Defining the physiological determinants of low nitrogen requirement in wheat

Wheat (Triticum aestivum L.) is a major crop grown worldwide, with >700 million tonnes produced per year (FAO stat 2018), and accounting for 23% of the protein in people's diet [1]. Grain protein content (GPC), of which gluten protein accounts for 70–75%, is one of the main criteria determining grain quality. Sufficient nutrient inputs, nitrogen (N) in particular, are required by the developing plants to achieve both high yield and GPC. Limited N availability is a major issue in many agro-ecosystems as wheat plants do not establish symbiotic association with N₂ fixing microbes. This limitation has been addressed by the development and widespread adoption of synthetic N fertilisers arising from the Haber–Bosch process. However, these are costly to farmers in developed countries, and remain unaffordable unless subsidised in developing countries, although their application is widespread and continues to increase [2]. The extensive use of synthetic fertiliser in wheat production, and lack of equilibrium between N supply and demand (Figure 1), has contributed to many environmental issues such as eutrophication of aquatic ecosystems [3], volatilisation of N₂O (a potent greenhouse gas, [4,5]) and increases in N deposition in natural terrestrial ecosystems contributing to the loss of biodiversity [6]. There is, therefore, a pressing need to identify means (e.g. agronomic practises or new varieties) to produce wheat that meets quality requirements for human consumption while lowering the overall application and loss of synthetic N fertiliser. The development of new wheat genetic material either through breeding or genetic modification can offer opportunities to lower wheat N requirement, which will reduce the need for N application. However, these approaches require an understanding of the genetic and physiological determinants of low N requirement in wheat. In this review, we dissect the drivers of N requirement at the plant level and the temporal dynamics of supply and demand. We also discuss the potential for achieving plasticity in wheat N response and present evidence supporting tractable reductions in N requirement. Further opportunities to optimise microbial interactions to boost N supply are also discussed. Together, this provides a novel perspective on the ability to use physiological and genetic mechanisms to reduce wheat N inputs while maintaining productive potential.

As the harvested and consumed component, wheat grains represent the ultimate N sink relevant to food production. Wheat grains are arranged on inflorescences (also known as spikes) with one inflorescence per tiller. Inflorescences are composed of an unbranched central rachis surrounded by single spikelets on either side, with each spikelet typically holding 3–4 fertile florets, and each floret potentially generating a grain (Figure 1). The number of spikes per plant, grains per spike and individual grain weight (see [7] for genetic determinant of wheat grain weight), determine the overall grain yield and potential N sink. Protein bodies located throughout the starchy endosperm of the wheat grain accumulate throughout grain development, starting from about 5 days post-anthesis. Endosperm proteins, particularly high molecular mass glutenins, are important for breadmaking characteristics. The germ (or embryo) appears to hold a higher N percent than the endosperm but represents an overall smaller sink [8]. Together, yield (number and size of grain) and GPC set the minimum N requirement for wheat production. However, these are negatively correlated, and increasing yield levels in newly selected varieties have been found to be associated with lower GPC in the U.K. [9], Germany [10], and the U.S. [11]. Recommendations of greater N application to achieve both high yield and GPC in varieties with greater N sink and lower predicted GPC have driven increases in N fertiliser use until the end of the last century [12]. While recent breeding efforts have increased yield at the expense of GPC, heritable variation in the positive grain protein deviation (GPD) from this negative correlation exists [13] and is associated with post-anthesis N uptake [14,15]. Therefore, selection for GPD may enable high yields with decreased N requirements.

Structural tissues maintaining the grains (palea, lemma, glumes, rachis in the spike, Figure 1 inset), leaves, stems, and roots hold the remainder of the N and constitute temporary N sinks throughout wheat plant development. During the vegetative stages (e.g. seedling development, tillering) leaf tissues represent a large N sink, which switches when the developing inflorescence emerge from the flag leaf sheath and especially after flowering (or anthesis) during the grain filling period. At anthesis, on average 23% of N (from above-ground tissues) is found in the ear with the remaining 35% in leaf blades, 14% in leaf sheath, and 28% in stem [16]. At maturity, an average of 89% of N (from above-ground tissues) is found in the ear with the remaining 5% in leaf blades, 6% in leaf sheath, and 4% in stem [16]. Some amount of N is also locked in roots, which are more difficult to assess in soil-grown plants. For example, wheat roots hold ∼2% N in younger plants (tillering stage), which decreases as the plant matures [17]. While lowering the N requirements relating to grain yield and protein content is problematic and could jeopardise productivity, manipulating N requirement related to these transitional N sinks offer an attractive alternative for improvements. These alternatives could either relate to the amount of N used to achieve specific development of transitional roles, or to the efficiency of the remobilisation to the grains at maturity.

One of the improvements brought about by the Green Revolution was an increase in the ratio of grain biomass to above-ground biomass (i.e. harvest index, HI), a result of selection for semi-dwarf traits for which underlying genes have been identified ([18], and references therein). Shorter wheat varieties partition less assimilates (carbon, C, and N) to the stem, but more to the grain and are more resistant to lodging (the physical displacement of the stem that can impair efficient harvest). In more recent varieties HI has reached ∼50–55% for wheat (for review [19]), with little improvement in the last 20 years, while recent increases in yield have been achieved through improved source development during the pre-anthesis foundation phase [20]. New reduced height (Rht) alleles are being sought [21]. Besides providing structural support to the spike, the stem may also hold significant amount of assimilates especially during the grain filling phase [16]. Thus, simply decreasing the size of transitory sink tissues is not sufficient, and the processes that lead to uptake, assimilation, and remobilisation from source tissues to sink tissues must be efficient.

N requirement varies with wheat developmental stage [22]. While wheat seedlings rely on the seed N store in the early developmental stages, high N levels are necessary for seedling growth, tiller production, and the establishment of an efficient canopy that can ensure sufficient C and N assimilation. Nitrogen taken up at the root level tends to be transported to the shoot for assimilation into amino acid through the glutamine synthetase-glutamine oxoglutarate aminotransferase (GS-GOGAT) pathway (for review [23]).

Wheat takes up N primarily in the form of nitrate (⁠$NO3−$⁠) and ammonium (⁠$NH4+$⁠) from the soil solution (amino acids can also be taken up but contribute less to the total N nutrition especially under high N conditions, [24]). These two N molecules can be provided as NH₄NO₃ fertiliser or be derived from soil organic matter (if urea is provided as N fertiliser it is reduced in the soil, a process that also leads to some amount of NH₃ volatilisation). A suite of low and high-affinity transporters are necessary to take up N from the soil solutions and the genes encoding these have now been identified in wheat [25,26]. At the uptake level, nitrate and ammonium transporters need to show high affinity, being present in abundance (and continuing to show high affinity even under higher N conditions). At the root level, N can also be released in the form of root exudates (as amino acid, [27]), thus there is some level of N recycling at the root level.

Investigating the physiological underpinning of high GPD led Bogard et al. [14] to identify a positive correlation between post-anthesis N uptake and GPD. Post-anthesis N uptake can contribute from 5% to 40% of the total grain N in winter wheat and is dependent on N being available in the soil during the grain filling period as well as favourable climatic conditions [28]. So far, the physiological determinants of post-anthesis N uptake are not well understood and merit further study given the potential opportunity to improve GPC without affecting yield. Taulemesse et al. [15] showed that under high N conditions post-anthesis N uptake per g dry weight was reduced compared with low N availability conditions. The physiological functioning of post-anthesis N uptake remains uncovered as it is not clear whether the N taken up by the roots post-anthesis is directly transported to the developing grain, thereby bypassing the flag leaf. Some level of nitrate reductase activity (a necessary step in the process of assimilating nitrate into amino acid) has been detected in both the grain and surrounding tissues (glumes, [29]). Therefore, there is evidence to suggest value in further work to elucidate the role of different tissues (root, shoot, leaves) in post-anthesis N uptake and transport and in further understanding the associated molecular responses.

Leaf blades represent the largest N sink during vegetative development before becoming a N source during the grain filling period and leaf senescence. Within the leaves, the most abundant protein is rubisco (ribulose 1,5-bisphosphate carboxylase/oxygenase), which carboxylates ribulose 1,5-bisphosphate (RuBP) a key step in photosynthetic C fixation. Rubisco represents 51% of leaf total N in wheat [21,22], though the total N cost of the entire photosynthetic apparatus is higher as light-harvesting complexes and photosystems also represent a high proportion of the C3 leaf N budget [30]. Given that rubisco has a low catalytic capacity (though on par with other similar catalytic enzymes), high N level is necessary to ensure high rubisco activity and therefore high CO₂ fixation. Leaf N content is correlated with leaf-level CO₂ assimilation rate; however, the slope of the correlation declines as N availability increases [31]. There is some level of variation in enzymatic characteristics of the rubisco protein from different members of the Triticeae family, suggesting that for a unit of N stored in the leaves greater photosynthetic fixation could be achieved [32].

In addition to carboxylation, rubisco can also oxygenate RuBP, which leads to the production of 3-phosphoglycerate (3-PGA) and 2-phosphoglycolate (2-PG). Due to the inhibitory activities of 2-PG, this metabolite is quickly converted through the photorespiratory pathway to 3-PGA. This process also generates CO₂ and $NH4+$ in the mitochondria through the activity of glycine decarboxylase. $NH4+$ is subsequently assimilated in the chloroplast by the plastidic isoform of glutamine synthetase (GS2). Photorespiration has generally been considered a wasteful process and it has been proposed that productivity increases could be achieved if it can be by-passed in crop [33,34]. However, Busch et al. [35] showed that N assimilation through the photorespiratory pathway can provide an increase in CO₂ fixation. An efficient assimilation in the leaf is important to prevent loss of NH₃ through volatilisation, which can account for 1–4% of the applied N in a wheat field [36]. NH₃ volatilisation occurs when the mole fraction of NH₃ in the atmosphere is lower than the mole fraction of gaseous NH₃ above the water film in the mesophyll cell walls in the substomatal cavity (NH₃ compensation point), and NH₃ is absorbed if it is higher [37–39]. Lower N availability [36] as well as low GS activity leads to higher NH₃ compensation point, and volatilisation [38]. Dynamic processes associated with photosynthesis drive C fixation and N demand during plant development. Optimising N requirement requires further understanding of the specific coordination of processes.

Besides being a necessary macronutrient, N (especially in the form of nitrate $NO3−$⁠) plays a role as a signalling molecule [40]. For example, external availability of N to a plant otherwise depleted in N leads to over-proliferation of roots while overall high N availability leads to a more compact and restrained root architecture. In Arabidopsis thaliana, AtNRT1;1 has been proposed as a transceptor (protein that plays the role of both receptor and transporter, [41]), though the wheat ortholog of AtNRT1;1 has yet to be characterised. This poses the question as to whether a similar mechanism of N perception is conserved in wheat and highlights the importance of conducting similar work to better understand the signal transduction pathway associated with N requirement in wheat.

In addition to its role in shaping root architecture, N can also influence shoot architecture. Increased N availability can induce tiller number and tiller height [42]. Higher availability early in the season will induce tiller production and the potential size of the N sink at maturity. This can be problematic as tillers that emerge later during development compete for light and nutrients [43,44], and surplus late tillers tend to die back or be less productive than early tillers. This may ultimately lead to a lower yield per unit of N (also defined as N use efficiency, NUE) when plants are grown under high vs low N conditions [10,45,46].

To achieve low N requirement, many aspects of wheat plant metabolism, including uptake, assimilation and remobilisation steps must be co-regulated. Regulatory elements that can affect all steps in a co-ordinated manner are of interest. For example, Grf4 was shown to be a N-responsive transcriptional regulator promoting both $NH4+$ uptake and growth in response to N supply (and counteracting the inhibitory effects of the DELLA protein SLR1 in rice [47]). More recently, N-dependent chromatin modification was shown to regulate tillering in rice [48]. Similarly, phytohormones (e.g. strigolactones, cytokinins, brassinosteroids) have the capacity to modulate the N response and affect different aspects of growth, including sink capacity [49–51].

Studies at the biochemical and physiological level in wheat tend to be conducted on modern varieties, which have been selected for a narrow specification of high yield and sufficient GPC under current agronomic practises including high N availability [10,52]. Changes in the commonly used bread making process meant that lower GPC is required and breeding for greater proportions of high molecular mass glutenin protein rather than gliadin in the grain [53] enabled good bread baking performance at lower overall GPC levels. Modern varieties, therefore, occupy a narrow range of yield and GPC. Yet, there is wide genetic and phenotypic diversity amongst wheat germplasm when landraces and crop wild relatives are included as well as elite varieties [10,11]. Many of these are adapted to low input environments suggesting that interesting alleles that may provide an advantage for growth under low N availability may be available but not yet exploited or incorporated into modern varieties.

As part of the Green Revolution, short straw varieties were selected which provided some level of protection against lodging. These varieties show a reduced N response in terms of shoot architecture (e.g. height, tillering capacity), due to low sensitivity to gibberellins [47,48]. Historic varieties also showed an altered root system architecture compared with modern varieties [54] and it seems that in terms of root plasticity [55], high N response has been selected for [56].

The fact that NUE is much lower for wheat plants grown under high N conditions compared with low N conditions suggests that wheat can be more efficient with the available N and that there is an unattained potential for lowering N requirement whilst maintaining high yield and GPC. Screening collections of landraces such as those held in the Watkins collection [57] for yield and GPC under a range of N conditions may provide a source of functional variation leading to lower N requirement.

In addition to the improvement to the wheat genetic material being used careful consideration of agronomic practices may compound some of the benefits. Studies of single plants growing in pots are very useful to establish N budget on a per plant basis and at varied developmental stages, but wheat plants are cultivated in fields within high-density populations (in the U.K. 350 plants.m⁻² is quite common at establishment). It is well known that plants respond to the presence of neighbours both above and below-ground [58,59]. In particular, increasing planting density can regulate the tiller number and lead to lower N requirements [42] and together with a change in leaf angle can increase NUE (in rice [60]).

In the soil, wheat roots both compete with microbes for N uptake and also depend on these microorganisms for N availability. Wheat plants can manipulate or structure rhizosphere microbial communities, through the production of a myriad of root exudates [61,62]. Different rhizosphere communities can impact wheat yield through the production of hormones [63], and N availability. While efforts in developing cereals that can establish symbiotic association with N₂ fixing microbes are a long-term goal [64], wheat association with endophytic N₂ fixing microbes has shown that accessing alternative sources of N is feasible [65]. Selecting wheat varieties that can optimise their carbon usage to manipulate their associated microbial communities to maximise N availability would provide an avenue to lower the apparent N requirement for wheat production.

Improving the efficiency of N use for wheat production has been a long-term goal, and thus far small incremental steps linked to yield increases have been achieved. Considering wheat N requirement provides new avenues for targeted improvement. Wheat plants with low N requirement should be able to perceive low amounts of N, respond effectively, and continue with a high level of productivity per unit if N supply even as N concentration increases. Overall, we propose that physiological and genetic approaches outlined here hold the potential to optimise the N fertiliser use for wheat production. In addition, Medici et al. [66] showed that in wheat as in other species such as A. thaliana and rice (Oryza sativa), N level controls the phosphate starvation response. This hints at a wider role for N in regulating additional nutrient efficiencies. Lowering N requirement may ultimately provide additional benefits to lowering overall nutrient requirements.

• Application of synthetic nitrogen fertiliser for wheat production has significant economic and ecological cost. Understanding the determinant for low nitrogen requirement can provide new avenues for lowering the dependency on synthetic nitrogen fertilisers.

• N requirements are disproportionally high under high vs. low nitrogen availability suggesting that there is a potential yet unattained for lowering these N requirements.

• Understanding what regulates how responsive wheat plants can be to both external nitrogen availability and internal nitrogen status may lead to lowering nitrogen requirement.

The authors declare that there are no competing interests associated with the manuscript.

N.S.F., A.R.B. and S.M.S. co-wrote the manuscript. N.S.F. and S.M.S. prepared the figure.

Open access for this article was enabled by the participation of University of Cambridge in an all-inclusive Read & Publish pilot with Portland Press and the Biochemical Society under a transformative agreement with JISC.

We thank BBSRC for funding under the Virtual Joint Centres in Agricultural Nitrogen Use Initiative (BB/N013441/1:CINTRIN), and under the GCRF translation award (BB/T012412/1).

2-PG

2-phosphoglycolate

3-PGA

3-phosphoglycerate

GPC

grain protein content

GPD

grain protein deviation

RuBP

ribulose 1,5-bisphosphate