Introduction

Iron (Fe)-deficiency anemia is one of the most prevalent human micronutrient deficiencies in the world, affecting an estimated one-third of the world’s population and causing 0.8 million deaths annually worldwide1. To address this problem, biofortification (i.e., the breeding of micronutrient-fortified crops) is advantageous for people who experience difficulty in changing their dietary habits because of financial, cultural, regional, or religious restrictions. Biofortification is also advantageous for governments because it is inexpensive and sustainable compared to nutritional supplement programs2. Rice is a particularly suitable target for biofortification because Fe-deficiency anemia is a serious problem in developing countries where rice is a major staple crop1,3. In addition, Fe translocation and Fe homeostasis in rice has begun to be understood at the molecular level4.

Based on knowledge of Fe transportation and Fe homeostasis in rice, three approaches have been reported to produce Fe-biofortified rice. The first approach is enhancement of Fe accumulation in rice seeds by ferritin gene expression under the control of endosperm-specific promoters. Endosperm is the rice-seed tissue that accumulates a high concentration of starch and becomes the edible part of the seed after milling, at which point these seeds are known as polished or white seeds3. Ferritin is a ubiquitous protein for Fe storage and stores about 4,000 Fe atoms in a complex5. Goto et al. generated transgenic rice plants that expressed the soybean ferritin gene, SoyferH1, in the endosperm using the endosperm-specific 1.3-kb GluB1 rice promoter; the transformants showed higher Fe accumulation in brown rice seeds6. A few reports have described the production of Fe biofortification rice by endosperm-specific expression of ferritin7,8. Qu et al. expressed SoyferH1 under the control of both the OsGlb1 promoter and 1.3-kb GluB1 promoter to further increase seed Fe concentration9. However, enhancement of ferritin expression did not produce further increases in seed Fe content9. Therefore, in addition to increased Fe storage in seeds, enhanced Fe uptake from the soil and enhanced translocation within the plant body are thought to be required to further improve Fe biofortification in seeds.

The second approach involves increasing Fe transportation within the plant body by enhancing the expression of NAS genes. Nicotinamine (NA) is a chelator of metal cations such as Fe(II) and Zn(II) and it is biosynthesized from S-adenosyl methionine via NA synthase (NAS)10. All higher plants synthesize and utilize NA for the internal transport of Fe and other metals11,12. Takahashi et al. produced NA-deficient transgenic tobacco plants that showed young leaves with serious chlorosis and Fe and Zn concentrations in the leaves and flowers decreased as a result of disrupted internal metal transport. These reports suggest that NA plays an essential role in Fe translocation to seeds. In addition, overexpression of the barley NAS gene, HvNAS1, led to increased Fe and Zn concentrations in the leaves, flowers and seeds of tobacco plants12. Likewise, overexpression of the NA synthase gene increased the Fe concentration in polished rice seeds threefold with greenhouse cultivation13,14,15.

The third approach is enhancement of Fe flux into the endosperm by expression of the Fe(II)-NA transporter gene OsYSL2. Koike et al. identified the rice NA-Fe(II) transporter gene OsYSL2, which is preferentially expressed in leaf phloem cells, the vascular bundles of flowers and developing seeds, suggesting a role in internal Fe transport16. OsYSL2 knockdown mutant plants exhibit a 30% decrease in Fe concentration in the endosperm17. Simple overexpression of OsYSL2 by the 35S promoter did not increase Fe concentration in seeds. In contrast, enhancement of OsYSL2 expression under the control of the rice sucrose transporter promoter OsSUT1, which drives high expression in the panicle and immature seeds during the seed maturation stage, increased Fe concentration in polished rice seeds by up to threefold17.

Additionally, introduction of mugineic acid synthase gene was reported as another approach to increase Fe concentration in seeds. In graminaceous plants, NA is the precursor of mugineic acid family phytosiderophores (MAs), which are natural Fe(III) chelators used in Fe acquisition from the rhizosphere18,19. Graminaceous plants synthesize and secrete MAs into the rhizosphere by TOM1 transporter20. They form Fe(III)–MAs complexes and are taken up into the root via YS1 and YSL transporters21,22. Rice biosynthesizes 2′-deoxymugineic acid (DMA), which facilitates Fe uptake and internal transport23,24. Barley biosynthesizes not only DMA but also mugineic acid (MA) by MA synthase, IDS325,26. However, rice lacks IDS3 gene and does not produce MA. We previously reported that Fe concentration in polished rice seed increased up to 1.25 and 1.4 times in calcareous and normal soil cultivation in field, respectively, by introduction of barley IDS3 genome fragment27,28.

Each of these approaches could increase Fe concentration in polished rice seeds. However, a higher Fe concentration in seeds was required to reduce the human Fe deficiency anemia health problem. The target Fe concentration in polished rice seeds is over 15 ppm in field cultivation29, but research has not yet achieved this yield. We hypothesized that a combination of these transgenic approaches would further increase the Fe concentration in seeds.

Here, we combined these three approaches and produced new transgenic rice lines with enhanced Fe accumulation in seeds using the Soybean ferritin gene under the control of two endosperm-specific promoters, the OsGlb1 and 2.3-kb OsGluB1. Furthermore, these seeds exhibited enhanced Fe transportation within the plant body due to overexpression of HvNAS1 and enhanced Fe translocation to seeds due to OsYSL2 expression under the control of the OsSUT1 promoter and OsGlb1 promoter ( Supplementary Fig. 1 ). Gene insertion, ferritin accumulation in seeds and higher expression of OsYSL2 and HvNAS1 were confirmed. The Fe concentration in polished T2 seeds increased by up to sixfold in plants grown in soil in a greenhouse. In field cultivation, the Fe concentration in T3 polished seeds increased up to 4.4-fold. This is the first report of the combination of three approaches to increase Fe accumulation in seeds.

Results

Selection of Fer-NAS-YSL2 transgenic rice lines

We produced 45 independent transgenic rice lines (Oryza sativa cultivar Tsukinohikari; Fer-NAS-YSL2 line). These included the OsGlb1 promoter–SoyferH2, OsGluB1 promoter–SoyferH2, OsActin1 promoter–HvNAS1 and OsSUT1 promoter–OsYSL2 and the OsGlb1 promoter–OsYSL2 cassettes ( Fig. 1 ). We used soybean ferritin, SoyferH2, as the ferritin gene. For comparison, single gene-introduced lines for OsActin1 promoter–HvNAS1 (AN; ref.13) and OsSUT1 promoter–OsYSL2 (SY; ref.17) were also used. T0-regenerated plants were cultivated in a greenhouse and those lines with a high Fe content in their T1 polished seeds were selected ( Supplementary Fig. 2 ). These lines carried all the introduced genes, as confirmed by genomic polymerase chain reaction (PCR) ( Fig. 2 ). Ferritin accumulation in T2 seeds was detected by Western blot analysis ( Fig. 3 ). Enhanced expression of OsYSL2 and HvNAS1 in immature T3 seeds of selected transgenic lines was confirmed by real-time RT-PCR analysis ( Fig. 4 ). The Fe concentration in the T2 seeds had increased up to sixfold in the Fer-NAS-YSL2 lines, threefold in the AN lines and twofold in the SY lines compared to the NT line ( Fig. 5 ). Thus, the introduction of multiple genes is a more effective method of increasing Fe concentrations in greenhouse-cultivated rice grains than single introduction of either HvNAS1 or OsYSL2.

Figure 1
figure 1

The gene cassette introduced into rice to produce the Fer-NAS-YSL2 lines.

Arrows show the direction of transcription. RB, right border; LB, left border; NP, Agrobacterium tumefaciens nopaline synthase gene (AF485783) promoter region; NPTII, neomycin phosphotransferase II gene (AF485783); Tnos, A. tumefaciens nopaline synthase gene terminator (AF485783); OsSUT1P, promoter region of the rice sucrose transporter gene OsSUT1 (D87819); OsYSL2, rice nicotianamine–metal transporter gene (AB126253); OsGluB1P, 2.3-kb promoter region of the OsGluB1 gene (AY427569)41; SoyferH2, soybean Ferritin gene (AB062754)42; OsAct1P, promoter region of the rice OsActin1 gene (Os03g0718100); HvNAS1, barley nicotianamine synthase 1 gene (AB010086); OsGlbP, promoter region of the 26 kDa OsGlb1 gene (AY427575)41; iGUS, β-glucuronidase gene with an intron (AF485783); 35SP, cauliflower mosaic virus 35S promoter (U28417); HPT, hygromycin phosphotransferase gene (K01193).

Full size image
Figure 2
figure 2

Detection of transgene insertion in the transgenic lines.

Rice genomic DNA was extracted from young leaves of selected T1 plants cultivated in a greenhouse and used as the template for PCR. NT, non-transgenic rice; numerals, Fer-NAS-YSL2 transgenic lines No. 15, 16, 19 and 32, respectively; Vec, the vector containing the Fer-NAS-YSL2 cassette (vector o in Supplementary Fig. 5 ; positive control).

Full size image
Figure 3
figure 3

Ferritin accumulation in T2 brown seeds.

Six mature T2 brown seeds harvested from a greenhouse were subjected to Western blot analysis. NT, non-transgenic rice; AN, OsActin1 promoter–HvNAS1 line 813; SY, OsSUT1 promoter–OsYSL2 line 1817; numerals, Fer-NAS-YSL2 transgenic lines No. 15-8, 16-2, 19-2, 19-4, 19-5 and 32-3, respectively; soybean, protein extracts from non-transformed soybean cotyledons (positive control).

Full size image
Figure 4
figure 4

Quantitative real-time RT-PCR analysis of OsYSL2 and HvNAS1.

(a) OsYSL2 and (b) HvNAS1 expression levels. T2 plants were cultivated in commercially supplied soil (Bonsolichigou) in a greenhouse. Total RNA was extracted from immature T3 seeds (seeds at an early milky stage, 10 days after fertilization) from each line (n = 3). Bars represent the means ± standard errors of three independent real-time RT-PCR reactions.

Full size image
Figure 5
figure 5

Fe concentration of T2 polished seeds.

T2 seeds from the ear of the main tiller were harvested in a greenhouse and polished using a Multi-beads Shocker® (Yasuikikai, Osaka, Japan) and the Fe concentration was measured by ICP-AES (SPS1200VR; Seiko Instruments Inc., Chiba, Japan), as described in ref.13. Bars represent the Fe concentration in polished seeds from independent transgenic or non-transgenic plants (n = 1). Numerals indicate the line numbers of independent T1 lines. The arrows and numerals above the graphs show lines with a high Fe content selected for the field trial. NT, non-transgenic rice; AN, OsActin1 promoter–HvNAS1 transgenic rice line No. 813; SY, OsSUT1 promoter–OsYSL2 transgenic rice lines17; Fer-NAS-YSL2, transgenic rice lines carrying the OsGlb1 promoter–Ferritin, OsGluB1 promoter–Ferritin, OsSUT1 promoter–OsYSL2, OsGlb1 promoter–OsYSL2 and OsActin1 promoter–HvNAS1.

Full size image

Field trial of Fer-NAS-YSL2 transgenic rice lines

In greenhouse cultivation, no environmental stress exists that reflects real cultivation in the paddy field. For practical usage, we therefore investigated whether Fer-NAS-YSL2 lines set seeds with increased Fe concentration under actual paddy field conditions. For this purpose, selected T2 lines with high Fe content in their polished seeds ( Fig. 5 , arrows) were cultivated in an isolated paddy field in Jinju, Gyeongsang Province, Korea ( Supplementary Figs. 3 and 4 ). The mean Fe concentrations in T3 polished seeds from the NT, AN and Fer-NAS-YSL2 lines were 0.9, 1.5 and 4.0 µg/g dry weight, respectively ( Fig. 6a ). Notably, the Fe concentration in the Fer-NAS-YSL2 seeds was 4.4-times higher compared to NT seeds and 2.7-times higher compared to AN seeds ( Fig. 6a ). Our results demonstrate that multiple introductions of Fer-NAS-YSL2 are better than a single introduction of either OsActin1 promoter–HvNAS1 or OsSUT1 promoter–OsYSL2 under field conditions. In addition, the Fer-NAS-YSL2 seeds also accumulated zinc (Zn), another important micronutrient for humans, up to 1.6-fold more ( Fig. 6b ). Fer-NAS-YSL2 seeds did not accumulate the toxic heavy metal cadmium (Cd) more than seen in the NT line ( Fig. 6e ). In brown seeds, Fe and Zn concentrations in the Fer-NAS-YSL2 and AN lines were similar, having increased by up to 1.6- and 1.4-fold, respectively, compared to NT lines ( Table 1 ). These increases in Fe and Zn concentration in brown rice might have been due to the OsActin1 promoter–HvNAS1 gene. In contrast, the Fe concentration in polished seeds increased to a greater degree in Fer-NAS-YSL2 lines than in the AN or SY lines. Moreover, in the field experiment, no negative effect on yield was found in terms of the panicle length, number of panicles per hill, number of total grains per panicle, number of total grains per hill, percentage of filled grain, total weight of grains per hill, or 1,000-grain weight in lines 19-2, 19-4 and 19-5, which had a high seed Fe concentration ( Supplementary Table 1 ).

Table 1 Metal concentration in T3 brown seeds obtained from the paddy field T3 seeds from the ear of the main tiller were harvested from the middle four hills in each plot and the Fe, Zn, Mn and Cu concentrations in brown seeds were measured. Values are µg/g dry weight (DW). The data represent the means ± standard errors of four block replications for each line (n = 4) (Supplementary Fig. 3). NT, non-transgenic rice; AN, OsActin1 promoter–HvNAS1 line; 15-8, 16-2, 19-2, 19-4, 19-5 and 32-3, Fer-NAS-YSL2 transgenic lines
Full size table
Figure 6
figure 6

Metal concentration in T3 polished seeds obtained from the paddy field.

(a) Fe concentration; (b) Zn concentration; (c) manganese (Mn) concentration; (d) copper (Cu) concentration; (e) cadmium (Cd) concentration. T3 seeds from an ear of the main tiller were harvested from the field and polished and the metal concentrations were measured (indicated as the average of the middle four hills in each block). ANOVA with Tukey–Kramer HSD test was used for each four-block data set (n = 4). Letters above the bars indicate significant differences (P < 0.05); (a, P < 0.0204; b, P < 0.0428; c, P < 0.0314; d, P < 0.0396). NT, non-transgenic rice; AN, OsActin1 promoter–HvNAS1 line 813; numerals, Fer-NAS-YSL2 transgenic lines.

Full size image

In addition to NA, DMA, which is synthesized from NA, is known as a natural chelator of metal cations in graminaceous plants23,24. The NA and DMA concentrations in T3-brown seeds of Fer-NAS-YSL2 lines increased by up to six- and threefold, respectively, compared to NT brown seeds and to a similar level in the AN line ( Fig. 7 ).

Figure 7
figure 7

NA and DMA concentrations in T3 brown seeds harvested from the paddy field.

Letters indicate significant differences (P < 0.05) by ANOVA with Tukey–Kramer HSD test used for each four-block data set (NA, P < 0.0406; DMA, P < 0.0042); n = 4.

Full size image

Discussion

To produce Fe-biofortified rice, we combined three transgenic approaches. First, we enhanced Fe storage in grains via expression of the Fe storage protein ferritin using endosperm-specific promoters. Second, we enhanced Fe translocation through overproduction of the natural metal chelator nicotianamine. Third, we enhanced Fe flux into the endosperm through expression of the Fe(II)–NA transporter OsYSL2 under the control of an endosperm-specific promoter and sucrose transporter promoter ( Supplementary Fig. 1 ). As a result, Fe concentration of polished seeds increased up to sixfold in greenhouse cultivation and 4.4-fold in paddy field cultivation ( Figs. 5 and 6a ).

In this article, three approaches were logically combined to enhance Fe(II)–NA translocation into seeds more effectively than previous methods. Overexpression of the NAS gene increases the NA concentration in the plant body. Abundant NA facilitates formation of Fe(II)–NA, which is stable under higher pH conditions, such as in phloem sap (pH 8.0)30,31. Consequently, Fe(II) transport in the plant body, including the phloem, is improved by NAS overexpression13,14. For effective translocation of enhanced Fe(II)–NA in phloem sap, we engineered the Fe(II)–NA transporter OsYSL2 to be under the control of the OsSUT1 promoter, which expresses in the companion cell of phloem in flag leaves and rachis32. The rice antisense mutant of OsSUT1 markedly reduces the sucrose uptake ability and filling rate of rice seeds33. Therefore, OsSUT1 is a key transporter of sucrose from the phloem to seeds and the OsSUT1 promoter is suitable for the expression of OsYSL2 to enhance Fe(II)–NA translocation in seeds17. Additionally, OsYSL2 expressed in endosperm cells under the control of the OsGlb1 promoter may enhance transport of Fe(II)–NA into endosperm cells. Moreover, ferritin expressed under the control of the OsGlb1 promoter and the OsGluB1 promoter accumulates Fe in seed endosperm cells. Fe incorporation into ferritin requires it to be in the Fe(II) form5. Therefore, enhancement of Fe(II)–NA transportation to seeds is important for Fe accumulation in seed ferritin, rather than Fe(III)–citrate, Fe(III)–DMA, or other Fe(III)–chelator complexes. As a result, Fe concentration in the seeds of Fer-NAS-YSL2 rice was markedly elevated compared to the result of a single application of each approach ( Figs. 5 and 6a ).

The Japanese rice variety used in this research, Tsukinohikari, contained a low Fe concentration, 0.9 µg/g in field-cultivated polished seeds ( Fig. 6a ). In some rice varieties, the Fe concentration of polished rice is higher than that of this variety34. Thus application of this transgenic approach to higher Fe concentration varieties has the potential to achieve the target Fe concentration for human health.

In addition to the increased Fe concentration in seeds, many advantages exist in using Fer-NAS-YSL2 rice. Fe-biofortification of polished rice by overexpression of NAS mitigated Fe-deficiency anemia in mice to a greater degree than the seeds of NT rice14. Increasing the NA concentration by enhancing NAS expression may improve the bioavailable mineral content of rice grains. Additionally, Fe is well absorbed by the human gastrointestinal tract from soybean ferritin35. These results support the notion that Fe-fortification of rice by endosperm expression of ferritin and overexpression of NAS represents an effective human dietary Fe source. Moreover, increased NA in rice will likely reduce the rates of high-blood-pressure disease36.

Note that the Zn concentration also increased in Fer-NAS-YSL2 lines ( Fig. 6b ). Some reports show that higher NA production increases the Zn concentration in seeds of rice plants13,15,37. Endosperm-specific ferritin expression also contributes to the increased Zn concentration in rice seeds8. Zn deficiency is one of the most critical micronutrient deficiency problems in human health1, therefore representing an advantage of Fer-NAS-YSL2 rice in terms of practical applications.

We conclude that the introduction of multiple genes, including ferritin, under the control of endosperm-specific promoters, NAS overexpression and OsSUT1 and OsGlb1 promoter-driven OsYSL2 expression led to an increased concentration of bioavailable Fe in rice and will assist in mitigation of Fe-deficiency anemia globally.

Methods

Plant materials

japonica rice (O. sativa L.) cultivar Tsukinohikari was used as the non-transgenic control and for transformation. The OsActin1 promoter–HvNAS1 and OsSUT1 promoter–OsYSL2 transgenic lines were originally from refs.13 and 17, respectively.

Vector construction and rice transformation

Construction of the Fer-NAS-YSL2 transformation vector ( Fig. 1 ) is shown in Supplementary Fig. 5 . Agrobacterium tumefaciens (strain C58) was used to introduce the construct into O. sativa L. cv. Tsukinohikari using the method outlined in ref.38. Forty-five independent Fer-NAS-YSL2 lines were obtained as T0 transgenic plants.

Greenhouse cultivation

T0 transgenic plants, T1 selected plants and T2 selected plants were cultivated in commercially supplied soil used for cultivation of rice nurseries in Japan (Bonsolichigou; Sumitomo Chemicals, Tokyo, Japan) with slow-release fertilizers (LongTotal-70 and -140; JCAM AGRI. Co. Ltd., Tokyo, Japan) in a greenhouse. Using the seeds obtained, high-Fe lines were selected based on the Fe concentration in T1- ( Supplementary Fig. 2 ) and T2- ( Fig. 5 ) polished seeds. T3-immature seeds were used for gene expression analysis by real-time RT-PCR.

Detection of ferritin accumulation in T2 brown seeds

Six mature T2 brown seeds harvested from a greenhouse were homogenized with a mortar and pestle, soaked in extraction buffer [4% sodium dodecyl sulfate (SDS), 5% 2-mercaptethanol, 20% glycerol, 20 mM Tris–HCl, 8 M urea and 0.1% bromophenol blue, pH 6.8] and shaken for 30 min. The resulting extracts were centrifuged at 13,000 rpm for 20 min and supernatant fractions were collected. Protein separation by SDS-polyacrylamide gel electrophoresis, transfer to polyvinylidene fluoride membranes and detection with antibodies were performed as described in ref.6.

Quantitative real-time RT-PCR analysis

Total RNA was extracted from immature T3 seeds (seeds at an early milky stage, 10 days after fertilization) of each line harvested from a greenhouse. Seeds were crushed using a Multi-beads Shocker® (Yasuikikai, Osaka, Japan). Next, RNA was extracted using an RNeasy Plant Mini Kit (Qiagen KK, Tokyo, Japan). First-strand cDNA was synthesized using ReverTra Ace (Toyobo, Osaka, Japan) with oligo-d(T)30. Real-time RT-PCR was carried out using the 7300 Real-Time PCR System (Applied Biosystems, Tokyo, Japan) with SYBR Green I (Takara, Shiga, Japan) and ExTaq™ Real Time-PCR Version (Takara, Tokyo, Japan). The primers used were as follows: OsYSL2 forward (5’-GAG GGA CAA CGG TGT CAT TGC TGG T-3’) and OsYSL2 reverse (5’-TGC AGA AAA GCC CTC GAC GCC AAG A-3’) for OsYSL2 expression, HvNAS1 forward (5’-GGA CGT CGC CGA CCT CAC CCA G-3’) and HvNAS1 reverse (5’-CAG GGA CGC CCC CTC CAC C-3’) for HvNAS1 expression. Transcript levels were normalized to the expression levels of alpha-Tubulin, as determined using the primers alpha-Tubulin forward (5’-TCT TCC ACC CTG AGC AGC TC-3’) and alpha-Tubulin reverse (5’-AAC CTT GGA GAC CAG TGC AG-3’). The sizes of the amplified fragments were confirmed by agarose gel electrophoresis.

Detection of transgene insertion in the transgenic lines

Total DNA was prepared from leaves of T2-transgenic lines and the non-transgenic line, according to the method described in ref.39. The OsGlb1 promoter–SoyferH2 cassette was detected using the OsGlb1 promoter forward primer (5’-GAG CTA AGA GTT ATC CCT AGG-3’) and SoyferH2 reverse primer (5’-CCA CAT TGA TCT GCT CGT TGA TTG C-3’). The 2.3-kb OsGluB1 promoter–SoyferH2 cassette was detected using the 2.3-kb OsGluB1 promoter forward primer (5’-GCT TGC TAC GCA AAA TGA CAA CAT GC-3’) and SoyferH2 reverse primer. The OsActin1 promoter–HvNAS1 cassette was detected using the OsActin1 promoter forward primer (5’-GCA GCG CAT GGG TAT TTT TTC TAG-3’) and HvNAS1 reverse primer (5’-GCA CTA GTA TTC CCG ATC TAG TAA CAT AGA TG-3’). The OsGlb1 promoter–OsYSL2 cassette was detected using the OsGlb1 promoter forward primer and OsYSL2 reverse primer (5’-GCC CAA CGA AGC TAA TCG CAA AGA GG-3’). The OsSUT1 promoter–OsYSL2 cassette was detected using the OsSUT1 promoter forward primer (5’-CCC GTG AAA TAA TCG CAC CGT CTA G-3’) and OsYSL2 reverse primer. HPT was detected using the HPT forward primer (5’-CGG CAT CTA CTC TAT TCC TTT GC-3’) and HPT reverse primer (5’-GTC TCC GAC CTG ATG CAG CTC-3’). NPTII was detected using the NPTII forward primer (5’-GAT GGA TTG CAC GCA GGT TCT C-3’) and NPTII reverse primer (5’-GCC AAC GCT ATG TCC TGA TAG C-3’). iGUS was detected using the iGUS forward primer (5’-CTG TGG AAT TGA TCA GCG TTG G-3’) and iGUS reverse primer (5’-CGC AAG TCC GCA TCT TCA TGA C-3’).

Cultivation in an isolated paddy field

Selected T2 lines with a high Fe content in their polished seeds ( Fig. 5 , arrows) were used for field experiments. Seeds were germinated on 20 May 2009, as follows. First, seeds were surface-sterilized with 70% ethanol and then with 10% sodium hypochlorite and 0.1% Triton X-100 with shaking at 37°C for 30 min. Sterilized seeds were sown on half-strength Murashige and Skoog medium (sucrose 30 g/l, NH4NO3 1.65 g/l, KNO3 1.9 g/l, 440 mg/l, 370 mg/l, KH2PO4 170 mg/l, Fe–EDTA 42.1 mg/l, H3BO3 6.2 mg/l, 22.3 mg/l, , 8.6 mg/l, KI 0.83 mg/l, 250 ng/l, 25 ng/l, 25 ng/l, thiamine–HCl 100 ng/l, nicotinic acid 500 ng/l, pyridoxine-HCl 500 ng/l, glycine 2 mg/l, myo-inositol 100 mg/l, agar 8 g/l, pH 5.8) with 30 mg/l hygromycin (for transgenic lines) or without hygromycin (for the NT line). Germinated nurseries were grown in soil in a greenhouse until transplantation. The paddy field was established in an isolated field (6.0×10.65 m) at Gyeongsang University, Jinju, Korea (35°N, 128°E) ( Supplemental Fig. 3 ). The soil type was SiL (silt loam) with a soil pH of 6.45 (1:5 soil–water extract) and soil EC 0.47 (sd/m). The Zn, Mn and Cu concentrations of the soil were 1.27, 46.9 and 1.91 (mg/kg), respectively (0.1 N HCl extraction). Cd was not detected. Twenty-seven-day-old seedlings were transplanted to the paddy field on 16 June 2009. A commercial fertilizer (Toruzou-kun, N:P:K = 14:20:14; Zennou, Tokyo, Japan) was applied at 60 kgN/ha. The paddy field was submerged in water until it was drained 2 weeks before harvest. The plants were harvested on 29 October 2009 (135 days after transplantation). The plants entered the heading stage around 18–23 August 2009 (65–70 days after transplantation). After harvest, the inner four hills from each population were measured to determine the grain yield and concentrations of metals, NA and DMA.

Paddy field plant height and yield analysis

Plant height was measured on the harvest date. The yield in each block was calculated as the average of the middle four hills in each plot. The data represent the means ± standard errors of four block replications for each line ( Supplemental Fig. 3 ). Filled grains were selected by soaking in NaCl solution (1.06 g/cm3 gravity). The number of total grains per panicle was counted based on the ear of the main tiller.

Metals concentrations of seeds

T1 and T2 seeds from the ear of the main tiller were harvested from a greenhouse and polished using a Multi-beads Shocker® (Yasuikikai), as described previously13. The Fe concentration in the polished seeds was measured by inductively coupled plasma atomic emission spectroscopy (ICP-AES) (SPS1200VR; Seiko Instruments Inc., Chiba, Japan), as described in ref.13. T3 seeds from an ear of the main tiller were harvested from the middle four hills in each plot in the field and used for metal concentration analysis. Seeds were polished using a Multi-beads Shocker®. The Fe, Zn, Mn and Cu concentrations of polished seeds were measured as described in ref.13. The Cd concentration of polished seeds was measured using a Thermo Fisher X series2 ICP-MS (Thermo Fisher Scientific Inc., Waltham, MA, U.S.A.). Fe, Zn, Mn and Cu in brown seeds were digested as described in ref.13 and measured with an ICPS8100 (Shimadzu Co., Kyoto, Japan).

Determination of the NA and DMA concentrations in T3 seeds

T3 seeds from the ear of the main tiller were harvested. The NA and DMA concentrations in brown seeds were analyzed as described in refs.13 and 40. The NA and DMA concentrations were calculated as the average of the middle four hills in each block.

Statistics

Analysis of variance (ANOVA) with the Tukey–Kramer HSD test was used to compare data with JMP8 (SAS Institute, Cary, NC, U.S.A.). The level of significance was set at P < 0.05.