Concomitant Activation of OsNAS2 and OsNAS3 Contributes to the Enhanced Accumulation of Iron and Zinc in Rice

Nicotianamine (NA) is produced by NA synthase (NAS), which contains three genes in rice and is responsible for chelating metals such as iron (Fe) and zinc (Zn), as well as preserving metal homeostasis. In this study, we generated a transgenic plant (23D) that shows simultaneous activation of OsNAS2 and OsNAS3 by crossing two previously identified activation-tagged mutants, OsNAS2-D1 (2D) and OsNAS3-D1 (3D). Concomitant activation of both genes resulted in the highest Fe and Zn concentrations in shoots and roots of the 23D plants grown under normal conditions and Fe and Zn limited growth conditions. Expression of genes for the biosynthesis of mugineic acid family phytosiderophores (MAs) and Fe and Zn uptake were enhanced in 23D roots. Additionally, 23D plants displayed superior growth to other plants at higher pH levels. Importantly, 23D seeds had NA and 2′-deoxymugineic acid (DMA) concentrations that were 50.6- and 10.0-fold higher than those of the WT. As a result, the mature grain Fe and Zn concentrations of the 23D plant were 4.0 and 3.5 times greater, respectively, than those of the WT. Furthermore, 23D plants exhibited the greatest resistance to excess metals. Our research suggests that simultaneous activation of OsNAS2 and OsNAS3 can enhance Fe and Zn accumulation in rice grains while also increasing plant tolerance to growing situations with metal deficiency and excess metal availability.


Introduction
For proper growth and development of plants and humans, iron (Fe) and zinc (Zn) are essential micronutrients and accumulate in edible parts of crops that are primary sources of micronutrients for humans [1]. In plants, Fe is required for electron transfer reactions in mitochondria, chlorophyll biosynthesis, and photosynthesis [2,3]. Zn is a structurally important component for almost 10% of all plant proteins; affects proteinprotein interactions in plant cells; and is required as a cofactor for the functional activity of various enzymes such as Cu/Zn superoxide dismutases, carbonic anhydrase, and alkaline phosphatase [4,5]. In the human body, Fe takes part in various cellular functions, such as immunity, oxygen transport, cellular respiration, energy metabolism, electron transfer reactions, and cell division and differentiation [6,7]. Zn plays a critical role as an essential including a construct containing AtIRT1 that encodes an Iron-Regulated Transporter1, that was driven by its native promoter and another construct containing PvFer that was driven by OsGlB-1 promoter and AtNAS1 driven by CaMV35S promoters, concentrations of Fe and Zn were increased up to 3.8-fold and 1.8-fold, respectively, in polished grains compared to the WT [28]. When a single construct containing the tissue-specific expression of the ZmYS1 transporter driven by the OsHMA2 promoter and the TOM1 transporters by OsFRDL1 was combined with endosperm-specific expression of ferritin and the constitutive expression of NAS, Fe and Zn concentrations were increased up to 9.3-fold and 1.2-fold, respectively, compared with WT [29]. However, none of the studies that employed the gene stacking method included multiple OsNAS genes.
In this study, we generated a transgenic rice line (23D) with simultaneous activation of OsNAS2 and OsNAS3 by crossing two rice activation-tagged mutants, OsNAS2-D (2D) and OsNAS3-D (3D) [17][18][19]. Mature grains of 23D contained increased levels of NA, DMA, Fe, and Zn compared to the 2D, 3D, and WT. In addition, we showed that 23D plants exhibited the highest tolerance to Fe and Zn deficiencies and excess metal toxicities in comparison with the 2D, 3D, and WT plants.

Isolation of Double-Activation-Tagged Mutant and Its Enhanced Tolerance to Fe and Zn Deficiencies
Previously, two activation-tagged mutants of rice NAS genes, OsNAS2-D1 and OsNAS3-D1, were studied to increase micronutrient content, specifically Fe and Zn [17][18][19]. Because simultaneous manipulation of multiple genes to increase micronutrient content is a good practice to enhance the nutritive value of grains, we attempted to increase OsNAS2 and OsNAS3 expressions together. For this aim, we crossed the two activation-tagged mutants, OsNAS2-D1 (2D) and OsNAS3-D1 (3D), and isolated a transgenic line containing both activation-tagged transgenes (23D) by PCR. Then, we analyzed the expressions of OsNAS2 and OsNAS3 in seedling shoots and roots, and flag leaves of 2D, 3D, and 23D along with the WT ( Figure S1). Compared to WT, significant expression increases of both genes were observed in all tissues, shoots, and roots of seedlings and flag leaves, tested of 23D, whereas OsNAS3 expression did not alter in 2D and OsNAS2 expression did not alter in 3D ( Figure S1).
We raised the mutant seedlings in Fe-and Zn-deficient medium to examine whether the elevated expression of OsNAS2 and/or OsNAS3 could have any effect on growth and metal distribution (Figures 1 and S2). Seedlings of all mutants did not show any significant change in shoot length and chlorophyll content on the control media, Yoshida solution, and a half-strength MS medium (Figures 1a,d,e and S2a,d,e). However, significant differences were observed in all mutant seedlings grown under Fe-and Zn-deficient media compared to those of the WT (Figures 1b-e and S2b-e). 23D seedlings outperformed all other seedlings, with lengths of 69.4 and 56.4% higher than the WT seedlings on Fe-and Zn-deficient media, respectively, while the lengths of 2D seedlings were 21 and 35.6% more, and 3D seedlings were 28.6 and 34.2% more on Fe and Zn deficient media, respectively, compared to the WT seedlings. Moreover, all transgenic seedlings had less chlorosis when Fe was limited, as compared to the WT (Figures 1e and S2e). For example, chlorophyll concentrations for 23D, 2D, and 3D were increased to 159.4%, 153.8%, and 157.6%, respectively, compared to the WT.
To evaluate whether simultaneous activation of OsNAS2 and OsNAS3 affected metal distribution, we measured Fe and Zn levels in shoots and roots of the mutants and WT seedlings (Figure 1f,g). We found that both metals were increased significantly in the mutants, irrespective of external metal supply, compared to that of the WT (Figure 1f,g). It is noteworthy that 23D seedlings had the highest Fe and Zn contents, compared with those of 2D, 3D, and WT (Figure 1f,g). Under the control growth conditions, the Fe content in the shoots of 2D, 3D, and 23D seedlings was increased to 198.6%, 218.9%, and 254.2%, respectively, compared with WT, while the corresponding root levels increased to 168.3%, 172.4%, and 203.5%, respectively (Figure 1f). Under Fe deficiency, Fe content in shoots of 2D, 3D, and 23D seedlings increased to 191.5%, 205.2%, and 232.9%, respectively, compared with WT, whereas the corresponding root values rose to respective values of 154.8%, 167.5%, and 203.7% (Figure 1f). Under Zn deficiency, Fe concentration in shoots and roots of 23D seedlings were highest out of the plants (Figure 1f). That is, the Fe contents in the shoots of 2D, 3D, and 23D seedlings were increased to 206.3%, 216.1%, and 232.9%, respectively, compared with WT, while the corresponding root levels increased to 168.9%, 171.8%, and 203.7%, respectively (Figure 1f). To evaluate whether simultaneous activation of OsNAS2 and OsNAS3 affected metal distribution, we measured Fe and Zn levels in shoots and roots of the mutants and WT seedlings (Figure 1f,g). We found that both metals were increased significantly in the mutants, irrespective of external metal supply, compared to that of the WT (Figure 1f,g). It is In addition, compared to WT, Zn concentrations in the shoots of 2D, 3D, and 23D seedlings grown on Zn sufficient medium increased by 187.1%, 208.6%, and 245.8%, respectively, while the corresponding root values increased by 170.3%, 180.5%, and 197.2%, respectively ( Figure 1g). Zn contents of 2D, 3D, and 23D seedling shoots increased under Zn deficiency to 179.3%, 188.6%, and 241.6%, respectively, compared to WT, whereas the corresponding root values increased to 187.3%, 179.4%, and 222.4%, respectively (Figure 1g). Even under Fe deficiency, Zn concentrations in shoots and roots of 23D seedlings were higher than that of other plants same as normal and Zn deficiency conditions (Figure 1g).
The above data prompted us to investigate the transcript abundance of various genes involved in the homeostasis of Fe and Zn ( Figure 2): the nicotianamine aminotransferase gene OsNAAT1 and 2 -deoxymugineic acid synthase gene OsDMAS1, which act downstream of OsNAS genes and are involved in the synthesis of MAs [30,31]; OsIRT1 and OsYSL15 (Yellow Stripe1-Like 15), which uptake Fe from the soil as membrane transporters [32][33][34][35]; OsYSL2, which acts as NA-metal transporter and facilitates Fe translocation [36,37]; rice ferritin1 (OsFer1), which acts as a principal Fe storage protein [38]; and OsZIP4 and OsZIP5, which encode a membrane protein that transports Zn [39,40]. Transcript levels of these genes were found to be significantly higher in the roots of all the mutants compared to the WT ( Figure 2). Overall, we found that the roots of 23D had the highest level of all the transcripts we looked at in comparison to others ( Figure 2). pared with WT, whereas the corresponding root values rose to respective values of 154.8%, 167.5%, and 203.7% (Figure 1f). Under Zn deficiency, Fe concentration in shoots and roots of 23D seedlings were highest out of the plants (Figure 1f). That is, the Fe contents in the shoots of 2D, 3D, and 23D seedlings were increased to 206.3%, 216.1%, and 232.9%, respectively, compared with WT, while the corresponding root levels increased to 168.9%, 171.8%, and 203.7%, respectively (Figure 1f).
In addition, compared to WT, Zn concentrations in the shoots of 2D, 3D, and 23D seedlings grown on Zn sufficient medium increased by 187.1%, 208.6%, and 245.8%, respectively, while the corresponding root values increased by 170.3%, 180.5%, and 197.2%, respectively ( Figure 1g). Zn contents of 2D, 3D, and 23D seedling shoots increased under Zn deficiency to 179.3%, 188.6%, and 241.6%, respectively, compared to WT, whereas the corresponding root values increased to 187.3%, 179.4%, and 222.4%, respectively ( Figure  1g). Even under Fe deficiency, Zn concentrations in shoots and roots of 23D seedlings were higher than that of other plants same as normal and Zn deficiency conditions (Figure 1g).
The above data prompted us to investigate the transcript abundance of various genes involved in the homeostasis of Fe and Zn ( Figure 2): the nicotianamine aminotransferase gene OsNAAT1 and 2′-deoxymugineic acid synthase gene OsDMAS1, which act downstream of OsNAS genes and are involved in the synthesis of MAs [30,31]; OsIRT1 and OsYSL15 (Yellow Stripe1-Like 15), which uptake Fe from the soil as membrane transporters [32][33][34][35]; OsYSL2, which acts as NA-metal transporter and facilitates Fe translocation [36,37]; rice ferritin1 (OsFer1), which acts as a principal Fe storage protein [38]; and OsZIP4 and OsZIP5, which encode a membrane protein that transports Zn [39,40]. Transcript levels of these genes were found to be significantly higher in the roots of all the mutants compared to the WT ( Figure 2). Overall, we found that the roots of 23D had the highest level of all the transcripts we looked at in comparison to others ( Figure 2).
, OsFer1 (f), OsZIP4 (g), and OsZIP5 (h) in the roots of WT and activation-tagged mutants. Significant differences among WT and mutant plants are indicated with an asterisk (* p < 0.05 by Student's t-test).

Expression Analysis of Three OsNAS Genes and the Comparison of Seedling Growth under High pH Conditions
pH influences the solubility and ionic form of several elements in soil and affects nutrient availability for plant growth [41]. For instance, ferric ion precipitates as the utterly , OsFer1 (f), OsZIP4 (g), and OsZIP5 (h) in the roots of WT and activation-tagged mutants. Significant differences among WT and mutant plants are indicated with an asterisk (* p < 0.05 by Student's t-test).

Expression Analysis of Three OsNAS Genes and the Comparison of Seedling Growth under High pH Conditions
pH influences the solubility and ionic form of several elements in soil and affects nutrient availability for plant growth [41]. For instance, ferric ion precipitates as the utterly insoluble ferric hydroxide in an aerated solution with a pH greater than 8 [19]. Zn solubility reduces by 100-fold for each unit increase in pH [42]. Therefore, we speculated that high pH could affect the expression of OsNASs even under Fe or Zn sufficiency. Then, we analyzed the expression levels of OsNAS genes in shoots and roots of the seedlings grown under high pH (pH 8.5) solution, along with or without Fe or Zn ( Figure 3). Transcript levels of OsNAS1 and OsNAS2 were upregulated under alkaline growth solution, as compared to normal growth conditions (pH 5.5) (Figure 3a,b). Even in seedling roots supplemented with Fe and Zn, high pH could promote their expression, indicating that the bioavailability of Fe and Zn decreased due to the alkaline media. High pH without Fe led to the highest levels of OsNAS1 and OsNAS2 expression, respectively, relative to other growth conditions. However, the shoots and roots of the seedlings grown under both growth conditions showed significantly lower OsNAS3 expressions in Fe-deficient media as compared to the seedlings cultivated with Fe ( Figure 3c). Under Zn deficiency, no significant expression changes were observed for OsNAS1 in seedling shoots and roots ( Figure 3a). In seedling shoots and roots, OsNAS2 expression increased upon Zn exclusion, but high pH raised it more ( Figure 3b). Regardless of pH, OsNAS3 rose in Zn-deficient seedlings (Figure 3c). According to these results, Fe or Zn deficiency combined with high pH causes significant expression changes in seedling shoots and roots, and OsNAS gene expression responded differently to high pH. Then, we investigated the growth response of WT, 2D, 3D, and 23D seedlings under high pH Yoshida solution and a half-strength MS medium, with or without Fe or Zn (Figures 4 and S3). In comparison to WT seedlings, all mutant seedlings of 2D, 3D, and 23D displayed a considerable increase in shoot length when grown in a high-pH environment, irrespective of Fe or Zn supply. Significantly, the 23D seedlings outperformed all others. For instance, in high-pH conditions, 2D, 3D, and 23D seedlings increased their shoot lengths by 25.9, 21.0, and 48.8% more than WT seedlings, respectively (Figures 4a,d and S3a,d). At high-pH conditions without Fe, shoot lengths for 2D, 3D, and 23D seedlings increased by 41.2%, 41.2%, and 75.6%, respectively, in comparison to WT seedlings (Figures 4b,d and S3b,d). Moreover, in high-pH conditions without Zn, 2D, 3D, and 23D seedlings' shoot lengths increased by 21.1%, 20.0%, and 43.7%, respectively, in comparison to WT seedlings (Figures 4c,d and S3c,d). Compared to WT, 2D, and 3D, our results suggest that the double-activation-tagged mutant 23D, which simultaneously expresses higher levels of OsNAS2 and OsNAS3, is better able to counteract the detrimental effects of high pH and high pH without Fe or Zn conditions on plant growth and development.

Increased Expression of Two OsNAS Genes Boosted Fe and Zn Contents in Mature Seeds
To test if the double activation of OsNAS2 and OsNAS3 might further increase Fe and Zn concentrations, we measured Fe and Zn concentrations in the flag leaves of WT, 2D, 3D, and 23D plants. We found that the mutant flag leaves had higher concentrations of Fe and Zn than the WT flag leaves ( Figure S4). Importantly, flag leaves of 23D contained the highest levels of Fe and Zn in comparison with other mutants. For instance, compared to the WT, Fe concentrations were 1.49-, 1.51-, and 1.87-fold elevated in the 2D, 3D, and 23D mutants, respectively ( Figure S4a), while Zn concentrations were 1.49-, 1.59-, and 1.94-fold elevated in the 2D, 3D and 23D mutants, respectively ( Figure S4b). Elevated amounts of Fe and Zn in flag leaves were correlated with increased amounts of NA and DMA in mature seeds, increasing the accumulation of Fe and Zn [18,19]. Therefore, we measured NA, DMA, Fe, and Zn concentrations in mature seeds ( Figure 5). NA concentrations in mature seeds of 2D, 3D, and 23D were found to be 20.2, 30.9, and 50.6 times higher, respectively, compared to WT (Figure 5a). In addition, DMA concentrations in 2D, 3D and 23D seeds were also increased to being 7.2-, 7.4-, and 10.0-fold higher, respectively, compared to the WT (Figure 5b). Following the enhanced NA and DMA amounts, both Fe and Zn were observed to be enhanced (Figure 5c,d). Relative to mature WT seeds, Fe levels in 2D, 3D, and 23D were 2.7, 3.1, and 4.0 times higher, respectively. (Figure 5c). We also performed Perls' Prussian staining to visualize the accumulation and intensity of the Fe in the mature seeds and embryos ( Figure S5). An enhanced blue color was observed in stained 2D, 3D, and 23D mature seeds and embryos compared to the WT ( Figure S5a,b). Notably, the strongest color observed visually was evident in mature 23D seeds ( Figure S5a) and embryos ( Figure S5b), suggesting the highest accumulation of Fe. Furthermore, concentrations of Zn in mature 2D, 3D, and 23D seeds were found to be 2.1, 2.4, and 3.5 times higher than those of WT, respectively ( Figure 5d). These data suggest that dual activation of OsNAS2 and OsNAS3 could accumulate more Fe and Zn in rice grains than single-activation mutants of OsNAS2 or OsNAS3.
However, negative impacts on agronomic traits were observed for the matured mutant plants compared to WT plants when grown in paddy fields (Figure 6a,b). 2D exhibited similar plant height and biomass as compared to the WT, but less seed yield (84.5% of WT) than the WT. Plant height, aboveground dry weight, and seed yield of 3D were decreased to 74.1%, 61.2%, and 47.5% of WT, respectively. 23D's agronomic characteristics fell in between 2D and 3D. When the plants were grown in alkaline soil, the single activationtagged mutants displayed similar plant height and biomass, but less seed yield as compared to WT (Figure 6c,d). Importantly, 23D plants exhibited a highly significant increase in plant height, aboveground mass, and seed yield compared to all others (Figure 6c,d). and OsNAS3 (c) grown under normal (pH 5.5) or high pH (pH 8.5) Yoshida solution wi (Con), or without Fe (Fe-) or Zn (Zn-). Plants were grown in normal Yoshida solution fo thereafter were transferred to each growth condition. The seedlings were grown for a three days, and shoots and roots from the seedlings were sampled for RNA isolation differences between the control (pH 5.5) and each growth condition are indicated wit (* p < 0.05 by Student's t-test).  OsNAS2 (b), and OsNAS3 (c) grown under normal (pH 5.5) or high pH (pH 8.5) Yoshida solution with Fe and Zn (Con), or without Fe (Fe-) or Zn (Zn-). Plants were grown in normal Yoshida solution for 7 days and thereafter were transferred to each growth condition. The seedlings were grown for an additional three days, and shoots and roots from the seedlings were sampled for RNA isolation. Significant differences between the control (pH 5.5) and each growth condition are indicated with an asterisk (* p < 0.05 by Student's t-test).

Increased Expression of Two OsNAS Genes Boosted Fe and Zn Contents in Mature Seeds
To test if the double activation of OsNAS2 and OsNAS3 might further increase Fe and Zn concentrations, we measured Fe and Zn concentrations in the flag leaves of WT, 2D, 3D, and 23D plants. We found that the mutant flag leaves had higher concentrations of Fe and Zn than the WT flag leaves ( Figure S4). Importantly, flag leaves of 23D contained the highest levels of Fe and Zn in comparison with other mutants. For instance, compared to the WT, Fe concentrations were 1.49-, 1.51-, and 1.87-fold elevated in the 2D, 3D, and 23D mutants, respectively ( Figure S4a), while Zn concentrations were 1.49-, 1.59-, and 1.94-fold elevated in the 2D, 3D and 23D mutants, respectively ( Figure S4b). Elevated amounts of Fe and Zn in flag leaves were correlated with increased amounts of NA and DMA in mature seeds, increasing the accumulation of Fe and Zn [18,19]. Therefore, we measured NA, DMA, Fe, and Zn concentrations in mature seeds ( Figure 5). NA concentrations in mature seeds of 2D, 3D, and 23D were found to be 20.2, 30.9, and 50.6 times higher, respectively, compared to WT (Figure 5a). In addition, DMA concentrations in 2D, 3D and 23D seeds were also increased to being 7.2-, 7.4-, and 10.0-fold higher, respectively, compared to the WT (Figure 5b). Following the enhanced NA and DMA amounts, both Fe and Zn were observed to be enhanced (Figure 5c,d). Relative to mature WT seeds, Fe levels in 2D, 3D, and 23D were 2.7, 3.1, and 4.0 times higher, respectively. (Figure 5c). We also performed Perls' Prussian staining to visualize the accumulation and intensity of the Fe in the mature seeds and embryos ( Figure S5). An enhanced blue color was observed in stained 2D, 3D, and 23D mature seeds and embryos compared to the WT ( Figure S5a,b). Notably, the strongest color observed visually was evident in mature 23D seeds ( Figure S5a) and embryos ( Figure S5b), suggesting the highest accumulation of Fe. Furthermore, concentrations of Zn in mature 2D, 3D, and 23D seeds were found to be 2.1, 2.4, and 3.5 times higher than those of WT, respectively ( Figure 5d). These data suggest that dual activation of Os-NAS2 and OsNAS3 could accumulate more Fe and Zn in rice grains than single-activation mutants of OsNAS2 or OsNAS3. However, negative impacts on agronomic traits were observed for the matured mutant plants compared to WT plants when grown in paddy fields (Figure 6a,b). 2D exhibited similar plant height and biomass as compared to the WT, but less seed yield (84.5% of WT) than the WT. Plant height, aboveground dry weight, and seed yield of 3D were

Increased Tolerance to Excess Metals during the Seedling Growth
NA is thought to participate in the detoxification of excess metals. Therefore, we verified the growth response of the mutants under excessive concentrations of various heavy metals (Figures 7 and S6). 2D, 3D, and 23D seedlings exhibited better growth under excessive amounts of Zn (5 mM), Cu (0.3 mM), Fe (1 mM), Ni (0.5 mM), and Co (0.5 mM) compared to the WT. Interestingly, the 23D mutant seedling was the best performer under all the treatments of excess heavy metal concentrations (Figures 7 and S6). For instance, shoot lengths in 2D, 3D, and 23D were 12.4%, 22.9%, and 31.4% higher than the WT, respectively, at a higher Zn level (Figure 7a,d); shoot lengths of 2D, 3D, and 23D were 40.6%, 57.8%, and 89.1% more than the WT, respectively, at an excess Cu level (Figure 7b,d); shoot lengths of 2D, 3D, and 23D were 8.8%, 10.5%, and 13.8% more than the WT, respectively, at an excess Fe level (Figure 7c,d); shoot lengths of 2D, 3D, and 23D were 21.5%, 30.0%, and 35.5% more than the WT, respectively, at an excess Ni level ( Figure S6a,c); shoot lengths of 2D, 3D, and 23D were 14.5%, 25.9%, and 42.5% more than the WT, respectively, at an excess Co level ( Figure S6b,c). These data suggest that concomitant activation of Os-NAS2 and OsNAS3 can improve plant tolerance at elevated heavy metal toxic conditions.

Increased Tolerance to Excess Metals during the Seedling Growth
NA is thought to participate in the detoxification of excess metals. Therefore, we verified the growth response of the mutants under excessive concentrations of various heavy metals (Figures 7 and S6). 2D, 3D, and 23D seedlings exhibited better growth under excessive amounts of Zn (5 mM), Cu (0.3 mM), Fe (1 mM), Ni (0.5 mM), and Co (0.5 mM) compared to the WT. Interestingly, the 23D mutant seedling was the best performer under all the treatments of excess heavy metal concentrations (Figures 7 and S6). For instance, shoot lengths in 2D, 3D, and 23D were 12.4%, 22.9%, and 31.4% higher than the WT, respectively, at a higher Zn level (Figure 7a,d); shoot lengths of 2D, 3D, and 23D were 40.6%, 57.8%, and 89.1% more than the WT, respectively, at an excess Cu level (Figure 7b,d); shoot lengths of 2D, 3D, and 23D were 8.8%, 10.5%, and 13.8% more than the WT, respectively, at an excess Fe level (Figure 7c,d); shoot lengths of 2D, 3D, and 23D were 21.5%, 30.0%, and 35.5% more than the WT, respectively, at an excess Ni level ( Figure S6a,c); shoot lengths of 2D, 3D, and 23D were 14.5%, 25.9%, and 42.5% more than the WT, respectively, at an excess Co level ( Figure S6b,c). These data suggest that concomitant activation of OsNAS2 and OsNAS3 can improve plant tolerance at elevated heavy metal toxic conditions.

Discussion
Transgenic approaches, especially multiple gene expression, have already been proven effective to biofortify enhanced amounts of Fe and Zn in rice grains. Much research on the biofortification of Fe and Zn in rice grains has introduced one of the three OsNAS genes alone or in combination with other genes involved in metal homeostasis, but none of them have introduced two OsNAS genes at once. In this study, we generated a transgenic line (23D) with increased OsNAS2 and OsNAS3 expression to see if simultaneous activation of OsNAS2 and OsNAS3 would improve the accumulation of Fe and Zn in rice grains more than the single-activation-tagged mutants. Compared to the WT and singleactivation-tagged mutants, NA and DMA levels were considerably higher in 23D, and as a result, Fe and Zn levels were significantly raised in mature grains. In addition, 23D had much higher concentrations of Fe and Zn in its leaves and roots, suggesting that NA and DMA had also accumulated there. Hence, the mutants were more resistant to Fe and Zn deficiency and high-pH conditions. In particular, 23D plants outperformed all others by demonstrating more tolerance.
Combinatorial effects of OsNAS2 and OsNAS3 resulted in 23D's improved properties as compared to 2D or 3D. This might have been because OsNAS2 and OsNAS3 have differing physiological functions [15]. In other words, OsNAS2 expression was present in Fesufficient root, and under Fe deficiency, its levels significantly increased in roots and leaves. OsNAS3 expression was enhanced in roots but inhibited in leaves in response to Fe deprivation, although the OsNAS3 transcript was present in both Fe-sufficient roots and leaves. In addition, Zn deficiency increased the expression of OsNAS3 in roots and

Discussion
Transgenic approaches, especially multiple gene expression, have already been proven effective to biofortify enhanced amounts of Fe and Zn in rice grains. Much research on the biofortification of Fe and Zn in rice grains has introduced one of the three OsNAS genes alone or in combination with other genes involved in metal homeostasis, but none of them have introduced two OsNAS genes at once. In this study, we generated a transgenic line (23D) with increased OsNAS2 and OsNAS3 expression to see if simultaneous activation of OsNAS2 and OsNAS3 would improve the accumulation of Fe and Zn in rice grains more than the single-activation-tagged mutants. Compared to the WT and single-activationtagged mutants, NA and DMA levels were considerably higher in 23D, and as a result, Fe and Zn levels were significantly raised in mature grains. In addition, 23D had much higher concentrations of Fe and Zn in its leaves and roots, suggesting that NA and DMA had also accumulated there. Hence, the mutants were more resistant to Fe and Zn deficiency and high-pH conditions. In particular, 23D plants outperformed all others by demonstrating more tolerance.
Combinatorial effects of OsNAS2 and OsNAS3 resulted in 23D's improved properties as compared to 2D or 3D. This might have been because OsNAS2 and OsNAS3 have differing physiological functions [15]. In other words, OsNAS2 expression was present in Fe-sufficient root, and under Fe deficiency, its levels significantly increased in roots and leaves. OsNAS3 expression was enhanced in roots but inhibited in leaves in response to Fe deprivation, although the OsNAS3 transcript was present in both Fe-sufficient roots and leaves. In addition, Zn deficiency increased the expression of OsNAS3 in roots and shoots while marginally decreasing the expression of OsNAS2 in roots [31]. OsNAS2 was found to be expressed in companion cells and pericycle cells close to the protoxylem in Fe-sufficient roots, according to promoter-GUS analysis. This suggests that OsNAS2 has a role in DMA secretion for Fe absorption in roots. However, because OsNAS3 is only expressed in the root pericycle and companion cells, it is possible that it is not involved in DMA secretion [15]. Moreover, OsNAS3 expression was seen in the entire leaf tissues of the control plants [15]. Therefore, OsNAS3 may work to produce NA or following parts such as DMA for the distribution of Fe and Zinc in rice plants.
NA is the precursor of DMA, which is synthesized by NAS, and NA is processed by NAAT and DMAS for synthesizing DMA [30,43]. We found that OsNAAT1 and OsDMAS1 were highly expressed in the mutants, with 23D showing the highest expression. This may account for the increased levels of NA and DMA in 23D. In rice, NA chelates Fe(II) and Zn(II) and carries them into the rice grains, and DMA chelates Fe(III) and transports it into root cells [2,13]. Through activating Fe(III)-DMA and Zn(II)-NA chelating processes and, likely, utilizing the YSL and Zn transporter genes, respectively, enhanced NA and DMA promote Fe and Zn accumulation in grains [26]. Moreover, mRNA levels of the Zn transporters (OsZIP4 and OsZIP5), as well as the YSL transporters (OsYSL2 and OsYSL15), were found to be much greater in mutants, which likely explains why transgenic plants and grains accumulate more Fe and Zn. The increased expression of the Fe transporter OsIRT1 and the Fe storage protein OsFer1 in 23D may also be a contributing factor to the elevated levels of Fe and Zn. The mature grains, roots, and leaves of transgenic OsIRT1 overexpressing plants were shown to have increased Fe and Zn concentrations [34].
However, the rice ferritins (OsFer) were not reported to be used for rice biofortification; rather, OsFer genes were found to be necessary for a positive response against Fe-induced oxidative stress [44]. Accordingly, 23D seedlings were more tolerant to Fe toxic conditions, probably due to the elevated expression of OsFer1 in 23D than in 2D, 3D, and WT. Similarly, OsNAS3 knockout plants were sensitive to excess Fe, whereas a transgenic rice line overexpressing barley HvNAS1 with elevated NA production was tolerant of excess Fe [45]. While 23D plants were the best performers under Zn, Cu, Ni, and Co toxicities, 2D and 3D plants also performed better under these conditions. This is likely because 23D plants have improved abilities to chelate more of these metal ions than other plants, which reduces their toxic effects on plant growth.
Because of enhanced accumulation of Fe and Zn in shoots and roots due to higher levels of NA and DMA, transgenic plants were more resistant to Fe-and Zn-deficient conditions as well as high-pH conditions with or without Fe and/or Zn. As a result, 23D plants would grow more successfully than single-activation mutants in alkaline growth conditions where high pH induces nutritional limitation. To maintain the solubility and availability of micronutrients to plants, soil pH is a crucial feature. Alkaline soils have reduced micronutrient availability, which has a detrimental effect on plant growth and yield [46]. For instance, Fe is abundant in soil but not easily soluble, especially in calcareous soils, which make up 30% of the earth's surface [47]. In addition, the uptake of other metal micronutrients, such as Zn and Mn, can also be inhibited in alkaline soils [48]. The 23D plants exhibited superior tolerance due to the combinatorial effects of OsNAS2 and OsNAS3, as evident from the significant expression enhancement of OsNAS2 and OsNAS3 in the shoots and roots of rice seedlings under normal and high pH environments with or without Fe or Zn. Elevated expressions of OsNAS genes under high pH with or without Fe or Zn also explain the better tolerance of the mutants in alkaline conditions. Although we measured Fe and Zn concentration only in the dehusked grains of the transgenic rice plants, we speculate that Fe and Zn amounts would be higher in the milled grains as well, with higher bioavailability. Intensified blue colors were observed for the whole transgenic grains and their embryos after Perls' staining, where the strongest blue color in the 23D grain was obvious. Furthermore, previous studies on the singleactivation-tagged mutants, 2D and 3D, showed that the increased Fe and Zn in the milled grains were more bioavailable, as more NA and DMA would bind to more Fe and Zn and transport them into the grain endosperm [17][18][19]. In line with previous reports, 23D seeds may accumulate more Fe and Zn bound to NA and/or DMA than 2D, 3D, or WT seeds, resulting in increased bioavailability. Mouse-feeding experiments can be used to test this in the future. An important point to consider while biofortifying rice is avoiding the negative impacts on normal plant growth and yield. Unfortunately, activation-tagged mutants displayed decreased growth and seed yield. Because they contained the higher amounts of Fe and Zn throughout all growth stages, they might have suffered from the oxidative stress, resulting to poor growth and reduction in yield. Although we did not observe any difference during the seedling stage, the influence increases over time and becomes more pronounced throughout the mature stage, making the negative effect more obvious. As a result, shorter plants with reduced yield were observed for 2D and 3D plants, probably for excess Fe-induced oxidative stress conditions in planta [19]. Similarly, 23D plants showed almost resembling agronomic traits as observed in 2D, but better than 3D. However, when growth under high-pH soil, 23D exhibited the greatest agronomic traits among WT and mutants, indicating the synergistic effect of OsNAS2 and OsNAS3.
Future studies on 23D plants would benefit from their raised biofortified Fe and Zn levels as well as their increased tolerance to alkalinity and metal toxicity. By combining OsNAS2 and OsNAS3 with additional genes, future studies may further raise the Fe and Zn contents in rice grains.

Plant Materials and Growth
The Korean rice Dongjin (Oryza sativa L. ssp. japonica) was used as the experimental cultivar in this study. Two activation-tagged mutants (OsNAS2-D1 and OsNAS3-D1) have been described previously [17,18]. Wild-type segregants from each progeny are referred to as wild-type (WT) progeny hereafter. Thirty-day-old seedlings of WT and activation-tagged mutant plants were transplanted into the paddy fields located at the Daegu Gyeongbuk Institute of Science and Technology (36 • N) in 2020 under natural conditions. For the growth evaluation in calcareous soil, four-leaf-stage seedlings of WT and activation-tagged mutants were raised in pots (diameter 17 cm, height 18 cm) with nursery soil containing soluble CaCO 3 to adjust to pH 8.0 under outdoor growth conditions until harvest. To prevent the soil from drying out, watering was performed frequently.

Hydroponic Culture for the Growth Test under Fe and Zn Deficiencies
The rice seeds were surface-sterilized, submerged in deionized water for 24 h in the dark, and then allowed to sprout for a further 24 h on wet filter paper. The uniform seedlings were selected and transferred into the containers containing Yoshida

Seedling Growth on MS Solid Medium under Fe and Zn Deficiencies
WT and transgenic rice seeds were germinated on a standard MS agar medium supplemented with 0.1 100 µM Fe (III)-EDTA and 30 µM ZnSO 4 as micronutrients [50]. Seedlings were grown for 10 days at 28 • C under continuous light. For our micronutrient deficiency tests, seeds were germinated and grown on an MS medium devoid of ZnSO4 or Fe (III)-EDTA for 10 days.

RNA Isolation and Quantitative Real-Time PCR
The total RNA was extracted from each tissue type with WelPrep total RNA isolation reagent (WELGENE, Gyeongsan, Republic of Korea), according to the manufacturer's instructions, and treated with RNase-free DNase I (Takara Bio, Shiga, Japan) to remove the genomic DNA. First-strand cDNA was synthesized from 2 µg of total RNA in a 25 µL reaction mixture using the ImProm II Reverse Transcriptase system kit (Promega, Madison, WI, USA), followed by quantitative real-time PCR (qRT-PCR) analysis to determine gene expression levels (Bio-Rad, CFX96 Touch Real-Time PCR Detection System, USA) using a SYBR premix ExTaq kit (Takara Bio, Shiga, Japan). The rice ubiquitin gene (LOC_Os06g46770) was used to normalize the expression ratio for each gene, and changes in expression were calculated via the ∆∆ Ct method. The gene-specific primers for PCR are listed in Table S1.

Measurement of Chlorophyll Concentrations
Chlorophyll was extracted from 0.1 g of leaf samples using 1 mL of 80% acetone [35]. After homogenization, the samples were incubated for 15 min on ice and spun for 10 min at 15,000× g, and then an aliquot of the supernatant fraction was collected for spectrophotometer measurements of A 663 and A 643 . Chlorophyll concentrations, including chlorophyll a and b, were determined according to the method of [51].

Element Analysis in Plant Tissues
Samples from each tissue were dried at 70 • C until the weight remained constant, and then samples were digested in 1 mL of 11 N HNO 3 for 3 days at 180 • C. After dilution, metal concentrations in the solutions were determined by inductively coupled plasma mass spectroscopy (ICP-MS, Varian 820-MS, Varian, Australia).

Perls' Staining of the Mature Seeds
After the husk was manually removed, seeds were immersed in ultrapure water for four to five hours before being cut in two lengthwise through the embryo in a Petri dish using a razor blade. The cut seeds were submerged in freshly prepared Perls' Prussian blue [52] solution (2% (v/v) HCl and 2% (w/v) potassium ferrocyanide), vacuum infiltrated for 15 min (500 mbar), and incubated at room temperature for 30 min. The seeds were then rinsed three times in distilled water and mounted in ultrapure water for visualization on a stereomicroscope.

Determination of NA and DMA Levels
Mature seeds were crushed into a fine powder with a Multi-Beads Shocker (Yasui Kikai, Osaka, Japan), and then 400 µL of 80% ethanol was added to the samples. The samples were centrifuged at 12,000× g for 5 min and supernatants were collected. Two more centrifugations were performed using the supernatants collected each time. Concentrations of NA and DMA were analyzed using the final supernatants by LC/ESI-TOF-MS, as previously described [53].

Excess Metal Tolerance
For the excess-metal treatments, seeds were germinated and grown for 10 days on solid media containing half-strength MS salts and supplemented with 5 mM ZnCl 2 , 0.3 mM CuCl 2 , 1 mM FeCl 3 , 0.5 mM NiCl 2 , or 0.5 mM CoCl 2 .