Metabolic origin of δ15N values in nitrogenous compounds from Brassica napus L. leaves

Authors


P. P. G. Gauthier. Fax: +61 2 6125 5095; e-mail: paul.gauthier@anu.edu.au

ABSTRACT

Nitrogen isotope composition (δ15N) in plant organic matter is currently used as a natural tracer of nitrogen acquisition efficiency. However, the δ15N value of whole leaf material does not properly reflect the way in which N is assimilated because isotope fractionations along metabolic reactions may cause substantial differences among leaf compounds. In other words, any change in metabolic composition or allocation pattern may cause undesirable variability in leaf δ15N. Here, we investigated the δ15N in different leaf fractions and individual metabolites from rapeseed (Brassica napus) leaves. We show that there were substantial differences in δ15N between nitrogenous compounds (up to 30‰) and the content in (15N enriched) nitrate had a clear influence on leaf δ15N. Using a simple steady-state model of day metabolism, we suggest that the δ15N value in major amino acids was mostly explained by isotope fractionation associated with isotope effects on enzyme-catalysed reactions in primary nitrogen metabolism. δ15N values were further influenced by light versus dark conditions and the probable occurrence of alternative biosynthetic pathways. We conclude that both biochemical pathways (that fractionate between isotopes) and nitrogen sources (used for amino acid production) should be considered when interpreting the δ15N value of leaf nitrogenous compounds.

Abbreviations
EA

elemental analysis

EA-IRMS

elemental analyser coupled to isotope ratio mass spectrometer

GABA

γ-aminobutyric acid

GC-C-IRMS

gas chromatography coupled to isotope ratio mass spectrometry

SF

deproteinated soluble fraction

V-PDB

Vienna Pee Dee Belemnite

INTRODUCTION

It is now more than 50 years since the first isotope fractionation between the nitrogen isotopes 14N and 15N associated with a biological reaction (N2 fixation by Azotobacter) was measured (Hoering & Ford 1960). Since then, considerable advances have been made in understanding nitrogen isotope composition (δ15N) in plants, with clear isotopic patterns along nitrogen assimilation or symbiotic N2 fixation. As a matter of fact, δ15N values are now considered as useful tools for investigating the physiology of plant nitrogen assimilation and nitrogen use efficiency. It is now well recognized that 14N/15N fractionation occurs during nitrate absorption and assimilation so that plant organic matter is on average 2–3‰15N depleted compared to inorganic soil nitrogen (for a review see Evans 2001; Tcherkez & Hodges 2008). Such a 15N depletion is nevertheless variable since it depends upon soil N availability (Mariotti et al. 1982; Evans 2001) and correlates to transpiration efficiency of net N uptake (Cernusak, Winter & Turner 2009).

The δ15N value of total plant organic matter hides disparities among plant metabolites. For example, it has been shown that leaf nitrates are 15N enriched (Yoneyama & Tanaka 1999) while secondary metabolites such as alkaloids are 15N depleted (Weilacher, Gleixner & Schmidt 1996). This isotopic difference stems from isotope effects along metabolic pathways [for a review, see (Werner & Schmidt 2002)]. For example, Gln synthetase (EC 6.3.1.2), which fixes ammonia onto Glu to evolve Gln, fractionates against 15N by 16‰ (Yoneyama et al. 1993) and nitrate reductase (EC 1.7.1.1) fractionates against 15N by 15‰ (Ledgard, Woo & Bergersen 1985; Tcherkez & Farquhar 2006), thereby enriching in 15N nitrate molecules left behind and depleting the primary amino acids Glu and Gln. More generally, most enzymes associated with primary nitrogen metabolism (transaminases, Glu synthase, EC 1.4.1.14, Asn synthetase, EC 6.3.5.4, etc.) fractionate between nitrogen isotopes (Werner & Schmidt 2002). However, the isotope composition in metabolites is not only influenced by enzymatic isotope effects but also by metabolic fluxes and commitments (Schmidt & Kexel 1997; Tcherkez et al. 2011). Typically, metabolic reactions that run to completion do not fractionate between isotopes simply because all substrate molecules are consumed; by contrast, incomplete substrate turnover by a reaction implying an isotope effect, for example, at a metabolic branching point, leads to isotope fractionation. Using a flux-modelling approach, we have recently shown that the δ15N value in leaf amino acids depends on source nitrate, the isotope effect during nitrate reduction, reactions of amino acid biosynthesis and photorespiration (Tcherkez 2011). Experimental δ15N measurements in amino acids show a 15N depletion in Gly and Ser (relative to glutamate) and a 15N enrichment in others (Hayes 2001), suggesting indeed the 15N-depleting effect of photorespiration and the 15N-enriching effect of other reactions (such as transaminations). However, experimental data with simultaneous isotopic analysis of several amino acids, nitrate or secondary metabolites are scarce [with the exception of Hofmann et al. (1997) and Bol, Ostle & Petzke (2002), but these two papers do not explore the relationships between nitrates, amino acids and N content] and therefore, the means by which isotopic fractionations and metabolic fluxes are integrated into metabolite δ15N are still uncertain.

As an aid in clarifying 15N distribution among plant compounds, we took advantage of isotope ratio mass spectrometry (IRMS) techniques to measure the δ15N in free amino acids and several metabolic fractions from rapeseed leaves (Brassica napus L.) and explored the relationships between them. We show that the nitrate content had a major influence on δ15N in the leaf soluble fraction. Known metabolic pathways and fluxes satisfactorily explained the δ15N in metabolites on a steady-state basis in the light. We further show that although Glu and Gln content remained the same in the light and in the dark, their δ15N value was dissimilar, likely because of changes in source N. δ15N in other compounds mostly reflects the influence of precursors and biosynthetic/consuming reactions.

MATERIAL AND METHODS

Plant material

Seeds of canola (Brassica napus var oleifera cv Darmor) were germinated in Petri dishes on wet Whatman paper. After 72 h, seedlings were transferred to 500 mL pots filled with potting mix. Plants were grown in the glasshouse under 22/18 °C, 60/55% relative humidity, 16/8 h photoperiod (day/night) as described by Vartanian, Damerval & Vienne (1987). Plants were automatically watered three times a day with nutrient solution [Hydrokani C2 (HURELARC, Baillet en France, France), devoid of ammonium] in which nitrate had a δ15N value of +2.69 ± 0.61‰ with respect to atmospheric N2. Carbon dioxide in air was at natural 13C abundance [δ13C = −8.92 ± 0.55‰, where δ13C is the carbon isotope composition with respect to Vienna Pee Dee Belemnite (V-PDB)]. Sampling was done on 6-week-old plants: mature leaves (rank 5 or 6 from the apex) were cut and instant frozen with liquid nitrogen. Samples were collected in the light and in the dark on light- or dark-adapted plants, respectively. Since there was no significant δ15N variation within the dark or the light period, values indicated as average represent mean values obtained on a number of separate leaf samples (3 biological replicates × 6 close sampling time in Table 2; or 4 biological replicates in Fig. 3). In Fig. 2, all the individual leaf samples were plotted to appreciate inter-leaf variability. Sampled leaves were freeze-dried (lyophilized) and ground to fine powder for further analysis (except for DNA that was purified from fresh material, see below).

Soluble, protein and chlorophyll fractions

The soluble fraction (SF, which contains sugars, organic and amino acids and nitrates) and proteins were extracted as described in Nogués et al. (2004). 100 mg of leaf powder were resuspended in 2 mL of distilled water. After centrifugation (5 min, 10 000 g, 5 °C), the aqueous supernatant was transferred to another tube and the pellet was conserved at −80 °C for chlorophyll extraction. The supernatant was heated at 100 °C for 5 min for protein precipitation. After centrifugation (5 min, 10 000 g, 5 °C), the protein precipitate was frozen and lyophilized and the supernatant was used as the deproteinated SF. It was lyophilized and 1.4 mg was weighed in tin capsules for isotopic analysis. The solvent extraction of chlorophylls was carried out on the pellet obtained above, using ethanol 96% v/v. After 10 min agitation at ambient temperature and centrifugation (5 min, 10 000 g, 5 °C), the solvent phase was transferred to a tube. 200 µL were transferred to thick tin capsules adapted for solvents and ethanol was oven evaporated at 35 °C.

Nitrate content and purification

The nitrate content was measured in the SF using a nitrate-selective electrode (CI-6735; PASCO Scientific, Roseville, CA, USA) calibrated with standard nitrate solutions of known concentrations at ambient temperature. 1 mL of SF was first diluted 50-fold with distilled water and then measured under continuous agitation. For isotopic analyses, nitrates were purified from aqueous leaf extracts by collecting the HCl loading fraction through a cation-exchange column (see below, Amino acids extraction . . .).

Isotopic analyses

Dried SF, chlorophyll, protein, nitrate and DNA were analysed by elemental analysis-isotope ratio mass spectrometry (EA-IRMS) using an elemental analyser (Flash-EA; Thermo Fisher Scientific, Illkirch, France) and an isotope ratio mass spectrometer (Optima; Elementar, Villeurbanne, France). EA-IRMS settings were adjusted (split ratio and trap current) so as to have a sufficient mass 44 signal (1.0 10−9 A) in all instances including for small samples (proteins). Any possible δ15N offset was corrected using reference material of known isotope composition (glutamic acid USGS40, −4.5 ± 0.1‰ and caffeine IAEA600, +1 ± 0.2‰; International Atomic Energy Agency, Vienna, Switzerland) included in each sample trial. Similarly, δ13C values were corrected for any offset using standard glutamic acid (USGS40, −26.4 ± 0.04‰). In this paper, all δ15N and δ13C values are given with respect to atmospheric N2 and V-PDB as the standard reference, respectively.

Amino acids extraction, derivatization and isotopic analysis

The extraction and analysis of amino acids were carried out after Molero et al. (2011). Briefly, 250 µL of L-norleucine solution (1 mmol L−1) were added to 2 mL of SF as an internal reference (δ15N = 17.0 ± 0.4‰). Samples were then spin-dried and kept at −80 °C. The samples were resuspended in 1 mL HCl 0.1 mol L−1 and purified through a cation-exchange column (Dowex 50W X8 H+, 200–400 mesh size; Sigma-Aldrich, Saint-Quentin Fallavier, France). The amino acid-enriched fraction was obtained by elution with NH4OH and dried with an infrared lamp under a non-oxidizing atmosphere (N2 flow). Samples were then derivatized with 50 µL of N-methyl-N-(tert-butyldimethylsilyl)trifluoroacetamide and 50 µL pyridine and incubated at 70 °C for 1 h. The isotopic analysis of amino acids was carried out by gas chromatography-combustion-isotope ratio mass spectrometry (GC-C-IRMS). The GC-C-IRMS coupling consisted of a GC6890 gas chromatograph (Agilent Technologies, Palo Alto, CA, USA) coupled to a Delta-Plus spectrometer through the GC-C-III combustion interface (Thermo Fisher Scientific, Courtaboeuf, France). Chromatography was carried out with a PTE-5 column (30 m × 0.32 mm × 1 µm; Supelco, Schelldorf, Germany). Helium was used as the carrier gas for separation. A volume of 1 µL was injected in splitless mode at an injector temperature of 270 °C. The temperature programme used was 90 °C for 1 min, ramping at 8 °C min−1 to 140 °C for 5 min, then ramping at 3 °C min−1 to 220 °C and finally ramping at 12 °C min−1 to 285 °C, holding for 12.5 min. Water was trapped using a Nafion® membrane. CO2 was trapped with liquid N2. The chromatographic sequence of amino acid derivatives was checked by injecting the same samples into a GC8060 gas chromatograph (Fisons Instruments, Manchester, UK) coupled to a MD800 mass spectrometer (ThermoFinnigan, Bremen, Germany) using helium as the carrier gas. Amino acids were separated on a DB-5MS column (30 m × 0.25 mm × 0.25 µm; Agilent Technologies, Santa Clara, CA, USA). Amino acid derivatives were identified by their mass spectra (Mass Spectral Library: NIST 05, NIST, Gaithersburg, MD, USA). The absolute concentration of amino acids was obtained from GC-MS signals with calibration curves using a standard amino acid mix [AAS18, Sigma-Aldrich, in which norleucine, Gln, Asn and γ-aminobutyric acid (GABA) were added], after correction for the recovery of norleucine. The δ15N value of amino acids was corrected for any offset using norleucine as an internal isotopic standard.

DNA purification

1 g of leaf fresh material was extracted in a mortar at ambient temperature with 20 mL of extraction buffer [Tris-HCl 200 mmol L−1, NaCl 250 mmol L−1, ethylenediaminetetraacetic acid (EDTA) 25 mmol L−1, sodium dodecyl sulphate (SDS) 0.5%, pH 7.5]. After centrifugation (15 min, 12 000 g, 5 °C), the supernatant was mixed with phenol/chloroform/3-methylbutanol (25/24/1 v/v/v) and agitated to separate the different phases. The aqueous phase was collected and mixed with isopropanol to induce DNA flocculation. After centrifugation (15 min, 12 000 g, 4 °C), the supernatant was discarded and the DNA pellet was frozen with liquid nitrogen and lyophilized.

Modelling and theory

The δ15N in amino acids was computed using the model developed by Tcherkez (2011; Supporting Information Fig. S1). In this model, the reactions of nitrogen primary metabolism are considered and isotopic mass-balance equations are written for each compound. That is, the model assumes a steady state for isotope ratios. For example, if a compound i is (1) consumed by n reactions associated with isotope effects (denoted as αk, k ∈ [1, . . . ,n]), and (2) comes from m reactions associated with isotope effects (denoted as βj, j ∈ [1, . . . ,m]), we have (isotopic steady state):

image

where Fk and Gj are the fluxes associated with reactions consuming and producing i, respectively. Rjs are the isotope ratios of substrates of reactions producing i. The values of isotope effects are reviewed and discussed in detail in Tcherkez (2011) and recalled in Supporting Information Fig. S1.

With this ‘standard’ model, the steady 15N/14N isotope ratio in Glu (RGlu) in the light can be rearranged to (symbols defined in Table 1):

image(1)
Table 1.  Symbols used in eqns 1–4 of the current paper
SymbolMeaning
  1. The subscript of isotope effects α corresponds to the number of the reaction considered.

  2. IE, 14N/15N isotope effect; GOGAT, glutamine-2-oxoglutarate aminotransferase (glutamate synthase); GDC, glycine decarboxylase complex. The units of (iso)flux are µmol m−2 s−1.

α 1 N IE of the glutamine synthetase at the amido-N level (NH3)
α 1 G IE of the glutamine synthetase at the amino-N level (Glu)
α 2 N IE of the GOGAT at the amido-N level (amido-N of Gln)
α 2 G IE of the GOGAT at the amino-N level (amino-N of Gln)
α 5 IE of the nitrate reductase
α 6 N IE of the GDC for NH3 liberation
α 8 IE of the aspartate-2-oxoglutarate aminotransferase
e Flux of Gln accumulation
γ Flux equal to i + (vo – µg)/2 – t
i Influx of reduced N
µ g Accumulation of Ser escaping from photorespiratory cycle
R*Isotope ratio of leaf nitrate
r Flux of Asp consumption
τ Gln-N Consumption isoflux of the amido-N of Gln
τ Gln-G Consumption isoflux of the amino-N of Gln
τ Gly Consumption isoflux of Gly
τ Glu Consumption isoflux of Glu
t Rate of NH3 escape from the mitochondrion
v o Rubisco-catalysed oxygenation rate

Where α1N (1.016), α1G (1.000), α2N (1.022), α2G (1.000), α5 (1.015), α6N (0.995) are the isotope effects associated with glutamine synthetase (subscript 1), glutamate synthase (subscript 2), nitrate reductase (subscript 5) and glycine decarboxylase complex (EC 1.4.4.2, 2.1.2.1, 1.8.1.4; subscript 6). For glutamate synthase and glutamine synthetase, isotope effects at the amido and amino-N atom levels are distinguished with superscripts N and G, respectively. e is the rate of Gln accumulation/export in the light (within the 0.05–0.2 µmol m−2 s−1 range), vo is the ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco)-catalysed oxygenation rate (typically 5 µmol m−2 s−1), µg the rate of Gly escape from photorespiratory recycling to Ser (0.018 µmol m−2 s−1), i the reduced N input rate (within the 0.05–0.35 µmol m−2 s−1 range), t the rate of NH3 escape from the leaf (0.0004 µmol m−2 s−1) and γ is equal to i + (vo – µg)/2 – t. R* is the isotope ratio in leaf nitrate. Parameters denoted as τ are homogeneous to consumption isofluxes (linear combination of leaving fluxes and inverse isotope effects, 1/α). In this model, the rate of Asp production from Glu by transamination is fixed by mass balance on the Glu content.

However, the Asp content decreased in the light (Fig. 3a), suggesting that Asp could have been consumed rather than produced by transamination from Glu. Under the assumption that Asp is a nitrogen source (with a fixed isotope ratio RAsp), we have

image(2)

Where r is the rate of Asp consumption and α8 the equilibrium isotope effect associated with Asp transaminase (EC 2.6.1.1) in the direction of Asp synthesis (0.9985).

In eqns 1 and 2, it should be noted that the influence of α1N on the isotope ratio in Glu is extremely small because t is negligible (when t = 0, α1N disappears in the equations). That is, NH3 is fully committed to recycling when t = 0 such that the isotope effect associated with glutamine synthetase is of negligible importance (furthermore, note that α1G = 1). In its simplified form (α1G = α2G = 1, t = 0), Eqn 1 gives

image(3)

And Eqn 2 gives

image(4)

In the present paper, r can be roughly estimated with the dark-to-light decrease of Asp content (Fig. 3), which gives 0.22 µmol m−2 s−1. The minimal i-value can be obtained from the N content (excluding nitrates) in leaf organic matter, that is, 0.025 µmol m−2 s−1. This value is underestimated, however, simply because leaf-assimilated N is redistributed to other organs. The nitrate reduction rate in rapeseed shoots has been shown to be 0.29 µmol m−2 s−1 (Leleu & Vuylsteker 2004). In the two scenarios investigated here (Asp production, Asp consumption), the total amino acid accumulation (sum of Gly, Ser, Glu, Gln, Asp if applicable) matched the total input (nitrate reduction + Asp utilization). To facilitate comparisons, the two scenarios considered here are symmetrical for Asp metabolism: when Asp is assumed to accumulate, the associated rate is fixed at 0.22 µmol m−2 s−1, that is, identical to the rate of Asp remobilization when Asp is assumed to be consumed (see above). Under both scenarios, input and output isofluxes of the whole metabolic system were similar (steady-state condition), at 2.9 (Asp production scenario) and 5.2 (Asp consumption scenario)‰µmol m−2 s−1. Note that the isoflux with Asp consumption was slightly 15N enriched because Asp was more enriched (δ15N = +12.5‰) than nitrogen evolved by nitrate reduction (+9.8‰).

RESULTS

Comparison of δ15N in leaf fractions

The average δ15N value in total organic matter and leaf fraction is reported in Table 2. Leaf nitrate appeared to be substantially 15N enriched. Soluble proteins were isotopically very close to total organic matter while amino acids (weighted average) and chlorophyll were 15N enriched. By contrast, DNA was 15N depleted (nearly 30‰ compared to nitrates). The apparent isotope fractionation between leaf nitrates and organics was thus comprised between 15 and 30‰.

Table 2.  The nitrogen isotope composition (δ15N) of leaf fractions in B. napus (mean ± SD, n = 18)
Leaf fraction δ 15N (‰)
  1. The large standard deviation values come from substantial δ15N variations between plants. The δ15N of nitrate used in the nutrient solution during growth was +2.7 ± 0.6‰. The weighted average of amino acids (see Fig. 3) in the light and in the dark is also indicated. There is no significant day/night δ15N difference in other fractions.

Total organic matter+1.4 ± 1.6
Soluble fraction+8.6 ± 3.9
Heat-precipitated proteins+1.6 ± 2.9
DNA–4.8 ± 2.0
Chlorophylls+6.3 ± 3.4
Leaf nitrate+25.7 ± 1.4
Amino acids (day/night)+9.1 ± 2.4/+5.8 ± 2.4

The natural carbon and nitrogen isotope composition in leaf fractions is represented in Fig. 1, as a deviation from total organic matter. Quite clearly, leaf fractions are more widely distributed along the Δδ15N axis (with a difference of nearly 25‰ between extreme values) than along the δ13C axis (10‰ difference between extreme values). Proteins appeared to be close to total organic matter though slightly 13C enriched by 0.8‰ on average. Chlorophyll and DNA showed substantial scattering along both Δδ15N and Δδ13C axes; nevertheless, chlorophyll was on average 13C depleted by 3‰ compared to total organic matter and DNA was 13C enriched by 1‰. As a result, there was limited overlapping between compounds in the {Δδ13C, Δδ15N} space. Such a pattern likely reflected contrasted metabolic precursors or biosynthetic pathways (see Discussion below).

Figure 1.

Bidimensional isotopic distribution (Δδ13C, Δδ15N) of leaf fractions in B. napus: deproteinated soluble fraction (closed circles), heat-precipitated proteins (open circles), DNA (triangles) and chlorophylls (stars). Isotope composition is expressed relative to total organic matter (TOM) so as to account for plant-to-plant isotopic variations, that is, Δδ13Cfraction = δ13Cfraction – δ13CTOM (and the same for 15N). Continuous lines represent envelope curves.

Relationship between nitrates and the leaf soluble fraction

As seen in Fig. 1, the leaf soluble fraction was always 15N enriched compared to other fractions, but the nitrogen isotope composition was quite variable, with δ15N values between +2 and +15‰. Due to the substantial enrichment in leaf nitrates (Table 2), the leaf soluble fraction was certainly influenced by the nitrate content. In fact, there was a positive relationship between δ15N and leaf nitrate content (Fig. 2a). Furthermore, when expressed on a nitrogen mole fraction basis (fraction of leaf soluble N represented by nitrates, denoted as n), there was a linear relationship between δ15N and nitrate content (Fig. 2b), with a regression coefficient of 0.7. In other words, leaf nitrate content explained 70% of the δ15N variation in the soluble fraction. However, the linear relationship did not coincide with the predicted mixing line between the two major soluble N compounds: nitrates (+25.7‰) and amino acids (average δ15N of +7.5‰). That is, there was a systematic depletion of 5 to 8‰ in the soluble fraction. δ15N in proteins was also related to that in the soluble fraction (Fig. 2c), although with a lower regression coefficient and a slope of nearly 0.6.

Figure 2.

Relationship between the δ15N of the leaf soluble fraction and the nitrate content expressed as mmol NO3 per g dry weight (a) or as soluble nitrogen mole fraction denoted as n (dimensionless, mol NO3 per mol total soluble N) (b), and relationship between soluble fraction and proteins (c) in B. napus leaves. In a, the continuous line stands for the hyperbolic trend of the plot. In b, the thick line represents the mixing line between average amino acid fraction (day/night average, +7.45‰) and nitrate (+25.7‰). The thin line is the linear regression (R2 = 0.70, P < 0.05) that gives y = +1.93 + 15.12x. In c, the thick line is the 1:1 axis and the thin line represents the linear regression (R2 = 0.59, P < 0.05) that gives y = −3.15 + 0.59x.

δ15N in free amino acids

Individual free amino acids were extracted and analysed by GC-C-IRMS (Fig. 3). There were clear differences between amino acids, with Thr and Asp being the most 15N enriched of them and Gly and Asn the most 15N depleted (Fig. 3b). In addition, there were day/night differences, Glu, Gln and most amino acids being more 15N enriched in the light than in the dark. Such a 15N enrichment was statistically significant in Gln only. Except for Asp that decreased in the light, there was little change in leaf amino acid content in the light compared to the dark (Fig. 3a). Ala, Asn, Gly and Val were not very abundant.

Figure 3.

Content (a) and δ15N values (b) in free amino acids in B. napus leaves in the light (open bars) and in the dark (closed bars). Observed values are mean ± SD (n = 4). The nitrogen isotope composition in amino acids was obtained by GC-C-IRMS. The δ15N value and the content indicated for Pro is associated with Pro and GABA taken together since they co-eluted under our chromatographic conditions.

Comparison of modelled versus observed values

The δ15N value in amino acids has been computed using the steady-state model of Tcherkez (2011), modified to account for changes in parametrization (see Material and methods) and the scenario of Asp consumption in the light suggested by the measured Asp content (Fig. 3). The comparison of observed values and modelled values is shown in Fig. 4, in which the two scenarios have been represented (Asp production, Asp consumption). There was a rather good agreement between actual and computed δ15N values in Asp, Glu, Gly and Ser but modelled values were less satisfactory in Gln and GABA. In the latter case, this may have come from the fact that GABA and Pro co-eluted in the GC-C-IRMS profiles, such that the observed δ15N also integrated the contribution of Pro – not accounted for in the model. In addition, the direct precursor of both GABA and Pro is Glu, which was indeed slightly underestimated. The computation of δ15N in amino acids in the dark was not carried out. In fact, dark metabolism does not appear to involve neither Asn accumulation nor Glu degradation (Fig. 3a) and furthermore, the typical 15N enrichment in Gln caused by its consumption (that fractionates against 15N) to sustain Asn synthesis in the dark was not observed (Fig. 3b). It is rather clear therefore that the dark amino acid conversion did not involve Asn metabolism under our conditions.

Figure 4.

Comparison of δ15N values in main amino acids (i.e., amino acids for which isotope effects associated with biosynthesis are documented) obtained experimentally in the light (dark grey bars) or from the steady-state model: with Asp production in the light (dark bars, Eqn 1) or Asp consumption (the δ15N of Fig. 3) in the light (light grey bars, Eqn 2). For further modelling details, see the text (Modelling and theory). In this figure, the isotope composition in GABA is assumed to represent that of Pro + GABA (see legend of Fig. 3).

DISCUSSION

Nitrate has a major influence on leaf δ15N

Nitrate purified from leaves has been shown to be 15N enriched compared to organic matter or other compounds (Yoneyama & Tanaka 1999) and in fact, nitrate was on average enriched by 24‰ compared to total organic matter under our conditions. Such an enrichment is believed to come from the nitrate allocation pattern (for a review, see Tcherkez & Hodges 2008): while nitrate absorption per se may fractionate against 15N by a few per mil (Mariotti et al. 1982), nitrate molecules left behind after reduction by root metabolism are partly exported to shoots. As nitrate reduction fractionates against 15N, shoot nitrates are naturally 15N enriched. However, nitrogen assimilation in roots is believed to be modest, as suggested by 15N-labelling studies and kinetics of N pools: the redistribution velocity of nitrogen from roots is within 3 and 10% of that from leaves (Malagoli et al. 2005) and the N content assimilated by roots represents about 10% of total assimilated N (Rossato, Lainé & Ourry 2001). Likely, the pool of leaf nitrate that is used as the nitrogen source for assimilation and reduction by leaf metabolism is eventually enriched due to the fractionation against 15N by nitrogen assimilation (isotopic Rayleigh effect). The apparent isotope fractionation between nitrate and Glu (or Gln) was 15.8‰[= 25.7 (Table 1) – 9.9 (Fig. 3)], thus matching the isotope fractionation associated with nitrate reduction (Ledgard et al. 1985). Nitrate had a visible influence on leaf soluble fraction and total organic matter and under our conditions, it represented nearly 15% of total leaf nitrogen, that is, had an isotopic impact on δ15N in total organic matter of 0.15 × 25.7 = +3.4‰. Provided nitrates are also naturally 15N enriched in natural samples, this contribution might be of importance in the field since many plants have leaf-accumulated nitrate (Gebauer, Rehder & Wollenweber 1988). However, the extrapolated δ15N in the leaf soluble fraction was not equal to the value in nitrate (+17.1‰ instead of +25.7‰) when nitrate-N mole fraction tended to 100% (n = 1 in Fig. 2b). This discrepancy was possibly explained by (1) the isotopic contribution of increasingly 15N-depleted compounds (such as free nucleotides, NAD+, etc.) when nitrate is abundant and (2) the inherent variability of δ15N values in our data.

Day and night patterns

There was a visible effect of light/dark conditions on the δ15N in amino acids, with a general 15N enrichment in most of them in the light. As nitrogen assimilation is believed to occur mainly in the light (Reed & Canvin 1982; Pilgrim et al. 1993; Delhon et al. 1995; Stitt et al. 2002), this enrichment probably came from the use of 15N-enriched nitrate (see above). The 15N enrichment derived from 15N-enriched nitrate was balanced by a 15N-depleted source in the dark, such as the recycling of proteins or other nitrogenous compounds. Protein recycling is of importance for metabolism under stressful conditions (Araújo et al. 2011) and has also been suggested to occur in ordinary metabolism (Bouma et al. 1994; Zerihun, McKenzie & Morton 1998). In fact, there is a systematic 15N depletion (of 3 to 10‰) in proteins compared to amino acids (Table 1 and Fig. 2) and nitrates (soluble fraction) explain 60% or less of the isotopic signal in proteins (Fig. 2c). The clear 15N depletion in Glu (and Gln) in the dark (Fig. 3) further suggest the involvement of depleting reactions, for example, the production of Asp from Glu in the dark (Fig. 3a) by Asp aminotransferase that fractionates against 14N during Glu utilization (equilibrium isotope effect,(equilibrium isotope effect, Rishavy, Cleland & Lusty 2000).

δ15N in amino acids reflects leaf metabolic fluxes

Here, we have analysed free amino acids rather than amino acids obtained from protein hydrolysis. We thus assume that the δ15N values obtained here reflect dynamics of nitrogen metabolism in leaves. In the light, most amino acids have a δ15N value of around 8‰, with significant 15N depletion in Gly, Ser and Asn. In Ser and Gly, this is likely caused by (1) the fractionation against 15N associated with NH2-transfer from Glu to produce Gly (Tcherkez 2011) and (2) the loss of 15N-enriched ammonia during conversion of Gly into Ser by the Gly dehydrogenase complex (for a rather similar reaction Rodriguez, Angeles & Meek 1993) found a fractionation against 14N of 5‰; Fig. 5). Our calculations based on steady-state equations applied to day metabolism and parameterized with such enzymatic isotope effects satisfactorily predict δ15N in Gly and Ser (Fig. 4).

Figure 5.

Biochemical scheme depicting the metabolic fractionations (in‰) involved in δ15N of leaf compounds. The present figure is simplified in that it does not include all metabolic interactions and simply represents the origin of N atoms. Question marks stand for uncertain fractionation values (see the text). The sign of fractionation values is positive when against 15N and negative otherwise. δ-AL, δ-aminolevulinate; Carb-P, carbamyl-phosphate; Carb-Asp, carbamyl-Asp. The present numerical values were previously reviewed in Tcherkez (2011).

Nevertheless, Glu and Gln are predicted to be near 7.5 and 16‰, respectively, while observed values are 9.9 and 11.6‰, respectively. The simultaneous underestimation of δ15N in Glu and overestimation of δ15N in Gln suggest that the accumulation/export rate of Gln may have been slightly overestimated in the model, thereby ‘trapping’15N-enriched glutamine subsequently not consumed by glutamate synthase. We recognize that accumulation or export fluxes are partly uncertain since they cannot be properly obtained from amino acid content. That is, some data on phloem composition would be necessary to determine the contribution of export to amino acid metabolism. For example, the lack of change in Glu and Gln content in the light compared to the dark (Fig. 3) may cloak a simultaneous increase of production and export, with no net effect on leaf pools.

It should be noted that the δ15N values in Gln and Asn reported here (Fig. 3) do not distinguish between amino (–NH2) and amido (–CONH2) groups, although the amino-N atom has been shown to be substantially 15N depleted compared to the amido-N atom (Sacks & Brenna 2005). The δ15N value in Asn was very different (nearly 14‰ depleted) from that in Asp, suggesting an apparent isotope effect associated with Asn synthesis from Asp (and Gln as an amido donor) as large as 36‰. This value is very high and much larger than the 22‰ obtained in vitro with the enzyme Asn synthetase (Stoker et al. 1996). The present value may be caused by (1) an unusually large isotope fractionation of the rapeseed enzyme compared to other species (although unlikely) and (2) the involvement of other fractionating reactions of Asn metabolism that may deplete Asn in 15N, such as asparaginase (EC 3.5.1.1, that hydrolyses Asn to Asp plus ammonia) or cyanoalanine hydratase (EC 3.5.5.4, that converts cyanoalanine to Asn or Asp) that has been found in Brassica (Ishikawa, Onoda & Hikosaka 2007). Further experiments on Asn/Asp metabolism are nevertheless needed to provide a clear explanation of the high isotopic offset between Asp and Asn.

δ15N in secondary metabolites reflect both precursor δ15N and 14N/15N fractionations

The δ15N value in chlorophyll was relatively close to the average value in amino acids or to the day/night average of Glu. Glu is the precursor of the tetrapyrrolic nucleus of chlorophyll via δ-aminolevulinate (Fig. 5). Chlorophyll biosynthesis likely fractionates against 15N but the associated enzymatic isotope effects are not known. Here, the fractionation against 15N between the precursor (Glu) and the product (chlorophyll) appeared to be small, probably because the intermediates of chlorophyll synthesis do not accumulate. That is, Glu utilization (conversion to δ-aminolevulinate) may be a committed step for chlorophyll biosynthesis and therefore there is no substantial isotope fractionation.

DNA showed a clear 15N depletion compared to amino acids, and this is consistent with isotope effects associated with synthesis of bases (Fig. 5). In fact, most enzymes of pyrimidine synthesis fractionate against 15N [carbamyl-phosphate synthase (Rishavy et al. 2000), Asp carbamoyl-transferase (Waldrop, Urbauer & Cleland 1992), dihydro-orotase (11‰Anderson et al. 2006), orotate phosphoribosyltransferase (probably near 25‰, Zhang, Luo & Schramm 2009)]. Purines are formed via a complex succession of reactions that use Glu, Gln, Asp and Gly as nitrogen sources. Little is known about associated fractionations, but presumably the combination of enriched (e.g. Gln) and depleted (e.g. Gly) nitrogen sources mean that purines are probably only slightly depleted compared to total organic matter. Taken as a whole, DNA is expected to be 15N depleted (as observed) but the magnitude of this depletion is not easily predictable.

CONCLUSIONS AND PERSPECTIVES

There are clear δ15N differences among leaf compounds (see also Werner & Schmidt 2002) and here, we argue that key metabolic pathways are involved (Fig. 5). Among them is photorespiration, which is associated with several isotope fractionations (aminotransferases, NH3 production by glycine decarboxylase). In addition to depleting Gly and Ser in 15N, photorespiration also enriches ammonia. In the framework of our model, the predicted δ15N in NH3 is 15 to 19‰, which is considerably enriched compared to total organic matter. Therefore, ammonia liberated by canopies (for rapeseed canopies, see Nemitz et al. 2000) certainly contributes to enrich atmospheric NH3 in 15N.

At the plant level, the contrasted δ15N value in the different nitrogenous compounds might have consequences on foliar δ15N. So seems to be the case of nitrates that are 15N enriched thereby contributing to the relative 15N enrichment in leaves compared to roots (see above). In the steady state, the δ15N in amino acids exported by leaves should reflect the isotopic input by nitrate assimilation. Still, the steady δ15N value in leaf metabolites is sensitive to physiological parameters (such as photorespiration) that may change along the plant life cycle. Furthermore, the redistribution and remobilization of amino acids (with particular δ15N values) within the plant may also contribute to isotopic differences between organs.

The isotopic difference between compounds is also of considerable importance for isotopic mass balance. For example, since Asn has a rather particular (low) δ15N value and is believed to be involved in phloem transfer (Gaufichon et al. 2010), Asn export from source leaves may contribute to δ15N differences between source and sink organs. This will be addressed in a subsequent study in which phloem amino acid composition and δ15N will be investigated.

ACKNOWLEDGMENTS

The authors acknowledge Ms. Morine Lempereur for technical assistance and Dr. Nathalie Nesi for providing B. napus seeds. This work was financially supported by the Agence Nationale de la Recherche through a Jeunes Chercheurs project (under contract no. 08–330055). P.G. was financed by a PhD Grant from the French Ministère de l'Education Nationale et de la Recherche. P.G thanks COST-SIBAE for its financial support to carry out GC-C-IRMS measurements. M.L and C.M. thank the Institut Fédératif de Recherche 87 for its financial support to carry out isotopic analyses and developing the quality management system. The authors wish to thank the anonymous referees and the associate editors for helpful comments to improve the paper.

Ancillary