Isotopic labelling reveals the efficient adaptation of wheat root TCA cycle flux modes to match carbon demand under ammonium nutrition

Proper carbon (C) supply is essential for nitrogen (N) assimilation especially when plants are grown under ammonium (NH4+) nutrition. However, how C and N metabolic fluxes adapt to achieve so remains uncertain. In this work, roots of wheat (Triticum aestivum L.) plants grown under exclusive NH4+ or nitrate (NO3−) supply were incubated with isotope-labelled substrates (15NH4+, 15NO3−, or [13C]Pyruvate) to follow the incorporation of 15N or 13C into amino acids and organic acids. Roots of plants adapted to ammonium nutrition presented higher capacity to incorporate both 15NH4+ and 15NO3− into amino acids, thanks to the previous induction of the NH4+ assimilative machinery. The 15N label was firstly incorporated into [15N]Gln vía glutamine synthetase; ultimately leading to [15N]Asn accumulation as an optimal NH4+ storage. The provision of [13C]Pyruvate led to [13C]Citrate and [13C]Malate accumulation and to rapid [13C]2-OG consumption for amino acid synthesis and highlighted the importance of the anaplerotic routes associated to tricarboxylic acid (TCA) cycle. Taken together, our results indicate that root adaptation to ammonium nutrition allowed efficient assimilation of N thanks to the promotion of TCA cycle open flux modes in order to sustain C skeleton availability for effective NH4+ detoxification into amino acids.

Metabolic networks are generally considered as more complex in plants compared to other organisms. This complexity is due to several aspects, which include the huge chemical repertoire of plants together with their sessile nature that makes them to be adaptable to a range of environmental conditions, including the nutritional heterogeneity of soils. Nitrogen (N) is an essential nutrient for plant growth and development, and plants need to rearrange their metabolism in function of both the availability and the form of N present in the soil 1,2 . The typical N deficiency of agricultural soils is behind the need of the supply of nitrogenous fertilizers to maintain optimal crop yield. In this context, sustainable agriculture pursues to maximise nitrogen use efficiency (NUE) to reduce the cost of fertilization and its environmental impacts. NUE depends, among others, on N uptake and assimilation efficiency, which both rely on the N form available in the soil 3,4 .
In general, the most common available N forms are nitrate (NO 3 − ) and ammonium (NH 4 + ). Their proportions are dependent on environmental conditions, on soil microbiota and on soil use and management. Although plants are able to use both N forms, the preference to absorb one or the other varies between species and even within the same species. In any case, when NH 4 + is present in high concentrations it provokes a number of physiological and morphological disturbances collectively known as ammonium syndrome 5 . The symptoms associated to this syndrome include alterations in pH homeostasis, ionic imbalance, alterations in plant development and a profound adaption of cell metabolism. Ultimately, ammonium syndrome may entail a severe stressful situation provoking a dramatic biomass reduction, chlorosis and even plant death 5,6 . One of the main strategies that plants deploy to avoid excessive NH 4 + accumulation in tissues is its assimilation into organic molecules, principally through GS/GOGAT pathway 7 . The induction of GS activity is a common response of plants when facing ammonium nutrition 8,9 and Arabidopsis thaliana mutants lacking GLN1:2 isoform display ammonium accumulation and hypersensitivity to ammonium stress 10 . Indeed, GS enzyme has been considered as a marker to predict the N status in many plant species including wheat 11,12 . NH 4 + is a highly mobile molecule and it is present in the xylem sap, therefore it can be assimilated in leaves 13,14 . However, in general, roots are the primary site of ammonium assimilation. Indeed, the root is the first organ facing high NH 4 + concentrations in the medium and acts as a physiological barrier to prevent its transport to the more sensitive shoot, where the excess of NH 4 + can impair photosynthetic apparatus 5,6 . In this line, the metabolic adjustment of roots to exclusive ammonium nutrition has even been shown to determine the capacity of the plants to cope with ammonium stress in many species including wheat and tomato 15,16 . Interestingly, in other species such as oilseed rape (Brassica napus) the metabolic adjustment seems to occur mostly at the leaf level 17 . Proper carbon (C) supply to maintain ammonium assimilation in the roots is considered as a key aspect to deal with an excess of NH 4 + 18 . In this sense, the provision of exogenous inorganic C to the root zone partially increased ammonium tolerance in cucumber and tomato [19][20][21] . Similarly, biomass reduction observed in tomato roots under ammonium stress did not happen when plants were grown at high atmospheric CO 2 conditions 15 , therefore evidencing the important interplay between N and C metabolism in roots exposed to ammonium nutrition.
The supply of C skeletons is mediated by the tricarboxylic acid (TCA) cycle, together with its associated anaplerotic routes, which becomes essential for N assimilation into amino acids 22 . Actually, Arabidopsis mutants with reduced PEPC or ICDH activity showed alteration in the synthesis of amino acids 23,24 . Moreover, several works have reported an increase of in vitro determined TCA-associated enzyme activities when plants deal with ammonium stress, suggesting that the provision of C skeletons linked to TCA is essential to maintain NH 4 + homeostasis 9,25-27 . Indeed, the high C demand in roots for primary ammonium assimilation has been put forward as a trade-off of the consumption of C resources responsible for the growth inhibition typically observed under ammonium stress 5 . Nevertheless, there is still a need to understand the in vivo adaptation of C and N metabolic fluxes when plants grow under ammonium nutrition compared to nitrate nutrition.
The use of isotope-labelled substrates to study metabolic fluxes is an excellent strategy to obtain a dynamic in vivo picture of cell metabolic activity and has been useful to understand different aspects of plant metabolism, including N assimilation and C allocation, mainly through 13 C and 15 N labelling 28 . In fact, the sometimes lack of correlation between the information provided by in vitro enzyme activities and metabolite data and the poor understanding of the role and the regulation of some enzymes in different cell metabolic contexts urge the use of labelled substrates to track the metabolic fluxes 29,30 .
To further advance in the in vivo metabolic strategies root cells deploy to control NH 4 + levels and to adapt to changing N sources, in this work we aimed at understanding 1) how the adaptation of wheat plants to the exclusive provision of nitrate or ammonium as N source determines the efficiency to assimilate one or another N source and 2) how the in vivo N assimilation dynamics are linked to TCA cycle activity in the roots. To do so, roots were incubated with either 15 NH 4 + , 15 NO 3 − , or [ 13 C]Pyr from thirty minutes to a maximum of six hours; and the enrichment of 13 C or 15 N amino acids and organic acids was evaluated by gas chromatography coupled to mass spectrometry.

Results
Wheat physiologic and metabolic response to the growth under ammonium or nitrate nutrition. Most  . Thus, as expected, wheat plants grown under 10 mM ammonium nutrition showed lower biomass production compared with nitrate nutrition ( Supplementary  Fig. S1), being this effect evident in both shoot and root (Table 1). Nonetheless, the higher chlorophyll contents in the leaf (Table 1) indicated that plants were experiencing a mild stress degree in response to NH 4 + supply. Consistently, ammonium nutrition led to an increased NH 4 + , amino acids, protein and C contents in the root compared to nitrate nutrition (Table 1). Asn, Gln and Ala were the major amino acids in wheat roots, their contents being superior in RA with respect to RN ( Fig. S2; RN and RA stand for root of plants grown for six weeks under nitrate or ammonium nutrition, respectively). The organic acids contents, in contrast, were higher in RN, and malate and citrate comprised the majority of the organic acids accumulated ( Supplementary Fig. S2). As expected, it was observed a general induction of in vitro determined root enzyme activities associated with ammonium assimilation (GS, NADH-GOGAT) as well as NADH-GDH activity in RA (Fig. 1). As expected, nitrate reductase (NR) showed an opposite behaviour, being more active under nitrate nutrition. The enzyme activities of the TCA cycle (CS, NADP-ICD, MDH) and its associated anaplerotic routes (PEPC, NADP-ME) were also enhanced in RA (Fig. 1). 15 Table S2). The labelling pattern in the amides, in contrast, did not show differences between both incubation assays ( Supplementary Fig. S4).

Isotopic labelling of roots with
The  Fig. S6). As previously mentioned, GDH enzyme activity was highly induced in RA (Fig. 1). GDH is a reversible enzyme that in vitro can both assimilate NH 4 + to form Glu or deaminate Glu to form 2-OG. To ascertain the in vivo role of GDH, its capacity to incorporate 15 NH 4 + to form [ 15 N]Glu was evaluated. As shown in Fig. 4 and Supplementary S7, the incubation of root segments with GS/GOGAT pathway inhibitors MSX and AZA highly reduced or even abolished the incorporation of 15 NH 4 + into amino acids. Importantly, the newly synthesized [ 15 N]Glu was almost undetectable (Fig. 4).

Isotopic labelling of roots with [ 13 C]pyr.
To gain insight on how TCA metabolic flux is modulated in vivo in roots adapted to nitrate or ammonium nutrition, RA and RN were supplied with 10 mM [ 13 C]Pyr, plus the corresponding N source (10 mM of unlabelled NH 4 + or NO 3 − ), and 13 C incorporation into organic acids and amino acids was followed. The formation of [ 13 C]amino acids showed a logarithmic increasing trend under both N nutritions; however, total [ 13 C]amino acid content was higher in RA, reaching a maximum of 15.4 µmol g −1 DW at 6 h compared to 5.1 µmol g −1 DW in RN (Fig. 5). Regarding total [ 13 C]organic acids, they accumulated following a linear regression, which increased from ca. 2 µmol g −1 DW at 30 min to ca. 10 µmol g −1 DW at 6 h in RA, while in RN only reached a maximum of 4.3 µmol g −1 DW at 6 h (Fig. 5).
Observing [ 13 C]molecules individually (Supplementary Table S3), around 80% of the [ 13 C]Pyr was labelled in both RN and RA during the whole incubation period, whilst the 13 C enrichment for the rest of organic acids increased with time. The most abundant amino acids also showed an increasing enrichment pattern over the incubation period. Looking deeper into the labelling pattern of the organic acids (Supplementary Table S4), practically all the [ 13 C]Pyr was labelled in its three carbons (m + 3) during the whole incubation period with 10 mM [ 13 C]Pyr; while the rest of [ 13 C]organic acids were, in general, more labelled in two carbons (m + 2). By the end of the incubation period, the second most labelled mass peak was m + 3 (molecule labelled in three C) for [ 13 Table 1. Wheat plant biomass, leaf chlorophyll and root content of N, C, NH 4 + , amino acid and soluble protein from wheat plants grown under nitrate or ammonium nutrition. Values represent mean ± SE (n = 6; for biomass n = 18). Asterisk (*) indicates significant differences between N nutritions (p < 0.05).
www.nature.com/scientificreports www.nature.com/scientificreports/ and [ 13 C]GABA (Supplementary Table S4). Since Ala directly derives from Pyr, most of [ 13 C]Ala present in the roots at 30 min was labelled in its three carbons (m + 3) regardless the previous N nutrition. In contrast, the other five amino acids were mostly labelled in two carbons (m + 2); followed by one-carbon labelling (m + 1) for [ 13 Fig. S8).

Discussion
The induction of root metabolic pathways for amino acid synthesis is essential to face ammonium stress. Ammonium nutrition entails a growth constraint in most plants including cereals such as barley 32 and wheat 16,31,33 , as also observed in the present study, mainly as a consequence of reduced root biomass (Table 1; Supplementary Fig. S1). The wheat root acts as a physiological barrier that accomplished two main strategies to face ammonium stress: to increase NH 4 + assimilation into amino acids and protein; and to accumulate the excess free NH 4 + ( Table 1). The enhanced assimilation of NH 4 + demands a consequent C skeleton supply 25,33 . Thereby the induced activities related to the TCA cycle, together with GS/GOGAT pathway, would be in the charge of such an enhanced NH 4 + assimilation in RA (Fig. 2), as also previously observed in different species 15,16 . Particularly, PEPC activity has been shown to increase under ammonium nutrition 18,26 and to correlate with GS activity 34 . The investment of C skeletons in the synthesis of amino acids is further reflected by organic acids content decrease in RA compared to RN (Supplementary Fig. S2).
In order to optimize the storage of N compounds in the cell, the main amino acids accumulated in RA were those with higher N:C ratio (Asn and Gln; Supplementary Fig. S2). Asn stands out as a long-distance N transporter on whole plant basis 35 and it is usually the main amino acid accumulated in monocots under ammonium nutrition 16,36 . Therefore, evidencing that AS may act as a complement of GS/GOGAT pathway in NH 4 + detoxification 37 . Additionally, the observed induction of GDH activity (Fig. 1), due to its reversible nature, may be contributing to the synthesis of Glu or 2-OG. Nonetheless, in vitro determined activities do not always reflect the dynamic of metabolic fluxes and thus, an in vivo isotopic-labelling approach was engaged to provide a dynamic depiction of root N assimilation and TCA cycle tuning in response to N nutrition. 15 (Table 2; Fig. 3). In RA, the assimilated 15 NH 4 + was ultimately diverted to [ 15 N]Asn, the major storage amino acid in wheat plants ( Supplementary Fig. S2). Interestingly, the formation of [ 15 N]GABA, that would be sinthetized by decarboxylation via glutamate decarboxylase (GAD) activity, also followed an increasing trend in RA-15 NH 4 + (Fig. 3). The induction of GABA production is common under stress conditions 38 and was shown to contribute to the alleviation of ammonium toxicity in rice 39 . Therefore, its synthesis could be an indicator of the stressful situation undergone in response to high NH 4 + contents in wheat roots. Alternatively, GABA could also take part in the GABA shunt, being transaminated with Pyr by GABA transaminase (GABA-T), producing Ala and succinate 40 .

In vivo
Additionally, [ 15 N]Glu could be also converted to 2-OG by GDH. Indeed, one of the common responses of ammonium nutrition is the induction of GDH activity, as observed in Arabidopsis 9 or wheat 16,41 , which is also reported in the present work (Fig. 1). In vitro GDH catalyses both the reductive amination of 2-OG and the oxidative deamination of Glu but in vivo its role remains controversial. Studies using tobacco overexpressing GDH isoforms 42 and Arabidopsis gdh mutants 43,44 demonstrated in vivo that GDH central role is Glu deamination. Alternatively, it has also been proposed that under certain conditions GDH could be participating in the direct assimilation of NH 4 + , such as during stressful situations where NH 4 + is accumulated 45,46 . In this line, very recently, in vivo GDH aminating activity was observed in tomato roots exposed to excess 15 NH 4 + provision 47 . Similarly, Ferraro et al. 48 , using GDH knock down tomato lines, also suggested in fruits that GDH could be working in vivo in its aminating sense. In the present work, in line with the works advocating for GDH role in Glu deamination, the nearly absence of 15 N label incorporation into [ 15 N]Glu when GS/GOGAT pathway is inhibited and the slight incorporation of the label into other amino acids (Figs 4, S7) led us to discard a significant participation of GDH in the in vivo assimilation of 15 NH 4 + in wheat roots. Therefore, the enhanced GDH activity observed in RA (Fig. 1) could mean increased 2-OG provision in wheat roots, essential for the detoxification of NH 4 + into Gln via GS/GOGAT pathway or into Asp and Asn via OAA formation in the TCA cycle. The results presented in this paper together with the available literature suggest that in vivo GDH role would be different depending on the plant organ, on plant growth and environmental conditions and, importantly, also on the plant species.
Importantly, although roots adapted to both N nutritions showed a similar immediate response to 15 NH 4 + supply, RA-15 NH 4 + was able to maintain higher 15 NH 4 incorporation rates along time, as illustrated by the raised accumulation of 15 N label into Asn and Ala at 6 h (Fig. 3). Thereby, this result demonstrates that the activation of www.nature.com/scientificreports www.nature.com/scientificreports/ enzymes responsible for ammonium assimilation in plants adapted to an ammonium-based nutrition conferred the root improved capacity to assimilate 15 NH 4 + into amino acids. However, the diversion of plant's C resources to ammonium detoxification may be a trade-off for plant growth. Thus, highlighting that the fine regulation of C:N balance is key for the plant performance under ammonium nutrition.

NO 3
− labelling underscores the relevance of ammonium assimilation machinery for efficient nitrate incorporation into amino acids. Wheat root metabolic fluxes were also determined in response to 15  , with a more active NR enzyme (Fig. 1), was able to use nitrate more efficiently than RA-15 NO 3 − . The reduction of NO 3 − is highly dependent of adequate reductant (Fd and NAD(P)H) availability, and in fact, NR activity has long been considered to be the rate-limiting step in most plants 3 . In this work, the formation of [ 15 N]amino acids from 15 NO 3 − was much lower compared with that www.nature.com/scientificreports www.nature.com/scientificreports/ of direct 15 NH 4 + supply both in RN and RA (Fig. 3). Therefore, these results reinforce that the energetic cost of nitrate primary assimilation supposes a bottleneck that slows down the use of 15 (Figs 2b, 3). So, the fact that the roots of plants previously adapted to an exclusive ammonium-based nutrition favoured 15 NO 3 − assimilation compared to plants adapted to nitrate nutrition would be indicating that the pre-activation of nitrate uptake and reduction machinery would not be an advantage to use 15 NO 3 − more efficiently. Alternatively, RA-15 NO 3− clearly benefits from the previously induced GS/GOGAT activities and more active TCA cycle (Fig. 1) that provide C and metabolic energy to efficiently reduce and assimilate 15 NO 3 − into amino acids, particularly [ 15 N]Asn (Fig. 3).
[ 13 C]Pyr highlights the flexibility of TCA cycle to meet de demand of C skeletons for efficient ammonium detoxification. In order to explore the rearrangement of the TCA cycle in response to the N source in vivo, [ 13 C]Pyr was supplied to wheat roots. [ 13 C]Pyr is decarboxylated by pyruvate dehydrogenase and thus, the 13 C label enters into TCA cycle as a two-carbon acetyl CoA molecule. In one 'turn' of the cycle, two carbons are lost as CO 2 but the entering labelled acetyl carbons remain present in the recycled OAA molecule 49 (Figs. 6 and 7a).
Although the initial uptake of [ 13 C]Pyr was equal for both N sources, the higher consumption of [ 13 C]Pyr in RA (Fig. 6) agrees with the higher synthesis of both total [ 13 C]organic acids and [ 13 C]amino acids (Fig. 5), evidencing that wheat plant adaptation to ammonium nutrition induces the acceleration of C metabolism through TCA cycle. Such response is principally illustrated by the enhanced [ 13 C]Citrate accumulation in RA (Fig. 6), sustained by the remarkable CS activity (Fig. 2). Citrate, as one of main C reservoirs in the cell, will assure the functioning of the respiratory cycle 22,50,51 . Indeed, the labelling study carried out by Gauthier et al. 52 showed that N assimilation in illuminated Brassica napus leaves was principally maintained by the citrate and malate stored during the night period. On the other hand, despite the higher ICDH activity in RA (Fig. 2), the content and accumulation dynamic of [ 13 C]2-OG was almost equal in both RA and RN due to its extremely rapid use for the synthesis of [ 13 C]Glu and [ 13 C]Gln (Fig. 6). Low levels of 2-OG are usually detected in the cell when compared to other organic acid (Supplementary Fig. S2) because it is rapidly incorporated into diverse N-containing molecules. Indeed, the predominance of [ 13 C]amino acids labelled in two of their carbons (Table S4) underlined that [ 13 C]2-OG was the principal C skeleton donor from [ 13 C]Pyr, and therefore essential in detoxification of high ammonium contents in wheat root. Likewise, part of the 13 C flux continued alongside the TCA cycle leading to high [ 13 C]Malate contents, especially in RA (Fig. 6), that lead to the improved formation of two-carbon labelled [ 13 C]Asp and [ 13 C]Asn via OAA (Table 3, Fig. 7a). 2-OG has been also suggested to function as a signalling molecule, among others, for the regulation of primary metabolism in eukaryotic algae and cyanobacteria 53,54   www.nature.com/scientificreports www.nature.com/scientificreports/ intermediaries of the cycle by the anaplerotic routes. Several alternative flux modes were proposed in which the initial substrate of the cycle will be PEP instead of being Pyr 22,49,55 . In this line, in the late period of the incubation with [ 13 C]Pyr, a substantial part of [ 13 C]Succinate, [ 13 C]Fumarate and [ 13 C]Malate showed to be labelled only in one carbon (Table 4). The most plausible explanation is that PEPC will be reassimilating the 13 CO 2 from [ 13 C] Pyr decarboxylation into one-carbon labelled OAA that will be fuelling the left branch of the cycle. This reassimilation was also observed in leaves after the provision of 13 C labelled substrates ([ 13 C]Pyr, [ 13 C]Glucose, or 13 CO 2 ) 51,56 . Based on these results, an open flux model is proposed for ammonium-adapted wheat roots (Fig. 7b) where the TCA-related anaplerotic enzymes accomplish a pivotal role. In accordance with this model, [ 13 C]OAA will be diverted both to the synthesis of one-carbon labelled [ 13 C]Asp by AAT or one-carbon labelled [ 13 C]Malate through the induced reversible activity of MDH (Fig. 2). Finally, [ 13 C]Malate can be decarboxylated by NADP-ME in the cytosol to produce unlabelled Pyr, which will enter again into the mitochondrial TCA cycle (Fig. 7b). This highlights the dual role of malate, together with citrate, in the TCA cycle to produce energy or carbon skeletons. Therefore, the coordinated operation of PEPC, MDH and ME in different organelles provides TCA cycle with the flexibility to function in an open-mode according to the metabolic demands of the cell 22,29,57 .
In conclusion, in vivo metabolic flux analysis evidenced that the enhanced ammonium assimilation machinery together with the increased C flux through non-cyclic TCA pathways are essential to sustain the needed amino acid synthesis for the detoxification of the NH 4 + . 15 N labelling together with inhibitor assays asserted that NH 4 + is exclusively assimilated via GS/GOGAT in wheat roots, discarding the participation of GDH. Besides, the adaptation of the root to ammonium nutrition is an advantage that promotes a more efficient assimilation of the supplied N, both in form of NO 3 − or NH 4 + . Such response would mean a greater use of the different N sources available in the soil by ammonium-adapted plants that could be a key strategy to increase their assimilation NUE. Thus, the control of the nitrogen source availability, notably through promoting the use of ammonium-based fertilizers, merits to take a bigger place in fertilization management strategies, since it could be of interest to maximize NUE by the plant and thus to reduce the impact of nitrogen fertilization on the environment.

Materials and Methods
Growth conditions. Seeds  Free ammonium content was determined using the colorimetric method based on the phenol hypochlorite assay (Berthelot reaction) in aqueous extracts obtained by the homogenization of 25 mg of frozen material with 500 µL of ultrapure water. The homogenates were incubated 5 min at 80 °C and centrifuged at 4000 g and 4 °C. Chlorophyll content was determined following the protocol established by Arnon et al. 58 . To do so, 50 mg of frozen leaf tissue was extracted in 1 mL of 80% aqueous acetone, the homogenates were centrifuged at 12 000 g and 4 °C and the absorbance from the supernatant measured at 645 and 663 nm.
Protein extraction and enzyme activities determination. Protein was extracted from 150 mg of frozen root powder with 1.5 mL of extraction buffer described in Sarasketa et al. 9 . For soluble protein quantification a Bradford based dye-binding assay (Bio-Rad, Hercules, CA, USA) was employed, using bovine serum albumin as standard.
Every enzyme was determined with a 96-well microplate reader (BioTek Instruments). For every enzyme except glutamine synthetase (GS), nitrate reductase (NR) and citrate synthase (CS), 20 µL of extract were incubated during 20 min at 30 °C with 280 µL of reaction buffer and the evolution of NAD(P)H monitored at 340 nm. For NADH-dependent glutamate dehydrogenase (GDH) NADH-dependent glutamate synthase (GOGAT), phosphoenolpyruvate carboxylase (PEPC), NADP-dependent malic enzyme (NADP-ME), malate dehydrogenase (MDH) and NADP-dependent isocitrate dehydrogenase (ICDH) the reaction buffers are described in Sarasketa et al. 9 . For aspartate aminotransferase (AAT) activity, the reaction buffer contained 50 mM MBM (pH 8), 0.2 mM NADH, 10 mM aspartic acid, 1 mM 2-OG, 3 mM pyridoxal phosphate and 3.6 U of MDH mL −1 . GS activity was determined following the formation of γ-glutamylmonohydroxamate (γ-GHM) at 540 nm and NR activity following the formation of KNO 2 at 546 nm as described in Sarasketa et al. 59 . For CS enzyme activity, no DTT was added to the extraction buffer described above. CS activity was measured in extracts 20 µL of extract incubated during 20 min at 30 °C with 280 µL of reaction buffer described in Srere et al. 60 and the formation of 2-nitro-5-thiobenzoic acid (TNB) was monitored at 412 nm.
www.nature.com/scientificreports www.nature.com/scientificreports/ Isotopic labelling. Six plants per tank were harvested and the roots from two tanks were pooled as a sole sample. The roots were washed three times in deionised water and were cut in 3 cm long segments. Root segments (1 g) were pre-incubated during 30 min in 10 mL buffer solution (10 mM MES + KOH, pH 6.5), in order to acclimate the root pieces to the medium.
Labelling was provided by the addition of 5 mM ( 2 Ca or (NH 4 + ) 2 SO 4 , respectively. For every condition and time point three independent samples were analyzed. Root segments were collected at 0, 0.5, 2 and 6 hours after the substrates addition, washed three times with buffer, immediately frozen in liquid nitrogen and stored at −80 °C until extraction for GC-TOF-MS analysis.
As a complementary experiment, in order to inhibit the GS/GOGAT enzyme activities, RN and RA root segments were pre-incubated during 1 hour with 2 mM L-methionine sulfoximine (MSX, Ref.  15 In order to distinguish in labelled samples the proportion of M + n due to natural abundance and that due to the incorporation of the 15 N during the experiments, a correction factor (cf n ) is calculated for each unlabelled amino acid as: Standard pure amino acids are used for this analysis. This correction factor is determined experimentally for each fragment ion of each amino acid and will be used to quantify the amount of naturally enriched amino acids in labelled samples as: This value will be subtracted from the total amount of the labelled amino acid. So, the 15 N enrichment (%) was finally calculated as: The total amount of each amino acid is calculated by the sum of the areas of all the fragment ions (unlabelled and labelled). Calibration curves were obtained for each amino acid by using a mixture of commercial pure amino acids. For individual [ 15 N]amino acid contents, the 15 N enrichment (%) value of each amino acid was multiplied by the total content of the respective amino acid. Data are given on a dry weight basis.
Quantification of 13  www.nature.com/scientificreports www.nature.com/scientificreports/ for 90 min. Then, a second derivatization with 50 µl of N-methyl-N-tert-butyldimethylsilyl-trifluoroacetamide (MBDSTFA, Macherey Nagel, France) was carried out at 70 °C for 30 min. 13 C analyses were performed by GC-MS (436 GC-MS Scion; Bruker) as described in Cukier et al. 61 for 15 N analyses. Importantly, for an optimal separation of derivatized 13 C compounds, the temperature of the oven was regulated according to the following program: 5 min at 70 °C followed by an increase of the temperature at 5 °C min −1 until reaching 300 °C and finally 5 min at 300 °C. The 13 C enrichment (%) calculations were performed in the same way as for [ 15

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.