Disruption of OsYSL15 leads to iron inefficiency in rice plants.

Uptake and translocation of metal nutrients are essential processes for plant growth. Graminaceous species release phytosiderophores that bind to Fe3+; these complexes are then transported across the plasma membrane. We have characterized OsYSL15, one of the rice (Oryza sativa) YS1-like (YSL) genes that are strongly induced by iron (Fe) deficiency. The OsYSL15 promoter fusion to β-glucuronidase showed that it was expressed in all root tissues when Fe was limited. In low-Fe leaves, the promoter became active in all tissues except epidermal cells. This activity was also detected in flowers and seeds. The OsYSL15:green fluorescent protein fusion was localized to the plasma membrane. OsYSL15 functionally complemented yeast strains defective in Fe uptake on media containing Fe3+-deoxymugineic acid and Fe2+-nicotianamine. Two insertional osysl15 mutants exhibited chlorotic phenotypes under Fe deficiency and had reduced Fe concentrations in their shoots, roots, and seeds. Nitric oxide treatment reversed this chlorosis under Fe-limiting conditions. Overexpression of OsYSL15 increased the Fe concentration in leaves and seeds from transgenic plants. Altogether, these results demonstrate roles for OsYSL15 in Fe uptake and distribution in rice plants.

Iron (Fe), an essential nutrient for plants, plays a crucial role in a variety of cellular functions. Because plants are the primary source of food for humans, their nutritional value is important to health. The most widespread dietary problem in the world is Fe deficiency (World Health Organization, 2003). Such a deficiency also causes a metabolic imbalance deleterious to plant growth (e.g. impairing chlorophyll biosynthesis, chloroplast development, and photosynthesis). Therefore, Fe availability is directly correlated with plant productivity.
Despite its abundance in soils, Fe is present as oxihydrates with low bioavailability. To avoid a deficiency, two distinct strategies are possible for Fe acquisition (Marschner et al., 1986). In strategy I, used by dicotyledonous and nongraminaceous monocotyledonous plants, Fe 2+ transport is coupled to an Fe 3+ chelate reduction step. Under Fe deficiency, protons are released via H + -ATP pumps into the rhizosphere to lower the soil pH, and subsequently, Fe 3+ -chelate reductase is induced to reduce Fe 3+ to the more soluble Fe 2+ , which is then absorbed by a specific transporter. The plasmalemma root ferric-chelate reductase, FRO2, reduces soil Fe 3+ (Robinson et al., 1999) and provides Fe 2+ for IRT1, a major metal transporter that takes up Fe 2+ into the root epidermis. This is evidenced by the lethal chlorotic phenotypes of IRT1 knockout mutants (Eide et al., 1996;Henriques et al., 2002;Varotto et al., 2002;Vert et al., 2002).
With strategy II, low Fe availability in the soil is overcome in grasses such as maize (Zea mays) and rice (Oryza sativa). In response to Fe deficiency, these crops synthesize and release nonproteinogenic amino acids in the mugineic acid family of phytosiderophores (MAs) to fix Fe 3+ in the soil (Takagi et al., 1984). All MAs are synthesized from L-Met, sharing the same pathway from L-Met to 2#-deoxymugineic acid (DMA; Bashir et al., 2006). L-Met is adenosylated by S-adenosylmethionine synthetase (Takizawa et al., 1996). Nicotianamine synthase (NAS) catalyzes the trimerization of S-adenosylmethionine molecules to form nicotianamine (NA; Higuchi et al., 1999), which is then converted into a 3$-keto intermediate via the transfer of an amino group by nicotianamine aminotransferase (NAAT; Inoue et al., 2008). Genes encoding NAS and NAAT for MA biosynthesis have been isolated (Higuchi et al., 1999;Takahashi et al., 1999;Inoue et al., 2003Inoue et al., , 2008. The subsequent reduction of the 3$-carbon in the keto intermediate produces DMA by DMA synthase (Bashir et al., 2006). In response to Fe deficiency, these grasses increase their production and secretion of MAs into the rhizosphere, where they chelate various metals, including Fe 3+ . The Fe 3+ -MA complexes are then taken up by an Fe 3+ -MA transporter. The Yellow Stripe1 (YS1) gene has been isolated and characterized in maize, where it encodes the high-affinity Fe 3+ -MA transporter (Curie et al., 2001). YS1 functions as a proton-coupled symporter for phytosiderophore (PS)-chelated metals (Curie et al., 2001;Roberts et al., 2004;Schaaf et al., 2004).
Among the 18 YS1-like (YSL) genes in rice, OsYSL2 is up-regulated by Fe deficiency in the leaves, particularly in the phloem, and is also expressed in developing seeds (Koike et al., 2004). OsYSL2:GFP is localized to the plasma membrane. Electrophysiological measurements using Xenopus laevis oocytes have shown that OsYSL2 transports Fe 2+ -NA and Mn 2+ -NA but not Fe 3+ -DMA (Koike et al., 2004). OsYSL15 encodes a functional Fe 3+ -DMA transporter whose expression pattern strongly indicates its involvement in Fe 3+ -DMA uptake from the rhizosphere and in phloem transport of Fe (Inoue et al., 2009). OsYSL15 knockdown seedlings are severely arrested in their germination and early growth but are rescued by a high Fe supply, demonstrating that OsYSL15 plays a crucial role in Fe homeostasis during the first stages of growth (Inoue et al., 2009). It is generally believed that most plants depend largely on a single strategy (I or II) for Fe acquisition. In graminaceous plants, such as barley (Hordeum vulgare) and maize, the inducible Fe 2+ transporter system either is absent or is expressed at very low levels (Zaharieva and Romheld, 2001). Rice and its wild relatives are well adapted for growth under submerged conditions, in which Fe 2+ is more abundant than Fe 3+ . Isolation of OsIRT1 and OsIRT2, which encode functional Fe 2+ transporters (Bughio et al., 2002;Ishimaru et al., 2006), suggests that rice plants can take up Fe 2+ even though they do not have inducible Fe 2+ chelate reductase activity (Ishimaru et al., 2006). Analysis with the positron-emitting tracer imaging system has confirmed that, in addition to taking up Fe 3+ -DMA, rice plants can absorb Fe 2+ from the root environment (Ishimaru et al., 2006).
The presence of YSL genes is not restricted to strategy II plants. In fact, eight orthologs (AtYSL1-AtYSL8) have been found in Arabidopsis (Curie et al., 2001), a genus that neither synthesizes nor uses MAs. NA is the biosynthetic precursor to MAs and is also a strong chelator of various transition metals (von Wirén et al., 1999). NA is ubiquitous in the plant kingdom, and these YSL proteins mediate the internal transport of metals bound to NA (Walker, 2002). AtYSL1 is a shoot-specific gene whose transcript levels increase in response to high-Fe conditions (Le Jean et al., 2005). It is also expressed in young siliques, suggesting a role in the Fe loading of seeds. Indeed, mutations in that gene reduce seed Fe content and delay germination, but those can be rescued by exogenous Fe. AtYSL2 transcript levels are decreased under Fe-deficient conditions; they also respond to copper (Cu) and zinc (Zn;DiDonato et al., 2004;Schaaf et al., 2005). Whereas single mutants of ysl1 and ysl3 have no visible phenotypes, the ysl1/ysl3 double mutants exhibit Fe deficiency symptoms. Likewise, their mobilization of Zn and Cu from the leaves is impaired during senescence, suggesting that the physiological roles for YSL1 and YSL3 are their delivery of metal micronutrients to and from vascular tissues (Waters et al., 2006). In the heavy-metal hyperaccumulator Thlaspi caerulescens, TcYSL3, a member of the YSL gene family, encodes a Ni/Fe-NA transporter (Gendre et al., 2007).
Nitric oxide (NO) is a bioactive molecule implicated in a number of physiological functions, including the intracellular and intercellular mediation of some animal responses (Anbar, 1995). It also serves as a cellular messenger in controlling growth, development, and adaptive responses to multiple stresses in higher plants (Durner and Klessig, 1999;Beligni and Lamattina, 2001;Neill et al., 2002;Lamattina et al., 2003;Graziano and Lamattina, 2005;Lopez-Bucio et al., 2006). NO stimulates the accumulation of both ferritin mRNA and protein, suggesting that it is a signaling molecule for modulating Fe homeostasis in plants (Murgia et al., 2002). NO probably acts through the regulation of a trans-acting factor, because its effect relies on the Fedependent regulatory sequence present on the Atfer1 promoter. NO appears to play a role in improving Fe availability within a plant, as demonstrated by an Fedeficiency phenotype that is recovered by NO without changing the internal levels of total Fe (Graziano et al., 2002). In tomato (Solanum lycopersicum), the NO level is increased in deprived roots to improve Fe availability by controlling the expression of uptake-related genes and by regulating the physiological and morphological adaptive responses to Fe-deficient conditions (Graziano and Lamattina, 2007).
Here, we have determined the physiological roles of OsYSL15 in Fe homeostasis using two independent T-DNA insertional mutants and overexpression lines. supply (Fig. 1). Seedlings were grown for 7 d on Murashige and Skoog (MS) medium with 0, 1, 10, 100, or 500 mM Fe. Standard MS medium contains 100 mM Fe. At the highest concentration, transcript levels were similar to those measured in plants grown on the standard. However, when Fe was limited, transcripts of OsYSL2, OsYSL9, and OsYSL15 were increased. The degree of induction was greater in shoots than in roots. In contrast, a reduction in Fe concentration did not significantly affect the expression of OsYSL16. Previous analysis with northern blots showed that OsYSL15 transcripts are not detected in the leaves of Fe-deficient rice plants (Koike et al., 2004;Inoue et al., 2009). This experimental inconsistency may have been due to differences in growing conditions, age of the plants, and genetic background. Here, we germinated rice seeds on agar medium containing different Fe concentrations and then grew the seedlings for 7 d prior to sampling. In contrast, Koike et al. (2004) and Inoue et al. (2009) reared their seedlings on an Fe-sufficient nutrient medium for 3 weeks, then transferred the plants to an Fe-deficient solution for 3 more weeks before sampling. Therefore, it is possible that their plants were not fully Fe limited at the sampling time. Furthermore, we used a different rice cultivar (Dongjin) whose genetic variation may have affected OsYSL15 expression under Fe deficiency. However, the expression of OsYSL2, which was induced by Fe deficiency in root and shoots, was similar to that described in a previous report (Koike et al., 2004).
We also tested the expression of the other 14 OsYSL genes at different Fe concentrations. The expression of OsYSL5, OsYSL6, OsYSL12, OsYSL13, and OsYSL14 was relatively constant irrespective of Fe status (Supplemental Fig. S1). By comparison, no expression of OsYSL1, OsYSL3, OsYSL4, OsYSL7, OsYSL8, OsYSL10, OsYSL11, OsYSL17, or OsYSL18 was observed under our experimental conditions. Expression Analysis of OsYSL15 Using the Promoter-GUS Fusion Molecule To investigate the spatial and temporal expression patterns of OsYSL15, we generated transgenic plants carrying the 1.0-kb OsYSL15 promoter region that was fused to the uidA reporter gene. Histochemical GUS staining of 5-d-old seedlings showed that GUS activity was stronger in seedlings grown under Fe-deficient conditions ( Fig. 2A, right) than in those in an Fesufficient scenario ( Fig. 2A, left). Cross sections of the seminal roots cultured under the latter exhibited activity that was present mainly in the vascular cylinder ( Fig. 2B). In response to Fe deficiency, GUS activity increased throughout the root tissues, including epidermis, exodermis, endodermis, cortex, and vascular cylinder (Fig. 2C). In leaves, activity was hardly detectable when the plants were grown on an Fe-supplemented medium (Fig. 2, D and F). When Fe was deficient, the OsYSL15 promoter became active in all tissues except the epidermal cells, suggesting a role for OsYSL15 in Fe distribution within the leaves (Fig. 2, E and G). These observations coincide with our previous results from quantitative real-time PCR analysis (Fig. 1C). However, results from our promoter-GUS analysis contrast with those previously reported by Inoue et al. (2009), where the OsYSL15 promoter was weakly OsYSL genes under different Fe concentrations. Seedlings were grown for 7 d on MS medium containing 0, 1, 10, 100, or 500 mM Fe. Each value is the average of three independent experiments. Transcript levels are represented by the ratio between mRNA levels of OsYSL2 (A), OsYSL9 (B), OsYSL15 (C), and OsYSL16 (D) and those of rice Actin1. The nomenclature of OsYSL genes in this study is the same as presented by Koike et al. (2004). Error bars indicate SD. Gene-specific primers are listed in Supplemental Table S2.
active in companion cells and in some of the xylem parenchyma of Fe-sufficient leaves. Fe deficiency treatment did not change expression levels or localization. In our experiments, we used 5-d-old seedling plants grown in sufficient or deficient medium to determine OsYSL15 promoter activity. By comparison, Inoue et al. (2009) used 6-week-old plants to analyze GUS expression. This difference may explain these contradictory results.
Because reproductive organs are the major sinks for Fe, we examined OsYSL15 expression in the flowers and seeds. In the developing spikelets, GUS activity was detectable mainly in the vascular bundles of the palea and lemma but not in the lodicule, anther, and ovary (Fig. 2H). After pollination, activity was found in the upper part of the carpel, including the style and stigma, and also in the embryo (Fig. 2I). During seed development, GUS expression remained unchanged, suggesting that OsYSL15 also functions in the translocation of Fe into rice grains.

OsYSL15 Is a Plasma Membrane-Localized Transporter
To determine the subcellular localization of OsYSL15, we prepared a DNA construct containing a fusion between OsYSL15 and GFP (Fig. 3). This OsYSL15-GFP construct was expressed transiently in onion (Allium cepa) epidermal cells. The PSORT program (http:// psort.nibb.ac.jp) predicted that OsYSL15 would localize to the plasma membrane with high probability. We also used a control construct, which expressed a fusion protein between red fluorescent protein (RFP) and the Arabidopsis (Arabidopsis thaliana) proton ATPase2 (AHA2) that is localized to the plasma membrane ( Fig. 3A; Kim et al., 2001). The DNA constructs, encoding OsSYL15-GFP and AHA2-RFP under the control of the cauliflower mosaic virus 35S promoter, were simultaneously delivered via particle bombardment (Fig. 3B). After 12 h of incubation, expression of the introduced genes was examined with a fluorescence microscope. The green fluorescent signal was Figure 2. Spatial expression patterns of the OsYSL15-GUS construct. A, Fiveday-old seedlings grown on Fesufficient (Fe+) or Fe-deficient (Fe2) medium. B, Cross section of a 5-d-old seedling root grown under Fe-sufficient conditions. The analyzed region is at the differentiation zone (arrowhead in A). C, Cross section of a 5-d-old seedling root grown on Fe-deficient medium. The analyzed region is at the differentiation zone (arrowhead in A). D and E, Cross sections of midrib regions from leaves under Fe-sufficient (D) or Fe-deficient (E) conditions. The analyzed regions are indicated as arrowheads in A. F and G, GUS expression in leaf blades under Fe-sufficient (F) or Fe-deficient (G) conditions. H, Temporal and spatial expression patterns of OsYSL15-GUS fusion in spikelets at various developmental stages. Samples are as follows: 1, tetrad; 2, early young microspore; 3, late young microspore; 4, vacuolated pollen; 5, late pollen mitosis; 6, dehulled flower at late pollen mitosis stage. I, GUS activity in developing seeds at various stages. Samples are as follows: 1, 3 d after pollination (DAP); 2, 5 DAP; 3, 7 DAP; 4, 10 DAP; 5, 20 DAP. A, Anther; AR, aerenchyma; C, cortex; CE, chalazal end; E, embryo; ED, endodermis; EP, epidermis; EX, exodermis; G, glume; LO, lodicules; M mesophyll; P, palea; V, vascular bundle. Bars = 0.5 cm in A, 50 mm in B to G, 1 mm in H, and 0.5 mm in I. detected at the plasma membrane coincident with the red signal driven by the AHA2-RFP (Fig. 3B). When GFP alone was expressed as a control, the protein was localized in the cytoplasm and nucleus (Fig. 3B). These results suggest that OsYSL15 is located at the plasma membrane.
Transporter Activity of OsYSL15 We tested whether OsYSL15 is capable of transporting Fe using the fet3fet4 yeast strain that is defective in Fe uptake (Dix et al., 1994). In this assay, the mutant strain was transformed with either the OsYSL15-expressing plasmid or the empty pYES6/CT vector, which served as a negative control. We utilized a b-estradiol-regulated expression system to control the level of OsYSL15 expression in yeast (Gao and Pinkham, 2000). Via this system, we achieved a tightly regulated off state. Viability of the strains was proved by growing them under permissive conditions (50 mM Fe citrate; Fig. 4, A and G). To examine whether OsYSL15 can transport Fe 3+ , the strain containing the vector alone or else OsYSL15 was grown on a medium with FeCl 3 (Fig. 4B) or on one with FeCl 3 plus DMA (Fig. 4C). This experiment showed that OsYSL15 complemented yeast growth in the presence of FeCl 3 and DMA but not with FeCl 3 alone, thereby suggesting that OsYSL15 is able to transport Fe-PS complexes. When b-estradiol was withheld from the medium, the strain was unable to grow, demonstrating its dependence upon OsYSL15 expression (Fig. 4D). Functional complementation still occurred when a strong Fe 2+ chelator, 4,7-biphenyl-1,10-phenanthroline-disulfonic acid (BPDS), was used to remove any residual Fe 2+ from the Fe 3+ -DMA medium (Fig. 4E), thus supporting that OsYSL15 transports Fe 3+ .
When Fe 2+ was provided as FeSO 4 , OsYSL15 failed to restore growth ( Fig. 4H), but when NA was added along with FeSO 4 , OsYSL15 complemented fet3fet4 (Fig. 4I). Growth of the strain depended on the presence of b-estradiol (Fig. 4J). These results indicate that OsYSL15 is capable of utilizing both Fe 2+ -NA and Fe 3+ -DMA.

Disruption of OsYSL15 Results in Chlorotic Phenotypes under Fe Deficiency
To examine the role of OsYSL15 further, we isolated mutants in which the OsYSL15 gene was disrupted. From our rice flanking sequence tag database Jeong et al., 2006), we identified two independent T-DNA knockout alleles. T-DNA was inserted into the second intron and second exon in osysl15-1 and osysl15-2, respectively (Fig. 5A). T2 progeny were genotyped by PCR to obtain homozygous knockout plants and segregated wild-type siblings using gene-specific primers and a T-DNA primer (Fig. 5A). Reverse transcription (RT)-PCR analysis revealed the disruption of OsYSL15 expression in T-DNA homozygous plants, demonstrating that both are null alleles (Fig. 5B).
To study the roles of OsYSL15 in Fe transport, we germinated seeds of the osysl15 homozygous progeny and their wild-type segregants, then grew their seedlings on solid MS medium in the absence or presence of Fe (100 mM). When they were supplied with a sufficient amount of micronutrients in the control MS medium, the mutant plants did not differ in phenotype from the wild-type segregants (Fig. 5C). Growth rates, based on heights and fresh weights, also did not vary significantly between the two (Supplemental Fig.  S2, A and B), and chlorophyll concentrations in the knockout plants were not significantly different from that of the wild type (Supplemental Fig. S2C). However, the osysl15 mutants showed impaired growth on the Fe-deficient medium (Fig. 5D), differing from the wild type in their heights, fresh weights, and total chlorophyll concentration (Supplemental Fig. S2). For example, respective heights for osysl15-1 and osysl15-2 were reduced to 63% and 69%, fresh weights to 78% and 77%, and chlorophyll concentrations to 54% and 57%, relative to the wild type. We also tested the growth of knockout plants under Zn deficiency and found no distinction between the osysl15 knockout mutants and the wild type in their appearance, fresh weights, and chlorophyll concentrations ( Fig. 5E; Supplemental Fig. S2). These results are contradictory to the ones from OsYSL15 knockdown seedlings grown in the standard MS medium, which showed severe arrest in germination and early growth and died less than 21 d after sowing (Inoue et al., 2009). This discrepancy may have been caused by different genetic backgrounds, resulting in a variation in sensitivity to Fe deficiency between rice cultivars.
To evaluate whether the disruption of OsYSL15 affects Fe distribution, we measured Fe concentrations in shoots and roots at the seedling stage. When plants were grown in an Fe-sufficient medium, concentrations from osysl15-1 and osysl15-2 were reduced to 79% and 77% that of the wild type in the shoots and to 84% and 84% that of the wild type in the roots (Fig. 6A). Under Fe deficiency, relative concentrations in osysl15-1 and osysl15-2 also were decreased to 79% and 75% in shoots and to 78% and 79% in roots, respectively (Fig.  6A). However, Zn concentrations in osysl15-1 and osysl15-2 were not significantly different from those in the wild-type plants (Fig. 6B). Levels of Cu and manganese (Mn) were also unchanged in shoots and roots (Supplemental Fig. S3, A and B).
Whereas Fe concentrations in osysl15 mutants were reduced to 80% of normal, their chlorophyll concentrations were decreased to 50% that of the wild type under Fe deficiency. To evaluate Fe distribution in plants, we measured Fe concentrations in mesophyll protoplasts and chloroplasts from 10-d-old wild-type and osysl15-1 seedlings (Fig. 6, C and D). Under both sufficient and deficient conditions, concentrations in the protoplasts were reduced to 80% in osysl15-1 compared with the wild type. Whereas the Fe concentration from osysl15-1 chloroplasts was reduced to 80% that of wild-type chloroplasts under Fe sufficiency, the concentration from osysl15-1 chloroplasts was only 66% that of wild-type chloroplasts under Fe deficiency. This was consistent with the severe chlorosis of osysl15-1 under Fe deficiency. However, Zn levels in the chloroplasts were not affected by disruption of OsYSL15 (data not shown).
Promoter-GUS analysis showed that OsYSL15 was also active during seed development. Therefore, we postulated that disruption of OsYSL15 would affect Fe loading into the grains. In fact, those from osysl15-1 and osysl15-2 had 83% and 87% as much Fe, respectively, as seeds measured from the wild type (Fig. 6E), while Zn concentrations were similar for both homozygous knockout plants and the wild type (Fig. 6F). Levels of Cu and Mn were unchanged in mature seeds (Supplemental Fig. S3, C and D).

Effect of NO on Reversing the Chlorotic Phenotype of osysl15
NO is able to reverse the chlorotic phenotypes of two Fe-inefficient maize mutants, ys1 and ys3, both impaired in their Fe uptake (Graziano et al., 2002). Because our osysl15 mutants also showed a chlorotic phenotype in response to Fe deficiency, we evaluated whether NO could likewise rescue the deficiency symptoms of the mutants. Rice seeds were germinated and seedlings were grown for 10 d on Fe-deficient media containing various concentrations of sodium nitroprusside (SNP), an NO donor. NO-mediated increases in chlorophyll concentration were more prom- inent in osysl15-1 than in wild-type plants ( Fig. 7A; Supplemental Fig. S4C). Although 10 mM SNP was effective in elevating chlorophyll levels in the wild type, 50 mM SNP was needed to reverse chlorophyll deficiencies in the mutant (Fig. 7C).
We evaluated the effect of NO depletion using an NO-specific scavenger, 2-(4-carboxy-phenyl)-4,4,5,5tetramethyllimidazoline-1-oxyl-3-oxide (CPTIO). Treatment with 100 mM CPTIO almost abolished the protective effect of NO on plant growth and chlorophyll accumulation in osysl15-1 mutants (Fig. 7, A and C; Supplemental Fig. S4). The chlorophyll concentration was reduced to 83% in treated plants compared with untreated mutants. However, CPTIO had no effect on wild-type plants (Fig. 7). Under Fe-sufficient conditions, CPTIO did not influence seedling growth in either the wild type or the mutant.
NO treatment did not change the whole plant Fe concentration or enhance translocation of Fe from one organ to another (Supplemental Fig. S4D). Our results are consistent with the previous suggestion that NO improves the internal availability of Fe.
Expression Analysis of Fe Homeostasis-Related Genes in osyls15-1 NO plays a role in many different signaling pathways and affects the expression of numerous genes. To evaluate its influence on Fe homeostasis, we examined three NAS genes and two ferritin genes. Transcript levels of OsNAS1 and OsNAS2 were not detectable when Fe was sufficient but were markedly increased in response to Fe deficiency (Fig. 8A). In wild-type plants, NO treatment abolished gene expression even at a low dose (i.e. 10 mM SNP). In the osysl15-1 mutant, however, expression of those two genes could be detected even when 25 mM SNP was supplied (Fig.  8A). Under NO treatment, CPTIO induced their expression in the mutants but not in the wild type (Fig.  8B). OsNAS3 was expressed in plants grown on an Fesufficient medium but was suppressed when Fe was limited. NO treatment upon Fe deficiency increased OsNAS3 expression in both wild-type and mutant plants (Fig. 8A), although to a relatively lesser extent in the latter. NO can stimulate the accumulation of both ferritin mRNA and protein, indicating that it is a key signaling molecule for regulating Fe homeostasis in plants (Murgia et al., 2002). Two ferritin genes in rice, OsFer1 and OsFer2, show high sequence similarity to each other but distinct expression patterns (Gross et al., 2003). We observed here that transcript levels of both were reduced when Fe was deficient but were elevated by SNP treatment in the wild type and the osysl15 mutant (Fig. 8A). However, CPTIO diminished this NO effect (Fig. 8B). These results suggest that the mutation in OsYSL15 did not affect those ferritin genes.

Overexpression of OsYSL15 Using the Rice Actin Promoter
We placed OsYSL15 cDNA in a sense orientation under the control of the rice Actin1 promoter, resulting in pGA2875 (Fig. 9A). After generating transgenic plants, we studied constitutive expression of that gene using RNA samples prepared from leaves (Fig.  9B). Based on our quantitative real-time PCR analysis, we selected lines 2 (OX-2) and 6 (OX-6) for further examination. Fe and Zn concentrations were measured in their seeds (Fig. 9, C and D) via an atomic absorp-tion spectrometer. Although Fe concentrations in seeds from both transgenic lines were increased compared with the wild type (Fig. 9C), Zn concentrations were not changed by overexpression of OsYSL15 (Fig. 9D). The levels of Mn and Cu in mature seeds of transgenic plants were similar to those of the wild type (Supplemental Fig. S3, C and D).

Phenotype Analyses of osysl15-1 Knockout and OsYSL15 Overexpression Plants Grown under Different Fe Concentrations
Expression of OsYSL15 was strongly induced by Fe deficiency and was decreased as the Fe concentration increased (Fig. 1). In testing the phenotypes of osysl15-1 knockout and OsYSL15-overexpressing plants, we observed that as the level of Fe rose, chlorophyll concentrations were increased in a dose-dependent manner (Fig. 10). Visual differences were documented by measuring chlorophyll concentrations (Fig. 10B). Under Fe-deficient conditions, the osysl15-1 knockout plants showed greater chlorosis, but that phenotype disappeared when higher Fe concentrations (at least 100 mM) were supplied. In OX-2 and OX-6 plants, this chlorotic phenotype was diminished at an Fe concentration of 10 mM or greater. These results indicate that OsYSL15 functions primarily when plants have a low availability of Fe.

Disruption and Overexpression of OsYSL15 Affects Plant Architecture
To examine how the disruption or overexpression of OsYSL15 might influence plant architecture and grain yields, we cultivated transgenic seedlings along with their segregated wild-type siblings in the field. All knockout plants as well as OX-2 and OX-6 plants flowered about 10 d later than their wild-type segregants. Whereas transgenic plant heights were significantly reduced (Fig. 11), tiller numbers did not change (Supplemental Table S1). Moreover, although fewer total spikelets were counted on the knockout mutants and overexpression plants, their grain yields were not significantly different from those of the wild type (Supplemental Table S1). We also measured Fe, Zn, Mn, and Cu concentrations in wild-type and mutant flag leaves, sampling for uniformity after flowering. Disruption of OsYSL15 resulted in reduced Fe levels in flag leaves, while its overexpression increased those values in flag leaves (Fig. 11E). However, Zn, Mn, and Cu concentrations in the flag leaves were unaffected by either disruption or overexpression of OsYSL15 ( Fig. 11F; Supplemental Fig. S3, F and G).

DISCUSSION
Here, we report the functional roles of OsYSL15 for Fe homeostasis in rice. This was manifested by reduced Fe levels in knockout plants that showed chlorotic phenotypes under Fe deficiency and by increased Fe levels in overexpressors. OsYSL15 expression was strongly induced by Fe deficiency, suggesting that OsYSL15 is needed when plants grow under such conditions. In roots, this gene was strongly induced in all cell types, including the epidermis, implying that it is involved in Fe uptake from the rhizosphere. The gene was also induced in almost all shoot cell types, except the epidermis, which implies that this transporter functions primarily in distributing Fe. Using a yeast system, we demonstrated that OsYSL15 transports Fe 3+ -DMA and Fe 2+ -NA, strongly supporting the possibility of dual roles for OsYSL15: Fe uptake from soil and its distribution in the plant. OsYSL15 was also expressed in developing seeds, suggesting a role in the translocation of Fe into grains, as confirmed by the reduced Fe concentration in mutant seeds. Furthermore, overexpression of OsYSL15 resulted in higher Fe concentrations in leaves and seeds, supporting that OsYSL15 is an Fe transporter. Disruption or overexpression of OsYSL15 affected the concentration of Fe, but not Zn, Mn, or Cu, in our rice plants. Therefore, OsYSL15 appears to be an Fe-specific transporter. Although OsYSL15 is highly homologous to OsYSL2, they appear to have different substrate specificities. OsYSL2 is capable of mediating transport of Fe 2+ -NA and Mn 2+ -NA but not Fe 3+ -DMA and Mn 2+ -DMA (Koike et al., 2004).
Disruption of OsYSL15 resulted in a 20% reduction in Fe concentration under both sufficient and deficient conditions. Although no altered phenotypes were visible when Fe supplies were adequate, severe chlorosis occurred in the osysl15 mutants when Fe was limited. This suggests that OsYSL15 is important for distribution into the chloroplast, as was correlated with a great reduction in chloroplast Fe concentrations. Therefore, how OsYSL15 affects Fe distribution into chloroplasts needs to be investigated.
The disruption or overexpression of OsYSL15 was manifested by shorter plants and alterations in their architecture. Therefore, because only the concentration of Fe varied, Fe homeostasis must play an important role in growth and development. Because the Fe concentration was altered by such disruption or over- expression, the physiological balance of metal ions was disturbed, resulting in defective growth in the field. When IRT1 is ectopically expressed, transgenic plants show no visible morphological changes (Connolly et al., 2002). Although IRT1 mRNA is expressed constitutively, IRT1 protein is present only in Fe-limited roots, indicating posttranscriptional regulation of IRT1 (Vert et al., 2002). Because our OsYSL15 overexpressors accumulated more Fe and had altered growth, we can assume that OsYSL15 protein levels are higher in the over-expression lines than in the wild types. However, we do not know that posttranscriptional regulation is occurring as AtIRT1. Unfortunately, we were unable to measure the level of OsYSL15 protein due to a lack of available antibodies.
Seedlings of the maize ys1 mutant, which is defective in the uptake of Fe 3+ -PS complexes, experience severe Fe deficiency chlorosis (yellowing between the veins) and, ultimately, mortality, indicating that such uptake is an essential process for that species (Walker Figure 8. Expression analysis for Fe homeostasis-related genes. RNA was prepared from 10-d-old shoots. A, Real-time PCR analysis of three rice NAS genes and two rice ferritin genes after SNP treatment. B, Expression analyses of OsNAS and Osferritin genes in plants grown on medium supplemented with 100 mM CPTIO, an NO inhibitor. Transcript levels are represented by the ratio between mRNA level for OsNAS or Osferritin genes and that for rice Actin1. Error bars indicate SD. WT, Wild type. and Connolly, 2008). Here, however, mutation in OsYSL15 did not produce lethal phenotypes, although those plants were shorter and contained less Fe, probably because of gene redundancy. Three additional rice genes are highly homologous to maize YS1. Among them, induction patterns for OsYSL2 and OsYSL9 by Fe deficiency are quite similar to those for OsYSL15. Rice also has efficient Fe 2+ uptake systems, in contrast to other grasses (Ishimaru et al., 2006).
NO is a bioactive molecule, playing important roles in many physiological processes, including determining Fe availability within a plant (Graziano et al., 2002). Here, we used osysl15 mutants to evaluate the effect of NO in rice. By treating with SNP, an NO donor, we were able to partially reverse their chlorotic phenotype. An NO scavenger, CPTIO, abolished that protective effect. However, NO treatment did not increase total Fe content within a plant. These results support that NO increases Fe availability in rice, as observed from other organisms. NO stimulates the accumulation of both ferritin mRNA and protein in Arabidopsis (Murgia et al., 2002). We observed similar positive effects on rice ferritin mRNA levels. In contrast, expression of two NAS genes that was induced by Fe deficiency was inhibited by NO. Therefore, we conclude that there are alternative ways to alleviate the stresses associated with diminished supplies of Fe.
Quantitative real-time PCR and promoter-GUS analyses have indicated that OsYSL15 is strongly induced by lower Fe levels. Two cis-acting elements, IDE1 and IDE2, synergistically mediate Fe deficiencyinduced gene expression in tobacco (Nicotiana tabacum; Kobayashi et al., 2003). Sequences similar to IDE1 and IDE2 are found in the promoter regions of OsIRT1, OsNAS1, and OsNAS2, which are induced by Fe deficiency (Kobayashi et al., 2003(Kobayashi et al., , 2005. The rice basic helix-loop-helix protein OsIRO2, an essential regulator  of the genes involved in Fe uptake under deficient conditions, also contains putative IDE sequences in its promoter region (Ogo et al., 2006). Therefore, the regulatory networks mediated by IDE elements during Fe deficiency are apparently conserved between dicot and monocot species. The promoter region of OsYSL15 has two putative IDE sequences, at -249 and -593 bp from the translation initiation site. Our OsYSL15 promoter-GUS construct contained these two putative IDE-like elements, and GUS expression was induced by Fe deficiency. Further research is necessary to verify whether these elements are indeed responsible for that deficiency response. OsYSL15 is expressed strongly in OsIRO2-overexpressing transgenic plants (Ogo et al., 2007), suggesting that the former functions downstream of the latter.
OsYSL15 overexpression was positive in raising the Fe concentration in our seeds and vegetative tissues, albeit with some side effects. This presents the possibility that OsYSL15 can be used for enhancing Fe levels in rice grains, perhaps via targeted expression with seed-specific promoters.

Plant Growth
Wild-type transgenic rice (Oryza sativa 'Dongjin') and seeds were surface sterilized and germinated on an MS solid medium supplemented with 0, 1, 10, 100, or 500 mM Fe 3+ -EDTA. Shoot and root samples from 7-d-old seedlings were frozen with liquid nitrogen. SNP (10-100 mM) was used as an NO donor, and 100 mM CPTIO served as an NO scavenger. Transgenic plants were transplanted and grown to maturity in paddy fields located at Pohang University of Science and Technology (36°N). The field tests were performed twice, in 2007 and 2008.

RNA Isolation and RT-PCR Analysis
Total RNA was obtained from each tissue type with an RNA isolation kit (Tri Reagent; MRC). For cDNA synthesis, we used 2 mg of total RNA as template and M-MLV reverse transcriptase (Promega) in a 25-mL reaction mixture. RT-PCR was performed in a 50-mL solution containing a 1-mL aliquot of the cDNA reaction, 0.2 mM gene-specific primers, 10 mM deoxyribonucleotide triphosphates, and 1 unit of rTaq DNA polymerase (Takara Shuzo). PCR products were separated by electrophoresis on a 1.2% agarose gel. Quantitative real-time PCR was performed with a Roche LightCycler II as described previously (Han et al., 2006). The Actin1 mRNA levels were used to normalize the expression ratio for each gene. Changes in expression were calculated via the DD Ct method. The gene-specific primers are listed in Supplemental Table S2.
For complementation assays, synthetic dextrose (SD)-Trp medium was made with an Fe-free yeast nitrogen base buffered with 25 mM MES at pH 5.7 for plates containing Fe 2+ or at pH 6.0 for plates containing Fe 3+ . To prepare our Fe 2+ assay, the following were added, in order, to the center of each empty plate: 125 mL of 200 mM ascorbic acid, 7.5 mL of freshly prepared 10 mM Fe 2 SO 4 , and 20 mL of 10 mM NA. This solution was mixed briefly and then incubated at room temperature for 10 min to allow complex formation. Afterward, 25 mL of molten SD-Trp with 10 mg mL 21 blasticidin was added, which was then allowed to solidify. To prepare for our Fe 3+ assay, 34 mL of 7.4 mM FeCl 3 and 25 mL of 10 mM DMA were placed in the center of each empty plate and incubated at room temperature for 10 min. We then added 25 mL of molten SD-Trp with 10 mg mL 21 blasticidin to the plates before solidification. To prepare plates with 10 mM BPDS, 25 mL was incorporated just prior to the addition of the molten medium.
Suspensions were prepared from 3-d-old yeast colonies, which were removed from the plates and suspended in sterile water before the optical density at 550 nm of the resulting suspension was measured. After that value was brought to 0.1, serial dilutions (1:10, 1:100, 1:1,000, and 1:10,000) of the suspension were prepared, and 7 mL of each dilution was spotted on the plates. They were then grown at 28°C for 3 d.

Generation of the OsYSL15 Promoter-GUS Fusion Molecule and GUS Assay
Genomic sequences (-1,000 to -1 bp from the translation initiation site) containing the promoter region of OsYSL15 were amplified by PCR using two primers (pf, 5#-AAAAGCTTAGCATGTCTCCAGATTCTCCAT-3#; and pr, 5#-AAGGATCCGCGGCGGCGGCGGCGTCGATCTC-3#). This fragment was connected to a GUS-NOSt cassette (derived from pBI101.2) and ligated into pCAMBIA1302, resulting in pGA2866. This plasmid was transferred to Agrobacterium tumefaciens strain LBA4404 by the freeze-thaw method (An et al., 1988), and transgenic plants carrying the above construct were generated via Agrobacterium-mediated cocultivation (Lee et al., 1999). Histochemical GUS staining of those transgenic plants was performed according to the method reported by Dai et al. (1996). Ten-micrometer sections were prepared as described previously by Jung et al. (2005) and were observed with a microscope (Nikon) under bright-field illumination.

Subcellular Localization of the OsYSL15-GFP Fusion Protein in Onion Epidermal Cells
Full-length OsYSL15 cDNA was PCR amplified with the primer pair gf (5#-AATCTAGAGTTTCTTTCTTGTCCTCGTGGT-3#) and gr (5#-AAGGATC-CAGCTTCCAGGCGTAAACTTCATGC-3#). These primers contained XbaI or BamHI sites (underlined sequences) to facilitate cloning of the amplified cDNA. After sequence analysis, the OsYSL15 cDNA was cloned into the XbaI and BamHI sites of the 326-GFP vector (Lee et al., 2001). The AHA2-RFP fusion molecule under the control of the cauliflower mosaic virus 35S promoter was obtained from Inhwan Hwang (Pohang University of Science and Technology). Constructs were introduced into onion (Allium cepa) epidermal cells by particle bombardment using the Biolistic PDS-1000/He particle delivery system (Bio-Rad). At 12 h after transformation, expression of GFP and RFP was monitored with a fluorescence microscope and two filters (Axioplan2; Carl Zeiss).

Generation of the OsYSL15-Overexpressing Construct
To create our OsYSL15-overexpressing construct, the full-length cDNA sequence of OsYSL15 was amplified by a primer pair (FL, 5#-AATCTA-GAGTTTCTTTCTTGTCCTCGTGGT-3#; and RL, 5#-AACTCGAGACCTCT-TAGCTTCCAGGCGTAA-3#). The PCR product was cloned into XbaI and XhoI sites between the rice Actin1 promoter (McElroy et al., 1990) and the T7 terminator of the binary vector pGA1671, thereby generating pGA2875 (Jeon et al., 2000). Transformation of this plasmid into Agrobacterium and the generation of transgenic plants were as described previously (Lee et al., 1999).

Preparation of Protoplast and Chloroplast
Mesophyll protoplasts were prepared as described previously by Moon et al. (2008). Briefly, the third leaves were harvested and dissected from 10-dold seedlings grown on either MS or Fe-deficient medium. The materials were digested in an enzyme solution (1.5% cellulase RS, 0.3% macerozyme, 0.1% pectolyase, 0.6 M mannitol, 10 mM MES, 1 mM CaCl 2 , and 0.1% [w/v] bovine serum albumin) for 4 h at 26°C with gentle agitation (50-75 rpm). KMC solution (117 mM KCl, 82 mM MgCl 2 , and 85 mM CaCl 2 ) was added afterward. Protoplasts were sorted from the debris through a nylon mesh (20 mm), collected by centrifuging at 1,000 rpm for 2 min, and resuspended in EP3 solution (70 mM KCl, 5 mM MgCl 2 , 0.4 M mannitol, and 0.1% MES, pH 5.6). Chloroplasts were isolated according to the published protocol (Triboush et al., 1998). Five grams of leaves from 10-d-old seedlings was homogenized in a mortar with 20 mL of STE buffer (400 mM Suc, 50 mM Tris, pH 7.8, 20 mM EDTA-Na 2 , 0.2% bovine serum albumin, and 0.2% b-mercaptoethanol). After the homogenate was filtered through a 50-mm nylon mesh, the extract was centrifuged at 1,000 rpm. The supernatant was then centrifuged at 4,000 rpm, and this resuspended pellet was centrifuged again at 4,000 rpm to prepare the chloroplast fraction.

Measurement of Plant Growth and Metal Concentrations
Seeds of the wild type and mutant were germinated, and plants were then grown for 10 d on a solid medium containing MS salts supplemented with different concentrations of Fe 3+ -EDTA. Their chlorophyll concentrations were measured as described previously . Briefly, 0.1 g of leaf samples was harvested and the chlorophyll was extracted with 1 mL of 80% acetone. After homogenization, the samples were incubated for 15 min and spun at 15,000g for 10 min before an aliquot of the supernatant fraction was taken to measure A 663 and A 643 with a spectrophotometer. Chlorophyll concentrations, including chlorophyll a and b, were determined according to the method of Arnon (1949), and metal concentrations were measured as described previously by Kim et al. (2002). Seeds and shoot and root portions were dried for 2 d at 70°C before those samples were weighed. Afterward, they were digested in 1 mL of 11 N HNO 3 for 3 d at 200°C. Following dilution, their metal concentrations were determined by an atomic absorption spectrometer (SpectrAA-800; Varian).
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers number AB190923.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Expression analyses of OsYSL genes that were constitutive under different iron concentrations.
Supplemental Figure S3. Mn and Cu concentrations in wild-type and transgenic plants.
Supplemental Figure S4. Characterization of wild-type and osysl15-1 mutant plants grown with various concentrations of SNP.
Supplemental Table S1. Characterization of osysl15 mutants grown in the paddy field.
Supplemental Table S2. Primers used for RT-PCR analysis.