Adjustment of growth and central metabolism to a mild but sustained nitrogen-limitation in Arabidopsis


M. Stitt. Fax: +49 331 567 8101; e-mail:


We have established a simple soil-based experimental system that allows a small and sustained restriction of growth of Arabidopsis by low nitrogen (N). Plants were grown in a large volume of a peat–vermiculite mix that contained very low levels of inorganic N. As a control, inorganic N was added in solid form to the peat–vermiculite mix, or plants were grown in conventional nutrient-rich solids. The low N growth regime led to a sustained 20% decrease of the relative growth rate over a period of 2 weeks, resulting in a two- to threefold decrease in biomass in 35- to 40-day-old plants. Plants in the low N regime contained lower levels of nitrate, lower nitrate reductase activity, lower levels of malate, fumarate and other organic acids and slightly higher levels of starch, as expected from published studies of N-limited plants. However, their rosette protein content was unaltered, and total and many individual amino acid levels increased compared with N-replete plants. This metabolic phenotype reveals that Arabidopsis responds adaptively to low N by decreasing the rate of growth, while maintaining the overall protein content, and maintaining or even increasing the levels of many amino acids.


Nitrogen (N) is essential for plant growth (Marschner 1995; Coruzzi 2003; Epstein & Bloom 2005; Miller et al. 2007). It is required for the synthesis of nucleotides and amino acids, which are the building blocks for the synthesis of nucleic acids and proteins, and for the synthesis of phospholipids and many secondary metabolites that have diverse roles in signalling, structure and adaptation. Plants obtain N as nitrate and ammonium, with amino acids also making a contribution (Miller et al. 2007). Nitrate is converted to ammonium by nitrate reductase (NR) and nitrite reductase. Ammonium is converted to Glu via glutamine synthase (GS) and glutamine:oxoglutarate aminotransferase (GOGAT). Low inorganic N typically results in a large decrease in Gln, which is the first amino acid formed during ammonium assimilation a decrease of the Gln/Glu ratio (Stitt & Krapp 1999; Foyer, Parry & Noctor 2003; Lemaitre et al. 2008), decreased levels of many other amino acids, decreased levels of protein and other N-containing structural components like chlorophyll, accumulation of starch and an inhibition of growth (Scheible et al. 1997a; Scheible, Krapp & Stitt 2000; Wang et al. 2000, 2003; Scheible et al. 2004; Gutiérrez et al. 2007). There are changes in plant architecture and phenology, including preferential root growth (Scheible et al. 1997b; Stitt & Krapp 1999), increased lateral root growth (Zhang & Forde 1998; Zhang et al. 1999; Signora et al. 2001; Forde 2002), accelerated senescence (Wang et al. 2000; Diaz et al. 2006; Vanacker et al. 2006) and early flowering (Dickens & van Staden 1988; Bernier et al. 1993).

Some of the responses to N are triggered by nitrate (Crawford 1995; Scheible et al. 1997a,b,c; Forde 2002; Wang et al. 2004). Nitrate induces genes that are required for nitrate uptake and reduction, ammonium assimilation, glycolysis, the oxidative pentose pathway and organic acid synthesis (Crawford 1995; Scheible et al. 1997a, 2000), and represses phenylpropanoid metabolism (Fritz et al. 2006). Nitrate is implicated in systemic signals that regulate shoot–root allocation (Scheible et al. 1997b; Stitt & Feil 1999; Signora et al. 2001; Forde 2002), and in local signals that stimulate the proliferation of lateral roots (Zhang & Forde 1998; Zhang et al. 1999; Walch-Liu et al. 2005; Remans et al. 2006). Transcript profiling has identified large sets of genes that are rapidly induced or repressed by nitrate, including genes encoding transporters and pathway enzymes, transcription factors, protein kinases and protein phosphatases (Wang et al. 2003, 2004; Scheible et al. 2004; Gutiérrez et al. 2007). Many of these responses are adaptive, in the sense that they promote nitrate transport and metabolic processes that are required during nitrate assimilation, and initiate adaptive changes in composition and phenology. Other responses to N require the metabolism of nitrate to organic N metabolites. Downstream signalling regulates large sets of genes involved in cellular growth, including genes encoding components of the protein synthesis apparatus (Scheible et al. 2004; Wang et al. 2004; Gutiérrez et al. 2007). There is increasing evidence for the role of glutamate (Lam et al. 1998; Li et al. 2006; Walch-Liu et al. 2006; Forde & Lea 2007) and the PII protein (Moorhead & Smith 2003; Ferrario-Mery et al. 2006) in regulating responses to N. Nitrate-dependent induction of IPT3 in the roots leads to increased synthesis and export of cytokinins to the shoot (Sakakibara et al. 1998; Takei et al. 2004). Recently, an E3 ubiquitin-ligating protein was identified that is involved in responses to low N (Peng et al. 2007a,b).

The ability to grow in low N is of major importance for agriculture. Over the last decades, the availability of cheap inorganic N fertilizer from the Haber–Bosch process has allowed large increases in crop yield (Smil 1999; Ladha et al. 2005). It is important to decrease the dependence of yield on heavy fertilization. This releases large amounts of N into natural ecosystems (Ladha et al. 2005; Kulkarni, Groffman & Yavitt 2008). Furthermore, as the Haber–Bosch process consumes large amounts of energy, decreased use of inorganic N fertilizer would substantially improve the CO2 balance of food, feed and bioenergy production (Smil 1999).

It is therefore increasingly important to understand the molecular, metabolic and physiological bases of nitrogen use efficiency (NUE) in plants (Marschner 1995; Glass 2003; Hirel et al. 2007). NUE can be subdivided into two processes (Hirel et al. 2007): the efficiency with which N is captured from the soil, and the efficiency with which the absorbed N is used to produce plant biomass. Depending on the context, the latter can be vegetative biomass or seed yield (Lea & Azevedo 2006, 2007). The ability of a plant to remove N from the soil depends on the capacity, affinities and regulation of the N-uptake transporters (Orsel et al. 2004, 2006; Lea & Azevedo 2006), allocation to root growth and root architecture and plasticity (Whu et al. 2005; Remans et al. 2006; de Dorlodot et al. 2007). The efficiency with which N is used for biomass formation depends on the N efficiency of photosynthesis (Garnier, Gobin & Poorter 1995; Zhu, de Sturler & Long 2007) and on appropriate allocation and recycling of N within the plant (Gastal & Lemaire 2002; Good, Shrawat & Muench 2004; Hirel et al. 2007; Lea & Azevedo 2007). There have been considerable efforts to improve NUE by altering the expression of single genes (Andrews et al. 2004; Yanagisawa et al. 2004; Martin et al. 2006; Lea & Azevedo 2007). A complementary approach is to use natural diversity to identify quantitative trait loci for increased growth in low N (Loudet et al. 2002, 2003; Coque & Gallais 2005; Flint-Garcia et al. 2005; Hirel et al. 2007).

Plants respond in a highly dynamic manner to changes in the N supply. While the expression and activity of enzymes, the levels of amino acids and the fluxes in central metabolism show strong diurnal changes (Scheible et al. 1997c, 2000; Matt et al. 2001a,b), adjustments of architecture, allocation and phenology require longer time periods. N responses are often studied in simple experimental systems where seedlings are grown in nutrient agar (e.g. Zhang et al. 1999; Signora et al. 2001). However, such set-ups can only be used for the seedling growth because N rapidly becomes limiting as the plants become larger. The conditions are also highly artificial. To achieve a controlled supply of N over a longer period of time, plants have been grown in hydroponic systems (Matt et al. 2001a,b; Urbanczyk-Wochniak & Fernie 2005), or on sand (e.g. Scheible et al. 1997a,b,c; Fritz et al. 2006) or soil (e.g. Loudet et al. 2003; Lemaitre et al. 2008) and supplied with nutrient solutions at regular intervals. However, it is not trivial to generate a sustained N limitation. As plants become larger, they need to acquire and assimilate an increasing amount of N each day to maintain a given growth rate. This requires supplying a calculated and increasing amount of N in each irrigation cycle (Buysse & Merckx 1995; Ingestad & Agren 1995; Ingestad 1997; Walter & Schurr 1999). This is rarely done, with the result that the plants often become increasingly N-limited with time. The concentrations of inorganic N that are supplied are higher than the Km of the high-affinity uptake systems for nitrate and ammonium (Orsel et al. 2004; Lea & Azevedo 2006). As the added N is rapidly absorbed, the plants alternate between periods when inorganic N is available and periods when it is exhausted (Scheible et al. 1997c; Matt et al. 2001a). Furthermore, as nutrient solution is repeatedly supplied to the entire container volume, including areas that the roots have already occupied, plants can access N without this requiring continued growth of the root system.

The aim of the following experiment was to set up a simple soil-based N-limited growth system for the reference species Arabidopsis that allows a small but sustained inhibition of growth compared with that in N-replete conditions, and where access to inorganic N requires continued root growth. The resulting metabolic phenotype was investigated using robotized assays for enzyme activities and central C and N metabolites (Gibon et al. 2004). This unexpectedly revealed that N-limited Arabidopsis rosettes contain similar levels of protein and higher levels of amino acids than N-replete plants. We propose that metabolism and growth are adjusted to the N supply to maintain internal N homeostasis, and discuss how this might be achieved.


Soil mixtures

Four types of soils were used within this study: (1) Low-N soil consisted of 50% (v/v) white peat (basic substrate, Gramoflor GmbH, Vechta, Germany) and 30% (v/v) fine and 20% (v/v) coarse grained vermiculite (AGRA – RHP, Kausek GmbH, Mittenwalde, Germany). In addition, 260 mg K2HPO4, 396 mg GRANUKAL 85 (80% CaCO3 & 5% MgCO3– Kreidewerke Dammann KG, Soehlde, Germany), 1.6 mg Fetrilon-Combi micronutrient fertilizer (BASF AG, Ludwigshafen, Germany) and 30 mL tap water were added to 100 mL of soil mix, which corresponds to the volume of an individual pot; (2) High-N soil was identical to low-N soil except that it was additionally supplemented with 90 mg solid NH4NO3 per pot. Prior to use, soils were stored for 2 weeks at 10 °C to allow homogenization of nutrients; (3) ‘Stender’Arabidopsis soil mix (Stender AG; Schermbeck, Germany) is based on white peat, clay and coco fiber, and contains a slow releasing fertilizer (Osmocot Start); and (4) Einheitserde type ‘GS90’ (Einheitserde; Wansdorf, Germany) is based on white peat and clay and has high levels of free ammonium and nitrate. As the latter two are the standard soils for Arabidopsis growth in the institute, they were used as reference.


Chemicals were purchased from Sigma (Taufkirchen, Germany), except NAD, NADH, NADP, NADPH, and phosphoenolpyruvate (Roche, Mannheim, Germany) and glucose, fructose, sucrose (Merck, Darmstadt, Germany). Xylulose-5-phosphate was purchased from W.D. Fessner (University of Darmstadt, Germany). Enzymes for analysis were purchased from Roche except phosphoglycerokinase and glycerokinase (Sigma-Aldrich, Hamburg, Germany). 13C-glucose was obtained from Campro Scientific (Berlin, Germany) and 13C-glycine from Cambridge Isotope laboratories Inc (CA, USA).

Plant material and growth

All experiments within this study were performed using the Arabidopsis thaliana accession Col-0. Although seeds germinate on low-N soil, seedling establishment and growth is more heterogeneous and there is a strong delay in vegetative development. For this reason, seedling establishment was performed in standard conditions on high-N soil in short-day conditions. Short-day conditions were used to delay flowering. Seeds were initially germinated on standard greenhouse soil (Stender AG; Schermbeck, Germany) in a 16/8 h day/night cycle (20/6 °C) with 75% relative humidity and 250 µE m−2 s−1 light intensity. After 1 week, plants were transferred into an 8/16 h day/night cycle (20/18 °C), 150 µE m−2 s−1 light intensity and a relative humidity of 60/75% for day/night, respectively. After a further week (14 d post germination), individual plants were pricked out into individual pots (volume 100 cm3) containing one of the soils (see above), a transparent lid added to maintain high humidity, and grown for another 7 d in the previous conditions. At 21 d after germination, the lids were removed and the day length extended to a 12/12 h light/dark cycle with 150 µE m−2 s−1 light intensity and 80% constant relative humidity. They were watered every other day with ca. 30 mL water per pot. The low and the high N treatments contained ca. 1.25 and 31.5 mg inorganic N per pot, respectively. Unless otherwise stated, plants were harvested 35–36 d after germination, equivalent to 14–15 d after transfer to the differing substrates. Soil analyses were carried out by LUFA Nordwest, Institute for Soil and Environment, Oldenburg, Germany. Soil pH was measured according to Scheffer et al. (1998).

Elemental analysis

Inductively Coupled Plasma–Optical Atomic Emission Spectroscopy was performed on dried rosette material from low N, high N and GS90 grown plants as previously described (Armengaud, Breitling & Amtmann 2004). Total N content of oven-dried fine powdered leaves was determined by combustion in excess of oxygen at 1100 °C in an Elemental Analyzer CN2000 (Leco Instruments GmbH, Moenchengladbach, Germany). Samples of 50 mg were used and their relative amounts were determined by a thermal conductivity detector with reference to standard compounds with known elemental composition according to Friis, Holm & Halling-Sorensen (1998).

Metabolite analysis

If not stated otherwise, metabolites were extracted twice with 80% ethanol and once with 50% ethanol as in Geigenberger et al. (1996). Starch and proteins were extracted as previously described (Hendriks et al. 2003). Enzymes were extracted according to Gibon et al. (2004). The protein content was assessed with the Bio-Rad Bradford reagent (Bio-Rad, Munich, Germany) according to the manufacturer's instructions. All free amino acids were assayed according to Bantan-Polak, Kassai & Grant (2001) in a final volume of 207 µL containing 0.038 mg of extract, 7.2 mm sodium borate (pH 8.0) and 0.043% fluorescamine. Individual amino acids were quantitatively determined by high-performance liquid chromatography after o-phthaldialdehyd derivation as previously described (Geigenberger et al. 1996). Nitrate detection was adapted to microplates from previously described coupled enzymatic assays (Mori 2000). A 100 µL reaction included 9.5 µg extract, 0.1 m potassium phosphate (pH 7.5), 0.25 mm NADPH and 0.005 U NR. After 30 min incubation at room temperature, 20 µL of 5 mm phenazine methosulfate were added and the microplates were incubated for further 10 min in the dark. Nitrite was then detected with the addition of 30 µL of 2% (w/v) sulphanilamide in 2 m H3PO4 and 30 µL of 0.04% (w/v) N-(1-naphtyl)-ethylenediamine dihydrochloride. Chlorophyll content was determined as previously described (Arnon 1949) with 1.5 mg of extract diluted in 120 µL of 98% ethanol. Malate and fumarate were measured as described in Nunes-Nesi et al. (2007) with 0.3 mg of extract in the presence of 100 mm Tricine/KOH buffer (pH 9.0), 1 mm thiazolyl blue, 0.125 mm phenazine ethosulfate, 3 mm NAD+ and 0.5% (v/v) Trition X 100 in a microplate reader at 570 nm after adding 1 U malate dehydrogenase (determination of malate) and once absorbance stabilized, by adding 0.1 U fumarase (determination of fumarate). 2-oxoglutarate (OG), pyruvate and phosphenolpyruvate (PEP) were assayed in perchloric extracts as previously described (Scheible et al. 1997c). Metabolite extraction for gas chromatography mass spectrometry (GC-MS) and further analysis was performed as described in Lisec et al. (2006).

Enzyme measurements

Absolute enzyme activity analyses were performed according to Gibon et al. (2004) and Sulpice et al. (2007) using 96 tip head robots (EP3 and Janus, Perkin Elmer; Zaventem, Belgium).

Statistical analyses

All statistical analyses were carried out in R Version 2.6.1 (The R Foundation for Statistical Computing, 2007, and SigmaPlot10 (Systat Software Inc, 2006,

Correlations (Pearson coefficients) between trait values were calculated with the least square means of accessions.


Establishment of the plant growth system

To establish a soil-based system where growth is subject to a sustained limitation by inorganic N, we provided a relatively large volume (100 cm3 per plant) of a soil-based substrate with low N content. To identify a suitable soil, we analysed the inorganic N content in different commercially available soils. The inorganic N content was rather variable, depending on the batch. Even low-quality soils sometimes contained substantial amounts of inorganic N. The most reproducible source of low-inorganic N substrate was commercial peat. After supplementation with vermiculite to increase water holding capacity and provide aeration, this peat-based substrate had a typical N content of 1.25 mg inorganic N per 100 cm3. Between five separate batches, the N content varied from 0.9 to 1.4 mg inorganic N per 100 cm3.

When Arabidopsis was grown in a peat–vermiculite mix supplemented with standard macronutrients and micronutrients, with the exception of N, biomass (estimated as fresh weight, FW) after 5 weeks was two- to threefold lower than in a standard nutrient-rich soil (‘Stender’). To check that this growth reduction was due to low N, we supplemented the low-N peat with varying amounts of solid NH4NO3 (Fig. 1). Biomass increased progressively when 30 and 60 mg NH4NO3 (corresponding to 10.5 and 21 mg N) were added per pot, remained high until 120 mg per pot (corresponding to 42 mg N) and declined slightly when higher amounts of NH4NO3 were added. The biomass in the presence of 21–42 mg additional inorganic N was similar to that on standard high-nutrient ‘Stender’ (Fig. 1) or GS90 (not shown) soils. For most future experiments, we added either 0 or 90 mg additional NH4NO3 per pot, resulting in a total of ca. 1.25 and 31.5 mg inorganic N per pot, respectively. These are referred to as the low and high N treatments (Fig. 2a). Soil nitrate and ammonia contents were also determined in the soil after plant harvest at 36 d. At this time, total inorganic N in low N (1.4 mg per pot) was similar to the starting level. Much higher levels (15.9 mg per pot) were found at this time in high-N soil.

Figure 1.

Effect of supplementation of solid NH4NO3 to the low-N soil mixture on shoot fresh weight compared with standard soil for Arabidopsis growth (Stender control, white bar). Data are given as mean ± SD (n = 3). Each sample consisted of three pooled plants. Harvest was performed 35 dpg.

Figure 2.

Comparison of low N (red), high N (blue) and Stender (white). (a) Basic recipe for soil mixtures. (b) Shoot nitrate content 35 dpg. (c) Shoot fresh weight 35 dpg. (d) Rosettes 35 dpg; low N (left), high N (center) and ‘Stender’ (right). (e) Relative growth rate over vegetative development. (f) Total amino acid content in shoots 35 dpg. (g) Soluble shoot protein 35 dpg. (h) Total N content 40 dpg. Data are given as mean ± SD (n = 3).

We checked that other nutrients were not limiting in the peat substrate. Compared with the high N treatment, plants contained significantly (* = significant at P < 0.05) elevated levels of K (4.40 ± 0.31*, 3.51 ± 0.17 and 3.18 ± 0.28,), P (0.85 ± 0.03*, 0.77 ± 0.06, 0.60 ± 0.00) and S (0.89 ± 0.06*, 0.63 ± 0.08 and 0.58 ± 0.05) in low N compared with high N or the ‘Stender’ control, and no significant differences for B, Zn, Fe Na and Mg (data not shown). This is equivalent to an increase of ca. 23, 3 and 8 µmol K+, SO42− and PO42−, respectively. The increase of K+ is double that of the two anions, presumably reflecting the fact that SO42− and PO42− carry two negative changes at physiological pH. These results, nevertheless, indicate that the decrease of nitrate is partly compensated for by an increase of other inorganic solutes. Soil pH was measured in all treatments (low N, high N and ‘Stender’ control) immediately before planting and after harvest. Soil pH increased from 5.4 (before use) to 6.1 (after harvest) in all soil mixes. In a separate experiment, we checked that the NH4+ levels in the low N treatment were not inhibitory. Addition of 8 mg NH4Cl (corresponding to 2.7 mg N) per pot in the absence of additional nitrate led to a 1.33-fold stimulation of growth. We also compared the performance of A. thaliana Col-0 on a purely nitrate-based nutrient solution according to Gibeaut et al. (1997) with that on ammonium nitrate-based nutrient solution (Loque et al. 2003), and found that Arabidopsis performed better on the ammonium nitrate-based nutrient solution (data not shown). Furthermore, in peat-based soil, ammonium will bind during the pre-incubation period to the ion-exchange capacity of peat. During cultivation, ammonium will be successively converted to nitrate by soil microorganisms, and the plant roots will not be exposed to high ammonium concentrations.

Figure 2b–h compares plants that were grown in low and high N regimes and harvested 35 d after germination. Plants in low N contained about 15 µmol nitrate gFW−1 (Fig. 2b), which is about fivefold lower than in high N or the ‘Stender’ control. In many other experiments, we have also seen that Arabidopsis Col-0 rosettes typically contain >100 µmol nitrate gFW−1 in N-replete conditions. The vast majority of this nitrate is located in the vacuole (Miller & Smith 1996; Miller et al. 2007). Rosette biomass was decreased by over 50% in low N (Fig. 2c). The rosette cross section, leaf area, leaf number and petiole length were decreased, but the plants showed no other visible symptoms; they were as green as the controls and showed no accumulation of anthocyanin (data not shown). Visual analysis of the soil revealed proliferation of roots through the entire pot. It was not possible, however, to quantitatively recover root material for analysis. Crucially, analysis of a more extensive experiment with successive harvests (see below for more data) revealed that low-N plants showed a stable 20% reduction of the relative growth rate (RGR) of the rosette across the entire period from day 22 to day 39, compared with plants growing in high N or the ‘Stender’ control (Fig. 2e). There is an indication of a slowing of growth during the last time interval (35–39 d), but even at this time, the plants are still accumulating biomass and (not shown) did not show any visual signs of stress.

Unexpectedly, low-N plants contained similar levels of protein and higher levels of total amino acids (assayed using fluorescamine) in their rosettes than the high-N plants (Fig. 2f–g). There was a small decrease of total N content in low-N plants compared with high N and ‘Stender’ controls [6.2, 7.5 and 7.3% of dry weight (DW), respectively; Fig. 2h]. This small decrease of total N is due to the 10-fold reduction in nitrate content (Fig. 2b; 90 µmol nitrate gFW−1 is equivalent to 1.25% N per g DW). Similar total N contents have been reported in earlier studies with Arabidopsis (e.g. Schulze et al. 1994; Orsel et al. 2006; Lemaitre et al. 2008). The DW/FW ratio was similar in low N, high N and ‘Stender’ (0.088, 0.083 and 0.088, respectively, data not shown.

A plant growing with a RGR of 0.25 mgFW mgFW−1 per day in high N requires about 1.5 µg N mgFW−1 per day. This integrates to a total of about 0.18, 0.54, 0.72, 1.44 and 3.0 mg N per plant after 26, 30, 32, 35 and 39 d, when the FW is about 30, 90, 120, 240 and 500 mg, respectively (see below for data). Thus, there is a >10-fold excess of inorganic N (about 31.5 mg per pot) in our high N regime. Our low-N growth system provides about 1.25 mg inorganic N per pot (see above). This is less than half that required to support the growth of a 40-day-old N-replete plant. It is barely enough to support the observed growth in the low N regime, where the plants are two- to threefold smaller than in high N.

Interaction with the photoperiod

We checked whether the inhibition of growth depended on the photoperiod length (Fig. 3). Changing the photoperiod will alter the amount of C fixed each day and the C/N balance plant. In this experiment, the high N treatment received only 45 mg inorganic N per pot, and would therefore still have been marginally N-limited (see Fig. 1). In a 6/18 h light/dark photoregime, a similar rosette biomass was found in high N and low N. In a 12/12 h light/dark photoregime, rosette biomass decreased by about 20% in low N compared with high N. This matches the inhibition expected from Fig. 1. In an 18/6 h light/dark photoregime, rosette biomass was twofold lower in low N than high N. In long days, the plants had developed a floral inflorescence in both treatments but had not yet bolted. Although longer photoperiods increased the response to high N, we used a 12 h photoperiod in future experiments to avoid possible complications due to the transition to flowering.

Figure 3.

Influence of photoperiod on shoot fresh weight in low-N (red) and high-N (blue) grown plants. Data are given as mean ± SD (n = 6–12). Each sample consisted of five pooled plants. Harvest was performed 36 dpg.

Age-dependent changes of protein, chlorophyll, enzyme activities and metabolites

Figure 4 shows a more extensive analysis of the changes in metabolites and enzymes from central C and N metabolites in plants growing in a 12 h photoperiod on white peat with 0 (low N) or 90 mg (high N) additional N or on standard ‘Stender’ soil. The plants were harvested at the end of the light period, 22, 26, 30, 32, 35 and 39 d after germination. This experiment was performed for three reasons: to check if the changes in composition noted in Fig. 2 are maintained across a longer time period, to exclude the possibility that the unexpected response of amino acids in low N is a plant-size effect and to provide more information about other changes in metabolism. It should be noted that the plants were maintained in short-day conditions during seedling establishment, and were not transferred to a 12 h photoperiod until day 21 after germination. Growth will not have been strongly reduced until this transfer (Fig. 3). The response at the first time points may therefore also be affected by transient changes due to changes in the C/N balance or other changes after lengthening of the photoperiod. The response at later time points may also be affected by the gradual exhaustion of N in the soil (see Discussion). The complete data set and abbreviations are provided in Supporting Information Table S1, and a diagram showing the location of the enzymes and metabolites in metabolism in Supporting Information Fig. S1.

Figure 4.

Developmental changes biomass, major metabolites and enzymes in shoots of low-N (red), high-N (blue) and Stender (white) grown plants. (a) Fresh weight. (b) Nitrate. (c) Chlorophyll. (d) Protein. (e) Total amino acids. (f) Starch. (g) Sucrose. (h) Malate. (i) Fumarate. (j) Glutamine. (k) Glutamate. (l) Aspartate. (m) Alanine. (n) Glycine. (o) Serine. (p) Asparagine. (q) Arginine. (r) Leucine. (s) GABA. (t) NR Vmax activity. (u) GS activity. (v) Fd-GOGAT activity. (w) ShikimateDH activity. (x) AlaAT activity. (y) GluDH activity. (z) PEPC activity. Data are given as mean ± SD (n = 3–6). Each sample consisted of three to 10 pooled plants. Harvest was performed at the end of the light period.

Rosette FW increased with time in all three N regimes but increased more slowly in low N than in high N or the control treatment (Fig. 4a). As already indicated, Ln-transformation of these data reveals a sustained 20% decrease in RGR (Fig. 2e).

Nitrate levels (Fig. 4b) remained steady at >150 µmol gFW−1 in the high N treatment, and 60–100 µmol gFW−1 in the control ‘Stender’ soil. The nitrate levels in the high N treatment are even higher than those in the experiment of Fig. 2. In the low N treatment, nitrate declined from an initial value of 20–25 µmol gFW−1 at days 22–26 to about 10 µmol gFW−1 at days 32–35 and 5 µmol gFW−1 at day 39. Chlorophyll (Fig. 4c) and total rosette protein content (Fig. 4d) were initially twofold lower in low N, but recovered by 25 d to levels that were similar to those in the high N and control treatments. This recovery was maintained until the end of the experiment. Total rosette amino acids in low N (Fig. 4e) were initially lower but rose after 26–30 d to similar levels and after 32–39 d to 30–100% higher levels than in high N and the control. Amino acid levels are negatively correlated with nitrate (R2 = 0.62).

Starch typically accumulates in low N (Scheible et al. 1997a; Stitt & Krapp 1999; Foyer et al. 2003; Lemaitre et al. 2008). Nitrate assimilation typically leads to the accumulation of malate, which acts as a counter-anion (Benzioni, Vaadia & Lips 1971; Smith & Raven 1979; Scheible et al. 1997a; Stitt & Krapp 1999). In Arabidopsis, large amounts of fumarate also accumulate (Chia et al. 2000). Starch levels were higher in low N (Fig. 4f) and sucrose (Fig. 4g) and reducing sugars (not shown) were broadly similar in low N and high N. Malate (Fig. 4h) and especially fumarate (Fig. 4i) levels were strongly decreased in low N, especially at the later time points when nitrate was very low and amino acids accumulated. Nitrate correlated strongly with malate (R2 = 0.71; Fig. 4) and less strongly with fumarate (R2 = 0.41) in the low N treatment (plots not shown). The changes in C metabolism are fully consistent with a N limitation of metabolism and growth, despite the fact that high levels of amino acids accumulate in the low N treatment.

With respect to osmotic balance, the large decrease of nitrate in low N (>100 µmol gFW−1) is not offset by changes in the levels of other central metabolites. Sugars show only small changes, amino acids increase by about 30 µmol gFW−1 and organic acids decrease by about 10 µmol gFW−1. As already noted, elemental analysis revealed an increase of K, P and S in the experiment in Fig. 2. Elemental analyses were not repeated in subsequent experiments. The increases of K, P and S noted for the experiment in Fig. 2 would be equivalent to an increase of ca. 23, 3 and 4 µmol K+, SO42− and PO42−, indicating that changes of other inorganic solutes partly compensate for the large changes in nitrate.

Figure 4j–r investigates the responses of individual amino acids. The temporal response of the individual amino acids was similar to that of the total amino acids, with many showing a transient decrease at 22–25 d, and a recovery later in the experiment. Summing the individual amino acids reveals strong agreement with the total levels measured by the fluorescamine assay (R2 > 0.9, data not shown).

Gln rose progressively to levels that were higher in low N than in high N (Fig. 4j). Glu (Fig. 4k) rose to higher levels from 25 d onwards. Asp (Fig. 4l) and Ala (Fig. 4m) were lower in low N than in high N. Except at the first time point, Gly was similar and Ser was considerably higher in low N than high N or the control ‘Stender’ treatment (Fig. 4n,o). This implies that there is a restriction on the transfer of amino groups from Glu to Asp and Ala, and from Ser to other amino acids. Asn was lower in low N than high N but similar to the ‘Stender’ control. Analysis of the minor amino acids revealed that many were increased (Arg, Lys, Phe, Tyr, Trp, Ileu, His) or unaltered (Val) in low N, with only one (Met) showing a decrease (Supporting Information Table S1). Arg and Leu are shown as examples of minor amino acids that increased in low N (Fig. 4p–q). GABA was also consistently two- to threefold higher in low N than in high N or the ‘Stender’ control (Fig. 4s).

Figure 4t–y shows the activities of several enzymes that are involved in N metabolism. Almost all enzyme activities decrease in low N at 22 d, and then recover. This resembles the initial decrease and recovery of the total protein content (see Fig. 4d). The following presentation concentrates on the values between days 25 and 39. NR was assayed in the absence of magnesium to detect total NR activity, and presence of magnesium to distinguish total and post-translationally activated (dephosphorylated) enzyme (Kaiser & Huber 1994; MacKintosh, Douglas & Lillo 1995). Total (Fig. 4t) and post-translationally activated (Supporting Information Table S1) NR activities were consistently decreased by 30–45%. GS activity was unaltered (Fig. 4u), Fd-GOGAT activity (Fig. 4v) and AlaAT (Fig. 4x) were slightly decreased, and ShikDH (Fig. 4w) was unaltered in low N. Apart from NR, the most marked change was for GluDH, which increased twofold in low N (Fig. 4y).

N assimilation depends on the provision of C-skeletons. The primary C-acceptor in the GOGAT pathway, 2-OG, is synthesized via combined action of PEPC, PK, pyruvate dehydrogenase, aconitase and ICDH (Stitt & Krapp 1999; Hodges 2002). Several of these enzymes were represented on our platform. PEPC (Fig. 4z) and NADP-ICDH (Supporting Information Table S1) showed a slight (20–25%) decrease, while PK, NAD-MDH and fumarase remained unaltered (Supporting Information Table S1). As already mentioned, malate and fumarate levels were decreased in low N.

RubisCO total activity were slightly reduced at day 22 and subsequently recovered to levels comparable with the high N and control treatments (Supporting Information Table S1). RubisCO activation state remained unchanged over the period of the experiment as well as between treatments (Supporting Information Table S1). This is consistent with the results obtained for total protein content, as RubisCO typically represents 30–50% of total leaf protein (Ellis 1979). There were no marked changes in the activities of several other enzymes in the Calvin cycle, and the pathways for starch and sucrose synthesis, except for sucrose phosphate synthase (SPS) which showed a small increase (Supporting Information Table S1).

Diurnal changes

The levels of nitrate, amino acids and carbohydrates and the activities of enzymes in central C and N metabolism show marked diurnal changes in many species (Stitt & Krapp 1999) including Arabidopsis (Gibon et al. 2004, 2006; Fritz et al. 2006; Usadel et al. 2008). Nitrate typically decreases during the light period and recovers at night, starch and amino acids accumulate in the light and decline at night, and NR activity rises in the first part of the light period, and then decreases again. These diurnal changes are partly driven by the need to accumulate starch in the light to act as a source of C during the night (Smith & Stitt 2007; Stitt et al. 2007) and by transient imbalances between the rate of nitrate uptake, nitrate assimilation and the supply of C-acceptors for ammonium assimilation and amino acid synthesis at different times during the diurnal cycle (Scheible et al. 1997c, 2000; Matt et al. 2001a,b).

The diurnal responses of metabolites and enzymes were investigated during a diurnal cycle in low N and high N to learn whether they are modified by low N, and to exclude the possibility that the unexpected increase in amino acids in low N was a chance effect due to choice of an atypical sampling time. The plants were harvested on day 36. Figure 5 provides an overview of the diurnal responses of some major parameters, and Fig. 6 shows the detailed responses of further individual parameters. A complete data set is provided in Supporting Information Table S2. A second replicate experiment gave similar results (data not shown). Metabolites in low-N and high-N plants were also profiled by GC-MS at the end of the day and the end of the night. The results broadly confirmed those seen by enzymatic and high-performance liquid chromatography analyses (Supporting Information Table S3). Table 1 summarizes metabolites that were only detected by GC-MS.

Figure 5.

Pathways integrating carbon and nitrogen metabolism to produce biomass. The scheme summarizes the diurnal metabolic phenotype found in leaves of low-N (red) compared with high-N (blue) grown plants. In accordance with previous N-limitation studies growth, nitrate content, organic acids, PEPC activity, NR Vmax and its activation decreased in low N while starch, sugars, GluDH activity, SPS and AGPase activity increased. Unexpectedly, protein content remained unchanged while total amino acids accumulated in low N. Data are given as mean ± SD (n = 6). Each sample consisted of five to seven pooled plants. Harvest was performed 35 dpg.

Figure 6.

Diurnal changes in amino and organic acids in shoots of low-N (red) and high-N (blue) grown plants. (a) Glutamine. (b) Glutamate. (c) Aspartate. (d) Alanine. (e) Glycine. (f) Serine. (g) Arginine. (h) Lysine. (i) Leucine. (j) Isoleucine. (k) Methionine. (l) Phenylalanine. (m) GABA. (n) 2-Oxoglutarate. (o) Pyruvate. (p) Phosphoenolpyruvate. Data are given as mean ± SD (n = 6). Each sample consisted of five to seven pooled plants. Harvest was performed 35 dpg.

Table 1.  Levels of metabolites determined by GC-MS
AnalyteLow N vs. high N ENLow N vs. high N ED
Mean ± SEMean ± SE
  1. The results are given as a ratio of the level in low N compared to that high N at the end of the night. The results are given as the mean ± SE (n = 5). A complete list of all metabolites measured by GC-MS can be found in Supporting Information Table S3.

Ornithine4.33 ± 0.566.21 ± 1.45
Pro1.04 ± 0.111.10 ± 0.14
Putrescine0.53 ± 0.050.45 ± 0.05
Succinate0.71 ± 0.030.39 ± 0.02
Citrate0.03 ± 0.000.05 ± 0.00
F6P2.48 ± 0.112.93 ± 0.19
G6P2.65 ± 0.093.81 ± 0.26
Maltose1.20 ± 0.051.00 ± 0.06
Isomaltose0.69 ± 0.160.30 ± 0.07
Trehalose0.91 ± 0.020.87 ± 0.06
Raffinose0.74 ± 0.140.54 ± 0.12
Rhamnose0.73 ± 0.080.44 ± 0.04
Glycerate0.47 ± 0.030.39 ± 0.04

As previously seen, total protein is similar in high N and low N, and does not show any marked diurnal changes (Fig. 5). The same applies for RubisCO activity (Supporting Information Table S2).

Nitrate remained high throughout the diurnal cycle in high N, and low throughout the diurnal cycle in low N (Fig. 5). This is in contrast to early studies with sand-grown tobacco, where nitrate showed a strong decrease in the light period and recovery in the dark when high N was supplied, and a transient peak following irrigation when low N was supplied (Scheible et al. 1997c; Matt et al. 2001a,b) In high N, maximum NR activity rose during the first part of the light period, declined at the end of the light period and started to recover during the night. This resembles the response previously reported in tobacco (see above) and Arabidopsis (Gibon et al. 2004), except that the recovery during the night starts earlier. While NR activity also rose during the day in low N, the overall activity was lower than in high N. To distinguish total and post-translationally activated (dephosphorylated) NR, activity was also measured in the presence of magnesium (see above). Activity in the presence of magnesium showed an even larger decrease in low N (Supporting Information Table S2). The ratio between activity in the presence and absence of magnesium was used to calculate the activation state (Kaiser & Huber 1994). NR activation was lower in low N than high N, and fell to especially low levels in the night (Fig. 5). The activities of other enzymes in N metabolism including GS, Fd-GOGAT, AspAT, AlaAT and ShikDH did not show any marked diurnal changes, and were unaltered between high N and low N (Supporting Information Table S2 and Supporting Information Fig. S2).

Total central amino acid (Gln, Glu, Asp, Ala, Gly, Ser, Asn) and total minor amino acids were higher in low N than high N, and showed a more marked accumulation during the light period (Fig. 5). The increase of major amino acids in low N during the light period in low N was mainly due to a consistently higher level of Gln (Fig. 6a) and an increase of Glu (Fig. 6b), and Ser (Fig. 6f) during the light period. While Gly (Fig. 6e) increased during the light period in both conditions, this was less marked in low N. Asp (Fig. 6c) and Ala (Fig. 6d) levels were similar in the dark in low N and high N but decreased in the light in low N, rather than rising as in high N. The increase of Gln and Glu and decrease of Asp and Ala in low N confirm the response already noted in the developmental series (Fig. 4).

A decrease of the Gln/Glu ratio is often taken as a marker for the N limitation (Stitt & Krapp 1999; Foyer et al. 2003; Fritz et al. 2006). The Gln/Glu ratio was about twofold lower in low N at early time points in the developmental series but at later time points was similar in low N and high N (Fig. 7a). During the diurnal cycle, the Gln/Glu ratio was higher in low N at the end of the night, declined towards the end of the day when it was similar to high N and increased again in the night (Fig. 7b). These results point to major changes in central amino acid metabolism in low N, which differ from those seen in previous studies of N-limited plants.

Figure 7.

Glutamine to glutamate ratio in low N (red), high N (blue) and Stender (white). (a) Over vegetative stages of development, harvested at the end of the light period. (b) Within the diurnal cycle, at 36 dpg. The data are from the expriments shown in Fig. 4 (panel a) and Figs 5 and 6 (panel b). They are given as mean ± SD (n = 3–6 for panel a, and 6 for panel b).

For the minor amino acids, low N led to increased levels of Arg (Fig. 6g), ornithine (Table 1), citrulline (Supporting Information Table S2), Lys (Fig. 6h), Leu (Fig. 6i), Ileu (Fig. 6j), Met (Fig. 6k), Tyr and His (Supporting Information Table S2), slightly higher levels of Thr, and did not alter the levels of Pro (Table 1), Trp, and Val (Supporting Information Table S2). In all cases, the increase of minor amino acids during the light phase was maintained or increased in low N. It is striking that amino acids with a high N/C ratio, like Arg and Lys, are higher in low-N conditions. It has earlier been shown that the level of Phe is extremely responsive to many physiological signals, including light, C and N (Fritz et al. 2006). Phe showed very similar responses in low N and high N (Fig. 6l). There was also an increase in GABA (Fig. 6m), which remained level in high N but rose during the day and fell during the night in low N. In contrast to amino acids, the polyamine putrescine was twofold reduced in low N (Table 1).

There are marked differences in the diurnal response of organic acid metabolism between low N and high N. PEPC activity decreased ∼30% in low N (Fig. 5). The levels of malate and fumarate were also strongly reduced (Fig. 5). Whereas in high N both organic acids rose three- to fourfold during the light and decreased in the dark, in low N, their levels were three- to fourfold lower at the end of the night and did not increase in the light period. This confirms the decrease of PEPC activity and the lower levels of organic acid noted in Fig. 4. It should however be noted that the relative changes in the diurnal fluxes to and from organic acids are far larger than the change in maximum PEPC activity. The activities of other enzymes involved in organic acid synthesis and metabolism including PK, NADP-ICDH, NAD-MDH and fumarase did not change between low N and high N (Supporting Information Fig. S2 and Supporting Information Table S2). As in the developmental series, low N led to an increase in GluDH activity (Fig. 5).

2-OG is the C-acceptor in the GOGAT pathway. It was slightly (20%) but non-significantly lower in low N than high N across the full day/night cycle (Fig. 6n). Pyruvate is the C-acceptor during synthesis of Ala from Glu. It rose >2-fold during the light period in high N but only 1.5-fold in low N (Fig. 6o). Consequently, pyruvate was twofold lower in low N at the end of the light period. This is qualitatively similar to, though smaller than, the decrease of Ala at this time. PEP is the terminal metabolite in plant glycolysis, and serves as a substrate for PEPC and PK. It rose slightly during the light period, and was not significantly different between low N and high N (Fig. 6p). The general trend of decreased organic acids was confirmed by GC-MS metabolite profiling, which also revealed that there is also a strong decrease in citrate and succinate (Table 1).

Starch accumulation was slightly increased in low N compared to high N (Fig. 5). This was accompanied by a small increase in AGPase activity (Fig. 5). The increase of starch was also seen in the developmental series but not the small shift in AGPase activity (Fig. 4). Sucrose (Fig. 5) and reducing sugars (Supporting Information Table S2 and Table 1) were slightly increased while sugars phosphates were strongly increased (Table 1) in low N. While maltose levels were unchanged, isomaltose, trehalose and raffinose were slightly decreased (Table 1).


Soil-based system that leads to a sustained decrease of growth and adjustment of metabolism to low N

The first aim of these experiments was to develop a soil-based growth system that allows a mild but sustained restriction of growth by N. To do this, plants were grown in a large volume of low N soil, or, as a reference, the same soil supplemented with inorganic N. Plants growing in the low N regime consumed almost all the supplied inorganic N, and may also use some N that is mobilized from organic N sources. Continual growth of roots will be required to access the N, which is distributed in a large rooting volume. The low N growth regime led to a sustained 20% reduction of RGR over a period of 2 weeks, resulting in a >2-fold reduction in rosette biomass in 35-day-old plants. However, there was no visual sign of stress, or accumulation of stress-related secondary metabolites, and no decrease of total protein or chlorophyll.

The low N regime led to widespread changes in central C and N metabolism. Similar changes were seen in plants sampled at different ages across a period of 2 weeks (from 25 to 39 d after germination) and at different times during the diurnal cycle in 35-day-old plants. Some of these responses resembled those reported in earlier studies of plant responses to low N, including lower nitrate content and NR activity, lower malate content and PEPC activity, and increased starch. Others were unexpected, including the unaltered protein content and, in particular, an increase of amino acid level and a higher Gln/Glu ratio. We transiently observed the ‘expected’ metabolic response on the first day after transfer from a short day to a 12/12 h light/dark photoregime, when the low-N plants had lower protein content, lower amino acids and a low Gln/Glu ratio. On the following days, amino acid levels and protein rose to higher levels than in the high-N plants, even though the RGR was 20% less in the low N treatment. These results point to a very close interaction between the N supply, metabolism and growth, which serves to maintain the internal pools of organic N metabolites when Arabidopsis is growing with a suboptimal N supply.

Changes in N assimilation and central C metabolism

Even though low-N plants had a >10-fold lower nitrate level than N-replete plants, their rosettes still contained about 5–10 µmol nitrate gFW−1. Furthermore, the nitrate level was rather stable during the diurnal cycle. This contrasts with earlier studies with tobacco growing in sand culture on low N (Scheible et al. 1997a,c), where nitrate peaked temporarily each day after irrigation and then fell to very low levels.

Nitrate assimilation is tightly regulated by NR, which is subject to a hierarchy of transcriptional, translational and post-translational regulation (Hoff, Truong & Caboche 1994; Lea et al. 2006). Whereas constitutive transcriptional activation has little effect, post-translational activation leads to a major decrease of nitrate and increase in the levels of amino acids (Lea et al. 2006). Low-N plants showed a large decrease of NR activity, due to a twofold decrease of the Vmax activity, a more marked diurnal decrease in activity, and weaker post-translational activation.

Malate and fumarate are synthesized as counter-anions for nitrate (Benzioni et al. 1971; Smith & Raven 1979). They accumulate in the light and decrease at night, reflecting the expected temporal distribution of nitrate assimilation (Scheible et al. 1997c; Matt et al. 2001a,b). The lower level of malate and fumarate and the abolition of the diurnal changes in low N are consistent with a lower rate of nitrate assimilation. Other organic acids also decreased, including citrate, succinate and, to a lesser extent, 2-OG. There was also a twofold decrease in PEPC activity, which is required for the net synthesis of organic acids. Nitrate leads to transcriptional induction (Wang et al. 2003; Scheible et al. 2004) and an increase in PEPC activity (Scheible et al. 1997a,c) The decrease in PEPC maximum activity in low N could contribute to the lower levels of organic acids.

In earlier studies, amino acids typically decreased when N was limiting (reviewed in Stitt & Krapp 1999; Foyer et al. 2003, see also Gloser 2005; Man et al. 2005; Tilsner et al. 2005; Urbanczyk-Wochniak & Fernie 2005; Fritz et al. 2006). Similar observations have been made in studies with Arabidopsis growing in liquid-fertilized soil systems (Loudet et al. 2003; Lemaitre et al. 2008) and liquid systems (Scheible et al. 2004). We were therefore surprised to find that the total amino acid content increases in our low N regime. This implies that the decrease in growth rate is not triggered by decreased levels of amino acids (see below for further discussion). There also were shifts in the relative levels of individual amino acids. In central metabolism, Gln, Glu and Ser increased, and Asp and Ala decreased, indicating that there are changes in the regulation of central amino acid metabolism. The Gln/Glu ratio increased instead of decreased, as is expected from previous studies of N-limited plants (see Stitt & Krapp 1999; Foyer et al. 2003; Scheible et al. 2004; Fritz et al. 2006; Lemaitre et al. 2008).

Low N led to a small increase in starch. This is expected when growth is decreased. However, there was also an increase in SPS activity and, in contrast to many previous studies of N-limited plants (Stitt & Krapp 1999), sucrose and reducing sugars remained relatively high. Furthermore, starch still showed a diurnal turnover, whereas this turnover is often abolished in published studies of N limitation (e.g. Scheible et al. 1997c). This indicates that there is still a high requirement for export of C in the low N treatment.

These changes of enzyme activities and metabolites need to be interpreted in the context of the changes in growth rate. Growth consumes sugars and amino acids that are produced by photosynthesis and N metabolism in source leaves. Although biomass was strongly decreased after 35–39 d in low N, this was due to the incremental effect of a relatively small (20%) decrease in the RGR. This implies that there is a ∼20% decrease in the fluxes in central C and N metabolism in low N. This highlights the magnitude of the two- to threefold decrease in NR maximum activity, and the 20–30% decrease in post-translational activation of this enzyme. The 20–30% decrease in PEPC activity, the 10–20% increase in SPS and 30–80% increase in GluDH activity are also noteworthy, as are the 10–20% changes in starch levels. Interestingly, while most other enzymes including those in the GOGAT pathway and the central amino transferases do not show any detectable change in their maximum activities, metabolites often show large changes of their levels, including malate, fumarate and many amino acids. This indicates that small shifts in fluxes are often accompanied by highly amplified changes of the levels of individual metabolites.

Central amino acid metabolism

The primary C-acceptor in the GOGAT pathway is 2-OG. Glu and 2-OG are usually rather stable and often change, when they do change, in parallel (Mueller et al. 2001; Novitskaya et al. 2002; Fritz et al. 2006). In circumstances where 2-OG synthesis is disturbed, for example, due to an altered redox state in the mitochondria, there are also major perturbations of the Gln and Glu pools (Dutilleul et al. 2005; Fritz et al. 2006). In our experiments, 2-OG was consistently about 20% lower in low N than high N. The increased level of Gln and the unaltered or increased level of Glu cannot be explained by this small decrease of 2-OG. In particular, the increase of Glu levels during the light period indicates that Glu is being utilized more slowly in low N.

Ala and Asp are synthesized via aminotransferase reactions, which use glutamate as the amino donor and pyruvate and oxaloacetate as C-acceptors, respectively. The twofold decrease of pyruvate levels in low N could contribute to the threefold decrease of Ala during the light period. Our data provide indirect information about the response of oxaloacetate. Oxaloacetate is equilibrated with malate, NADH and NAD in the reaction catalysed by NAD-MDH. Malate levels were strongly decreased in low N. The majority of the malate accumulates in the vacuole (Gerhardt, Stitt & Heldt 1987). However, it is possible that malate levels are also lower in the cytoplasm in low N, and that the resulting decrease of oxaloacetate contributes to the decrease of Asp in low N.

The Gly/Ser ratio was consistently decreased in low N. These two amino acids are formed during photorespiration. Two molecules of glyoxylate are converted to two molecules of Gly, which are converted by glycine decarboxylase to one molecule of Ser, which is then converted to hydroxypyruvate. The most parsimonious pathway for these transformations involves one internal glyoxylate:Ser aminotransferase reaction, and one glyoxylate:Glu amino transferase reaction that is coupled with the shuttling of 2-OG and Glu between the peroxisome and the plastid, where the GOGAT pathway reassimilates the ammonium released by glycine decarboxylation (Liepman & Olsen 2001; Liepman & Olsen 2003; Reumann & Weber 2006). Our metabolite measurements do not provide any obvious explanation why the activity of these aminotransferases is restricted in low N. An alternative explanation would be that glycine decarboxylation is favoured in low N.

The 20–30% increase in GluDH activity in low N was rather unexpected. GDH transcripts and GluDH activity typically increase in low C conditions, indicating that GluDH is involved in amino acid catabolism (Melo-Oliveira, Oliveira & Coruzzi 1996; Lam et al. 1998; Gibon et al. 2004; Blaesing et al. 2005; Usadel et al. 2008). In agreement, gdh1 gdh2 double knockout mutants are deficient in recycling glutamate formed during amino acid catabolism in prolonged C starvation (Miyashita & Good, 2008). The increase of GluDH activity in low N indicates that GluDH is also involved in the adjustment of amino acid metabolism to low N. Arabidopsis contains three genes encoding GluDH. While GDH1 and GDH2 are induced by low C, GDH3 is induced by low N (Scheible et al. 2004). In tobacco and grape, there is also evidence for a role of GluDH-encoding genes and protein in recycling ammonium in companion cells (Dubois et al. 2003; Tercé-Laforgue et al. 2004a; Tercé-Laforgue, Maeck & Hirel 2004b; Fontaine et al. 2006).

Possible reasons for the general increase of amino acids in low N

The large decrease in NR activity and activation (see above) indicates that the rate of nitrate reduction in the rosette is lower in low N than in high N. The low and relatively constant level of nitrate in the rosette in low N is also consistent with this assumption. This indicates that the amino acids must be derived from another source of N. One possibility is preferential uptake and assimilation of ammonium in the rosette. The relatively high level of Gln (see Matt et al. 2001b) and the abolition of the diurnal changes of organic acids are consistent with preferential use of ammonium. Nevertheless, this seems unlikely to be the main explanation, as ammonium is also strongly limiting in the low N regime. A second possibility would be that inorganic N is assimilated in the roots and amino acids exported to the rosette. The increased accumulation of starch and the increased activity of SPS are consistent with an increased requirement for sucrose elsewhere in the plant. However, this on its own still does not explain why a decreased supply of N results in elevated levels of amino acids. A third possibility would be that increased remobilization of amino acids from protein. However, this cannot be a major factor, as overall protein content remains high in low N. By default, the most likely explanation is that amino acids are being utilized more slowly for protein synthesis and growth in low N.

Reasons for the decrease in rosette growth

We did not analyse photosynthesis in these experiments. However, the low-N plants contained similar levels of chlorophyll and total protein, similar maximum activities of Rubisco and other enzymes required for photosynthesis, and similar or slightly higher levels of starch and sugars than high-N plants. This indicates that the decrease of RGR is not due to an inhibition of photosynthesis. Rather, the decreased supply of N has altered the utilization of C.

The estimated rate of consumption of N for growth is 1.5 mg N gFW−1 per day in high N, falling to about 1.2 mg N gFW−1 in low N where the RGR is decreased by 20% (see above). In high N, rosettes contain about 150 µmol nitrate gFW−1 and 40 µmol amino acids gFW−1 (Figs 2b,f & 4b,e), which is equivalent to about 2 and 0.56 mg N gFW−1, or 160% and 40% of the daily N requirement, respectively. This resembles the values in N-replete tobacco (Matt et al. 2001a). In low N, the analogous values are 5–10 µmol nitrate gFW−1 and 50–60 µmol amino acids gFW−1, or about 0.04 and 0.7 mg N gFW−1, respectively. The N in nitrate is negligible, and the N in the amino acid pool would suffice for about half a day of growth. This comparison highlights that maintenance of a steady rate of growth over a period of 2 weeks in low N will require a very tight coordination of growth processes with the N supply. This is unlikely to be exerted by amino acids, as they actually increase. Our results imply that there is a more direct effect of the N supply on protein synthesis and growth.

After re-supplying nitrate to N-starved Arabidopsis seedlings, one of the most striking transcriptional responses was a coordinated induction of hundreds of genes that are required for RNA synthesis and processing, and protein synthesis (Scheible et al. 2004). This was followed by a decrease of the level of many amino acids, indicating that protein synthesis has been stimulated. This might be triggered by nitrate, or a metabolite generated during its assimilation. Hormones may also contribute to the control of growth. It is known that nitrate induces IPT3 in the roots (Wang et al. 2004), leading to increased synthesis and export of cytokinins to the shoot (Sakakibara et al. 1998; Takei et al. 2004). A further possibility is suggested by a recent study of transcript profiles in Arabidopsis growing under a strong and a mild N limitation (Bi et al. 2005). A strong N limitation led to large changes in the expression of genes involved in N assimilation, C and N metabolism, photosynthesis and protein synthesis, as seen in earlier studies (Wang et al. 2003; Scheible et al. 2004). In contrast, mild N limitation led to only restricted changes in genes, mainly those related to stress responses. One consequence of mild stress responses might be a slight restriction of growth. It should however be noted that that the mild N limitation in Bi et al. (2005) led to only a 20% decrease in biomass after 3 weeks, which will reflect an even smaller decrease of RGR (see above). Related changes in metabolism and cellular growth processes might be small and difficult to detect.

Finally, it is likely that the decrease in rosette growth in low N is accompanied by a stimulation of root growth. It is known that a moderate N limitation not only alters the root : shoot ratio, but also increases the absolute rate of root growth (see Introduction). This will decrease shoot growth, due to competition for carbon. Allocation to root growth can be triggered by nitrate, and occurs independently of the net rate of nitrate assimilation and the levels of amino acids (Scheible et al. 1997b; Zhang & Forde 1998; Stitt & Feil 1999; Zhang et al. 1999; Signora et al. 2001; Forde 2002). The response of root growth could not be directly checked in our growth system, as it did not allow quantitative recovery of root material.

Concluding remarks

We have established a simple soil-based experimental system that allows a small and sustained restriction of growth of Arabidopsis by low N. The metabolic phenotype in this growth system differs from that seen in earlier studies, revealing that Arabidopsis responds adaptively to low N by decreasing the rate of growth while maintaining or even increasing the levels of many amino acids. This opens up many questions with respect to how the decreased N supply is sensed, and how this close coordination of growth and metabolism is achieved. The growth system established in this study can be used in future studies to investigate how modification of single candidate genes affects the ability to adjust to a lower N supply, and to screen large populations of accessions and inbred lines to identify genetic factors that influence the ability to grow in low N. It will also be important to develop further growth systems in which the root system can be accessed more readily, while retaining the metabolic phenotype in the rosette.


This work was supported by the Max Planck Society and the German Ministry for Education and Research (BMBF, GABI-Funcin).