Japonica rice (Oryza sativa L.) cultivar Tsukinohikari was used as the NT control and for transformation. The transgenic rice line gNAS1 (Fig. 2a) possesses a 13.0-kb HvNAS1 barley genome fragment and originated from line 24 of Higuchi et al. . The transgenic rice line gIDS3 (Fig. 2c) possesses a 20-kb IDS3 barley genome fragment and originated from line 1 of Kobayashi et al. .
To produce the transgenic rice line gNAS1 + gNAAT (Fig. 2b), a 7.6-kb barley HvNAS1 genome fragment  was excised using HindIII and inserted into the HindIII site of the vector pBIGRZ . Then, an 11.0-kb HvNAAT-A and HvNAAT-B genome fragment  was excised using NotI and inserted into the NotI site of pBIGRZ-NAS. The resultant vector was introduced into rice cultivar Tsukinohikari following the methods of Higuchi et al. .
Genomic insertion was confirmed using a polymerase chain reaction-based method and Southern blots. Gene expression was also confirmed using Northern blots under Fe-deficient conditions [8, 11] (data not shown).
The seeds of the three transgenic rice lines were harvested in an environmentally isolated paddy field that had Andosol soils at Tohoku University, Japan, in 2005. The seeds were sown on nursery beds for the experiment. The seeds were size-selected by soaking in NaCl solution. The specific gravity of the NaCl solution was 1.13 g/cm3 for the selection of NT, gNAS1, and gNAS1 + gNAAT seeds. Because there were not enough seeds from the gIDS3 line, gIDS3 seeds were selected using 1.06 g/cm3 NaCl solution. After size selection, the seeds were soaked overnight in fungicide solution (Momi-guard, Hokuko, Japan) and then soaked in water at 20°C for 6 days to break dormancy. The seeds were then incubated in water at 30°C overnight to accelerate germination. The seeds were sown in nursery bed soil by 3 April 2006. Previously germinated seeds were sown on brown soil (pH 5.5; Inaho Kumiai Muhiryou Baido, Agricultural Corporation, Japan) in small pots (3 ml) and then covered with the same soil. Three seeds were sown per pot, covered with soil, and grown for 45 days in a greenhouse at the environmentally isolated field.
Experimental design and transplantation
The paddy field was established at the Field Science Center of Tohoku University, Miyagi, Japan (39°N; 141°E). A 6.6 × 6.9 m paddy field was used for the experiment (Figs. 3a and 4). The soil type was Andosol with approximate pH of 6.0.
Seedlings were transplanted into the paddy field on 18 May 2006. A commercial fertilizer (Toruzou-kun; NPK = 14:20:14; Zennou, Japan) was applied at transplanting at 60 kg N ha−1. Each hill consisted of three plants. Column width of hill to hill was 30 and 15 cm (Fig. 3b). Each replication comprised eight (120 cm) × four (120 cm) hills (Fig. 3b). Each line was planted in four replications in the experimental area (Fig. 3a). To adjust the light and nutrient conditions between center and border replications of experimental area, NT plants were grown around the experimental area (Fig. 3a, gray zone). The paddy field was submerged in water until it was drained on 9 October 2006.
Measurement of plant growth and yield
The height of plants, number of tillers, and leaf color (SPAD value; Konika Minolta, Japan) of the largest leaves were measured for all plants in the inner six hills of each replication (Fig. 3b, black hills) on 3 June (16 DAT), 17 June (30 DAT), 29 June (42 DAT), 12 July (55 DAT), 26 July (69 DAT), and 27 August (101 DAT). Plants headed at around 23 August (97 DAT). The plants were harvested on 23 October (158 DAT). After harvest, the total number of seeds and grain weight for the inner six hills from each replication were measured to determine the grain yield.
Metal concentration analyses
Brown rice grains (specific gravity > 1.06 g/cm3) were collected randomly from all plants from the inner six hills of each replication (Fig. 3b, black hills) for metal concentration analyses. Ten seeds were digested with 1 ml of 13 M HNO3 and 1 ml of 8.8 M H2O2 (Wako, Japan) at 200°C for 20 min using MARS Xpress (CEM, USA); seeds were digested in six replicate subsamples from each replication. After digestion, the samples were collected and diluted to 5 ml and analyzed using a SPS1200VR ICP-AES (Seiko, Japan).
To obtain polished seeds, 60 brown seeds were placed into a 2-ml tube and shaken at 2,500 rpm for 150 s, at least four times, using a Multi-beads Shocker (Yasui Kikai, Japan). Approximately 25 well-polished seeds from each replication were digested and their metal concentration measured.
The aboveground parts other than ears (i.e., leaves and stems) were harvested. The plant materials were ground using a crushing machine (Fujiwara Seisakusho, Japan). Approximately 250–300 mg of the ground powder was digested and the metal concentration was measured.
Approximately 200 mg each of rachis and hulls were collected from each replication of NT plants, digested, and the metal concentration was measured. The total metal concentration of each plant part (i.e., leaves and stems, rachis, hulls, brown rice, and polished rice) was calculated.
Data were compared using analysis of variance or a Student–Newman–Keuls test; P < 0.05 was considered statistically significant.