Field experiments were performed in the paddy field of the Institute of Sustainable Agroecosystem Services, the University of Tokyo (hereafter referred to as Tokyo field) (35°44′19.5″N 139°32′31.6″E) and Experimental Farm Station of Graduate School of Life Sciences, Tohoku University (hereafter referred to as Miyagi field) (38°15′24.4″N 140°51′29.6″E) during rice cultivation season. For the F2 population (225 lines) between WT and 1095_k grown in the Yayoi campus of The University of Tokyo (hereafter referred to as Yayoi field) (35°43′01.2″N 139°45′45.4″E).

Five seeds for each genotype were selected to determine the Fe and Zn concentrations after removing the husk. After the measurement of dry weight, grains were applied to HNO₃ digestion in PYREX® tubes (Iwaki) as described. Two milliliters of HNO₃ (Wako) were applied to the tubes and heated for an initial 1 h at 70°C and then 120°C. After complete evaporation of HNO₃, 1 ml of HNO₃ was added, and the same temperature setting was applied. After complete evaporation of HNO₃, 1 ml of hydrogen peroxide (H₂O₂) was applied to the samples until they evaporated. Finally, the samples were dissolved in a 0.08 N HNO₃ solution and used for element concentration determination using the Inductive Coupled Plasma Mass Spectrometry (ICP-MS, Agilent 7800; Agilent Technologies).

For the polishing of rice, 15 seeds from WT and 1095_k were polished using a Pearlest grain polisher (Kett) for 1 cycle of 40 s.

For agronomic trait measurements, WT and 1095_k were grown in the Tokyo field. The harvesting of the WT and 1095_k plant was performed 3 cm above the ground level. The plant height was measured from the tip of the panicle to the harvested end of the plant using a regular scale. The panicle from each tiller has been removed right from the panicle base and the length was measured. The number of ripe panicles was counted.

For DNA isolation, plant leaves of 15-days seedlings were used. The leaves were crushed using Zirconia beads using 300 μl of TPS buffer (100 mM Tris-HCl pH 6.8, 1 mM EDTA pH 8.0, 1.2 M KCl) followed by centrifugation at 10,000 × g for 13 min at 4°C. A 200 μl of the sample was transferred into the fresh tubes, and an equal volume of isopropylalchol was added. The solution was mixed and centrifuged at 10,000 × g for 30 min at 4°C. The supernatant was removed, and 70% ethanol was added, followed by centrifugation at 10,000 × g for 15 min. Again, the supernatant was removed, and the tubes were dried for not more than 10 min. The pellet was dissolved with 100 μl of TE buffer (10 mM Tris-HCl pH 8.0, 0.1 mM EDTA pH 8.0).

To determine the genomic sequence of OsVIT1 (LOC_Os04g38940), the region containing the promoter (2,078 bp) and downstream sequence (1,002 bp), was amplified by using a forward primer 5′-CGTGCCGGAGAGAAGAACTA-3′ and a reverse primer 5′-GAATGGTTTCTGCAAGCTGGG-3′. The sequence of the OsVIT1 genome was determined by the primers in Table S1 (Primer ID 3–12) using Sanger sequencing. For the determination of the OsVIT2 (LOC_Os09g23300) genomic sequence, the region containing the promoter (2,035 bp) and downstream sequence (2,047 bp) was amplified by using a forward primer 5′-GACCGGTTAATTTCCCAACCG-3′ and a reverse primer 5′-AACCGACGGCACCGATCTAC-3′. The genome sequence is determined by the primers in Table S1 (Primer ID 15–23) using Sanger sequencing.

The genotype of the F2 generation was determined by derived cleaved amplified polymorphic sequences (dCAPS) marker. The DNA was isolated from the F3 bulk (6 seeds). The OsVIT2 genomic fragment was amplified with 5′-CCGCGGAGATCGCGGACATACTGTCGCACTA-3′ and 5′-AATTAGTTTTTTCCCCTTACTTCATC-3′ (Primer ID 30 and 31 in Table S1), which introduce a DdeI site. The DdeI digested PCR product was applied to 3% agarose gel.

For real-time quantification, plants were germinated for 1 week in tap water followed by another 1 week in Kimura B solution (Uraguchi et al., 2009). RNA extraction was performed by the NucloSpin RNA plant (TaKaRa Bio). The relative quantification of OsVIT2 was performed using a forward primer 5′-GGTATCTGGCGGCGAAGAG-3′ and a reverse primer 5′-CGACAGTATGTCCGCGATCT-3′ using the Thermal Cycler Dice Real-Time System III (TaKaRa Bio). Relative expression levels were calculated against the reference gene, Ubiquitin (OsUBQ10), amplified with a forward primer 5′-AACCAGCTGAGGCCCAAGA-3′ and a reverse primer 5′-ACGATTGATTTAACCAGTCCATGA-3′.

The EMS mutant 1095_k was isolated from the EMS mutant of Oryza sativa cv. Hitomebore population through ionome screening (Tanaka et al., 2016). The consistency of the mutant (1095_k) phenotype was assessed by growing both Hitomebore [wild type (WT)] and 1095_k in the paddy field condition. Both WT and 1095_k lines were grown in four independent years in the Tokyo field. After harvesting, ionome analysis was done using ICP-MS. It was found that the mutant 1095_k has a 1.4-fold higher grain Fe concentration compared with the WT (Figure 1A: Tokyo). To observe the consistency in the phenotype in a different field, WT and 1095_k were grown in the Miyagi field (Figure 1A: Miyagi). In addition, 1095_k showed a 1.5-fold higher grain Fe concentration compared with WT. Besides Fe, 1095_k has a 1.2-fold higher grain Zn concentration than WT (Figure 1B: Tokyo). In the Miyagi field, 1095_k showed a 1.5-fold higher grain Zn concentration compared with WT (Figure 1B: Miyagi). High Zn phenotype was observed in 2013, 2016, 2019, and 2021 in the Tokyo field. These results indicate that 1095_k has high Fe and Zn concentration in brown rice independent of field condition.

Hence, as commercial rice is available in the polished form, we determine the elemental concentrations of polished rice harvested in the year 2021. Polishing of brown rice resulted in ~11.37% decline in the weight of WT grains and ~12.01% decline in 1095_k grains. The polished 1095_k grains showed a 1.4-fold increase in Fe and a 1.2-fold increase in Zn concentration compared with WT (Figures 1B,D).

Apart from Fe and Zn concentration, little difference in toxic elements, such as Cd and As, was observed between WT and 1095_k in both brown rice and polished rice (Figure S1).

Agronomic characters are an essential factor when considering breeding. Both WT and mutant 1095_k were cultivated in the Tokyo field. In each line, 41 plants were cultivated. The following yield components were measured: number of panicles/plant, number of tillers/plant, panicle weight/plant, panicle length, plant height, and 100 grain weight. No significant difference was observed in yield components between WT and 1095_k (Figures 2A–D,F, Figure S2B). The plant height was significantly smaller in 1095_k than that of the WT (Figure 2E, Figure S2A). These results suggest that 1095_k is a potential mutant that can retain high Fe and Zn without compromising yield components.

Several genes have been identified to be involved in the transport of Fe and Zn to the grain of rice. Fe transporters, OsVIT1 and OsVIT2, have been characterized, whose T-DNA insertion lines show the increase in Fe and Zn concentration of grain (Zhang et al., 2012; Bashir et al., 2013). In 1095_k, there is an increase in Fe and Zn concentration in grains similarly to the mutant of OsVITs (Figure 1). Therefore, we hypothesized that the alteration in OsVIT genes might be a cause of the phenotype. To test this, the genomic sequence of OsVIT1 and OsVIT2 was determined: the promoter (2,078 bp for OsVIT1 and 2,035 bp for OsVIT2), gene body, and downstream sequence (1,002 bp of OsVIT1 and 2,047 bp for OsVIT2). There is no nucleotide change in the OsVIT1. In OsVIT2, there is a nonsense mutation in the third exon, where thymine (T) at the 360th position in CDS was replaced by adenine (A), resulting in the formation of the stop codon (TAA) (Figure 3A). The relative mRNA accumulation of OsVIT2 is reduced in 1095_k both in shoot and roots (Figure 3B), which might be due to nonsense-mediated mRNA decay. These data suggest that nonsense mutation in OsVIT2 of 1095_k is a possible candidate gene behind the high Fe and Zn phenotype.

To confirm if OsVIT2 is a causal gene of 1095_k, we observed the correlation between genotype and element concentrations in the F2 crosses between WT and 1095_k. The plants were grown in the Yayoi field, and the concentrations of the elements were determined by ICP-MS. The Fe and Zn concentration increase was observed in 1095_k compared with WT in this field, which confirms that the Yayoi field condition can be used for the F2 population phenotyping (Figure 3C). Fe and Zn concentrations were determined in F2 populations and there is a positive correlation (Pearson's r = 0.418) between Fe and Zn concentration (Figure 3D), suggesting that high Fe and Zn phenotype is caused by the same gene. Next, we observed the relationship between the OsVIT2 genotype and Fe or Zn concentration in the F2 population. The genotype of all the F2 (225 lines) was determined by the dCAPS marker (Figure 3E). To see the association between genotype and phenotype, a statistical analysis was performed for Fe and Zn concentration (Figure 3D). For both Fe and Zn concentrations, there is a statistical significance between the wild-type homozygous and mutant-type homozygous lines (WT Homo and 1095_k Homo in Figures 3D,E). These results strongly suggest that nonsense mutation of OsVIT2 is responsible for the high Fe and Zn phenotypes in 1095_k.

An EMS mutant, 1095_k, has an increase in Fe and Zn concentration in brown rice as well as in polished rice (Figures 1, 3C). Furthermore, there is an association between Fe and Zn concentration and OsVIT2 mutation in F2 crosses between WT and 1095_k. The OsVIT2 is a vacuolar localized transporter, which plays an important role in the vacuolar sequestration of Fe to regulate Fe homeostasis (Zhang et al., 2012; Bashir et al., 2013). It has been demonstrated that the disruption of OsVIT2 results in an increase in Fe and Zn concentration in rice grains. In addition to these elements, Cu and Mn change, although it varies from experiment to experiment: Cu increases in OsVIT2 mutant but no change in Mn (Bashir et al., 2013); no change both in Cu and Mn (Zhang et al., 2012). In terms of plant growth, the OsVIT2 mutant shows a decrease in shoot length at the seedling stage (Zhang et al., 2012). Although we did not perform the growth test in the seedling stage, in the field test, the plant height of 1095_k was lower than WT (Figure 2E). Taken together, these results suggest that OsVIT2 could be a causal gene for the high grain Fe and Zn phenotype of 1095_k.

In 1095_k, OsVIT2 mutation increased Fe and Zn content of grain in fields without increasing toxic elements, such as Cd and As, for humans (Figure 1 and Figure S1). In addition, we showed that the mutation does not affect the growth of rice, and furthermore, no reduction in the yield components was observed (Figure 2 and Figure S2). These results indicate that the mutation in OsVIT2 could be a candidate for DNA marker to breed high Fe and Zn. All of these characters are beneficial for breeding new cultivars.

OsVIT2 is localized to vacuole and sequester Fe into vacuole (Zhang et al., 2012). OsVIT2 mRNA is expressed throughout whole growth stages (Bashir et al., 2013) and upregulated by high Fe in the seedling stage (Zhang et al., 2012). These results suggest the function of OsVIT2 in excess Fe tolerance. Bashir et al. (2013) grew the T-DNA knockdown line under excess Fe conditions and found that under 100 and 500 μM Fe-EDTA conditions, the plant height is shorter than that of control plants, while root length and soil plant analysis development (SPAD) values are similar between them. Therefore, OsVIT2 may have some function in excess Fe tolerance but may not affect the growth severely. In our experiments, although the plant height is short at 1095_k, the yield of 1095_k is similar to WT. Therefore, the yield of the OsVIT2 mutant would not be affected so much. The other possible disadvantage is the high accumulation of toxic elements in humans, such as Cd and As. In our field condition and also in Cd-contaminated soil (Che et al., 2021), there is no significant difference between WT and mutants. Taken together with yield results, OsVIT2 is suitable as a breeding material.

There have been many efforts made in the past to increase the Fe and Zn concentration in the rice grains, but are generated by the transgenic means due to which they are restricted to use around the world (Takahashi et al., 2001; Lee and An, 2009; Zheng et al., 2010). In the case of 1095_k, an EMS mutant, the freedom of usage increases compared with genetically modified rice. Furthermore, if transgenic lines are accepted, breeders can use 1095_k as a material for transformation by overexpression of genes enhancing Fe and/or Zn concentration in grain as the mutant does not have any selection marker (e.g., hygromycin) gene. Hence, mutant 1095_k is a strong candidate to address the hidden hunger problem in the coming future.