Nitrogen (N), being the constituent of most biomolecules, viz. amino acids, nucleotides, proteins, chlorophyll, and many plant hormones, it is considered the major essential nutrient required for plant growth and development [1
]. Plants exhibit various changes in phenotype under N starvation, including reduced seed production (yield), leaf chlorosis, stunted growth, modulation in root architecture, etc., underlining the importance of N to plant growth and development [3
]. In the last few decades, the development of N-responsive varieties and the extensive use of N fertilizers has resulted in increased biomass and subsequently the yield of crop plants [1
]. However, of the total applied N fertilizer, plants are able to use only 30–40%, with the rest of the N fertilizers being lost to the atmosphere, groundwater, and rivers through various physicochemical processes, resulting in economic loss to farmers. The loss of N also results in the eutrophication of fresh water, the acidification of soil, and the release of greenhouse gases like nitrous oxide (around 300 times more toxic than CO2
), leading to adverse impact on the environment [5
]. The increase in the world’s population from 5 billion to 9 billion in the last 50 years demands increased production of staple foodstuffs. This in turn requires a huge quantity of N fertilizers, notwithstanding the fact that the production of N is an energy-demanding process. Thus, increased efficiency of N application in plants would not only result in higher crop yield under limited N supply, benefiting the farmers via higher net profit, but also mitigate the environmental risks arising due to an excess of fertilizers used on agricultural land. In this context, improving the nitrogen use efficiency (NUE) of rice, a dominant dietary source in almost every part of the world, would be worthwhile.
NUE of plants, in general, is defined as their efficiency at utilizing N from the soil. NUE has two major component traits, namely N uptake and N utilization. To be an N-efficient plant, both components are crucial. N uptake is mainly determined by various N transporters and probably also by the root architecture of the plant, whereas utilization is determined by assimilation, mobilization, and remobilization of the assimilated N for the purpose of economic yield. Agriculturally important crops take up N mainly in the form of nitrate (NO3−
) and ammonium (NH4+
) ions from well-fertilized soils. Among them, NO3−
ions act not only as a nutrient but also as signal molecules, inducing the expression of many genes including N transport and metabolizing genes, e.g., nitrate transporters (NRT1 and NRT2), nitrate reductase (NR), nitrite reductase (NiR), glutamine synthetase (GS), and glutamate synthase (GOGAT) [7
]. Under a low supply of N, high-affinity transporters NRT2 and NRT3 play a significant role in N uptake, as demonstrated in maize [10
Genome-wide expression analysis is an attractive approach to understanding complex traits like NUE. For genome-wide transcription profiling, RNA sequencing (RNA-seq), a high-throughput sequencing technology, is the best approach and has replaced microarray even in model plants like rice and Arabidopsis
, which have high-quality whole-genome sequence information available [11
]. In the last two decades, efforts have been made to understand the molecular and physiological basis of plants grown under N stress, which resulted in the identification of a large number of differentially expressed genes (DEGs) under limited N supply in many crop plants, including rice [13
], soybeans [15
], sorghum [16
], and tea [17
]. Most of these studies have concentrated on studying the global gene expression in a single genotype under low and optimal nitrogen (ammonia or nitrate), except the one on tea, where two genotypes of tea were compared for their responses under low and optimal ammonical nitrogen [13
]. Global gene expression and comparative analysis of genotypes contrasting for NUE would aid in narrowing down the candidate genes. Moreover, a huge volume of literature is available on quantitative trait loci (QTL) affecting NUE in rice [18
]. Integration of these two datasets (QTL and DEGs in contrasting genotypes) has the potential to identify robust candidate genes that can be directly deployed in crop improvement for NUE [21
]. In the current study, an exhaustive analysis has been conducted to identify differentially expressed genes in two rice varieties, IR 64 (IR64) and Nagina 22 (N22), and compared with the available QTL data to identify the robust candidate genes for NUE in rice.
Though nitrogen starvation-responsive genes have been explored in rice using medium-/high-density gene chip (microarray) and RNA-seq approaches [13
], these studies confined themselves to studying the root tissues of one rice genotype at a time, which are popular high-yielding cultivars, namely Minghui 63 (indica), Dongjin, and Hejiang (japonica), and either short-term (<1 h) or medium-term (five days) responses to N starvation conditions. With huge genetic variability available in rice for all traits including response to N fertilizer application, we explored the N-responsive genes in two rice genotypes (indica and aus type), contrasting their response to chronic N starvation after confirming their response by phenotyping and enzyme studies. Unlike previous studies that sampled only root tissues for genome-wide expression analyses [13
], except a single one where shoot tissues were studied under macronutrient (N, P, and K) deficiency [39
], we sampled both root and shoot tissues in our study.
In general, N starvation affected the overall growth of both rice genotypes; however, below ground, part of IR64 was either more tolerant or was non-responsive to N stress compared to N22, i.e., IR64 kept its biomass allocation almost constant in root tissue even in N stress conditions, while N22 increased its foraging ability of nitrogen. Since in the present study a nutrient-free media (vermiculite and perlite mixture) was used for seedlings growth, which caused heterogeneous nutrient distribution, these seedlings selectively altered the root growth patterns in nutrient-rich microsites by altering root biomass allocation [40
]. Interestingly, IR64 adapted to this alteration without any significant decrease in biomass parameters in N stress compared to N22.
The rice root system is mainly composed of nodal roots; however, it develops a seminal (radicle) root that emerges immediately after germination. Except for the seminal root, other RSA parameters showed a significant increase under N stress in IR64 compared to N22, wherein these root traits either showed no change or were reduced. Lateral root traits in terms of number (FOLRN, SOLRN, and LRD) were found to be modulated more profoundly than their lengths (LRS) in the case of IR64. These lateral roots play crucial roles in water and nutrient acquisition [41
] by exhibiting their foraging ability in a nutrient-heterogeneous environment. Thus, IR64 perceived the N-deprivation signal more efficiently than N22, which is exhibited by modulation of its root architecture.
Chlorophyll pigments play a major role in radiation interception and hence affect leaf and canopy photosynthesis, which ultimately decides the yield of the plant. With N being a major constituent of these pigments [42
], the availability of nitrogen in the leaves has a significant impact on the plant productivity. Our study demonstrated that IR64 has the unique capability to retain its total chlorophyll content even after chronic N starvation, as compared to N22. Though these pigments were estimated at a single stage, the results indicated that IR64 has a mechanism to protect their pigment system even after 15 days of N stress and hence would have better seasonal canopy apparent photosynthesis (CAP), which contributes in biomass and yield and consequently nitrogen use efficiency.
The 15-day nitrogen stress invariably caused reduced specific activity for almost all nitrogen- and carbon-metabolizing enzymes except NiR and CS (Figure 4
A), which is quite obvious in the absence of corresponding substrate during each enzyme-catalyzing step. The nitrogen acquisition, transport, assimilation mobilization, etc., are highly regulated processes that follow feedback inhibition depending on the nitrogen and carbon status of the plants. NR, which catalyzes the first step of nitrate assimilation, was found to be severely reduced under nitrogen stress in both genotypes, indicating the course regulation of this enzyme under chronic N-stress. GDH was found to be increased under N stress in both genotypes, indicating its role in glutamate homeostasis in a situation of reduced GS activity [43
To our knowledge, this is the first genome-wide expression profiling through RNA-seq report from both root and shoot tissues on chronic N starvation in rice, especially from two genotypes that are contrasting in their response to N starvation, though a similar study has recently been reported in tea [17
]. Subsequently, we found a larger number of DEGs in shoot than root tissues in both genotypes under N stress, suggesting that shoot tissues are equally or more significantly affected in response to N deficiency. A comparison between genotypes also showed a substantial number of DEGs under N starvation in shoot tissues, though the number of DEGs was higher in root tissues (Table 2
). Interestingly, the responses of these two genotypes were completely different to N starvation as only two genes in root tissues and eight genes in shoot tissues were found to be common between these two genotypes (Table 3
). Such a difference in the array of DEGs identified in the two genotypes under low N could be due to the inherent differences in the genotypes, where one was a high-yielding genotype (IR64) and the other was a traditional tall genotype (N22). Similar differences in starch metabolism-related enzymes’ activities (AGPase- ADP glucose pyrophophorylase and starch branching enzyme II) and their transcription profiles under different N supply were reported between an N-responsive japonica genotype (cv. Nipponbare) and N-unresponsive indica genotypes, Tetep and Johna [44
RNA-seq results supported the morphological/physiological observations: for example, chlorophyll metabolism-related genes were not differentially expressed in IR64, while 10 different genes known to function in chlorophyll metabolism, such as chlorophyllase, chlorophyll A-B binding domain-containing proteins, and many known chloroplast precursors such as lycopene epsilon cyclase, photosystem I and II related transport peptides, etc., were differentially regulated in N22. Further auxin biosynthesis was also differentially expressed in N22, unlike IR64. A comparison of differential expression between the two genotypes under N starvation also showed downregulation in these genes in addition to starch biosynthesis-related genes (Table S1
, sheet 6). As 75% of the plant’s N is present in the seat of photosynthesis, i.e., chloroplasts, of which 27% are bound to Rubisco, the major enzyme of carbon assimilation, the involvement of chloroplast, the genes identified in starch, and chloroplast metabolism can be explained [45
]. Furthermore, only a few genes (28 out of 279) were found to be upregulated in N22, of which 16 (57.14%) had unknown functions. Using microarray analyses of shoot tissues, similar observations wherein a few genes were upregulated and most of the differential expression occurring in basic plant development, chloroplast-related gene expression, and starch biosynthesis have been reported [39
]. Most of the studies have implicated N assimilation genes, starch synthesis-related genes, and gibberellin metabolism genes in nitrogen use efficiency of plants such as rice, maize, and tea [17
]. Interestingly, more of the LTPL lipases that are secretory proteins and signaling molecules and GDSL like lipases that are known to play a role in biotic [49
] and abiotic stress tolerance such as salt tolerance [50
], drought and biotic stress tolerance in transgenic plants [51
] have shown differential regulation in our study. As N starvation is closely linked to starch starvation [52
], a number of glycosyl hydrolases were found to be upregulated in both genotypes; however, IR64 had five different members of the glycosyl hydrolase family in shoots and one in roots highly upregulated, while N22 had just two and one members upregulated in shoot and root tissues, respectively. The N transporter genes were noticeable by their absence in any of the comparisons, probably because of the chronic N starvation of the seedlings. Still, we did observe the oxoglutarate and malate dehydrogenase and translocator genes to be N-responsive and differentially expressed in N22 but not IR64. Further studies on gene sequence comparison including promoter regions of these genes between the two genotypes could help in ascertaining their role in NUE.
Comparing the expression profiles of two contrasting genotypes for a specific trait under different treatments is supposed to help in identification of causal genes when genetic analysis (mapping) of such traits are undertaken [20
]. Here we have identified 87 such candidate genes (62 from root and 28 from shoot) in the major QTLs regions for NUE in rice. Based on their FPKM values, some of them have also been suggested as candidate genes. We expect this will serve as a major resource for validation of the NUE-related genes in rice.