Xiuqin Zho,Wensheng Wng,Ziyn Xie,Yongming Go,b,Chuncho Wng,Muhmmed Mhbubur Rshid,Mohmmd Rfiqul Islm,Binying Fu,b,*,Zhikng Li,b

aInstitute of Crop Sciences,National Key Facility for Crop Gene Resources and Genetic Improvement,Chinese Academy of Agricultural Sciences,Beijing 100081,China

bShenzhen Institute for Innovative Breeding,Chinese Academy of Agricultural Sciences,Shenzhen 518120,Guangdong,China

Keywords:Rice Low-nitrogen stress Metabolite Nitrogen metabolism

A B S T R A C T Identification of metabolites responsible for tolerance to low nitrogen availability(low-N)will aid in the genetic improvement of rice yield under nitrogen deficiency.In this study,a backcross introgression line(G9)and its recurrent parent Shuhui 527(SH527),which show differential responses to low-N stress,were used to identify metabolites associated with low-N tolerance in rice.Differences in metabolite contents in the leaves of G9 and SH527 at three growth stages under low-N stress were assessed by gas chromatography–mass spectrometry.Many metabolites,including amino acids and derivatives,were highly enriched in G9 compared with SH527 under the control condition,suggesting that the two genotypes had basal metabolite differences.Low-N stress induced genotype-specific as well as growth stagedependent metabolite changes.Metabolites induced specifically in G9 that were involved in glycolysis and tricarboxylic acid metabolism were enriched at the tillering and grain filling stages,and metabolites involved in nitrogen and proline metabolism were enriched at the booting stage.Enrichment of pyroglutamate,glutamate,2-oxoglutarate,sorbose,glycerate-2-P,and phosphoenolpyruvic acid in G9 suggests that these metabolites could be involved in low-N stress tolerance.The results presented here provide valuable information for further elucidation of the molecular mechanisms of low-N tolerance in crops.

1.Introduction

Rice(Oryza sativa L.)is a staple food crop worldwide,especially in developing countries.A predicted 26%increase in rice production will be required to feed the expanding population by 2035 without a corresponding increase in the area of agricultural land[1,2].

Rice cultivation requires higher inputs of nitrogen(N)fertilizers than other agricultural crops[3,4].However,incorporation of N into agricultural crops rarely exceeds 40%of the amount of applied N,resulting in severe N pollution that is becoming a threat to ecosystems[3,5–9].Moreover, nutrient malnutrition is aggravated by the human demand for increased food production,which has led to the development of nutrient-hungry crop cultivars[10].Therefore,the breeding of novel rice cultivars tolerant of low nitrogen availability(low-N)is urgently needed for sustainable agricultural production.

In higher plants,low-N stress results in physiological,morphological,and molecular changes.Plants show adaptive responses to low N availability,including extensive changes in primary and secondary metabolism, protein synthesis,cellular growth processes,expression of regulatory genes, and other cellular pathways [11]. For example, the plant root system is the most important organ for acquisition of soil N[12]and low-N stress leads to an increased root-to-shoot ratio [10, 13], reduced growth and photosynthesis rates [14, 15], accelerated uptake of N at early growth stages [16], and efficient internal recycling at terminal stages of development[17–19].Under low-N conditions,a variety of responsive genes with diverse functions are upregulated to support plant survival or maintain grain yield [20, 21]. For example, overexpression of nitrate transporter (NRT)genes [22], the Arabidopsis ammonium transporter AtAmt1.1 [23], Dof1 transcription factor [24], NADHglutamate synthase[25],and the alanine aminotransferase gene(AlaAT)[26,27]may increase plant growth or grain yield under low-N conditions.In addition,the low-N response in rice genotypes varies at different growth stages[10,20,28,29].

Genetic variation in low-N tolerance is associated with adaptation to low-input agriculture[30].This relationship is complex and depends on the crop and the environmental conditions.Although many physiological,phenotypic,and molecular analyses have been performed, the molecular mechanisms that govern genetic variation in low-N tolerance among cultivars remain poorly understood. To our knowledge,genetic manipulations for nutrient use in rice have been limited to experimental validation of few candidate genes.

Metabolite profiling is increasingly used to investigate metabolic regulation of the systemic response to the environment or to decipher gene function in plants[31–35].Integration of the results from metabolic profiling and morphological analyses is a powerful strategy for crop improvement[36–39].

As part of our rice breeding program involving introgression of diverse germplasm into elite cultivars [40], an introgression line(G9)that shows high yield under both normal growth and low-N conditions was selected from a backcross population(BC2F8)derived from the cross between Shuhui527(SH527)and Ye-Tuo-Zai.In the present study,a metabolomic analysis was conducted to investigate the impact of low-N stress on metabolite profiles of G9 and SH527 at the tillering,booting,and grain filling stages,with the aim of identifying metabolite differences in response to low-N stress between two genotypes with contrasting low-N tolerance.

2.Materials and methods

2.1.Plant materials and growth conditions

Two rice genotypes,Shuhui 527(SH527)and G9,were used.G9 is an introgression line selected from a BC2F8backcross population derived from the cross of SH527(indica)and Ye-Tuo-Zai(indica).SH527 is the recurrent parent of G9 and is a commercial elite indica restorer line in China[41].The G9 line was selected as a low-N-tolerant line because it showed stably higher grain yields than SH527 under low-N conditions in field experiments over several years.Genotype analysis using SSR markers shows that G9 differs from SH527 in 15 genomic segments from Ye-Tuo-Zai(Fig.S1).

Field experiments were conducted in 2014 and 2015 at the Langfang experimental station of the Chinese Academy of Agricultural Sciences,Hebei province,China.The soil chemical properties were pH 8.1,organic matter content 7.9 g kg?1,total N 0.57 g kg?1,alkali-hydrolyzable nitrogen 37.4 mg kg?1,Olsen-P 7.8 mg kg?1,and available K 70.7 mg kg?1.Forty days after sowing,the rice plants were transplanted into 10-row plots,consisting of 12 plants in each row(120 plants per plot)with spacing of 25 cm×15 cm and with three replications in 2014 and five replications in 2015.As fertilizer treatment,307 kg urea, 1029 kg calcium superphosphate, and 239 kg potassium sulfate were applied per hectare for the control(normal growth conditions).In the low-N treatment,102 kg urea per hectare(30%of the amount applied in the control)was applied and the amounts of P and K fertilizers applied were identical to those in the control.

2.2.Trait evaluation

Grain yield(GY),plant height(PHT),and tiller number per plant(TN)were recorded at plant maturity in 2014 and 2015.N concentration analysis was performed in 2015.The leaves and stems from three plants at maturity in each replicate were collected and dried in an oven for three days at 80°C and the N concentration in each dried sample was then measured following Lu[42].

The metabolite characteristics under low-N stress in G9 and SH527 were systematically evaluated at the tillering(65 days after sowing),booting(95 days after sowing),and grain filling(123 days after sowing)stages in 2015.The five topmost leaves of G9 and SH527 plants were collected in each replicate. All samples with five biological replicates were frozen in liquid nitrogen and stored at ?70°C until metabolite extraction.Metabolite extraction followed Bowne et al.[43]and Zhao et al.[44].The extracted samples were derivatized and analyzed by gas chromatography–triple quadrupole mass spectrometry(GCMS-TQ8040,Shimadzu Corporation,Japan).Chromatograms and mass spectra were processed using the search algorithm implemented in GC–MS Postrun Analysis software.Specific mass spectral fragments were detected in defined retention-time windows using the mass spectral libraries of the Smart Metabolites Database(Shimadzu Corporation). The gas chromatograph was equipped with a capillary column (SGE, BPX-5/30 m×0.25 mm×0.25 μm).The temperature program started at 60°C,held for 2 min,then ramped at 15°C min?1to 320°C,held for 3 min.The injector temperature was maintained at 250°C.The split ratio was 30.The temperature of the injection unit was 250°C.The column effluent was ionized by electron impact ionization at 70 eV at a source temperature of 200°C.

2.3.Data analysis

Statistical analysis of data for each agronomic trait was performed using mean values of five randomly selected plants per genotype in each replication. For subsequent statistical analyses of the metabolites, the relative signal intensities for the detected metabolites were normalized to the mean intensity of all reference samples. Analysis of variance(ANOVA)was performed with SAS 8(SAS Institute Inc.,USA)to determine the significance of differences in agronomic traits and metabolite contents between genotypes and between treatments. Differentially altered traits or metabolites were defined as those showing a significant increase or decrease between treatments or genotypes at P ≤0.05.

The normalized metabolite data were subjected to principal component analysis(PCA)using the SPSS 13.0 software(https://www.ibm.com/products/spss-statistics). The KEGG database(http://www.genome.jp/kegg/)was used to identify pathways in which targeted metabolites participated.

3.Results

3.1.Comparative analysis of grain yield and plant growth between G9 and SH527 under control and low-N conditions

No significant difference in grain yield(GY)was detected between G9 and SH527 under normal growth conditions in either year(2014 and 2015).However,low-N stress induced a significant reduction in GY of SH527 compared with that of the control in both years,whereas the GY of G9 was relatively stable in 2014 under both growth conditions,and was reduced by low-N treatment,with less decrease in G9 than in SH527 evident in 2015. Compared with the control, the low-N condition reduced GY by 32.0%in SH527 and 12.3%in G9 in 2015(Fig.1-A).

The yield-related traits plant height (PHT) and tiller number (TN) in both genotypes were also examined. No significant difference in PHT was detected between G9 and SH527 under the control condition in the successive years of the experiment.Low-N availability resulted in a significant reduction in PHT of SH527 in both years,whereas the PHT of G9 was decreased non-significantly under the low-N condition compared with the control in both years.The reduction in PHT under low-N availability in SH527 was significantly greater than that in G9:in the low-N treatment,the PHT of G9(SH527)decreased by 4.7%(12.6%)and 5.6%(15.9%)in 2014 and 2015,respectively(Fig.1-B).The TN was stable in both genotypes under the control and low-N conditions,except that low-N stress caused a significant reduction in TN of SH527 in 2015(Fig.1-C).Collectively,these results indicated that the growth and yield of G9 were less repressed than those of SH527 under low-N stress.

The N concentrations in the leaves and stems of each genotype at maturity were investigated in 2015.No significant difference was detected between the two genotypes under control conditions.However,low-N induced clear decreases in N concentration of 29.0%and 15.5%in the leaves and 57.0%and 41.8% in the stems of SH527 and G9, respectively,compared with their respective controls. However, the N concentration was significantly higher in the leaves and especially the stem of G9 compared with those organs of SH527 under low-N conditions.For example,the N concentration in the stem of G9 was 41.8%higher than that in SH527(Fig.1-D).These results showed that G9 accumulated more N in its aboveground tissues than SH527 under low-N conditions.

3.2.Overview of metabolite profiling of SH527 and G9 under control and low-N conditions

Gas chromatography–mass spectrometry (GC–MS) was used to profile the metabolites in the leaves of both genotypes at different developmental stages under low-N and control conditions. A total of 142 metabolites were characterized in all samples. These metabolites included 24 amino acids and derivatives, 63 organic acids, 24 sugars and derivatives, six nucleic acids, five phytochemical compounds, four lipids, and 16 other small molecular components (Table S1).

To obtain an overall picture of the metabolite profiles of SH527 and G9 in response to low-N stress at different developmental stages,all measured metabolite data were subjected to principal component analysis(PCA).As shown in Fig.2,the first principal component(PC1)accounted for 38%of the total detected metabolite variance and clearly separated the samples grown under the low-N condition from the corresponding controls at each growth stage. However,opposite patterns for the control and low-N samples were observed between the booting stage and the other two stages.For example,the low-N samples were located to the right of the controls at the tillering and grain filling stages,but to the left of the control samples at the booting stage.Loading analysis showed that the low-N condition induced increased accumulation of most metabolites at the tillering and grain filling stages,and decreased accumulation at the booting stage(Table S1).

PC2 accounted for 21.7%of the total variance,and clearly separated the metabolic phenotypes of both genotypes at the booting stage from those at the other two stages.Loading analysis showed that,compared with the tillering and grain filling stages, greater numbers of amino acids and fewer organic acids and sugars were accumulated in the leaves of both genotypes at the booting stage(Table S1),indicating that these metabolites play an important role in rice reproductive growth.PC3 accounted for 12.9%of the total variance,clearly separating the metabolic phenotypes at the tillering stage from those at the grain filling stage and contributing to the separation of low-N and control samples at the booting stage.Loading analysis showed that metabolites including erythrulose, arabinose, tartrate, malate, adipate, and dihydrouracil were highly accumulated in both genotypes at the tillering stage,whereas metabolites such as pantothenate,2-aminoadipate, phosphate, glutamate-5-methylester, 1-hexadecanol,3-aminoglutarate,and aspartate were accumulated in both genotypes at the grain filling stage(Fig.2,Table S1). Metabolites including threitol, 5-methoxytryptamine,tyramine,and kynurenate were significantly accumulated in samples of both genotypes at the booting stage under low-N treatment compared with the control.These data suggested that growth stage exerted a greater influence on metabolite performance than did low-N treatment.

Fig.1–Grain yield and yield-related traits of SH527 and G9 under control and low-nitrogen(low-N)conditions in 2014 and/or 2015.Grain yield(A),plant height(B),tiller number(C)and N concentration in leaf and stem(D)were assessed.An“a”indicates that the trait value of G9 is significantly higher than that of SH527 under the same treatment;*indicates that the trait value was decreased significantly by low-N stress compared with that of the control in the same experiment.

3.3.Comparison of metabolite contents between G9 and SH527 under the control condition

To investigate the intrinsic differences in metabolites between the two genotypes, a comparative analysis of the metabolite contents of SH527 and G9 under the control condition was performed(Table 1).Differentially expressed metabolites (DEMs) were assigned as those showing a significant difference in the content of a metabolite between two samples(P ≤0.05 based on ANOVA).

At the tillering stage,a total of 73 DEMs were detected between the two genotypes under the control condition.Of these DEMs,the contents of 67 in G9 were significantly higher than those in SH527,with an average 1.78-fold higher than that in SH527.At the booting stage,the contents of 47 DEMs in G9 were significantly different from those in SH527,with 31 metabolites enriched in G9 with an average 2.21-fold increase in content compared with that in SH527.At the grain filling stage under the control condition,62 DEMs were detected between G9 and SH527,with the contents of 20 metabolites in G9 significantly higher than those in SH527,and their average content was 2.64-fold higher in G9 than in SH527(Table 1).

The DEMs were comparatively analyzed at the three growth stages. Only five metabolites, namely glycerate-3-phosphate, valine, O-acetylserine, asparagine, and Nacetylserine,showed contents higher in G9 than in SH527 at all three stages(Table 1),indicating that these metabolites of G9 and SH527 showed stable differences prior to exposure to low-N stress.Metabolites that showed a higher content in G9 relative to that of SH527 decreased with the successive growth stages,showing that the basal metabolome difference between the two genotypes was dependent on growth stage.The contents of a larger number of amino acids and derivatives were higher in G9 than in SH527 across the three growth stages.

3.4.Metabolite alteration in G9 and SH527 under the low-N condition

To investigate the differences in the metabolomes of the two genotypes induced by low-N stress,metabolite changes in the two genotypes at the three growth stages under low-N treatment compared with the respective control were analyzed.

Fig.2–Plot of the first three principal components for the leaf metabolites in two rice genotypes,Shuhui 527(SH527)and the G9 introgression line,sampled at the tillering(T),booting(B),and grain filling(G)stages,under control(hollow shapes)and low-N(solid shapes)conditions.

At the tillering stage,100 and 120 DEMs were detected in G9 and SH527,respectively,under low-N stress relative to the respective control.Of these DEMs,the contents of 89(110)and 11(10)metabolites were increased and decreased,respectively,in G9(SH527)under the low-N condition.Venn diagram analysis(Fig.3)indicated that 82 increased and five decreased metabolites were commonly detected in both genotypes,indicating that the metabolome responses of the two genotypes to low-N stress were largely identical.However,the contents of 28(7)and 5(6)metabolites were increased and decreased specifically in SH527(G9),respectively(Table S2,Fig.3).The seven metabolites that increased specifically in G9 were phosphoenolpyruvic acid (PEP), oxalate, leucine,gluconate-6-phosphate, tetradecanoic acid, hydroquinone,and 4-hydroxyphenyllactate,which are involved mainly in glycolysis, the citrate (or tricarboxylic acid; TCA) cycle,pyruvate metabolism,and amino acid metabolism.Twentyfour of 28 metabolites were relatively stable in G9 but increased in SH527 under the low-N condition(Table S2).Among these metabolites,2-oxoglutarate,glutamate,glutamine,and methionine are associated mainly with N metabolism,arginine and proline metabolism,and metabolism of other amino acids.

At the booting stage,111 and 100 metabolites were differentially expressed in SH527 and G9,respectively,under low-N stress (Table S2, Fig. 3). Seven and 74 metabolites were increased and decreased,respectively,in abundance in both genotypes,whereas 6(12)and 23(7)metabolites were detected as specifically increased and decreased,respectively,in SH527(G9)(Fig.3).Of the genotype-specifically increased metabolites,12 metabolites were uniquely increased in G9:cadaverine,glycerate-2-phosphate,inositol,lauric acid,stearate,threitol,erythrulose,cholesterol,O-phosphoethanolamine,urea,benzoate,and trehalose.Of these metabolites,benzoate and urea are involved in nitrogen metabolism,arginine and proline metabolism,and biosynthesis of secondary metabolites.

At the grain filling stage,54 and 83 metabolites increased,whereas 16 and nine metabolites decreased in abundance in SH527 and G9,respectively,under low-N stress(Table S2,Fig.3).Comparative analysis showed that 38 and four metabolites were increased and decreased,respectively in both genotypes.A large number of increased metabolites(16 in SH527 and 45 in G9)were identified exclusively in SH527 and G9,respectively(Fig.3),showing that the metabolome response to low-N stress at the grain filling stage is highly genotype-specific.The 45 G9-specific metabolites increased in abundance by 95.7% on average. Most of them were organic acids and sugars,including pyruvate and 2-oxoglutarate,involved in glycolysis,the TCA cycle,and amino acid metabolism.In general, the number of metabolites whose content was increased by low-N stress exclusively in G9 increased at successive growth stages(7,12,and 45 metabolites at the tillering,booting,and grain filling stages,respectively)(Fig.3).Collectively,these results suggest that the response mechanisms differentiating the two genotypes at the metabolite level and the metabolites induced specifically in G9 by low-N stress may play an important role in nitrogen-deficiency tolerance.

Differences in metabolite content between G9 and SH527 under low-N stress were comparatively analyzed(Table S3).The content of 107 metabolites was significantly higher in G9 than in SH527 under low-N stress in at least one growth stage(30,87,and 35 metabolites at the tillering,booting,and grain filling stages,respectively).The majority of the increased metabolites at the booting stage were amino acids,organic acids, and sugars. Importantly, nine metabolites: PEP,pyroglutamate, 3-sulfinoalanine, malate, 2-oxoglutarate,glycerate-2-phosphate,glutamate 5-methylester,sorbose,and histidinol phosphate,were highly enriched in G9 relative to that of SH527 at all growth stages under low-N stress,indicating their crucial role in tolerance to low-N stress in G9.

Table 1–Metabolites differentially expressed between G9 and SH527 under the control condition.

Table 1(continued)

4.Discussion

Nitrogen is a major factor limiting rice yield under field conditions.Identification of molecular pathways associated with tolerance to N deficiency is imperative for the development of rice cultivars tolerant to low-N stress using molecular breeding[45].In the present study,an introgression line (G9) was selected from a backcross population that showed low-N tolerance in two successive years.G9 was much less affected by low-N stress than its recurrent parent SH527 with respect to growth traits and grain yield.Comparative analysis of the N contents of both genotypes suggested that G9 might have a higher N-uptake capacity than SH527 under low-N conditions,as reflected by a greater N concentration in the shoots of G9.N content in shoots has been reported to be positively correlated with grain yield[46],and the higher N content in the shoots of G9 under low-N conditions could help meet the N demand for grain development; accordingly, G9 has stable yield performance in response to low-N stress.

Fig.3–Venn diagram analysis of differentially expressed metabolites in the rice G9 introgression line and Shuhui 527(SH527)at the tillering(T),booting(B),and grain filling(G)stages under low-N stress.

The low-N stress response is a trait under strong genetic control that differs among crop genotypes [29, 47].Transcriptomic analysis has shown that low-N stress induces the transcription of hundreds of genes associated with various biological processes [9, 21, 48]. However, limited information is available on the global metabolite changes in plants exposed to low-N stress,especially for rice.To identify mechanisms of low-N tolerance at the metabolite level,metabolite profiling was performed for the low-N-tolerant G9 introgression line and the low-N-sensitive recurrent parent SH527 at three developmental stages under low-N and normal (N-sufficient) growth conditions. The results showed a wide range of metabolite differences among genotypes, experimental conditions and developmental growth stages.G9 had higher contents of a greater number of amino acids and derivatives than SH527 across the three growth stages under the control condition,suggesting that the two genotypes have intrinsic metabolomic differences,consistent with their contrasting phenotypes in response to low-N stress.Nitrogen is required for the synthesis of nucleotides and amino acids,which are the building blocks of nucleic acids and proteins[49].The enriched amino acid pool in G9 might play a basal role prior to exposure to low-N stress.Strikingly,five metabolites,namely glycerate-3-phosphate,valine,O-acetylserine,asparagine,and N-acetylserine,had consistently higher contents in G9 than in SH527 at all three developmental stages under the control condition. In a previous study [50], glycerate-3-phosphate increased in abundance in barley plants under phosphate deficiency,and the four amino acids listed above are extremely important in N assimilation in rice plants[51].Thus,these five metabolites may be crucial for the intrinsic tolerance of G9 to low-N stress.

The global metabolite alterations in the two genotypes under low-N stress were analyzed to assess differential responses to N deficiency.The results revealed a distinct genotype-dependent as well as growth stage-dependent metabolite alteration under low-N stress. The metabolite changes in the two genotypes in response to low-N stress differed among the three developmental stages.The overall trend was that low-N stress induced accumulation of much higher numbers of metabolites at the tillering and grain filling stages,whereas major metabolites decreased in abundance at the booting stage in both genotypes. These results are suggestive of a growth stage-dependent metabolite reprogramming pattern under low-N stress,consistent with the findings of a previous omics study of maize[52].The leaf is the major organ that produces photosynthate and energy for plant growth at the tillering stage,and the stored photosynthate in leaves is partially or completely translocated to grain at the grain filling stage. Thus, it is reasonable for leaf metabolites to increase in abundance at the tillering and grain filling stages under low-N stress.However,the booting stage is a critical period for yield determination[53–55]and the organs involved in spike differentiation are an active sink that shows a strong, competitive ability to attract large amounts of assimilates from source tissues[56].The heterotrophic organ systems in plants are dependent upon amino acid and sugar import for normal growth and development[57].Accordingly,the majority of the metabolites including amino acids and sugars in leaves are translocated to the young panicle at the booting stage to maintain normal development during low-N stress.

Further comparative analysis revealed that different sets of metabolites were uniquely increased in the low-N-tolerant G9 at the three developmental stages in response to N-deficiency exposure. Metabolites including PEP, oxalate, tetradecanoic acid, and gluconate-6-phosphate, which are involved in glycolysis and TCA metabolism, were highly increased in abundance at the tillering stage in G9 in response to low-N stress.Contents of metabolites associated with N metabolism and with arginine and proline metabolism were increased at the booting stage in response to N-deficiency stress in G9,whereas a large number of G9-specific metabolites involved in glycolysis,TCA metabolism,and amino acid metabolism were increased at the grain filling stage by low-N stress.Strikingly,no metabolite showed an increase in content at all three developmental stages, indicating that the metabolomic response to low-N stress is highly dependent on growth stage.

Nine metabolites showed higher contents exclusively in G9 relative to those of SH527 at all growth stages under low-N stress.The compound 2-oxoglutarate is a key regulator of C and N interaction in N metabolism[58,59]and plays an important role in regulation of C and N metabolism in rice[60];Malate is essential for N assimilation[61].Sorbose is reported to be widely involved in biotic and abiotic stress tolerance[62].Pyroglutamate is a reservoir of glutamate[63],and glutamate is an amino group donor as well as the main N transport molecule; glutamate content is fundamental in sensing plant nutritional status and in linking C and N metabolism [64–66]. In addition to the constantly higher content of pyroglutamate,the glutamate content was higher in G9 than that in SH527 at the booting and grain filling stages,a finding consistent with a positive role in low-N tolerance in G9. Glycerate-2-phosphate and PEP are both involved in glycolysis.Higher contents of 2-oxoglutarate,pyroglutamate,malate,sorbose,glycerate-2-phosphate,and PEP in G9 at all growth stages suggest that they play a critical role in low-N stress tolerance.

5.Conclusions

An introgression line G9 and its recurrent parent SH527 were evaluated under low-N stress and normal growth conditions for two consecutive years.G9 showed markedly improved tolerance to low-N.The metabolite differences in G9 and SH527 under low-N and control conditions were assessed by gas chromatography–mass spectrometry.The results indicated that low-N stress exerts extensive effects on rice plants,which show wide-ranging metabolite changes in a genotypedependent as well as growth stage-dependent manner.A set of metabolites,including amino acids and derivatives,were highly enriched in G9 relative to SH527, indicating their intrinsic role in tolerance to low-N stress.The metabolites that increased specifically in G9 under low-N stress are involved in glycolysis,the TCA cycle,and nitrogen metabolism,suggesting that these metabolites play an important role in tolerance to low-N stress.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2018.05.006.

Acknowledgments

This research was funded by the National Natural Science Foundation of China(31471429),the National Key Technology R&D Program of China(2015BAD02B01-2-1,2016YFD0100904),Fundamental Research Funds for Central Non-Profit of CAAS(Y2017CG21), a Bill & Melinda Gates Foundation Project(OPP51587),the CAAS Innovative Team Award(2060302-2-18)to B.Y. Fu's team, the Shenzhen Peacock Plan(20130415095710361), and the National High Technology Research and Development Program of China(2014AA10A604-8,2014AA10A601).