficiency Response Figure 3. Iron Supply-Dependent Co-Expression Networks in Arabidopsis Roots. Genes stably up-regulated upon iron deficiency presented in Table 1 (labeled here in yellow) were used as a basis for the generation of the co-expression network. The cluster containing the PYE BTS regulon is surrounded by a dashed red line. Networks were generated by the ATTED-II NetworkDrawer tool and the data were further processed using the Cytoscape Software. activity in leaves upon iron starvation. However, the resulting chlorosis may contribute to the enhanced susceptibility of DNA to UV irradiation (Triantaphylides and Havaux, 2009), or the depleted iron amounts may cause the misfunction of iron-dependent proteins, thus enhancing the probability of generation of reactive oxygen species (Andaluz et al., 2006; Brumbarova et al., 2008; Laganowsky et al., 2009). Interestingly, there have also been speculations based on computational predictions of protein protein interactions linking the irondeficiency-induced lysine-63-linkedubiquitinconjugase UBC13A to DNA repair (Li and Schmidt, 2010b). A cell division category significantly affected by iron deficiency was also identified in the study by Schuler et al. (2011). Inconclusion, the describedanalysisofgeneco-expressionnetworks presents a powerful basis for the investigation of iron-deficiency responses on the whole-plant level. It demonstrates the potential cross-talks betweenironhomeostasis and DNA damage repair, as well as sulfur metabolism in leaves, which may open further topics of iron homeostasis investigation. The shortlisted genes, presented in Table 1, can be used as reliable markers for iron deficiency in roots and leaves. The rootregulated genes have already been identified in previous studies, while the findings on gene regulation in leaves present novel aspects of responses to iron deficiency on the level of the whole plant. As post-translational events play a major role in the regulation of iron-deficiency responses, it will be of great interest to

11 Ivanov et al. d Regulation of Iron-Deficiency Response 37 Figure 4. Iron Supply-Dependent Co-Expression Networks in Arabidopsis Leaves. Genes stably up-regulated upon iron-sufficient (A) or deficient conditions (B D) presented in Table 1 (labeled here in yellow) were used as a basis for the generation of the co-expression networks. In (D), the cluster containing the PYE BTS regulon is surrounded by a dashed red line and that enriched in cell cycle and DNA repair genes by a dashed blue line. Networks were generated by the ATTED-II NetworkDrawer tool and the data were further processed using the Cytoscape Software.