How does nitrogen shape plant architecture?

Nitrogen regulation of the relationship between plant height and tillering and panicle structure occurs at different developmental stages through integration of the synthesis and distribution of multiple phytohormones.


Introduction
Nitrogen (N) is quantitatively the most important mineral nutrient in plants. N is acquired as nitrate (NO 3 − ) and/or ammonium (NH 4 + ) from soil . High-yield cultivation relies heavily on the use of N fertilizers. Excessive application of N fertilizers not only causes energy waste and increases production costs but also aggravates soil acidification and water eutrophication, as well as contributing to greenhouse gas emissions (Guo et al., 2010;Sutton et al., 2011). Therefore, there is an urgent need to breed crop varieties that use N efficiently in order to minimize N input for the sake of sustainable plant production.
For efficient acquisition of N from soil, plants have developed sophisticated regulatory mechanisms of root development and N transport. Several recent reviews (Forde et al., 2014;Giehl et al., 2014;Kiba et al., 2016;O'Brien et al., 2016;Xuan et al., 2017;Taleski et al., 2018;Yang et al., 2019) and two reviews in this Special Issue (Liu et al., 2020;Jia and von Wirén, 2020) have described root adaptation to altered N supply and the prospects for genetically engineering ideal root phenotypes. Meanwhile, the growth and development of aboveground plant parts are systematically regulated by N status (Chen et al., 2016;O'Brien et al., 2016;Xuan et al., 2017). Architectural features that are affected by N, such as plant height, branches, and panicles not only affect yield but also determine N distribution in various organs as well as the efficiency of N use (Hu et al., 2015;Chen et al., 2017 ). Although the basic regulatory mechanisms of plant architecture have been characterized (Wang and Li, 2008;Xing and Zhang, 2010;Wang et al., 2018a), reports concerning the N regulation of plant architecture, particularly by different forms of N at the cellular and molecular levels, are limited and dispersed in the literature. In this review, we summarize how the N supply shapes plant architecture and discuss the possible relationships between plant architecture, growth duration, and N use efficiency (NUE), mainly in cereal crops.

Nitrogen regulation of growth and development in different phases
The hormonal and genetic control of plant architecture, including shoot apical meristem (SAM) activity, axillary meristem formation and elongation, inflorescence structure, and plant height, has been characterized in Arabidopsis, rice, pea, maize, and tomato (Wang and Li, 2008;Xing and Zhang, 2010;Wang et al., 2018a). During developing phases, the plant architecture changes in several aspects, such as stem elongation, branch development, stem and leaf angle, and inflorescence development (Wang et al., 2018a). For cereal crops, the branch number and panicle structure (inflorescence) are two of the most important traits that directly determine the grain yield (Kyozuka et al., 2014). The number of branches (or tillers in rice and wheat) is determined by the initiation of axillary meristem and thereafter via the elongation of axillary buds (Bennett and Leyser, 2006). The mechanism controlling axillary bud outgrowth in apical dominance has been extensively studied; auxin is the main player involved in axillary bud regulation (Teale et al., 2006). The panicle structure is also determined by meristem activity. Floral meristem is the final phase wherein the meristem activity ceases. In grass species, the basic panicle structure is determined by spikelets, the small branches for producing flowers (Itoh et al., 2005;Kellogg et al., 2013). It has been shown that the inflorescence in rice is mainly determined by the floral meristem, which controls the timing of phase transition from the vegetative to the reproductive stage, thereby influencing panicle size (Itoh et al., 2005). Early transition decreases the number and length of branches and panicles, while delayed transition results in more and longer branches as well as larger panicles (Kyozuka et al., 2014).
The plant architecture is greatly influenced by aspects of the growth environment, particularly the duration and intensity of light, and supplies of nutrients and water (Kudoyarova et al., 2015;de Wit et al., 2016;Feng et al., 2016;Wang et al., 2018a). N is one of the major determinants of plant growth and development that affect the major components of plant architecture such as tiller number and panicle structure (Ladha et al., 1998;Zhang et al., 2009Zhang et al., , 2017Tian et al., 2017;Wang et al., 2018a;Yi et al., 2019;Yang et al., 2019). The plant N uptake rate varies during different growth and development stages. In rice, total N accumulation rapidly increases during the vegetative and early reproductive stages, then reaches a plateau before declining slightly during grain filling and ripening (Hashim et al., 2015). The root uptake rate and concentration of N during early growth stages are critical for forming effective tillers. During the ripening stage, the N that is used for grain formation and seed filling is transferred mainly from culms and leaves (Hashim et al., 2015). Therefore, varying the rate of N application at different stages (sowing, tillering, panicle initiation, and heading) can alter the yield components of rice. The N demand for forming effective tillers and for grain filling (weight) may vary among different varieties, probably due to their differences in growth and development (Thu et al., 2014).
Changes in plant architecture in response to N supply may vary among plant species and even accessions of the same species. We have observed natural variation in the response of the different components of plant architecture to N fertilization Fig. 1. Different responses of the major architecture components to nitrogen (N) fertilization in rice. A japonica rice cultivar (cv. 92-10geng) was grown in a paddy field with four different levels of N fertilizer applied and was transferred at the late grain-filling stage into pots for photographing and measurement of the plant structure. (A) The phenotypes of plants supplied with four different N levels. N1, lowest application (75 kg N ha -1 ); N2, low application (150 kg N ha -1 ); N3, moderate application (250 kg N ha -1 ); N4, high application (350 kg N ha -1 ). (B) Plant height and effective tiller number. (C) The shape of entire panicles. (D) Numbers of primary branches, secondary branches, and spikelets. Data in (B) and (D) are mean ±SD (n≥8).Means with different letters are significantly different (P<0.05).
among rice accessions in a core collection grown in a paddy field (data not shown). Nevertheless, N limitation suppresses rice growth, decreases height, and limits tiller number (Fig. 1A, B). Notably, the N demand for maintaining height and tiller number is not the same as that for the branch number of spikelets, and the growth of secondary branches rather than primary branches is sensitive to the supply of N (Fig. 1C, D).
In general, the effect of the level of N supply on plant height can be predicted, whereas its effect on other architecture components, such as tiller number, filled grains per panicle, 1000-grain weight, and grain yield, is complicated. Sufficient N supply stimulates shoot elongation and ensures that the plant will reach the expected height in both rice and wheat (Wu et al., 2020), while excess N prevents secondary cell wall formation, resulting in poor lodging resistance Zhang et al., 2017). The tiller number is affected by the N supply level and growth stage. The effective tiller number can be increased by increasing the N supply to an appropriate level and can be decreased by the excessive application of N (Haque et al., 2016). In rice, N deficiency suppresses bud elongation rather than initiation (Luo et al., 2017). The most critical time for N fertilization for rice grain yield is at the panicle initiation stage (Yoshida et al. 2006). The N supply affects inflorescence development, panicle length, and the number of flowers per panicle (Yoshida et al. 2006;Makino, 2011). In wheat, N accumulation at anthesis was found to be positively correlated with the onset of flag-leaf senescence, and thus total N accumulated at anthesis is an important trait for enhancing grain yield and NUE under low to moderate N supply (Nehe et al., 2018).
For efficient use of light, water, and nutrient resources, it is imperative that plants have the phenotypic plasticity to be able to adapt to varied environmental conditions. Interestingly, super-high-yield rice cultivars show high morphological acclimation in leaf dispersion and orientation to different agronomic practices, including N application . The efficient phenotypic adaptation of rice is coordinated with improved N uptake and assimilation; the shoot photosynthetic productivity of a given rice phenotype is closely and positively related to leaf N concentration and total N accumulation .

Nitrate in the regulation of shoot branching and flowering
The mechanism of N regulation of plant architecture has been partially elucidated during the past decades. In wheat, He et al. (2015) isolated a NO 3 − -inducible and cereal-specific NAC (NAM, ATAF, and CUC) transcription factor, TaNAC2-5A. Limited NO 3 − supply enhances the expression of TaNAC2-5A in shoots and roots. TaNAC2-5A can directly bind to the promoter regions of the genes encoding NO 3 − transporters and glutamine synthetase, consequently enhancing N acquisition and assimilation. Overexpression of TaNAC2-5A can increase tiller numbers, spikelet number, and 1000-grain weight, resulting in higher grain yield (He et al., 2015). In rice, OsMADS57, a MADS-box transcription factor whose expression is enhanced by NO 3 − supply, interacts with TEOSINTE BRANCHED1 (TB1) and targets Dwarf14 (D14) to control the outgrowth of axillary buds Huang et al., 2019a). OsMADS57 can also bind to the CArG motif (CATTTTATAG) within the promoter of OsNRT2.3a that functions in NO 3 − translocation; knockout of OsMADS57 suppresses the distribution of NO 3 − from root to shoot (Tang et al., 2012;Huang et al., 2019a). These results suggest that OsMADS57 may participate in NO 3 − -regulated tiller bud outgrowth of rice plants. Some NO 3 − transporters have been reported to participate in the modulation of plant architecture (tiller number and panicle architecture), mainly through changing N uptake and translocation to reproductive organs. Overexpression of OsNRT2.3b, but not of OsNRT2.3a, increases panicle size parameters including panicle length, number of primary and secondary rachises, number of seeds per panicle, and seedsetting rates under different N treatments (Fan et al., 2016). Chen et al. (2016Chen et al. ( , 2017 reported that overexpression of a high-affinity NO 3 − transporter gene, OsNRT2.1, driven either by the promoter of OsNAR2.1 (encoding a nitrate transport accessory protein) or by its own promoter can increase postanthesis N uptake and translocation from vegetative organs to grains, resulting in greater panicle length and seed set, more grains per panicle, and higher grain yield.
Several members of the NO 3 − and peptide transporter family (NPF) in rice have been characterized with regard to their functions in regulating shoot branching and panicle structure. OsNPF7.7 has two splicing variants, OsNPF7.7-1 and OsNPF7.7-2, that show similar expression responses to N in axillary buds . Enhanced expression of OsNPF7.7-1 and OsNPF7.7-2 increases NO 3 − and NH 4 + influx, respectively, while both OsNPF7.7-1 and OsNPF7.7-2 promote the outgrowth of axillary buds and increase the numbers of tillers, effective panicles, and filled grains per plant, resulting in higher grain yield . In addition, both OsNPF7.1 and OsNPF7.4 function in NO 3 − uptake, but they show opposite expression patterns in axillary buds (Huang et al., 2019b). Overexpression of OsNPF7.1 or knockout of OsNPF7.4 can increase axillary bud outgrowth, especially for the second bud, and subsequently tiller number in rice. Moreover, OsNPF7.2, a low-affinity nitrate transporter, can positively alter cell division in tiller buds to increase tiller number and grain yield (Wang et al., 2018b).
Very recently, the indica allele of the nitrate reductase gene OsNR2, which encodes a NADH/NADPH-dependent nitrate reductase, has been shown to promote NO 3 − uptake via feed-forward interaction with a NO 3 − transporter, OsNRT1.1B, thereby enhancing rice yield potential and NUE . Notably, effective tiller number is increased in Nipponbare plants expressing indica OsNR2 (cv. 9311) and decreased by reduced OsNR2 expression, probably via alteration of the expression of rice OsTB1, a gene controlling tiller bud formation and elongation . The feedforward interaction of OsNR2 and OsNRT1.1B may explain the effect of OsNRT1.1B in altering tiller number in rice (Hu et al., 2015).
Another aspect of nitrate-dependent regulation of plant architecture may be associated with flowering time. N fertilization influences the length of different growth phases and results in varied architecture, even within the same genotype (Leng et al., 2020;Hall et al., 2014). The transition from the vegetative to the reproductive phase is the end of leaf generation on the main stem; this influences the axillary meristem number. The rice branch number is commonly increased by late flowering and decreased by early flowering (Leng et al., 2020). Excess N application commonly causes a delay in the flowering time, resulting in later ripening. In Arabidopsis, the influence of N supply on flowering time has been well characterized (Castro Marín et al., 2011;Vidal et al., 2014). Both extreme deficiency and excess of N result in postponement of flowering time (Lin and Tsay, 2017). SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1) serves as a central integrator for multiple flowering pathways in the SAM (Srikanth and Schmid, 2011). It has been shown that NO 3 acts at the SAM to regulate flowering time, and that SOC1 is required for the regulation of N-dependent flowering (Olas et al., 2019). High nitrate can activate two AP2-type transcription factors, SCHLAFMUTZE (SMZ) and SCHNARCHZAPFEN (SNZ), via the gibberellin (GA) pathway to repress flowering time (Gras et al., 2018). High N levels inhibit the activity of ferredoxin-NADP oxidoreductase (FNR1), leading to the induction of the blue-light receptor cryptochrome 1 (CRY1) and FNR1 nuclear degradation, which act in the N signal input pathway to affect central circadian clock gene expression and flowering time (Yuan et al., 2016). It is an intriguing question whether there are common regulatory pathways of N-dependent alteration of flowering time and plant architecture in addition to their indirect effect on the plant architecture.

Amino acids in the regulation of plant architecture
Plant roots can directly acquire a large portion of NH 4 + and amino acids in addition to NO 3 − in the soil, particularly under highly N-fertilized or irrigated paddy conditions. After absorption, most of the NH 4 + and part of the NO 3 − are assimilated to amino acids in the roots; therefore, the transportation and distribution of N inside plants occur mainly in the form of amino acids . Although NH 4 + could regulate root architecture (Liu and von Wirén, 2017), it is not clear whether NH 4 + itself can directly regulate shoot architecture; however, several studies have shown that some transporters and synthetases of amino acids are directly involved in the regulation of plant architecture.
Amino acids are essential components of plant metabolism, not only as constituents of proteins but also as precursors of important secondary metabolites and as carriers of organic N between the organs of the plant (Dinkeloo et al., 2018). Amino acids in the roots are transported from the cortex or endodermis cells to the vasculature to circumvent the Casparian strip and then translocated to the aboveground tissues (Dinkeloo et al., 2018;Tegeder and Masclaus-Daubresse, 2018). Amino acid transporters play roles in amino acid uptake by roots, xylem and phloem loading, xylem-phloem transfer, and intracellular transport (Dinkeloo et al., 2018). It has been recently shown that amino acid transporters function in the regulation of tiller growth and the entire plant architecture in addition to altering N distribution and NUE. In rice, blocking amino acid permease 3 (AAP3) can stimulate bud outgrowth and effective tiller number, leading to higher grain yield; in contrast, overexpressing OsAAP3 results in an enriched amount of amino acids and inhibition of bud outgrowth . Amino acid permease 5 (OsAAP5) can also affect tiller number and grain yield through the regulation of cytokinin (CK) biosynthesis .
The synthesis of amino acids may be involved in regulating the plant architecture. We have shown that mutation of asparagine synthetase 1 (ASN1) in rice decreased the concentration of asparagine, while total N was unchanged (Luo et al., 2019). Knockout of OsASN1 suppressed tiller bud outgrowth and tiller number, suggesting that OsASN1 is involved in the regulation of rice development (Luo et al., 2019). Rice cytosolic glutamine synthetase OsGS1;2, which is involved in the primary assimilation of NH 4 + in roots, is also involved in the regulation of plant development (Funayama et al., 2013;Ohashi et al., 2015Ohashi et al., , 2017. It has been demonstrated that OsGS1;2 contributes to tiller bud outgrowth by regulating N-dependent CK biosynthesis (Ohashi et al., 2017).
In addition to their functions as basic compounds for growth and development, amino acids may function as signaling molecules (Dinkeloo et al., 2018). For example, serine acts as a signal in brain tissue and in mammalian cancer cells. The biosynthesis of serine is highly active and restricted to proliferating cells of the primary meristem (Häusler et al., 2014). Serine in the meristems has been suggested to regulate targets of rapamycin signaling in plants (Benstein et al., 2013;Cascales-Minana et al., 2013;Menand et al., 2002). Moreover, it is very likely that there are amino acid transceptors that are involved in the regulation of plant development (Dinkeloo et al., 2018).

Nitrogen-regulated small signaling peptides in plant development
Small signaling peptides have been identified in plant cell-tocell communication and plant growth regulation (Czyzewicz et al., 2013;Tavormina et al., 2015;Oh et al., 2018). Two small peptide families, CLAVATA3/EMBYO SURROUNDING REGION (CLE) and C-TERMINALLY ENCODED PEPTIDE (CEP), function in both local N-status-dependent signaling and systemic N signaling (Okamoto et al., 2016;de Bang et al., 2017;Taleski et al., 2018). CEPs, which are widely distributed among seed plants, are expressed in N-starved root parts and transported to the shoots, where they bind to the CEP receptor (CEPR) (Tabata et al., 2014). This signal triggers the expression of class III glutaredoxins as mobile signals transported from shoots to roots via phloem to induce the expression of the NO 3 − transporter NRT2.1 in the N-replete portion of the roots (Ohkubo et al., 2017).
Small signaling peptides have been identified to play important roles in the regulation of root morphology, while their functions in regulating shoot architecture have received less attention. In Arabidopsis, perturbed CEP expression leads to changes in plant height and leaf shape (Roberts et al., 2013). The CEP genes show different functions in regulating the shoot and root response to the growth conditions tested (Delay et al., 2013). For example, the overexpression (ox) line of CEP2 (CEP2ox) shows fewer rosette leaves, delay of flowering, and alteration of leaf morphology in comparison to the wild type (WT); CEP3ox and CEP4ox display a similar phenotype characterized by epinasty, leaf yellowing, and reduced rosette size; CEP6ox and CEP9ox show milder changes. These results indicate that CEPs may interact with different receptors and may play distinct roles in shoot development (Delay et al., 2013).
In rice, there are 17 OsCEP genes, and OsCEP6.1ox also has negative effects on rice shoot development (Sui et al. 2016). Compared with the WT line, OsCEP6.1ox transgenic lines exhibit reduced height, lower tiller number, shorter panicle length and smaller seed size (Sui et al., 2016). Further functional analysis demonstrated that the regulatory activity of CEPs on panicle development may be related to the alteration of cell size but not cell number (Sui et al., 2016). However, the downstream signaling components of OsCEPs and their response to N status remain largely undetermined in rice.

Nitrogen regulation of cell division and expansion for shaping architecture
The regulation of the cell cycle by N was reported decades ago. Limited N supply suppresses DNA synthesis, cell division, cell growth, and bud growth at similar rates (Rivin and Fangman, 1980). Long-term N starvation results in the cessation of cell division and associated growth of branches in rice (Luo et al., 2017). Increasing N supply levels accelerate cell division and expansion, resulting in greater biomass accumulation. Notably, the effect of the form of N supply on cell division and expansion is not significant in the short term. In Arabidopsis, the provision of either NO 3 or NH 4 + causes the same effects on shoot branching (de Jong et al., 2014). In rice, shoot branching is influenced significantly by the concentration but not the form of N during the vegetative stage (Luo et al., 2017). The NO 3 − supply level influences the synthesis and distribution of CKs and their downstream transcription factors, which further regulate cell division in plants (Landrein et al., 2018). Macadam et al. (1989) found that high N fertilization increased the rate of division of mesophyll cells and increased epidermal cell elongation of tall fescue leaf blades. In pea, the expression levels of cell-cycle-related genes (PCNA, cyclinB, cdc2, and histone H4) are enhanced in axillary tiller buds when the buds grow (Devitt and Stafstrom, 1995;Shimizu and Mori, 1998). N deficiency resulted in the dormancy of tiller buds, probably via altering the expression of cell-cycle-related genes (Luo et al., 2017). However, it is not known whether the suppression of cell division in response to N deficiency is a direct effect of N or rather results from a signal that is transmitted to the tiller bud.
Amino acids may also influence cell division. In human tumor cells, asparagine was found to be an important regulator of amino acid homeostasis, anabolic metabolism, and proliferation (Krall et al., 2016). The loss of function of asparagine synthetase (ASNS) resulted in the suppression of cell proliferation and inhibition of tumor growth in human gastric cancer cells, melanoma cells, and epidermoid carcinoma cells Yu et al., 2016). Silencing of ASNS arrested cell cycle progression at the G0/G1 phase, probably through regulation of the expression of cell cycle molecules such as CDK2 and cyclin E1 (Miao et al., 2013). These results all shed light on the possible relationship between amino acids and cell division in plants. Conducting such studies may uncover new mechanisms involved in the control of shoot architecture by N.
Nitrogen regulation of phytohormone synthesis and distribution for shaping architecture Plants integrate internal systemic signals, such as hormones, that provide information on the N status of organs to finely adjust the growth and development of shoots and roots (Wang et al., 2018a). Among these phytohormones, auxin, CKs, strigolactones (SLs), and GAs are of vital importance for regulating plant architecture. The dynamic balance between cell division and cell differentiation controls organ shape and size.

Auxin
Fluctuation of the N supply has a significant effect on auxin distribution. A decrease in N supply commonly increases indole-3-acetic acid (IAA) accumulation in the root of plants including Arabidopsis, soybean, durum wheat, and maize (Caba et al., 2000;Walch-Liu et al., 2006;Tian et al., 2008). The establishment of auxin distribution within plant tissues constitutes its function in plant morphogenesis, and this mainly depends on the function of auxin efflux facilitators of the PIN-FORMED (PIN) family (Friml et al., 2003;Woodward and Bartel, 2005).
Plant N status may also be related to auxin synthesis and/ or distribution in aboveground parts. For example, decreased NO 3 − supply to rice can down-regulate the expression of multiple OsPIN genes and decrease 3 H-IAA transport from shoots to roots, resulting in increased IAA content in the youngest leaves and decreased IAA content in the shoot base and roots (Sun et al., 2014). In addition, the provision of a mixture of NO 3 − and NH 4 + , in comparison to a single form of N, increases the IAA concentration in leaves and roots and increases the expression of both OsAUX1 and OsPIN genes (Song et al., 2011).
The molecular mechanisms of N-induced auxin distribution in shaping shoot architecture are still obscure. We have shown that N deficiency inhibits the expression of seven OsPINs (OsPIN1b/1c/2/5a/5b/9/10a) in the roots of rice (Sun et al., 2014). Since rice tiller numbers are increased by overexpression of OsPIN2 and OsPIN3 (Chen et al., 2012;Zhang et al., 2012) and are decreased by knockdown of OsPIN10a , these PIN members may be involved in N regulation of tiller bud outgrowth. However, transgenic plants overexpressing OsPIN1 and OsPIN5b show inverse aboveground phenotypes (Xu et al., 2005;Lu et al., 2015), probably due to the disturbance of auxin-mediated bud inhibition or other secondary messengers such as CKs (Barbier et al., 2019). It should be noted that auxin synthesis and distribution influenced by N supply largely depend on the plant species and N status; therefore, more evidence is required from future studies for elucidating the complex regulatory pathways of auxin for shaping N-controlled shoot architecture.

Cytokinins
Increasing evidence indicates that elevated CK content restricts root growth and promotes shoot growth, influencing plant height, shoot branching, flowering, and seed production . It has been shown that both the biosynthesis and distribution of CKs are closely linked to N availability during shoot and root development. Increasing NO 3 − supply to barley roots can rapidly stimulate the biosynthesis and acropetal transportation of zeatin riboside (ZR), a naturally occurring CK, while NH 4 + has less effect than NO 3 − on the increase of ZR (Samuelson and Larsson 1993). In rice, NO 3 − supply increases the concentrations of six CK forms in xylem sap, as well as leading to their high accumulation in both roots and leaves (Song et al., 2013). Notably, pretreatment with either nitrate reductase or glutamine synthetase inhibitor can prevent the NO 3 − -simulated increase of ZR level in barley (Samuelson and Larsson 1993), while a decrease of total N concentration in tiller buds reduces the active CK content (Liu et al., 2011;Ohashi et al., 2017). These results suggest that entire N status or N assimilation, rather than NO 3 − or NH 4 + alone, determine the synthesis and distribution of CKs.
Plants possess multiple regulatory pathways of N-dependent CK biosynthesis to modulate growth. The CK synthesis gene IPT3, encoding adenosine phosphateisopentenyltransferase, may play a critical role in mediating NO 3 − -induced CK synthesis in Arabidopsis and rice plants (Takei et al., 2004;Song et al., 2013) and possibly in N-controlled plant architecture. IPT3 is regulated by inorganic N sources in a NO 3 − -specific manner. Miyawaki et al. (2006) have shown that the phenotype of dramatically reduced shoot apical meristems and short, thin aerial shoots of atipt3/5/6/7 mutants can be complemented by expressing IPT3. Remarkably, IPT3 was mainly regulated by NO 3 − , and ipt3 mutants failed to sense NO 3 − signals to produce CKs (Takei et al., 2004). NO 3 − -induced expression of IPT3 is partly dependent on NRT1.1/CHL1 (Liu et al., 1999;Ho et al., 2009;Wang et al., 2009;Kiba et al., 2011). Recent findings indicate that a transcriptional regulatory system, NLP/NIGT1, controls IPT3 and CYP735A gene expression in Arabidopsis in response to NO 3 − (Maeda et al., 2018). Nevertheless, the function of IPT3 in regulating N-controlled plant architecture at different developmental stages still needs to be investigated. In addition to IPT3, other IPT members may also be involved in N-regulated CK biosynthesis. In rice, glutamine or a related metabolite rather than NO 3 − or NH 4 + can enhance the expression of OsIPT4, OsIPT5, OsIPT7, and OsIPT8, with accompanying accumulation of CKs. Repressing the expression of OsIPT4, the dominant IPT in rice roots, significantly reduces the N-dependent increase of CKs in the xylem sap and retards shoot growth despite a sufficient N supply (Kamada-Nobusada et al., 2013).

Strigolactones
SLs have been identified more recently as a group of plant hormones that modulate plant architecture (Umehara et al., 2008;Sun et al., 2014;Barbier et al., 2019). The function of SLs in altering shoot architecture, including involvement in plant stature, axillary tiller bud outgrowth, and tiller angle, has been partially characterized (Seto and Yamaguchi, 2014;Sang et al., 2014). Small sections of stem tissue are able to supply sufficient SLs to inhibit branching in mutant shoots that are unable to synthesize SLs (Dun et al., 2009), suggesting that SLs may act at very low concentrations.
Enhancement of the biosynthesis and exudation of SLs by N deficiency has been observed in several plant species (Yoneyama et al., 2007(Yoneyama et al., , 2012Xie et al., 2010;Sun et al., 2014). In sorghum plants, limited N or phosphorus largely increases the amount of 5-deoxystrigol in the root exudates (Yoneyama et al., 2007). In rice, N deficiency results in high endogenous SLs and degradation of D53 protein, a key repressor in the SL signaling pathway-the same effect as that caused by exogenous supply of the SL analogue GR24 (Sun et al., 2016). These results clearly demonstrate that SLs are involved in N-regulated rice development. However, the effect of N deficiency on the synthesis of SLs depends on the plant type and experimental conditions (Yoneyama et al., 2007(Yoneyama et al., , 2012Sun et al., 2014). For example, SL contents in the roots of red clover and alfalfa are not significantly affected by altering the N supply (Yoneyama et al., 2012).
The effect of N supply on branching in Arabidopsis is comparable between WT and mutants of SL biosynthesis (max1 and max3) and signaling (max4) (de Jong et al., 2014). Even though N limitation reduces branching in both SL mutants and WT, the mutants still produce more secondary shoots than WT under the same N-limiting condition. These results suggest that the ability to maintain N-regulated branching in Arabidopsis is at least partially dependent on SLs. In rice, the SL signaling gene D53 can repress ideal plant architecture 1 (IPA1), a key regulator of architecture, thereby functioning as a downstream transcription factor (Song et al., 2017). Thus, it is an intriguing question whether there are other targets that bind to D53 in rice plants and, if so, whether the target genes, including IPA1, are involved in SL participation in N-regulated plant development.

Gibberellins
GA is involved in the regulation of the inverse relationship between plant height and tiller number. Exogenous application of GA reduces tiller number in cereal plants (Zhuang et al., 2019). GA promotes plant height by stimulating the degradation of the DELLA protein SLR1 (SLENDER RICE 1) (Murase et al., 2008;Sasaki et al., 2003;Liao et al., 2019). Since the tiller number regulator MONOCULM 1 (MOC1) relies on binding to SLR1 to avoid degradation, GAs trigger both the degradation of SLR1, leading to stem elongation, and the degradation of MOC1, leading to a lower tiller number (Liao et al., 2019).
In current commonly cultivated reduced height (Rht) wheat, DELLAs are resistant to GA-stimulated destruction (Peng et al., 1999), whereas the semi-dwarfism rice sd1 allele reduces the abundance of bioactive GA (Itoh et al., 2002;Asano et al., 2011). Notably, growth-regulating factor 4 (GRF4) can bind to GRF-interacting factor 1 (GIF1) and activate the genes related to N uptake and assimilation, while DELLA protein inhibits the binding of GRF4 to GIF1; DELLA protein accumulation thus inhibits growth and N uptake and assimilation in rice and wheat (Li et al., 2018b). Moreover, N stimulation of tillering in rice is regulated by N-mediated tiller growth response 5 (NGR5) (Wu et al., 2020), an APETALA2 (AP2)-domain transcription factor previously known as SMOS1 (SMALL ORGAN SIZE1) and RLA1 (REDUCED LEAF ANGLE1) (Aya et al., 2014;Hirano et al., 2017;Qiao et al., 2017). NGR5 is a target of the GA receptor GID1; thus, NGR5 abundance is negatively associated with GA level. Mutation of NGR5 results in the insensitivity of tillering number to N supply. NGR5 regulates N-promoted H3K27me3 modification by recruiting PRC2 (POLYCOMB REPRESSIVE COMPLEX 2) to methylate the sites of D14 (encoding Dwarf14, an SL receptor protein) and OsSPL14 (encoding SQUAMOSA PROMOTER BINDING PROTEIN LIKE-14) and other tillering inhibition genes. Thus, in response to N supply, NGR5 inhibits the expression of the shoot-branchinginhibitory genes D14 and OsSPL14 and promotes tillering in rice (Wu et al., 2020).

Perspectives
N fertilization in the field has primary effects on plant growth and development. Based on the most recent findings, we have drawn an outline of N regulatory pathways in altering flowering time, shoot branching, and panicle size under varied NO 3 − and/or NH 4 + supply (Fig. 2). The genes directly or indirectly involved in the N regulation of plant architecture are summarized in Table 1. It should be noted that the N regulation of different components of plant architecture and yield is affected by environmental conditions and agricultural practices. The interaction effects of planting density and N fertilization on architecture and yield are worth further investigation from both physiological and molecular genetic perspectives. In addition, the relationship between N-regulated growth duration, suppresses the expression of ferredoxin-NADP oxidoreductase 1 (FNR1), which modulates cryptochrome 1 (CRY1) phosphorylation and delays flowering (Yuan et al., 2016). The transcription factors SCHLAFMUTZE (SMZ) and SCHNARCHZAPFEN (SNZ) can be activated by NO 3 via the gibberellic acid (GA) pathway to suppress flowering (Gras et al., 2018). The flowering time can directly or indirectly alter branching and panicle structure. In addition, N supply rapidly stimulates cytokine (CK) biosynthesis and acropetal transportation (Liu et al., 2011;Ohashi et al., 2017). Branching is regulated by the varied distribution of auxin, which is regulated by the expression of members of the PIN-FORMED family of auxin efflux transporters (PINs) via CKs and NO 3 - (Sun et al., 2014). The biosynthesis of strigolactones (SLs) is suppressed by a sufficient N supply, resulting in the outgrowth of branching (Yoneyama et al., 2007(Yoneyama et al., , 2012Xie et al., 2010). Moreover, the expression of indica-type nitrate reductase 2 (NR2; indicated with an asterisk) and the nitrate transporter 1.1B (NRT1.1B) can be induced by NO 3 and enhances the absorption of NO 3 and the regulation of branching (Hu et al., 2015;Gao et al., 2019). Expression of the nitrate transporter 2.1 (NRT2.1) gene and putative nitrate-peptide transporter family genes (NPFs) can enhance panicle size and branching (Chen et al., 2016;Huang et al., 2018Huang et al., , 2019b. Amino acids (AAs) influence shoot branching. The putative amino acid permease AAP3 can suppress branching, while AAP5 can alter branching by regulating the CK level Wang et al., 2019b). Glutamine synthetase (GS), such as OsGS1.2, and asparagine synthetase (AS) can mediate the synthesis of CKs for the regulation of branching (Funayama et al., 2013;Ohashi et al., 2015Ohashi et al., , 2017Luo et al., 2019). CRY1-P, phosphorylated CRY1; D14, Dwarf14 (an SL receptor); FT, flowering locus T; NGR5, nitrogen-mediated tiller growth response 5; SPL14, squamosa promoter binding proteinlike 14. Arrows represent enhancement of downstream target activity. Lines with a horizontal bar at the end represent suppression of downstream target activity. Dashed lines indicate that the evidence for the regulation is not strong. Phytohormones are highlighted in yellow, transporters in green, enzymes in blue, and transcription factors in orange. plant architecture, and NUE should be further investigated in crop production studies. The most important traits that influence the yield are the number of branches, panicle number, and seed size, while other aspects of plant architecture such as height, tiller angle, and leaf angle are also important for plant growth. Therefore, the trade-off among different architecture components regulated by different forms and concentrations of N should be considered for both high yield and NUE. Since the concept of "ideotype" was first put forward, the influence of environmental factors, including N fertilization, on the ideal plant architecture has received much less attention than expected, and has not been characterized in detail. To sustain the highest yield potential of the cultivars with ideotype, the architecture is expected not to be largely altered by varied N supplies in the field. Therefore, revealing the N-dependent mechanisms modulating plant architecture is helpful for molecular breeding of the ideotype with high NUE.
The influence of N fertilization on plant architecture can be monitored in real time at different scales in the field by the use of recently developed unmanned aerial vehicle (UAV)-based active canopy sensors. The modern UAV technique for providing phenotypic data shows great applicability and flexibility in the estimation of crop N status and in the analysis of plant architecture (Zaman-Allah et al., 2015;Watanabe et al., 2017;Elsayed et al., 2018;Li et al., 2018a;Buchaillot et al., 2019;Cen et al., 2019;Lu et al., 2019). As improvement of these real-time monitoring and data-modeling techniques continues, the remote-sensing technology may be extensively applied in the future to predict the responses of the plant, including plant architecture, to N application.
To better understand the direct N regulatory pathways affecting plant architecture, identification of key quantitative trait loci and the genes controlling N-sensitive or -insensitive responses of certain components of plant architecture is expected in the future. Principal component analysis in genomewide association studies is an effective means of extracting key information from phenotypically complex traits, and has been performed for analyzing rice architecture (Yano et al., 2019). This method has been broadly used for the analysis of N-related phenotyping in some other crops (Zhang et al., 2015;Monostori et al., 2017;Nigro et al., 2019;Steketee et al., 2019) and it can be applied to isolate the key genes involved in the N regulation of plant architecture.