Arabidopsis nitrate-induced aspartate oxidase gene expression is necessary to maintain metabolic balance under nitrogen nutrient fluctuation

Nitrate is a nutrient signal that regulates growth and development through NLP transcription factors in plants. Here we identify the L-aspartate oxidase gene (AO) necessary for de novo NAD+ biosynthesis as an NLP target in Arabidopsis. We investigated the physiological significance of nitrate-induced AO expression by expressing AO under the control of the mutant AO promoter lacking the NLP-binding site in the ao mutant. Despite morphological changes and severe reductions in fresh weight, the loss of nitrate-induced AO expression resulted in minimum effects on NAD(H) and NADP(H) contents, suggesting compensation of decreased de novo NAD+ biosynthesis by reducing the growth rate. Furthermore, metabolite profiling and transcriptome analysis revealed that the loss of nitrate-induced AO expression causes pronounced impacts on contents of TCA cycle- and urea cycle-related metabolites, gene expression profile, and their modifications in response to changes in the nitrogen nutrient condition. These results suggest that proper maintenance of metabolic balance requires the coordinated regulation of multiple metabolic pathways by NLP-mediated nitrate signaling in plants.

N itrate, one of the main sources of nitrogen (N) for plants, plays an important role in the regulation of gene expression, metabolism, growth, and development 1,2 . Application of nitrate to N-starved plants induces various responses, collectively referred to as nitrate responses, which include the promotion of nitrate assimilation 3 , increase in lateral root growth 4 , changes in the shoot to root weight ratio 5 , and induction of the expression of N assimilation-related genes 2 . Furthermore, genome-wide studies revealed that nitrate induces the expression of not only N assimilation-related genes but also diverse genes involved in other metabolic pathways, such as the pentose phosphate pathway 6 . However, with the exception of N assimilation-related genes, the physiological significance of nitrate-induced expression of metabolic enzyme genes has not yet been elucidated.
NODULE INCEPTION-LIKE PROTEIN (NLP) transcription factors with a conserved N-terminal region and the RWP-RK DNA-binding domain are the primary transcription factors regulating nitrate-responsive gene expression in plants [7][8][9][10] . NLPs are post-translationally activated through nitrate-induced phosphorylation of a serine residue in the conserved N-terminal region 11 and then accumulate in the nucleus 8 , where they interact with nitrate-responsive cis-elements (NREs) 10,12 in target gene promoters to activate their expression. Of the nine NLPs in Arabidopsis, NLP7 plays a dominant role in regulating genes involved in nitrate utilization 13,14 , while NLP6, most closely related to NLP7, supports the role of NLP7 15 . Repression of the activity of the entire NLP family by chimeric repressor gene-silencing technology has been shown to diminish most of the nitrateinduced gene expression 7,10 . Furthermore, genome-wide analyses revealed the binding of NLP7 to the vicinity of numerous nitrateinducible genes in the nucleus 8,16 .
NAD(H) and NADP(H), collectively referred to as pyridine nucleotides, are ubiquitous coenzymes required for a wide range of redox reactions in all living organisms 17 , because they deliver electrons via the reversible conversion of NAD + and NADP + to NADH and NADPH, respectively. NAD(H) is a redox carrier in the tricarboxylic acid (TCA) cycle and mitochondrial electron transport chain (ETC) in both plants and animals, whereas NADP(H) is a redox carrier in the pentose phosphate pathway and Photosystem I in plants. Thus, sufficient pools of NAD(H) and NADP(H) are necessary for maintaining the appropriate redox statuses in the cytosol, plastids, and mitochondria of plant cells 18 . In fact, partial defects of ETC have been shown to cause abnormalities in the redox status and reduction in N assimilation efficiency in Nicotiana sylvestris 19,20 . Defects in transporters comprising the malate valve, which delivers the reducing power of NADPH from the plastid to the cytosol in the form of malate, have also been shown to lower the efficiency of both carbon (C) and N metabolism, leading to the depletion of amino acids and accumulation of nitrate and organic acids 21 . Besides its primary role as a redox carrier, NAD + also acts as a substrate in protein poly-ADP ribosylation and deacetylation, alternative RNA cap formation, and pathogen-induced cell death signaling [22][23][24][25][26] . Thus, NAD(H) content may also be associated with various physiological processes in a redox-independent manner 18,27 .
In the de novo NAD + biosynthesis pathway in plants, aspartate is oxidized by aspartate oxidase (AO) to form iminoaspartate, which is converted to quinolinate by quinolinate synthase (QS) 28 . Then, quinolinate is converted into nicotinate mononucleotide (NaMN) by quinolinate phosphoribosyltransferase (QPT) 29 and subsequently to nicotinate adenine dinucleotide by nicotinamide mononucleotide adenyltransferase (NMNAT) 30 , finally leading to the synthesis of NAD + by NAD synthase (NADS) 31,32 (Supplementary Fig. 1). The AO-catalyzed production of iminoaspartate is likely the rate-limiting step in de novo NAD + biosynthesis because the NAD + content per unit tissue fresh weight increased in Arabidopsis plants overexpressing AO 33 and QPT overexpression enriched NAD + content only when supplied with quinolinate 34 . Besides this de novo pathway, the salvage pathway regenerates NaMN via a three-step process composed of the degradation of NAD + to form nicotinamide by poly (ADPribose) polymerase and sirtuins, deamidation of nicotinamide to nicotinate by nicotinamidase and conversion of nicotinate into NaMN by nicotinate phosphoribosyltransferase 35,36 (Supplementary Fig. 1). Thus, AO, QS, and QPT are exclusively responsible for de novo NAD + biogenesis, while NMNAT and NADS are involved in both de novo NAD + biosynthesis and NAD + salvage pathways ( Supplementary Fig. 1). Because AO, QS, and QPT function in plastids, whereas NMNAT functions in the cytosol 37,38 , it is assumed that the rate of NAD + biogenesis is regulated in the plastid, while NAD + synthesized in the cytosol is transported to various subcellular organelles 39,40 . AO, QS, QPT, and NMNAT are encoded by single-copy genes in Arabidopsis 31 , and knockout mutations in AO, QS, QPT, and NMNAT genes cause embryonic lethality 37,41 . On the other hand, NADP + and NADPH are generated by the phosphorylation of NAD + and NADH, respectively, by NAD phosphotransferases (NADK1 and NADK2) and NADH phosphotransferase in Arabidopsis 42,43 . Analysis of nadk2 knockout mutants and NADK2 overexpression lines shows that the pool size of NADP + and NADPH affects the regulation of amino acid metabolism and the Calvin cycle 44 .
In this study, we identified the AO gene as a direct target gene of the NLP family of transcriptional activators. Because the AOcatalyzed reaction is the key step in NAD + biogenesis, we further explored the physiological significance of NLP-mediated nitrate induction of AO expression. The results suggest that nitrateinduced expression of AO plays a critical role in maintaining metabolic balance and growth in response to N nutrition, emphasizing the importance of the coordinated regulation of multiple metabolic pathways by nitrate signaling.

Results
NLP-mediated and nitrate-induced AO expression in Arabidopsis. Our previous microarray data of NLP6-SUPRD transgenic lines, in which nitrate-induced expression of NLP target genes was diminished by expressing a dominant-negative chimeric repressor 7,10 , suggested that the expression of AO and QS is induced by nitrate-activated NLPs. Therefore, we performed a time-course analysis of AO and QS expression in Arabidopsis seedlings (Fig. 1a). Consistent with the microarray data, the expression of AO and QS was transiently induced by 10 mM KNO 3 but not KCl (control), although the induction was more pronounced for AO than for QS. Furthermore, the nitrateinduced AO expression was reduced in both NLP6-SUPRD transgenic lines and the nlp6 nlp7-1 double mutant (Figs. 1b, c). These results suggested that NLP6 and NLP7 directly or indirectly regulate nitrate-inducible AO expression, although a more evident reduction in the NLP6-SUPRD lines implied that other NLPs likely played redundant roles. The presence of cycloheximide, a protein synthesis inhibitor, did not affect the nitrateinduced expression of AO but induced the expression of IAA2 (a treatment control) 45 (Fig. 1d). Since NLPs are activated through nitrate-induced phosphorylation, de novo protein synthesis is unnecessary for increases in NLP activity. Therefore, this observation suggested that AO is directly regulated by NLPs. Consequently, AO was identified as a direct NLP target gene candidate.
NLP7 binds to an NRE in the AO gene promoter to induce its expression. To examine possible nitrate-and NLP-dependent activation of the AO promoter, co-transfection assays with ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-022-03399-5 protoplasts from N-starved plants were performed using the effector, reporter, and control plasmids (Figs. 2a, b). A 2137 bpgenomic sequence upstream of the transcription start site was employed as the AO promoter, considering that most functional cis-elements exist within 500 bp upstream of transcriptional start sites in the case of Arabidopsis genes 46 . In fact, this promoter succeeded in driving AO expression sufficient to complement the phenotype of the ao mutant, as described later. Reporter plasmids were constructed with the AO promoter and its derivatives truncated at approximately 300 bp intervals to facilitate the identification of NLP-binding sites. Protoplasts from N-starved plants co-transfected were incubated in the presence of 1 mM KNO 3 or KCl (control). The results indicated that NLP7 activated the AO promoter truncated at position −504 (relative to the transcription start site) and other longer truncations of the AO promoter in the presence, but not in the absence, of 1 mM KNO 3 . However, the AO promoter truncated at −164 was activated neither by the nitrate treatment nor by NLP7 (Fig. 2b), suggesting that nitrate-activated NLP7 activated the AO promoter by binding to a site located between −504 and −164.
To precisely identify the NLP-binding site(s), an electrophoretic mobility shift assay (EMSA) was performed using recombinant NLP7 and a variety of DNA probes corresponding to the sequence between −504 and −164 of the AO promoter (Fig. 2c, d). Only probe P2 produced a shifted band, as did an NRE-containing fragment from the NIR1 promoter 12 (positive control) (Fig. 2c). Since probe P2 contained three DNA sequences resembling the consensus NLP-binding sequence, TG(A/G)C(C/T)CTT or its reverse complement AAG(G/A)G(T/C)CA 10 , EMSA was further performed with three mutant probes (Mut1-3) harboring nucleotide substitutions in any of the three putative NLP-binding sites (Fig. 2d). NLP7 bound to Mut2 and Mut3 probes, but not to the Mut1 probe, thus identifying the TGGCCGTT sequence (−354 to −347 bp in the AO promoter) as an NLP-binding site. The relevance of the identified NLP-binding site to the nitrate-inducible expression of AO was investigated by co-transfection assays (Fig. 2e). Unlike the wild-type AO promoter, the mutant AO promoter (harboring the same mutations as the Mut1 probe) was barely activated by nitrate when NLP7 was not overexpressed by co-transfection of the NLP7 effector plasmid. Although overexpressed NLP7 caused small nitrate-dependent activation of the mutant AO promoter, the fold induction was much lower with the mutant AO promoter than with the wild-type AO promoter. Because the mutant AO promoter was not evidently activated in transgenic plants as described later, this small activation was probably due to a minor interaction between sequences somehow resembling the NLP-binding site and NLP7 overexpressed in the transient assay system. These results indicated that nitrateactivated NLP7 upregulates the AO promoter-dependent transcription by interacting with the TGGCCGTT sequence located between −354 and −347.
Effects of the loss of nitrate-dependent AO expression on plant growth and morphology. To determine the physiological significance of nitrate-induced AO expression, AO was expressed in the ao null mutant (fin4-2, SALK_013920) 47 under the control of the wild-type AO promoter or mutant AO promoter with mutations on the identified NLP-binding site (Fig. 3a). Because the ao null mutant shows embryonic lethality 37 , plants heterozygous for the ao null allele were transformed. Then, T3 progenies homozygous for both the introduced AO gene and the ao allele were selected. The generated transgenic plants showed comparable or slightly higher levels of AO transcripts in the absence of nitrate. Since the transgenic lines whose T2 progenies showed a 3:1 segregation ratio of the selection marker gene were selected (see Methods), the slight differences in AO expression levels in transgenic lines could be due to multiple tandem insertions at a single gene locus or chromosomal position effect. On the other hand, nitrate-induced AO expression was clearly detected in all three independent transgenic lines generated using the wild-type AO promoter (AOpro(wt)-AO plants), while it was barely detectable in the three independent lines generated with the mutant AO promoter (AOpro(mut)-AO plants) (Fig. 3b). These results revealed that the identified NLP-binding site is a cisacting element responsible for nitrate-inducible AO expression in planta.
Phenotypic analysis of the generated transgenic lines revealed that AOpro(mut)-AO plants show growth defects and morphological abnormalities (Fig. 4, Supplementary Fig. 2). Unlike AOpro(wt)-AO plants, AOpro(mut)-AO plants showed decreases in shoot fresh weight and shorter primary root length (about 50% and 70%, respectively, compared to wild-type plants) (Fig. 4c, d). , and in the wild-type (WT) and nlp6 nlp7-1 double mutant seedlings (c). Asterisks indicate significant differences (Welch's t-test; *P ≤ 0.05, **P ≤ 0.01). d Nitrate-induced expression of AO in plants treated with cycloheximide (CHX). In (a), 4-dayold WT seedlings were treated with 10 mM KNO 3 for the indicated time period. In (b-d), seedlings were treated with 10 mM KNO 3 or KCl (control) for 1 h. In (d), seedlings were treated with 100 µM CHX for 1 h before nitrate or control treatment. IAA2 was used as a positive control gene for the CHX treatment. All seedlings were grown under continuous light. Data represent mean ± standard deviation [SD; n = 3 in (a, d) and n = 5 in (b, c)]. Fig. 2 Nitrate-dependent activation of the AO promoter through the specific binding of NLP7. a Effector and internal control plasmids. The effector plasmid contains NLP7 cDNA cloned between the modified 35 S promoter (35Spro) and the NOS terminator (NOS), while the empty vector contained no DNA insert. The plasmid carrying the GUS gene under the control of the UBQ10 promoter (UBQ10pro) served as an internal control. b Co-transfection assay. Protoplasts from N-starved plants were co-transfected with a reporter plasmid (expressing the luciferase gene (LUC) under the control of a truncated fragment of the AO promoter), NLP7 expression vector or the empty vector, and the internal control plasmid, and then incubated in the presence of 10 mM KNO 3 or KCl (control). The transcription start sites are indicated by arrows. c, d Electrophoretic mobility shift assay (EMSA). DNA probes corresponding to different regions of the AO promoter (P1-P5) in (c) and the WT and mutant P2 probes in (d) were incubated in the presence (+) or absence (-) of recombinant NLP7. The mutant P2 probes (Mut1, Mut2, and Mut3) contained nucleotide substitutions (as indicated). A DNA fragment from the NIR1 promoter was used as a positive control. Black arrowheads indicate retarded bands. e Activation of the AO promoter, depending on both the NLP7binding site and NLP7. Reporter plasmids contained the WT or mutated AO promoter. The AO promoter truncated at -1,015 was used as a negative control. Nucleotide substitutions in the mutated AO promoter were the same as those in the Mut1 probe. In (b) and (e), data represent mean ± SD (n = 3), and relative LUC activity obtained from protoplasts co-transfected with the reporter plasmid (containing the full-length WT AO promoter) and the empty vector, and then incubated in the presence of 10 mM KCl, was set to 1. Asterisks indicate significant differences between seedlings treated with and without nitrate (Welch's t-test; *P ≤ 0.05, **P ≤ 0.01). n.s. not significant.
Furthermore, the AOpro(mut)-AO plants exhibited morphological abnormalities, such as narrower leaves with a disconnected vascular loop, shorter root tips, and shorter siliques (Fig. 4b, c, f, g). Defective lateral root development was also found in AOpro(mut)-AO seedlings grown on agar plates ( Supplementary Fig. 2). The growth defects of AOpro(mut)-AO plants were restored at least partially by supplementation with nicotinamide or nicotinate, intermediate metabolites of the NAD + salvage pathway (Supplementary Figs. 1 and 2). These results suggest that the loss of nitrateinduced expression of AO exerts pleiotropic effects on plant growth and development in Arabidopsis. We note that, consistent with these phenotypes, transgenic Arabidopsis plants expressing the βglucuronidase (GUS) gene under the control of the AO promoter showed strong GUS staining in leaves, root-shoot junctions, main root-lateral root junctions, root tips, and pollen grains (Supplementary Fig. 3).
Effects of the loss of nitrate-induced AO expression on NAD(H) and NADP(H) contents. The effects of the loss of the nitrate-dependent activation of AO expression on NAD(H) and NADP(H) contents were analyzed using the wild-type, AOpro(wt)-AO, and AOpro(mut)-AO plants, which were grown hydroponically in the N-free medium for 3 days to remove endogenous inorganic N and then in a 10 mM KNO 3 -containing medium for 1 day for de novo NAD synthesis requiring 5 enzymatic reactions ( Supplementary Fig. 1), according to the scheme shown in Fig. 5a. We note that DNA microarray analysis in a later section revealed that before N starvation treatment, the expression AO level was about 2.75-fold lower on the log 2 scale in AOpro(mut)-AO plants than AOpro(wt)-AO plants. The NAD + content of AOpro(mut)-AO plants was lower (approximately 20%) than that of wild-type and AOpro(wt)-AO plants before N starvation treatment with a statistically significant difference (Fig. 5b). The N starvation treatment decreased the contents of NAD(H) and NADP(H), while KNO 3 supply post-N starvation treatment increased these contents in plants of all genotypes used (Fig. 5b). However, the total contents of NAD(H) and NADP(H) in wild-type and AOpro(wt)-AO plants were significantly decreased by about 40% after the N starvation treatment and increased by about 20% after the KNO 3 supply post-N starvation treatment (Welch's t-test; P ≤ 0.01), whereas such changes in AOpro(mut)-AO plants were not statistically significant with only about 20% decrease and 10% increase, respectively. The total NAD(H) and NADP(H) contents after the KNO 3 supply were still lower than the contents before the N starvation, likely because the de novo NAD biosynthesis was still in progress to restore the original level. These results suggest that the loss of nitrate-induced expression of AO attenuated the modifications of NAD(H) and NADP(H) contents in response to the N nutrient condition to a limited extent. These observations are in sharp contrast to the substantially reduced growth of AOpro(mut)-AO plants (Fig. 4). Therefore, these results meant that the negative effects of the loss of nitrate-induced AO expression are mostly compensated for by the decline in growth rate, consistent with the lethal phenotype of the ao mutant. Considering fresh weight per plant (Fig. 4), the total amount of NAD(H) and NADP(H) synthesized by AOpro(mut)-AO plants per individual was estimated to be less than half of that in the wild-type and AOpro(wt)-AO plants.
Modified metabolite contents in AOpro(mut)-AO plants. To identify the metabolic pathways that are strongly affected by the loss of the nitrate-induced AO expression, we performed a metabolomic analysis of the plants grown hydroponically according to the scheme shown in Fig. 5a. Measurement of metabolite contents in wild-type plants and two independent lines of each AOpro(wt)-AO and AOpro(mut)-AO genotypes revealed the contents of 68 metabolites including various organic acids and amino acids. As a result, the effect of selective abolishment of nitrate-induced AO expression was found to be more pronounced in the contents of TCA cycle-related organic acids but less prominent in the contents for amino acids and other groups of primary metabolites (Supplementary Data 1). For instance, before N starvation treatment, fumarate content in AOpro(mut)-AO plants was about half of those of wild-type and AOpro(wt)-AO plants (0d samples in Fig. 6, Supplementary Data 1). Consistent with the fact that fumarate is one of the most abundant metabolites in the TCA cycle, the total amount of metabolites in the TCA cycle was significantly lower in AOpro(mut)-AO plants (approximately 30% of that in wild-type plants) (Supplementary Data 1, Fig. 6a). Furthermore, metabolomic analysis clarified that the nutritional environment-induced changes in contents of TCA cycle-associated metabolites were similar in wild-type and AOpro(wt)-AO plants but distinct in AOpro(mut)-AO plants (Fig. 6a). Most evidently, in AOpro(mut)-AO plants, KNO 3 supply post-N starvation treatment induced strong accumulation of citrate, another abundant metabolite in the TCA cycle, and succinate, which serve as substrates for the synthesis of 2-oxoglutarate (2-OG) necessary for N assimilation as carbon skeletons. Since the increases in citrate content in AOpro(mut)-AO plants (about 1000 nmol/g FW change) corresponded to about 15 % of the total organic acids in the TCA cycle, diminishing nitrate-inducible AO expression also exerted the pronounced effect on the composition of TCA metabolites, as b RT-qPCR analysis of nitrate-induced AO expression in 4-day-old seedlings of the WT and three independent transgenic lines (#1, #2, and #3). The seedlings were treated with 10 mM KNO 3 for 1 h. The expression level of AO in WT seedlings not exposed to the nitrate treatment was set to 1. Data represent mean ± SD (n = 3). Induction folds are indicated. Asterisks indicate significant differences between seedlings treated with and without nitrate (Welch's t-test; *P ≤ 0.05, **P ≤ 0.01).
shown by pie charts in Supplementary Fig. 4. These results revealed that nitrate-induced expression of AO is essential for proper maintenance of the TCA cycle in response to changes in the N nutrient environment.
The modified contents of TCA cycle-related organic acids in AOpro(mut)-AO plants may suggest attenuation of some enzymatic processes in the TCA cycle in AOpro(mut)-AO plants. Because fumarate is produced by mitochondrial complex II, a part of ETC, using NADH (Fig. 6b), we hypothesized that the low level of fumarate in AOpro(mut)-AO plants was caused by defects in the mitochondrial ETC. Consistent with this hypothesis, 1 µM antimycin A, an inhibitor of the ETC, inhibited the growth of AOpro(mut)-AO plants more effectively than that of wild-type and AOpro(wt)-AO plants (Fig. 6c). The nlp6 nlp7-1 double mutant and the NLP6-SUPRD transgenic line also showed high sensitivity against 1 µM antimycin A ( Supplementary Fig. 5), in line with the idea that NLP-mediated nitrate signaling impacts mitochondrial activity by modulating AO expression.
We note that, before N starvation treatment, the contents of arginine, proline, and several other urea cycle-related polyamines, such as putrescine and spermine, were different between AOpro(mut)-AO plants and the others, although the loss of nitrate-induced AO expression had little effect on the total amount of amino acids or the contents of most other amino acids (Supplementary Data 1, Fig. 7). Furthermore, these contents varied differently in WT and AOpro(wt)-AO plants and AOpro(mut)-AO plants. This result suggested another effect of the loss of nitrateinduced AO expression on primary metabolism.
Overall, compared to wild-type and AOpro(wt)-AO plants, AOpro(mut)-AO plants exhibited different accumulation of several TCA cycle-and urea cycle-related metabolites under any N nutrient conditions. These differences in metabolic balance, especially the differences under a normal steady growth condition before N starvation treatment, might be associated with the growth defect and morphological abnormality observed in Fig. 4 and Supplementary Fig. 2.
Limited effects of loss of nitrate-induced AO expression on nitrate reduction and N assimilation. Glutamine and glutamate contents were not notably different among wild-type, AOpro(wt)-AO, and AOpro(mut)-AO plants (Fig. 8a, Supplementary Data 1). Since glutamine and glutamate are the primary products of N assimilation and excellent metabolite markers for N assimilation activity 48 , this result suggested that the loss of nitrate-induced AO expression did not remarkably affect N assimilation and NADHdependent reduction of nitrate. Therefore, we examined decreases in nitrate concentration upon N starvation in wild-type, AOpro(wt)-AO, and AOpro(mut)-AO plants and then found that these plants show a similar decrease in nitrate concentration upon N starvation (Fig. 8b). The observed declines in nitrate concentration were dependent on nitrate reductase (NR) because the decrease in nitrate concentration was very slow in seedlings of G'4-3, a nia1 nia2 double mutant, which has only about 0.1% of the wild type NR activity in shoots and 5-10% of the wild type in roots 49,50 . Accordingly, these results suggest that the loss of nitrate-induced AO expression had no apparent effect on the nitrate reduction process as well as N assimilation.
Effect of the loss of nitrate inducibility of AO expression on gene expression profiles. Based on the different dynamics of primary metabolite contents in wild-type, AOpro(wt)-AO, and AOpro(mut)-AO plants, we speculated that the loss of nitrateinduced AO expression might also cause unique modifications in the expression profiles in AOpro(mut)-AO plants through metabolite-mediated regulations. This possibility was examined by DNA microarray analysis of AOpro(wt)-AO and AOpro(mut)-AO plants grown according to the scheme shown in Fig. 5a. The results showed that before the N starvation treatment, 401 and 268 genes were upregulated and downregulated in AOpro(mut)-AO plants, respectively (n = 3, adjusted P-value ≤ 0.05, log 2 fold change ≥ |1|) ( Supplementary Data 2 and 3). Subsequent GO analysis revealed that these genes were enriched with photosynthesis, disease response, and senescence-related genes (Supplementary Data 4). Interestingly, both N starvation and KNO 3 supply post-N starvation treatment modified the expression levels of a much smaller number of genes in AOpro(mut)-AO plants than in AOpro(wt)-AO plants (Supplementary Data 5-12). Furthermore, the slope of the linear regression lines representing the correlation between the changes in expression of each gene due to N starvation or KNO 3 supply in AOpro(wt)-AO plants and AOpro(mut)-AO plants was less than 1 (0.59 and 0.49) (Fig. 9a, b). These results indicated that in AOpro(mut)-AO plants, the reactions to N starvation and KNO 3 supply were attenuated in the gene expression profile, as were the modifications of NAD(H) and NADP(H) contents in response to the N nutritional change (Fig. 5b). GO analysis also supported this observation, showing that the GO terms enriched in AOpro(mut)-AO plants after N starvation or KNO 3 supply, such as nitrate and salicylic acid response, were more highly enriched in AOpro(wt)-AO plants and also that the range of GO term repertoire was broader in AOpro(wt)-AO plants than in AOpro(mut)-AO plants (Supplementary Data 13-16).
The further analysis was performed using the Upset Shiny apps (https://gehlenborglab.shinyapps.io/upsetr/) 51 to clarify overlaps among the lists of differentially expressed genes ( Supplementary  Fig. 6). As expected, many overlaps represented the genes similarly upregulated or downregulated by the same treatment in represent plants grown hydroponically on 1/10MS medium for 20 days and then on fresh 1/10MS medium for 3 or 4 days, respectively. b NAD + , NADH, NADP + , and NADPH contents of the whole shoots of WT, AOpro(wt)-AO, and AOpro(mut)-AO plants sampled according to the scheme in (a). Data represent mean ± SD (n = 5). Different letters denote statistically significant differences (Tukey's HSD test; P < 0.05).
AOpro(wt)-AO and AOpro(mut)-AO plants. Most of the other overlaps were found between the lists of N starvation-responsive genes and genes whose expression levels were modulated in the opposite direction by KNO 3 resupply. On the other hand, we also found that considerable overlaps between the list of differentially expressed genes in AOpro(mut)-AO plants before N starvation treatment and other gene lists barely existed, except for an overlap between the lists of the genes downregulated in AOpro(mut)-AO before N starvation treatment and the genes downregulated in AOpro(wt)-AO after KNO 3 resupply, which was enriched with defense and senescence-related genes. Therefore, it is likely that the unique modifications in transcriptional profiles of AOpro(mut)-AO plants do not necessarily coincide with short-term nitrate signaling responses, but may also partly reflect the regulation associated with metabolites in nitrate signaling-regulated metabolic pathways.
Detailed analysis of differentially expressed genes in AOpro(wt)-AO and AOpro(mut)-AO plants revealed that a set of genes encoding light-harvesting complex (LHC) proteins were upregulated in AOpro(mut)-AO plants. In contrast, typical pathogen-responsive genes were downregulated (Fig. 9c, Supplementary Data 2 and 3). Furthermore, N starvation treatment and KNO 3 supply resulted in relative increases in the expression of LHC genes only in AOpro(mut)-AO plants and only in AOpro(wt)-AO plants, respectively (column 2 vs. column 4 and column 3 vs. column 5 of the table inserted in Fig. 9c). Similarly, KNO 3 supply post-N starvation treatment reduced the activation of pathogen-responsive genes more clearly in AOpro(wt)-AO plants. Furthermore, N nutrient condition-dependent modulations of the expression levels of several N metabolism-related genes were evident only in AOpro(wt)-AO plants (Fig. 9c). Collectively, the detailed analysis of differentially expressed genes also indicated that the modifications in gene expression in response to changes in the nutrient condition are generally blunted in AOpro(mut)-AO plants. We note an additional finding of slight but constitutive upregulation of NMNAT and QS genes in AOpro(mut)-AO plants (Fig. 9c), which may be a mechanism compensating for the relatively lower expression level of AO in AOpro(mut)-AO plants.  Fig. 5a. Data represent mean ± SD (n = 5). b Schematic representation of the mitochondrial electron transport chain. Antimycin A inhibits the activity of Complex III. Q, ubiquinone; QH 2 , ubiquinol. c Effect of antimycin A on the shoot fresh weight. Seedlings were initially grown in the absence of antimycin A for 4 days and then in the presence of the indicated concentrations of antimycin A for 6 days (n = 7-12). Two independent lines (#1 and #2) of AOpro(wt)-AO and AOpro(mut)-AO plants were used. In (a, c), different letters denote statistically significant differences (Tukey's HSD test; P < 0.05).  Fig. 5a. Two independent lines (#1 and #2) of AOpro(wt)-AO and AOpro(mut)-AO plants were used. Data represent mean ± SD (n = 5). Different letters denote statistically significant differences (Tukey's HSD test; P < 0.05).  Fig. 5a. Two independent lines (#1 and #2) of AOpro(wt)-AO and AOpro(mut)-AO plants were used. Data represent mean ± SD (n = 5). b Decreases in the shoot nitrate content of wild-type, AOpro(wt)-AO, AOpro(mut)-AO, nlp7 and G'4-3 seedlings during the N starvation treatment. Seedlings were first grown in the presence of KNO 3 and NH 4 NO 3 for 8 days and then grown in the absence of N sources for 0, 1, or 3 days. To grow G'4-3 (nia1 nia2 double mutant) seedlings and other seedlings under the same nutrient condition, medium containing ammonium as N source was used in the pre-culture of all seedlings. Data represent mean ± SD (n = 4). In (a, b), different lowercase letters indicate significant differences among various genotypes (P < 0.05; Tukey's HSD test).

Discussion
In the present study, the AO gene, encoding a key enzyme in de novo NAD + biosynthesis, was identified as one of the genes directly regulated by nitrate-activated NLPs. Furthermore, by selectively abolishing the nitrate inducibility of AO expression in planta, we found that NLP-regulated nitrate-inducible AO expression is necessary to properly maintain metabolic balance and launch transcriptional adaptation of plants in response to changes in the N environment. Therefore, the present results provide direct evidence for the importance of coordinated regulation of multiple metabolic pathways by nitrate signaling in adaptation to the fluctuating N nutrient level, which is firstly ARTICLE COMMUNICATIONS BIOLOGY | https://doi.org/10.1038/s42003-022-03399-5 demonstrated so far as we know. Since a number of metabolic enzyme genes are probably under the direct control of NLPs, other metabolic enzyme genes may also play critical roles in nitrate responses. Hence, this study suggests that future analysis of the nitrate-induced activation of metabolic enzyme genes in other metabolic pathways, as was done in this study, may reveal overlooked critical points to fully understand the role of nitrate as a plant growth regulator.
Phenotypic and metabolic analyses suggested that AOpro(mut)-AO plants relieved the negative effects of the abnormal regulation of AO expression by reducing their size rather than by decreasing NAD(H) and NADP(H) contents per unit tissue fresh weight. Since NAD(H) and NADP(H) are essential for growth, AOpro(mut)-AO plants probably had no choice but to restrict their growth in proportion to amounts of de novo synthesized NAD(H) and NADP(H) to survive. This conclusion means that there are minimum NAD(H) and NADP(H) contents to support plant growth, consistent with the lethal phenotype of ao, qs, qpt, and nmnat knockout mutants lacking the ability for de novo NAD + biosynthesis 37,41 .
Metabolomic analysis revealed pronounced effects of the loss of the nitrate-induced AO expression on the TCA cycle rather than N assimilation (Supplementary Data 1, Fig. 6). Since an intermediate in the TCA cycle, 2-OG, is used as carbon skeletons required for N assimilation 52 , the nitrate responses may involve the enhancement of the TCA cycle as well as the N assimilation pathway. In fact, nitrate supply induced the expression of several genes encoding enzymes involved in the TCA cycle, leading to the accumulation of TCA cycle intermediates 53 . Therefore, a breakdown of nitrate-induced synchronized activation of enzymatic processes involved in 2-OG production and NAD + biosynthesis probably caused remarked increases in the citrate content upon KNO 3 supply (Fig. 6), emphasizing the importance of the nitrateinduced synchronized activation of enzymatic processes for the proper maintenance of the TCA cycle. Based on the low level of fumarate, low activity of mitochondrial ETC, and the irregular accumulation of intermediates and derivatives of the urea cycle in AOpro(mut)-AO plants (Fig. 7), we infer that nitrate-induced AO expression is tightly associated with the maintenance of the metabolic balance in mitochondria. On the other hand, selective abolishment of the nitrate-induced AO expression hardly affected the NADH-dependent reduction of nitrate in the cytosol, the first step in nitrate assimilation, as well as amino acid contents (Fig. 6). Therefore, we also speculate that the main reason for the induction of AO expression by nitrate may be maintaining the TCA cycle and other metabolic processes in mitochondria in response to nitrate supply.
At this stage, it is difficult to fully explain how the loss of nitrate-induced AO expression pronouncedly caused pleiotropic effects on metabolism and growth. However, considering that the most remarked effect of loss of nitrate-induced AO expression in the metabolite profile was the imbalance of metabolites related to the TCA cycle, which is central to primary metabolism, we assume that the loss of nitrate-induced AO expression first induces an imbalance in TCA cycle-related metabolites and then influences other metabolic pathways and growth via the imbalance. Although we cannot exclude the possibility that the loss of nitrate-induced AO expression also affects growth and development through other pathways, this hypothesis is consistent with the relatively weak variation in NAD(H) and NADP(H) contents and reduced responses to changes in the N nutrient condition in the gene expression profile in AOpro(mut)-AO plants. Furthermore, an imbalance in TCA cycle-related metabolites may cause indirect but widespread changes in gene expression profiles through metabolite-mediated regulations and other regulatory mechanisms. Further investigation of this hypothesis would be valuable in understanding how N nutrients promote plant growth.

Methods
Plant materials and growth conditions. Arabidopsis thaliana ecotype Columbia (Col) was used as the wild type. Seeds of the T-DNA insertion ao knockout mutant (fin4-2, SALK_013920) 46 and G'4-3 49 were obtained from the Arabidopsis Biological Resource Center (Columbus, OH, USA). The nlp6 nlp7-1 double mutant, transgenic NLP6-SUPRD line, and its parental line were generated previously 7,54 . All plants were grown under continuous light (60-90 µE m −2 s −1 ) at 23°C unless otherwise noted. To analyze nitrate-induced gene expression, seedlings were grown in N-free 1/10MS medium (N-free 1/10MS salts, 0.5 mM K 2 SO 4 , 0.1 g L −1 MES-KOH [pH 5.7]) supplemented with 0.5 mM ammonium succinate and 0.5 g L −1 sucrose for 4 days, and then treated with 10 mM KNO 3 or KCl for 1 h in the presence or absence of 100 µM CHX. Around 100 seedlings were collected together as a biological replicate. We note that K 2 SO 4 was added at a final concentration of 0.5 mM to prepare N-free 1/10MS medium because 1 mM K + is sufficient to support the growth of Arabidopsis 55 . To conduct phenotypic analysis with agar plates, 4-day-old seedlings grown on agar plates (N-free 1/2MS salts, 10 mM KNO 3 , 1% sucrose, 0.8% agar) were further grown on fresh agar plates for 6-10 days. To examine the effects of antimycin A on the mitochondrial ETC, 4day-old seedlings were transferred to agar plates with or without 1 or 5 µM antimycin A and grown for 6 days. To rescue the growth defects of AOpro(mut)-AO seedlings by supplementation with a metabolite, 4-day-old seedlings were transferred to agar plates containing 10 mM KNO 3 in the presence or absence of 200 µM nicotinamide or 2 µM nicotinate and grown for 6 days. For phenotypic analysis, plants were also grown hydroponically on N-free 1/10MS medium supplemented with 10 mM KNO 3 or on nutrient-containing peat moss (Jiffy-7, Sakata Seed Co., Kanagawa, Japan) under the long-day photoperiod (16 h light/8 h dark). To perform metabolic profiling and DNA microarray analysis, plants were grown hydroponically on 1/10MS medium for 20 days and then on N-free 1/10MS medium for 3 days. Subsequently, the plants were grown for 1 days on KNO 3supplemented medium (final concentration, 10 mM), while plants for reference samples were grown on 1/10MS medium for 3 or 4 days.
Reverse transcription-quantitative PCR (RT-qPCR). Total RNA for RT-qPCR was extracted from 4-day-old seedlings using the ISOSPIN Plant RNA Kit (NIP-PON GENE Co., Tokyo, Japan). First-strand cDNA was synthesized using oligo(dT) 15   Gray dots indicate genes that did not significantly respond to N starvation treatment or KNO 3 resupply, while open circles in red, blue, and purple indicate genes that significantly responded to N starvation treatment or KNO 3 resupply in only AOpro(wt)-AO, only AOpro(mut)-AO, and both, respectively (log 2 fold change ≥ |1|, adjusted P-value ≤ 0.05). Closed red, blue, and purple circles were applied when the adjusted P-value was less than 0.001. c Heatmap of typical genes whose expression levels were differentially modified in AOpro(wt)-AO vs. AOpro(mut)-AO plants during changes in the N conditions. Log 2 fold changes (relative to the expression levels in sample 0d of AOpro(wt)-AO plants) are shown with a color scale. A chart on the right side of the heatmap shows log 2 fold changes of expression levels in AOpro(mut)-AO to those in AOpro(wt)-AO plants before N starvation treatment (column 1), log 2 fold changes caused by N starvation treatment (column 2) or nitrate resupply (column 3) in AOpro(wt)-AO plants, and log 2 fold changes caused by N starvation treatment (column 4) or nitrate resupply (column 5) in AOpro(mut)-AO plants. Statistically significant differences are indicated in bold font (P < 0.05).
Plasmid construction. To construct reporter plasmids for the co-transfection assay, AO promoter fragments (from −2137 to +868, −1796 to +868, −1423 to +868, −1130 to +868, and −834 to +868, −504 to +868, and −164 to +868, relative to the transcription start site) were amplified by PCR using gene-specific primers. A mutated AO promoter fragment (−2137 to +868) carrying nucleotide substitutions within the NLP-binding site (−354 to −347 bp) was produced by overlap extension PCR. Then, the 35 S promoter and Ω enhancer region in the pJD301 vector were replaced by the obtained AO promoter fragments. Binary vectors for plant transformation were constructed by replacing the 35 S promoter and GUS gene in pCB302-HYG-35S-Ω-GUS 57,58 with the wild-type or mutated AO promoter fragment (−2,137 to +868) and a PCR-amplified genomic DNA fragment (ranging from the translation initiation site at position +870 to 868 bp downstream of the stop codon), respectively. Another binary vector for histochemical analysis was generated by replacing only the 35S promoter in pCB302-HYG-35S-Ω-GUS with the wild-type AO promoter (−2137 to +868). All primers used for plasmid construction are listed in Supplementary Data 17. All PCR products used for plasmid construction were verified by DNA sequencing.
Co-transfection assay. Mesophyll protoplasts were prepared from wild-type Arabidopsis treated with N-starvation by the Tape-Arabidopsis Sandwich method 59 . Co-transfection with the effector, reporter, and control plasmids was performed by the DNA-PEG-calcium transfection method 60 . Effector plasmids used were p35SC4PPDK-none-MYC and p35SC4PPDK-NLP7-MYC 7 , and pUBQ10-GUS was used as a control plasmid 7 . Measurement of luciferase (LUC) and GUS activity was performed using the Luciferase Assay System (Promega Co., WI, USA) and 4-methylumbelliferyl-β-D-glucuronide, respectively 54 .
EMSA. The His-tagged RWP-RK DNA-binding domain of NLP7 (recombinant NLP7 protein) was prepared by affinity chromatography using Ni-NTA resin 7 . Biotin-labeled DNA probes were generated by PCR using specific primers. The mutated DNA probes (Mut1-3), harboring point mutations in the putative NLPbinding site, were prepared by overlap extension PCR or by PCR with a primer harboring the Mut1 sequence. All primers used are listed in Supplementary Data 17. The recombinant NLP7 was incubated together with probe DNAs in the binding buffer (0.5 µg poly [d(I-C)], 20 mM HEPES-KOH (pH 7.6), 3 mM MgCl 2 , 2 mM dithiothreitol, 0.5 mM EDTA, 30 mM KCl, 30 mM NaCl, and 10% glycerol) at room temperature. After gel electrophoresis of the reaction mixtures, the probe DNAs were transferred from gels onto nylon membranes and detected using streptavidin-conjugated horseradish peroxidase 7 .
Generation of transgenic Arabidopsis plants. Plants heterozygous for the ao null allele were transformed by the floral dip method 61 using Agrobacterium tumefaciens strain GV3101 (pMP90) 62 harboring a binary vector. T2 lines showing a 3:1 (hygromycin B resistant: sensitive) segregation ratio were selected. Then, T3 progenies homozygous for both the introduced gene and the ao null allele were selected by PCR-based genotyping using gene-specific primers (Supplementary Data 17). T3 or T4 progenies were used for analyses.
Analysis of vascular patterns. Cotyledons were fixed in a mixture of ethanol and acetic acid (9:1), hydrated using a graded series of ethanol, and then mounted with the clearing solution 63 . Vascular patterns were analyzed using a stereomicroscope (MZ16F; Leica Microsystems, Germany).
Histochemical GUS staining. Wild-type Arabidopsis plants were transformed with a binary vector harboring the GUS gene under the control of the wild-type AO promoter. Homozygous T3 progenies were selected and used for GUS staining. Briefly, 10-day-old seedlings grown on 1/2MS agar plates and floral buds and stamens collected from 40-day-old plants grown in the soil were fixed in acetone at −20°C and washed with 0.1 M sodium phosphate buffer (pH 7.4). For GUS staining, samples were incubated in X-Glc solution (0.5 g L −1 5-bromo-4-chloro-3indolyl-β-D-glucuronide cyclohexylammonium salt, 0.5 mM potassium ferricyanide, 0.5 mM potassium ferrocyanide, and 0.1 M sodium phosphate, pH 7.4) at 37°C 12 .
Measurement of metabolite contents. Aerial tissues of each plant grown according to the scheme shown in Fig. 5a were harvested and immediately frozen in liquid N 2 . Two shoots (AOpro(wt)-AO plants) or three-five shoots (AOpro(mut)-AO plants) were pooled to form one independent biological replicate. NAD + , NADH, NADP + , and NADPH were extracted from frozen tissues and quantified by a plate reader method of Queval and Noctor 64 . Metabolite analysis was performed using the capillary electrophoresis coupled with triple quadrupole mass spectrometry (CE-LC/MS/MS) system (CE, 7100; MS, G6420A; Agilent Technologies, CA, USA) 65 .
Measurement of nitrate content. Nitrate content was measured using the shoots of seedlings that were grown on N-free 1/2MS agar plates supplemented with 3.3 mM KNO 3 and 3.3 mM NH 4 NO 3 for 8 days and then grown on N-free 1/2MS agar plates for 0, 1, or 3 days. The Nitrate Test Kit (NECi Superior Enzymes, MI, USA) was used according to the manufacturer's instructions.
DNA microarray analysis. Total RNA was extracted from the whole shoots of AO(wt)-AO (line #1) and AO(mut)-AO (line #1) plants grown according to the scheme shown in Fig. 5a, using the ISOSPIN Plant RNA Kit. Cyanine-3-labeled cRNA was prepared with the Low Input Quick Amp Labeling Kit (One-Color; Agilent Technologies). We used the shoot of two or three seedlings to make a biological replicate, and biological triplicates were used for DNA microarray analysis. The labeled cRNA was hybridized to the Arabidopsis (V4) Oligo Microarray (Agilent Technologies), and the DNA Microarray Scanner (G2565BA; Agilent Technologies) was used to scan the arrays. The scanned images were analyzed with Feature Extraction (v9.1) (Agilent Technologies). Data analysis was performed using a statistical R language 66,67 . The LIMMA package was used for linear modeling to acquire empirical Bayes statistics 66 . Differentially expressed genes with an adjusted P-value ≤ 0.05 and absolute log 2 fold change ≥ |1| were extracted for each comparison. Effects of genotypeenvironment interactions were examined by creating a contrast from two contrasts of each genotype before and after the treatments (P-value ≤ 0.001). Visualizing the intersections across differentially expressed gene lists was done with Upset Shiny apps (https://gehlenborglab.shinyapps.io/upsetr/) 51 . Gene ontology term enrichment analysis was done with the PANTHER classification system (http://pantherdb.org/). Microarray data were deposited in the Gene Expression Omnibus (GEO) repository of the National Center for Biotechnology Information (NCBI) database under the accession number GSE136542 on August 28, 2019.
Statistics and reproductivity. All data shown in graphs are presented as the mean with S.D. from more than three biological replicates. Statistical significance was tested by Welch's t-test (*P ≤ 0.05, **P ≤ 0.01) or Tukey's honestly significant difference [HSD] test (Different lowercase letters indicate significant differences, P < 0.05).