Enhanced NRT1.1/NPF6.3 expression in shoots improves growth under nitrogen deficiency stress in Arabidopsis

Identification of genes and their alleles capable of improving plant growth under low nitrogen (N) conditions is key for developing sustainable agriculture. Here, we show that a genome-wide association study using Arabidopsis thaliana accessions suggested an association between different magnitudes of N deficiency responses and diversity in NRT1.1/NPF6.3 that encodes a dual-affinity nitrate transporter involved in nitrate uptake by roots. Various analyses using accessions exhibiting reduced N deficiency responses revealed that enhanced NRT1.1 expression in shoots rather than in roots is responsible for better growth of Arabidopsis seedlings under N deficient conditions. Furthermore, polymorphisms that increased NRT1.1 promoter activity were identified in the NRT1.1 promoter sequences of the accessions analyzed. Hence, our data indicated that polymorphism-dependent activation of the NRT1.1 promoter in shoots could serve as a tool in molecular breeding programs for improving plant growth in low N environments. Here, the authors conduct a genome-wide association study from Arabidopsis thaliana accessions and demonstrate that the polymorphisms in the promoter of NRT1.1/NPF6.3, a dual-affinity nitrate transporter, are associated with different magnitudes of N deficiency in Arabidopsis shoot.

N itrogen (N) is an important macronutrient for plants and a major constituent of molecules essential for plant growth, such as amino acids, nucleic acids, and chlorophyll 1 . Thus, the availability of N is critical for many aspects of plant growth and development 2 . Because of the high N requirement of plants, huge amounts of N fertilizers are applied to the field to increase crop yield in modern agriculture. However, more than half of applied N is lost through leaching and volatilization, causing environmental pollution [3][4][5] . Therefore, identification of genes and their alleles capable of improving plant growth under low N conditions is critical for improving the current modern agriculture system and developing sustainable agriculture 3,6-9 . Although researchers have identified a few genes whose overexpression and/or ectopic expression enhances N uptake and assimilation [10][11][12][13][14][15][16][17][18] , the information available on genes that could potentially improve growth under N limiting conditions remains insufficient. Plants very frequently encounter N deficiency in the natural ecosystem and display N deficiency responses to overcome the stress [19][20][21][22] . N deficiency responses include modification of root architecture to promote the uptake of N sources 23 , degradation of N-containing macromolecules, such as protein and chlorophyll, to translocate N sources from older leaves to younger leaves to sustain growth 24,25 , and modulation of gene expression 26,27 . Plant factors controlling and/or affecting these N deficiency responses are strong candidates for tools in lessening the growth defects under N deficient conditions.
Nitrate is generally the most abundant form of N available in aerobic soils. Nitrate serves as a signaling molecule that modulates the expression of genes involved in N acquisition and assimilation, as well as in other aspects of primary metabolism, phytohormone biosynthesis and signaling, and cell cycle regulation [28][29][30][31][32][33] . Thus, nitrate uptake, the first step of N acquisition mediated by membrane transporters in roots, directly and indirectly affects plant growth and development. Higher plants possess two types of nitrate uptake systems with distinct kinetic properties, which enable the uptake of nitrate over a wide concentration range. The low-affinity transport system (LATS), comprising Nitrate Transporter1 (NRT1)/Peptide Transporter (PTR) family proteins, drives nitrate uptake at millimolar concentrations, whereas the high-affinity transport system (HATS), which consists of members of the NRT2 family, drives nitrate uptake at micromolar concentrations 28,34 . The NRT1/PTR and NRT2 families constitute a number of NRT proteins. In Arabidopsis thaliana, 53 and 7 genes encode NRT1/PTR and NRT2 proteins, respectively 35 . NRT1.1, which is also termed as NPF6. 3 36 , is a unique protein and its functions have been extensively studied [37][38][39] . Since NRT1.1 was initially identified as the product of a gene associated with chlorate sensitivity, it is also called CHLORINA1 (CHL1) 40 . Unlike other NRT1 proteins, NRT1.1 is a dual-affinity transporter that can facilitate nitrate uptake at concentrations ranging from micromolar to millimolar 37 , indicating that NRT1.1 can contribute to nitrate uptake in diverse natural ecosystems. NRT1.1 was also proposed to act as a nitrate sensor involved in the expression of other NRT genes including NRT2.1 41 . In addition to their role in nitrate uptake by roots, some of the NRT1 family proteins also facilitate nitrate loading into and unloading from xylem vessels for distribution to aerial organs and remobilization from old leaves 42 . Furthermore, some NRT1 family proteins transport not only nitrate but also a variety of structurally diverse compounds, including nitrite 43 , peptides 44 , amino acids 45 , and phytohormones including auxin 19 , abscisic acid 46 , and gibberellin 47 , implying their versatility.
To identify novel genes associated with N deficiency responses and alleles capable of improving growth under N deficient conditions, we conducted a genome-wide association study (GWAS) on 52 Arabidopsis accessions exhibiting natural variation in N deficiency-induced reduction of chlorophyll content. The results implied an association between N deficiency responses and NRT1.1 polymorphisms. This association was further examined using two Arabidopsis accessions, Gr-5 and Ullapool-8, both of which displayed high NRT1.1 expression and reduced N deficiency responses. Several lines of evidence indicate that polymorphism-dependent activation of the NRT1.1 promoter in aerial plant parts confers better growth under N deficient conditions in Arabidopsis.

Results
Arabidopsis accessions show variation in leaf yellowing phenotype under N deficiency. To evaluate differences in N deficiency responses among Arabidopsis accessions, the total chlorophyll content of all 52 accessions was quantified at the seedling stage (Supplementary Data 1) because yellowing of premature leaves, which is caused by the degradation of chlorophyll, is a typical N deficiency symptom. Seedlings were initially grown on agar plates containing half-strength Murashige and Skoog (1/2MS) medium and then on modified 1/2 MS agar plates that contained 0.1 mM KNO 3 and 0.1 mM NH 4 NO 3 (0.3 mM N) as N sources (low N treatment), or on modified 1/2 MS agar plates that contained 2 mM KNO 3 and 2 mM NH 4 NO 3 (6 mM N) as N sources (sufficient N treatment; control). The chlorophyll contents of all 52 accessions were almost comparable under the control N condition but were different under the low N condition (Fig. 1a, Supplementary Data 1), indicating natural variation in N deficiency responses. Thus, we used the ratio of the chlorophyll content under the low N condition (Chl low N ) to that under the control treatment (Chl control ) for assessing the N deficiency response of each accession (Fig. 1b, Supplementary Data 1). The Chl low N /Chl control ratio of Columbia (Col-0), which was used as a reference accession, was lower than that of most accessions but close to the average ratio.
Because GWAS is a powerful tool for finding loci associated with traits of interest 48 , we performed GWAS using the Chl low N / Chl control ratio to identify allelic genes that could explain the natural variation in the leaf yellowing phenotype of Arabidopsis accessions under the low N condition. Despite using only a small number of accessions, several peaks potentially associated with the observed differences in N deficiency responses were identified, although the height of each peak was not high (Fig. 1c). These peaks consisted of single nucleotide polymorphisms (SNPs) at the NRT1.1, AGAMOUS-LIKE 65 (AGL65, AT1G18750), ATP-BINDING CASSETTE G1 (ABCG1, AT2G39350), or INOSITOL 1,3,4-TRISPHOSPHATE 5/6-KINASE 3 (ITPK3, AT4G08170) locus, implicating a possible association of these genes with the natural variation in N deficiency responses. Among these loci, we focused on the NRT1.1 locus in this study.
Correlation between shoot NRT1.1 expression level and Chl low N /Chl control ratio. To investigate whether polymorphisms in NRT1.1 are associated with different magnitudes of N deficiency responses, we examined the nucleotide sequences (from the transcription start site to the stop codon) of NRT1.1 genes from 39 accessions, which were deposited in the Salk Arabidopsis 1,001 Genomes database (http://signal.salk.edu/atg1001/3.0/gebrowser. php). However, in comparison with the Col-0-type NRT1.1 sequence, no polymorphism leading to amino acid substitution was detected in the nucleotide sequences from 38 accessions ( Supplementary Fig. 1). On the other hand, only two or three SNPs were identified in the 1st, 2nd, and 3rd introns of NRT1.1, while a number of polymorphisms were identified in the 4th intron ( Supplementary Fig. 1). Because of this result and the results of previous studies showing that overexpression of NRT1.1  Table 1). Red bar indicates the Chl low N /Chl control ratio of Col-0, which was used as a reference accession in this study. The blue dotted line indicates the average Chl low N /Chl control ratio of all 52 accessions. Values of chlorophyll contents are shown in Appendix Table S1. c Genome-wide association study (GWAS) of Arabidopsis accessions using Chl low N /Chl control values. In Manhattan plots, horizontal green lines indicate the significant genome-wide threshold (p-value = 5 × 10 −6 ), while gray lines correspond to -log 10 (p-values) of 2, 4 and 6. Peaks consisting of single nucleotide polymorphisms (SNPs) located in the vicinity of the NRT1.1, AGL65, ABCG1 or ITPK3 locus are marked with a red rectangle. SNPs with a p-value smaller than 5 × 10 −6 are indicated by red dots. increases NRT1.1 activity and consequently promotes nitrate uptake in Arabidopsis 37,49 , we hypothesized that differences in the NRT1.1 expression level among the Arabidopsis accessions might be responsible for different magnitudes of N deficiency responses. To examine this possibility, we analyzed the expression levels of NRT1.1 in the shoots and roots of seedlings of all 52 accessions since NRT1.1 is known to express in both shoots and roots, although its expression is much stronger in roots 50 . The results showed a wide variation in the expression level of NRT1.1 in both shoots and roots (Fig. 2a, b). The coefficient of determination (R 2 ) between the Chl low N /Chl control ratio and NRT1.1 expression level was 0.3098 in shoots and 0.2378 in roots (Fig. 2c, d). The correlation coefficient was higher in shoots rather than in roots, implying that the higher expression level of NRT1.1 in shoots might affect leaf yellowing under the low N condition to a greater extent. Among the analyzed accessions, Gr-5 and Ullapool-8 showed the highest expression levels of NRT1.1, approximately 4.5-and 3.5-fold higher in shoots, and 1.9-and 2.8-fold higher in roots, respectively, compared with Col-0 ( Fig. 2b).
Growth of Gr-5 and Ullapool-8 seedlings under low N conditions. To scrutinize the possible association of NRT1.1 expression level with N deficiency responses, we compared the growth phenotypes of Col-0, Gr-5, and Ullapool-8 seedlings grown for 5 days on 1/2 MS agar plates and then for 7 days under three different N conditions (6 mM, 0.3 mM, and 0.03 mM N). Besides these accessions, we also employed transgenic Arabidopsis overexpressing NRT1.1 (NRT1.1-OX) lines. These transgenic lines were generated by the expression of NRT1.1 cDNA under the control of the cauliflower mosaic virus (CaMV) 35S promoter in the Col-0 background; thus the expression of NRT1.1 was approximately 10-fold higher in the NRT1.1-OX plants than that in wild-type Col-0 plants ( Supplementary Fig. 2). After 5 days of growth under the 0.03 mM N condition, leaves of Col-0 seedlings turned yellow, whereas those of Gr-5, Ullapool-8, and NRT1.1-OX seedlings remained green (Fig. 3a). Significant differences were also detected in the maximum quantum yield of photosystem II (Fv/Fm), fresh shoot weight, and chlorophyll content between Col-0 and the remaining three genotypes only under the 0.03 mM N condition, although the fresh shoot weight of Gr-5 and NRT1.1-OX seedlings was slightly greater than that of Col-0 seedlings even under the 6 mM N condition ( Fig. 3a-d).
N deficiency-induced expression of NRT1.1 in shoots. The N deficiency responses of Gr-5, Ullapool-8, and NRT1.1-OX seedlings showed great resemblance; however, the expression level of NRT1.1 in the NRT1.1-OX seedlings was much higher than those in Gr-5 and Ullapool-8 seedlings (compare Fig. 2 with Supplementary Fig. 2). Furthermore, it has been previously shown that the expression level of NRT1.1 in roots decreases under N deficient conditions 50,52 . Therefore, the results shown in Fig. 3 were not in complete agreement with our hypothesis that high expression levels of NRT1.1 reduce N deficiency-induced leaf yellowing and growth suppression. To resolve these conflicts, we performed a time-course expression analysis of NRT1.1 in the shoots and roots of 7-day-old Col-0 seedlings exposed to N deficiency stress. Consistent with the previous results, the expression level of NRT1.1 gradually decreased in roots; however it increased in shoots in response to N deficiency (Fig. 4a). Similar results were observed with 21-day-old Col-0 plants (Fig. 4b), indicating that the effect of N deficiency on NRT1.1 expression differs between shoots and roots.
Genetic confirmation of the association between the NRT1.1 locus and the modified N deficiency response. To obtain genetic evidence supporting the association between the NRT1.1 locus and the growth phenotype of Gr-5 and Ullapool-8 seedlings under low N conditions, segregation analysis was performed on F2 progenies of Col-0 × Gr-5 and Col-0 × Ullapool-8 crosses. Segregation ratios were calculated based on the number of Col-0like F2 seedlings vs. the number of Gr-5-like or Ullapool-8-like F2 seedlings when grown on 1/2 MS agar plates for 5 days, and then under the low N condition for 7 days (Table 1). Segregation ratios of Col-0-like to Gr-5-like progeny were very close to 1:3 in three independent experiments. However, segregation ratios of Col-0-like to Ullapool-8-like progeny were approximately 1:2.2 in each of the three independent experiments. These results suggest that a single locus is responsible for the modified N deficiency response in Gr-5, while more than two loci might be involved in the modified N deficiency response in Ullapool-8.
To confirm that the Gr-5-like phenotype of the Col-0 × Gr-5 F 2 progeny is associated with the Gr-5-type NRT1.1 locus, PCRbased genotyping was performed using an insertion-deletion (InDel) marker designed on the basis of a 7 bp deletion in the Gr-5-type NRT1.1 promoter (Supplementary Fig. 4A). The results showed that all F2 seedlings displaying the Gr-5-like phenotype were homozygous or heterozygous for the Gr-5-type NRT1.1 locus, whereas almost all of the analyzed Col-0-like F2 seedlings were homozygous for the Col-0-type NRT1.1 locus. A few of seedlings that were regarded as the Col-0-like F2 seedlings were heterozygous, probably because the heterozygous phenotype was obscured, compared with the homozygous phenotypes (Supplementary Fig. 4B, Supplementary Table 1).
Stronger activity of the Gr-5-type NRT1.1 promoter is responsible for the improved growth of Gr-5 under N deficient conditions. We expressed Col-0-type NRT1.1 cDNA under the control of the Col-0-, GR-5-, or Ullapool-8-type NRT1.1 promoter in the nrt1.1 null mutant, chl1-5, to verify that the enhancement of NRT1.1 expression improved the growth of GR-5 and Ullapool-8 under low N conditions. Nine independent lines were generated for each of three constructs (proNRT1.1 Col-0 : NRT1.1 Col-0 , proNRT1.1 Gr-5 :NRT1.1 Col-0 , or proNRT1.1 Ullapool-8 : NRT1.1 Col-0 construct). Among them, the expression level of NRT1.1 in shoots varied because of the positional effect; however, the Gr-5-type promoter was apparently stronger than the Col-0type promoter in planta. The Ullapool-8-type promoter was slightly stronger than the Col-0-type promoter (Fig. 6a, Supplementary Fig. 6a), although its activity did not appear to adequately explain the high expression level of NRT1.1 in Ullapool-8 plants (compare Fig. 6b with Fig. 2a), suggesting that elements outside the promoter fragment may also be responsible for the high expression level of NRT1.1 in Ullapool-8.
Subsequent phenotypic analysis of transgenic plants revealed that early leaf yellowing phenotype of the chl1-5 mutant under the low N condition could be rescued by the expression of Col-0-type NRT1.1 cDNA under any type of NRT1.1 promoter. However, quantitative analysis of the chlorophyll content and Fv/Fm ratio of plants indicated that Col-0-, GR-5-, and Ullapool-8-type promoter-driven NRT1.1 expression modified N deficiency responses to different extents. The expression of NRT1.1 was the most potent under the control of the Gr-5-type promoter, and Ullapool-8-type-driven and Col-0-type-driven NRT1.1 expression modified the N deficiency responses of chl1-5 to similar extents ( Fig. 6b-d, Supplementary Fig. 6b, c). Correlation analysis of transgenic lines revealed that NRT1.1 expression level in shoots showed a significant correlation with chlorophyll contents and Fv/Fm ratio under the low N conditions (Fig. 6e, f). Furthermore, fresh shoot weight of the seedlings of two independent proNRT1.1 Gr-5 :NRT1.1/chl1-5 lines (L1 and L2), which exhibited high NRT1.1 expression ( Supplementary Fig. 6a), was significantly higher than that of proNRT1.1 Col-0 :NRT1.1/chl1-5 plants under the low N condition in 10-day-old seedlings (Supplementary Fig. 7) and 24-day-old plants ( Supplementary Fig. 8). However, in another line of proNRT1.1 Gr-5 :NRT1.1/chl1-5 plants (L3), which exceptionally showed a lower level of NRT1.1 expression probably due to a positional effect ( Supplementary  Fig. 6a), chlorophyll contents, the Fv/Fm ratio under the low N conditions, and fresh shoot weight were comparable to those of proNRT1.1 Col-0 :NRT1.1/chl1-5 plants (Supplementary Figs. 6-8). Collectively, these results support our hypothesis that stronger activity of the NRT1.1 promoter in shoots leads to better growth under low N conditions.
Polymorphisms associated with strong activity of the Gr-5-and Ullapool-8-type NRT1.1 promoters. To identify polymorphisms that contribute to the strong activity of the Gr-5-and Ullapool-8type NRT1.1 promoters, we first determined the nucleotide sequences of Gr-5-and Ullapool-8-type promoters, which corresponded to the region from -2000 to -1 bp (relative to the translational start site) of the Col-0-type promoter, because no information on the nucleotide sequences of Gr-5-and Ullapool-8-type NRT1.1 promoters was available in any database. Since comparison of their nucleotide sequences revealed several polymorphisms in the NRT1.1 promoter (Fig. 7a, Supplementary  Fig. 9), activity of Col-0-, Gr-5-, and Ullapool-8-type NRT1.1 promoters was compared by transient expression assays using protoplasts prepared from leaves of Arabidopsis plants grown under N sufficient or N deficient conditions. The activity of all of NRT1.1 promoter types was stronger in protoplasts prepared from plants grown in N deficient conditions (Fig. 7b), consistent with the N deficiency-inducible expression of NRT1.1 in shoots (Fig. 4a). Furthermore, the Gr-5-type NRT1.1 promoter was stronger than the Col-0-type NRT1.1 promoter by approximately 2.6-and 4-fold in protoplasts prepared from plants grown in N sufficient and N deficient conditions, respectively. Similarly, the Ullapool-8-type promoter was slightly stronger than the Col-0type NRT1.1 promoter under both N sufficient and deficient conditions (Fig. 7b).    Table 2). To test whether these polymorphisms affect the activity of the NRT1.1 promoter, we introduced these polymorphisms into the Col-0-type NRT1.1 promoter. The Col-0-type NRT1.1 promoter harboring these polymorphisms showed significantly higher activity than the wild-type Col-0-type NRT1.1 promoter, especially in protoplasts from Arabidopsis plants grown under the N deficient conditions (Fig. 7b). Thus, these results suggest that polymorphisms detected in the NRT1.1 promoter increase its activity, particularly under low N conditions.

Discussion
GWAS of Arabidopsis accessions and further analyses of selected accessions presenting reduction in chlorophyll degradation under low N conditions revealed that polymorphism-dependent enhancement of NRT1.1 expression in shoots contributes to improved plant growth under low N conditions. These findings suggest that the function of NRT1.1 in aerial plant parts is closely associated with plant growth under N deficiency, although its importance has been overlooked so far. This finding also suggests the importance of SNPs in the NRT1.1 promoter as a tool in molecular breeding to improve plant growth under low N conditions. Enhancement of NRT1.1 expression in shoots improves plant growth in N deficient environments. We demonstrated that polymorphism-dependent enhancement of NRT1.1 expression in shoots modulates a variety of N deficiency responses in Arabidopsis. NRT1.1 is a dual-affinity nitrate transporter whose function in roots has been extensively studied in last few decades 53 . However, functions of NRT1.1 in the shoot are still largely unknown. Although NRT1.1 in guard cells promotes the opening of stomata through its nitrate influx ability 54 , the biological significance of NRT1.1 expression in other types of cells in the lamina and petiole remains unknown. NRT1.4 mediates the translocation of nitrate from the leaf petiole to other parts of the leaf 55 , whereas NRT1.7 in the phloem of minor veins in older leaves is involved in the translocation of nitrate from older to younger leaves 25 . Considering the functions of these NRT1 proteins, it is possible to speculate that NRT1.1 plays similar roles in nitrate translocation for maintaining plant growth under low N conditions. Alternatively, since NRT1.1 has been proposed to function as a nitrate sensor 41,56 , its sensor activity might be associated with differences in growth rate under N deficiency. Further analysis of the functions of NRT1.1 in shoots would help to reveal why the enhancement of NRT1.1 expression in shoots improves plant growth under N deficient conditions. A rice (Oryza sativa L.) homolog of Arabidopsis NRT1.1, OsNRT1.1B, which is considered as both a nitrate transporter and a nitrate sensor in rice 57,58 , is divergent between indica and japonica cultivars, and indica-type OsNRT1.1B rather than japonica-type OsNRT1.1B is more closely associated with the enhancement of nitrate uptake and the upregulation of nitrateresponsive genes. Therefore, the overexpression of indica-type OsNRT1.1B rather than japonica-type more effectively improved the growth and yield in japonica variety Zhonghua11 59 . Furthermore, a recent study showed that 35S promoter drivenexpression of OsNRT1.1A, another homolog of Arabidopsis NRT1.1 in rice, improved the grain yield of transgenic rice plants and increased the seed yield and biomass of transgenic Arabidopsis plants 49 . These results suggest the intimate relationship between the function of NRT1.1 and plant growth and further emphasize the importance of revealing how the function of NRT1.1 in shoots is associated with the modulation of plant growth under low N conditions. Fig. 6 Differential effects of Col-0-, Gr-5-, and Ullapool-8-type NRT1.1 promoter-driven NRT1.1 expression on N deficiency responses in chl1-5 mutant seedlings. a Levels of NRT1.1 transcripts in the shoots of 1/2 MS-grown 10-day-old chl1-5 seedlings expressing Col-0-type NRT1.1 under the control of Col-0type NRT1.1 promoter (proNRT1.1 Col-0 :NRT1.1/chl1-5), Gr-5-type NRT1.1 promoter (proNRT1.1 Gr-5 :NRT1.1/chl1-5), or Ullapool-8-type NRT1.1 promoter (proNRT1.1 Ullapool-8 :NRT1.1/chl1-5). Transcript levels of NRT1.1 were normalized relative to those of ACT2. b Images of seedlings and Fv/Fm ratios of Col-0, chl1-5, Gr-5, Ullapool-8, proNRT1.1 Col-0 :NRT1.1/chl1-5 (line 1), proNRT1.1 Gr-5 :NRT1.1/chl1-5 (line 1), and proNRT1.1 Ullapool-8 :NRT1.1/chl1-5 (line 2) seedlings grown with continuous light for 5 days on 1/2 MS agar plates and then for 5 days under the control N (6 mM) or low N (0.03 mM) condition. Scale bar = 1 cm. c, d Total chlorophyll contents (c) and Fv/Fm ratios (d) of the shoots of Col-0, chl1-5, Gr-5, Ullapool-8, proNRT1.1 Col-0 :NRT1.1/chl1-5, proNRT1.1 Gr-5 :NRT1.1/chl1-5, and proNRT1.1 Ullapool-8 :NRT1.1/chl1-5 seedlings grown with continuous light for 5 days on 1/2 MS agar plates and then for 5 days under the control N (6 mM) or low N (0.03 mM) N condition. In a, c, and d, nine independent transgenic lines, with four biological replicates per line, were analyzed for each construct. The horizontal line in the middle of each box represents the median, and upper and lower ends of each box indicate the upper and lower quantiles, respectively. Values for Col-0, Gr-5, and Ullapool-8 are indicated with dotted lines. Data represent mean ± SD of four biological replicates. Asterisks indicate significant differences between proNRT1.1 Col-0 :NRT1.1/chl1-5 plants and other genotypes (*p < 0.05, **p < 0.01; Student's t-test). e, f Correlation analysis of NRT1.1 expression with chlorophyll content (e) or Fv/Fm (f). Green, blue, and red circles represent proNRT1.1 Col-0 :NRT1.1/chl1-5, proNRT1.1 Ullapool-8 :NRT1.1/chl1-5, and proNRT1.1 Gr-5 :NRT1.1/chl1-5 lines, respectively. Black filled circles represent the chl1-5 mutant. b Transient expression assay using Arabidopsis protoplasts. Reporter plasmid containing the LUC reporter gene under the control of the Col-0-, Gr-5-, or Ullapool-8-type NRT1.1 promoter was transfected into protoplasts isolated from rosette leaves of Col-0 plants exposed to N deficient conditions (-N) or N sufficient conditions (+N). P1-P4 promoters represent Col-0-type NRT1.1 promoters harboring one of the four polymorphisms shown in a. LUC activity was normalized relative to GUS activity derived from an internal control plasmid (pUBQ10-GUS). Relative LUC activity obtained using the Col-0-type NRT1.1 promoter in protoplasts prepared from Col-0 plants exposed to N sufficient conditions (+N) was set to 1. Data represent mean ± SD of four biological replicates. Different lowercase letters indicate significant differences (p < 0.05; Tukey's multiple comparison test). N deficiency-induced expression of NRT1.1 expression in the shoot. Time-course analysis revealed opposite regulation of NRT1.1 expression in shoots and roots under N deficient conditions (Fig. 4), indicating that mechanisms underlying NRT1.1 expression in shoots are different from those in roots. This finding is consistent with the results of previous studies showing that NRT1.1 expression in roots is immediately induced by nitrate supply 50 but suppressed by N deficiency 52 . This finding is also consistent with the microarray data showing that the expression level of NRT1.1 in shoots is elevated under N deficient conditions 60 .
The mechanism underlying NRT1.1 expression in roots has been studied previously, and a few transcription factors involved in NRT1.1 expression have been identified. NLP7 binds to and activates the NRT1.1 promoter in roots 61 , and the NITRATE REGULATORY GENE2 (NRG2) transcription factor enhances NRT1.1 expression probably through a direct interaction with NLP7 62 . However, although NRT1.1 is also expressed in aboveground tissues, such as emerging and immature rosette leaves and flower buds 63 , the mechanism underling NRT1.1 expression in shoots remains unknown. In the current study, polymorphisms detected in the NRT1.1 promoter sequence in Gr-5 (7 bp deletion; −1319 to −1313 bp) and Ullapool-8 (A→C SNP; −1840 bp) were found to be associated with the activation of the NRT1.1 promoter under N starvation conditions. Since theses polymorphisms are located outside the two regions that directly interact with NLP7 (−1537 to −1334 bp and −686 to −465 bp) 61 , these polymorphisms provide clues for the identification of new transcription factors and a mechanism controlling N deficiency-induced NRT1.1 expression in shoots. Because the polymorphisms made the expression level of NRT1.1 higher, these polymorphisms may disrupt the binding sites of a transcriptional repressor. Alternatively, these polymorphisms may alter the chromatin structure of the NRT1.1 locus and promote an interaction with a transcriptional activator, thus up-regulating NRT1.1 expression in shoots. Further research is needed to test these possibilities.
We examined the association between expression levels of NRT1.1 and haplogroups, which were identified using 4th intron sequences of NRT1.1 that contained a number of polymorphisms ( Supplementary Fig. 10a). We note that 4th intron sequences were used, because NRT1.1 promoter sequences of accessions used in this study were barely available in the database and only two or three SNPs were found in the 1st, 2nd, and 3rd introns of NRT1.1 ( Supplementary Fig. 1). Since we did not find any significant difference in the expression level of NRT1.1 among different haplogroups ( Supplementary Fig. 10b), it is possible that spontaneous variation specific to each accession contributes to differences in the level of NRT1.1 transcripts and growth under low N conditions in each accession. However, this possibility would need to be carefully evaluated after grouping of haplotypes using the entire sequences for the NRT1.1 locus, including the promoter sequences.
Polymorphism-based plant biotechnology. Current plant biotechnology is generally based on the generation of transgenic plants harboring an artificially introduced transgene, limiting the utilization of transgenic plants in the agriculture sector. In the current study, we identified naturally existing polymorphisms in the NRT1.1 promoter that improve the N use efficiency of plants under low N conditions. Furthermore, although GWAS was performed with a small number of accessions in the current study, additional peaks were also detected (Fig. 1c). Characterization of the genes associated with these peaks and other genes that would be found in more intensive GWAS may identify other polymorphisms that improve the N use efficiency under low N conditions. With the advent of the CRISPR/Cas9 technology, which allows the modification of genomes without leaving behind any trace of foreign DNA 64,65 , genetic engineering for crop improvement may be possible by utilizing these polymorphisms.

Methods
Plant materials and growth conditions. Seeds of all Arabidopsis thaliana accessions, except Columbia (Col-0) (Supplementary Data 1), and the nrt1.1 null mutant, chl1-5 40 , were obtained from the Arabidopsis Biological Resource Center (ABRC), Columbus, OH, USA. Col-0 seeds were maintained as a stock in our laboratory. The Col-0 × Gr-5 and Col-0 × Ullapool-8 crosses were generated in the current study.
Arabidopsis seeds were sterilized, cold-stratified at 4°C for 4 days, and sown on 1/2 MS agar plates (1/2 MS salts, 0.8% agar, 0.5% sucrose, and 3 mM MES-KOH [pH 5.8]). To induce germination, the plates were transferred to a growth chamber maintained at 23°C with continuous light (cool white fluorescent bulbs; 70 μmol m −2 s −1 light intensity). Nutrient conditions for seedling growth are indicated in each subsection of Materials and Methods (below) and in the figure legends.
Phenotypic analysis. The leaf yellowing phenotype under low N conditions was investigated using 52 Arabidopsis accessions that displayed different nutrient responses in our previous study 66 . Seedlings were grown for 5 days on 1/2 MS agar plates and then for 5 or 7 days on modified 1/2 MS agar plates (N-free 1/2 MS salts, 0.8% agar, 0. Quantification of total chlorophyll contents. Arabidopsis seedlings were homogenized with zirconia beads using 80% ice-cold acetone, and chlorophyll pigments were extracted as described previously 67 . The absorbance of extracts was measured at 647 and 664 nm, and chlorophyll contents were calculated as described previously 68 . RNA extraction and RT-qPCR analysis. Total RNA was isolated from the shoots and roots of Arabidopsis seedlings using the ISOSPIN Plant RNA Kit (NIPPON GENE Co., Ltd., Tokyo, Japan), according to the manufacturer's instructions. Firststrand cDNA was synthesized using total RNA, SuperScript™ II reverse transcriptase, and oligo(dT) 15 primer (Invitrogen, Carlsbad, CA, USA), and qPCR was conducted on the StepOnePlus™ instrument (Applied Biosystems, Foster City, CA, USA) using the KAPA SYBR FAST qPCR Kit (KAPA Biosystems Inc., Wilmington, MA, USA) and gene-specific primers (Supplementary Table 3). Transcript levels of each gene were normalized relative to those of ACTIN2 (ACT2).
Measurement of Fv/Fm. Chlorophyll fluorescence and Fv/Fm of Arabidopsis seedlings were measured using a kinetics multispectral fluorescence imaging system (FluorCam 800 MF; Photon System Instruments, Brno, Czech Republic), according to the manufacturer's instructions.
Plasmid construction and plant transformation. All binary vectors for plant transformation were constructed using PCR-amplified DNA fragments and the Gateway ® cloning system. To generate transgenic plants overexpressing NRT1.1, the Col-0-type NRT1.1 cDNA was inserted between the CaMV 35S promoter (at the 5' end) and a sequence encoding four copies of the MYC epitope tag (4 × MYC; at the 3' end) in the pGWB17 Gateway binary vector. To generate transgenic plants expressing the Col-0-type of NRT1.1 cDNA under the control of the Col-0-, Gr-5-, or Ullapool-8-type NRT1.1 promoter, the Col-0-type NRT1.1 promoter (−2000 to −1 bp, relative to the translational start site) and the corresponding sequence in Gr-5 and Ullapool-8 were amplified from the genomic DNA of Col-0, Gr-5, and Ullapool-8, respectively. Then, the amplified NRT1.1 promoters, together with Col-0-type NRT1.1 cDNA, were cloned upstream of the 4 × MYC epitope tag in the pGWB16 Gateway binary vector. All constructs were verified by sequencing. Primers used for cloning are listed in Supplementary Table 3.
Each binary vector was introduced into Agrobacterium tumefaciens strain GV3101 (pMP90). Then, Agrobacterium-mediated transformation of Col-0 and chl1-5 plants was conducted using the floral-dip method 69 . T2 progenies with T-DNA insertion(s) at a single locus were selected, and homozygous T3 plants were used for all subsequent analyses.
Micrografting. Hypocotyls of 5-day-old seedlings grown on 1/2 MS agar plates were cut using a sterile sapphire knife to prepare scions and rootstocks. Then, scions and rootstocks were placed on a 1/2 MS agar plate and aligned using an optical microscope. After 4 days, the grafted seedlings forming connections between the scion and rootstock were transferred onto modified 1/2 MS agar plates (N-free 1/2 MS salts, 0.8% agar, 0.5% sucrose, and 3 mM MES-KOH [pH 5.8]) supplemented with KNO 3 and NH 4 NO 3 (2 or 0.01 mM each), and then grown for an additional 7 days under continuous white light (70 μmol m −2 s −1 ) at 23°C.
Measurement of nitrate uptake. Grafted seedlings generated using 5-day-old Col-0 and Gr-5 seedlings were grown for 7 days on modified 1/2 MS agar plates (N-free 1/2 MS salts, 0.8% agar, 0.5% sucrose, 3 mM MES-KOH [pH 5.8]) supplemented with 2 mM KNO 3 and 2 mM NH 4 NO 3 . Nitrate uptake was measured using the 15 N stable isotope, as described previously 31 . Briefly, roots were submerged in 0.1 mM CaSO 4 for 1 min and then incubated in 0.2 mM K 15 NO 3 for 5 min. After two 1 min washes with 0.1 mM CaSO 4 , the roots were collected and dried. The N content and 15 N/ 14 N isotopic composition of roots were analyzed by SI Science Co., Ltd. (Tokyo, Japan).
Transient expression assay using Arabidopsis leaf protoplasts. To construct reporter plasmids that contained the LUC reporter gene encoding luciferase downstream of the Col-0-, Gr-5, or Ullapool-8-type NRT1.1 promoter, the Col-0-, Gr-5-, and Ullapool-8-type NRT1.1 promoters were cloned into the pJD301 vector 70 . Mesophyll protoplasts were isolated from Col-0 plants grown hydroponically for 2 weeks in N-free 1/10 strength MS medium supplemented with 2 mM KNO 3 and 2 mM NH 4 NO 3 , and then for 5 days in fresh N-free 1/10 MS medium supplemented with or without 2 mM KNO 3 and 2 mM NH 4 NO 3 , as described previously 71 . Polyethylene glycol (PEG)-mediated transfection was performed using 2 μg reporter plasmid, 1 μg internal control plasmid (pUBQ10-GUS), and 1 × 10 5 protoplasts, as described preciously 72,73 . Transfected protoplasts were incubated in protoplast culture solution (0.4 M mannitol, 15 mM MgCl 2 , and 4 mM MES-KOH [pH 5.8]) in the dark at room temperature for 16 h. LUC activity was determined using the Luciferase Assay System Kit (Promega, Madison, WI, USA) and normalized relative to the ßglucuronidase (GUS) activity derived from the internal control plasmid.
Haplogrouping. Information on nucleotide sequences in the 4th intron of NRT1.1 genes from 39 Arabidopsis accessions was obtained from the Salk Arabidopsis 1001 Genomes database (http://signal.salk.edu/atg1001/3.0/gebrowser.php). A haplogroup tree was constructed using MEGA X software 74 by the neighbor-joining method with 1000 bootstrap replicates.
Statistics and reproductivity. All data shown in graphs are presented as the mean with S.D. from more than four biological replicates. Statistical significance was tested by two-tailed Student's t-test (*p < 0.05, **p < 0.01) or ANOVA followed by Tukey's post-hoc test (Different lowercase letters indicate significant differences, p < 0.05).
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
GWAS summary statistics data is in Supplementary Data 2. Information on nucleotide sequences in the NRT1.1 genes (Accession code: At1g12110) from 39 Arabidopsis accessions was obtained from the Salk Arabidopsis 1001 Genomes database (http://signal. salk.edu/atg1001/3.0/gebrowser.php). All other materials are available from the corresponding author upon reasonable request.