Exploring ammonium tolerance in a large panel of Arabidopsis thaliana natural accessions

Summary Ammonium nutrition is toxic to many plants. Arabidopsis displays high intraspecific variability in ammonium tolerance (shoot biomass), and ammonium accumulation seems to be an important player in this variability.


Introduction
Plants have a fundamental dependence on inorganic nitrogen (N), and intensive agriculture requires the use of N compounds to supplement the natural supply from the soil. Indeed, >100 Mt of nitrogenous fertilizers are added to the soil worldwide annually (Good and Beatty, 2011). In part because of the intense use of fertilizers, agriculture is now a dominant force behind many environmental threats, including climate change and degradation of land and fresh water (Foley et al., 2011;Tilman et al., 2011). Moreover, recent studies suggest that agricultural output would need to roughly double to meet the expected demand associated with the increase in the world's population (FAO, 2009). Nitrate (NO 3 − ) and ammonium (NH 4 + ) are the main forms of N available for plants. There is a serious concern regarding NO 3 − loss in the field, giving rise to soil and water pollution. Moreover, incomplete capture and poor conversion of nitrogen fertilizer also causes global warming through emissions of nitrous oxide. Due to these detrimental effects of adding high NO 3 − concentrations to ecosystems (Gruber and Galloway, 2008), the potential of NH 4 + as an N source for agriculture has been reconsidered alongside the search to improve N use efficiency (NUE) while mitigating the impact of agriculture (IPCC, 2007). Similarly, lowering fertilizer input and breeding plants with better NUE without affecting yield is a main goal for research in plant nutrition (Xu et al., 2012). Plants have differential N source preference, but this depends not only on their genetic background but also on a wide and dynamic range of environmental variables including soil pH, temperature, etc. Thus, a robust classification of plants species adapted to NO 3 − or NH 4 + does not exist. However, it appears that most non-bred plants preferentially take up NH 4 + (Bloom et al., 1993;Kronzucker et al., 2001). Moreover, crop species have traditionally been bred under nitric or combined N nutrition, provoking a negative selection pressure towards NH 4 + assimilation, and this undoubtedly is one of the reasons they prefer NO 3 − , although NO 3 − must be taken up against an electrochemical gradient and then be reduced to NH 4 + with the consequent energy cost (Kronzucker et al., 2001). In this sense, NH 4 + nutrition has been generally considered as toxic for plants, particularly when NH 4 + is supplied as the sole N source. Indeed, NH 4 + is also toxic to animals and fungi when present in excess amounts (Britto and Kronzucker, 2002). Ammonium toxicity syndrome in plants includes several symptoms, among others leaf chlorosis, ion imbalance, hormone deregulation, disorder in pH regulation, decrease in net photosynthesis, and changes in metabolite levels including amino acids, organic acids, and carbohydrates. At the whole-plant level, a reduction in plant growth with increasing external NH 4 + concentrations, as compared with NO 3 − nutrition, is a common effect of NH 4 + nutrition (Cruz et al., 2006). Biomass reduction has been associated with carbohydrate limitation for growth due to excessive sugar consumption for NH 4 + assimilation and to the energy costs of futile transmembrane NH 3 /NH 4 + cycling in root cells (Coskun et al., 2013). Indeed, plant growth is probably the best indicator of NH 4 + stress as it is a comprehensive measure of the physiology of the plant as a whole (Cruz et al., 2006;Dominguez-Valdivia et al., 2008;Ariz et al., 2011).
Substantial variations in NH 4 + tolerance have been observed amongst closely related species (Monselise and Kost, 1993) and even within species (Rauh et al., 2002;Cruz et al., 2011;Li et al., 2011), suggesting the evolution of highly distinct mechanisms to cope with this stress. The strategies plants deploy to avoid NH 4 + toxicity include enhancing NH 4 + assimilation and increasing the efflux outside the cell or into the vacuole. Nevertheless, at present there is no consensus as to which traits confer NH 4 + tolerance or sensitivity to plants. Ammonium assimilation mainly occurs via the glutamine synthetase/glutamate synthase cycle (GS/GOGAT). However, it seems that other alternative pathways could be involved in ammonium assimilation when NH 4 + is supplied as the sole source of N. Although controversial, under these conditions, glutamate dehydrogenase (GDH), that catalyses the reversible deamination of glutamate to 2-oxoglutarate, might be collaborating in NH 4 + assimilation (Skopelitis et al., 2006;Setien et al., 2013).
Arabidopsis thaliana and the Brassicaceae family are considered to be a species, and a family, sensitive to NH 4 + . Most of the works focused on NH 4 + toxicity in Arabidopsis have compared plants fed with NO 3 − versus plants fed with a combined nutrition of NO 3 − supplemented with increasing concentrations of NH 4 + . Studies where Arabidopsis has been grown under long-term ammonium supply as the sole N source are scarce and have shown how NH 4 + causes a retardation of seedling growth or a dramatic reduction in plant biomass (Rauh et al., 2002;Hoffmann et al., 2007;Helali et al., 2010). Also, recent genetic approaches have been useful to identify new molecular players involved in the signalling pathways that lead to NH 4 + sensitivity, for example a GDPmannosepyrophosphorylase enzyme (Qin et al., 2008) or the ammonium transporter AMT1:3 (Lima et al., 2010).
Overall, the evolutionary trade-off between high productivity, adaptation to low-nutrient environments, and the use of ammonium as fertilizer is a challenge to most plant cultivars that have been selected under non-limiting NO 3 − or combined NH 4 + /NO 3 − fertilization (Presterl et al., 2003;Xu et al., 2012). Approaches based on natural variation have become an important means to study plant adaptation to the environment. In Arabidopsis, it has already been reported that a plant's response to N availability is dependent on both the genotype and the N fertilization level (Loudet et al., 2003), and natural variation has been observed for N remobilization during seed filling, among others (Masclaux-Daubresse and . Thus, the present work compares the natural intraspecific variability of A. thaliana grown under a low NO 3 − or NH 4 + supply, focusing on the importance of N assimilation mechanisms in relation to the differential N source provided.

Determination of ammonium and total amino acids content
Aliquots of ~25 mg of frozen material were ground to powder with liquid nitrogen and homogenized with 800 μl of ultrapure water.
Samples were then incubated at 80 ºC during 5 min and centrifuged at 4000 g and 4 ºC for 20 min, and supernatants were recovered.
Total free amino acids were determined by the ninhydrin method (Yemm and Cocking, 1955). Ammonium content was determined by using the colorimetric method based on the phenol hypochlorite assay (Berthelot reaction).
Nitrate reductase (NR) activity was measured at 30 ºC. The reaction medium consisted of 50 mM HEPES-KOH, pH 7.6, 5 mM KNO 3 , 0.2 mM NADH, 10 μM FAD phosphate, 1 mM DTT, 20 mM EDTA. The reaction was started by adding 50 μl of protein extract to 250 μl of reaction medium and stopped by adding 32 μl of 50 mM zinc acetate. Then, samples were centrifuged, 100 μl of supernatant was recovered, 8 μl of 50 mM phenacin metosulphate added, and the samples incubated for 20 min at room temperature. Finally, 80 μl of 1% sulphanilamide in 3 M HCl and 80 μl of 0.02% N-(1-naphthyl) ethylenediamine dihydrochloride were added and the absorbance determined at 546 nm.

Data analysis
Data analyses were carried out using SPSS 17.0 (Chicago, IL, USA). Statistical differences between nitrate and ammonium nutrition for each accession and variable were assessed comparing the mean values by paired t-test. To test the connectivity between variables, Pearson's correlation coefficient was calculated for P≤0.05. Multiple regressions provided a view of the relationship between a trait and shoot biomass independent of other correlated traits. Multiple regression estimations can suffer from multicollinearity wherein highly correlated traits might act redundantly. Thus, to help in interpretation, Akaike's information criterion (AIC) was also used to determine the 'best' model by rewarding added explanatory power but penalizing the inclusion of additional terms. This provides the simplest model with the least collinearity and, thus, supposedly, the best estimation of selection (Shaw and Geyer, 2010).

Results
To evaluate NUE with ammonium as the sole N source, Arabidopsis rosette biomass was compared after 3 weeks of growth under 1 mM NH 4 + [0.5 mM (NH 4 ) 2 SO 4 ] or 1 mM NO 3 -[0.5 mM Ca(NO 3 ) 2 ], and the ratio between shoot biomass under NH 4 + and NO 3 conditions (SB NH 4 + /NO 3 -) was used to estimate ammonium tolerance as it has been previously used in other studies (Cruz et al., 2006;Ariz et al., 2011). In general, Arabidopsis is a species sensitive to NH 4 + and nearly every ecotype analysed showed shoot biomass inhibition in response to NH 4 + . Twenty-four out of the fortyseven accessions analysed experienced a significant growth inhibition upon NH 4 + nutrition (Fig. 1A). The accession Te-0 was the one showing the lowest SB NH 4 + /NO 3 ratio (<0.4), which was significantly lower than that of the next most sensitive accession to NH 4 + (Rubenzhnoe-1; SB NH 4 + /NO 3 -0.56). Only three accessions had an SB NH 4 + /NO 3 ratio >1, but without significant differences between both types of nutrition (Akita, Enkheim-T, and Gre-0; Fig. 1B). Overall, intraspecific shoot growth variability under a contrasting N source is evident by the use of this collection of accessions (Fig. 1B).
The content of ammonium and free amino acids (Supplementary Table S1 available at JXB online) as well as NR, GS, and GDH enzyme activities (Supplementary Table  S1) were determined. GDH activity was measured both in the aminating (GDHam) and in the deaminating (GDHdeam) direction. Regarding NH 4 + content, overall, plants under NH 4 + nutrition contained significantly more NH 4 + compared with plants fed with NO 3 -. Eight accessions (Enkheim-T, Gre-0, Ishikawa, Jea, Ms-0, Ran, Ta-0, and Tsu-0) did not show significant differences between both treatments (Supplementary Table S1). Amino acid content followed a similar trend to NH 4 + content ( Supplementary Fig. S1) and every accession under NH 4 + nutrition contained significantly more amino acids compared with under NO 3 nutrition (Supplementary Table S1, Fig. S1). Concerning the enzyme activities, as expected, every accession under NO 3 nutrition had a higher NR activity ( Fig. 2B; Supplementary Table S1). GS activity was similar for every accession under both forms of nutrition, except for Mt-0 and Ct-1 that showed a slightly higher GS activity under NO 3 nutrition and for Rld-2, N7, and N14 that experienced a small increase under NH 4 + nutrition ( Fig. 2A; Supplementary Table S1). GDHam activity was higher under NH 4 + nutrition in 35 out of the 47 accessions.
In contrast, GDHdeam activity was higher under NO 3 nutrition in every accession except for Akita, Ishikawa, Rld-2, Pa-1, and Sah-0, which did not show significant differences between both forms of nutrition (Fig. 2C, D; Supplementary  Table S1).
To investigate the connectivity between the different parameters, a Pearson correlation analysis was performed for each pair of parameters. Values are given for the correlation coefficient (r 2 ) and the significance (P). The results are presented separately for the plants grown under NH 4 + ( Table 1) and NO 3 nutrition (Table 2). Shoot biomass under both ammonium and nitrate nutrition was negatively correlated with NH 4 + and free amino acid content (Tables 1, 2; Fig. 3A), which is reasonable because it could mean that part of the absorbed N is not being used for growth, and ammonium accumulation inside plant tissues is known to be deleterious for plant performance (Britto and Krontzuker, 2002;Ludewig et al., 2007). None of the parameters determined in NH 4 + -fed plants showed any correlation with the SB NH 4 + /NO 3 ratio (Table 1). In contrast, in NO 3 --fed plants, NH 4 + and amino acid content, together with GDHam activity, were positively correlated with the SB NH 4 + /NO 3 ratio (Table 2).
Regarding the enzyme activities, in NH 4 + -fed plants, neither GS nor NR activity showed any correlation with any of the parameters determined (Table 1). GDHam and GDHdeam activities were positively correlated with each other, suggesting that when a genotype shows high GDH activity, it occurs in both the aminating and deaminating directions. Both GDHam and GDHdeam activities were positively correlated with amino acid content; however, only GDHam activity was positively correlated with NH 4 + content (Table 1). In NO 3 -fed plants, NR activity was positively correlated with NH 4 + content and with GS activity (Table 2). In addition, GS activity was also correlated with GDHdeam activity. Interestingly, and similarly to NH 4 + -fed plants, in NO 3 --fed plants GDHam activity was also correlated with ammonium and amino acid content (Table 2).
In order to better understand the relationships between the SB NH 4 + /NO 3 ratio and the different determined parameters, a multiple regression full model and AIC best model (AIC-selected) were applied. The full model only indicated a significant selection for the ammonium content in NO 3 -fed plants (Table 3) and explained 23% of the variance in SB NH 4 + /NO 3 -. In the best model, the percentage of the variance in SB NH 4 + /NO 3 explained increased up to 38%. From    The same analysis was performed for the shoot biomass under both forms of nutrition. For NH 4 + -fed plants, the models only indicated selection for ammonium content, and both the full and best models only explained 19% of the variance in shoot biomass (Supplementary Table S3 at JXB online). For NO 3 --fed plants, both the full and the best model explained 39% of the variance in shoot biomass. The full model indicated selection for ammonium and amino acid content (Supplementary Table S2), and both models significantly retained the ammonium and amino acid content (Supplementary Table 2).
According to the importance given by both Pearson correlations and the multiple regression models, the correlation of ammonium content both with shoot biomass and with SB NH 4 + /NO 3 was illustrated (Fig. 3). As shown by Pearson analysis (Tables 1, 2), ammonium content was negatively correlated with shoot biomass under both NH 4 + and NO 3 nutrition (Fig. 3A). Interestingly, and as suggested by the multiple regression model, only the ammonium content in NO 3 --fed plants was correlated with the SB NH 4 + /NO 3 ratio (Fig. 3B).
To understand further the behaviour of the N-assimilating enzymes determined, the enzyme activities were illustrated and western blotting analysis was performed for the accessions Te-0 and Gre-0, the most sensitive and tolerant accessions to ammonium, respectively (Fig. 3). This analysis did not show any difference for any of the three enzymes under both forms of nutrition. However, it was useful to ascertain that although there were no significant differences in GS activity, the GS1 isoform content was clearly accumulated upon ammonium nutrition (Fig. 2E). NR protein content, in agreement with NR activity, was dramatically induced in NO 3 -fed Te-0 and Gre-0 plants. Finally, GDH content increased in NH 4 + -fed plants, according to the increase in GDHam activity (Fig. 2C). In contrast, although GDH was induced upon ammonium nutrition, as described above, GDHdeam activity increased in NO 3 --fed plants (Fig. 2D). However, it must be noted that under NH 4 + nutrition, the average GDHam activity was around eight times higher than the GDHdeam activity, whilst under NO 3 nutrition GDHam activity was about three times higher than GDHdeam activity.

Discussion
Plant response to N availability depends on the genotype, the N source, and N fertilization level, and the limiting steps in N metabolism are different at low and high N supply (Chardon et al., 2012;Xu et al., 2012). Overall, NUE is higher when N supply is limiting. In general, adaptation to low N environments is challenging to most cultivars, because they have been selected under high-nutrient environments but plants in natural field conditions are faced with environmental changes where N availability varies and the better NUE under low N conditions is a competitive advantage (Kant et al., 2011). Moreover, reducing N fertilizer input in the soil while maintaining productivity is an unavoidable strategy to reduce agricultural impact on the environment. Thus, and taking into account that Arabidopsis and the Brassicaceae family have been described as very susceptible to ammonium nutrition, in this work, a low N dose (1 mM) was used. Because of this high sensitivity, most of the studies related to ammonium toxicity in Arabidopsis have been performed with mixed nutrition, and thus long-term ammonium-based nutrition studies involve the use of a low ammonium concentration. Approaches based on intraspecific natural variation have become an important means to study plants adaptation. Regarding nitrate nutrition, studies based on natural variation have already been used in several species including maize (Coque and Gallais, 2007) and rice (Namai et al., 2009). Arabidopsis natural variation has also been studied under limiting and ample nitrate supply (North et al., 2009;Chardon et al., 2010) and to evaluate the capacity of different genotypes for N remobilization during seed filling (Masclaux-Daubresse and . In contrast, studies focused in intraspecific variation of N use with ammonium as the sole N source are more scarce, although examples exist, studying, among others, four maize cultivars (Schortemeyer et al., 1997), a collection of rice inbred lines (Obara et al., 2010), and four pea cultivars . In this work, data from 47 natural accessions of Arabidopsis were collected and several traits related to N metabolism were measured to determine the natural variation of Arabidopsis growth and N metabolism (ammonium and amino acid content, and NR, GS, and GDH enzyme activities) under two different N sources (nitrate or ammonium). Biomass is considered as the best indicator of plant performance because it integrates every aspect of plant metabolism, from nutrient uptake to its assimilation, and the ratio of the shoot biomass under ammonium versus nitrate nutrition was considered here as an indicator of the plant's tolerance/sensitivity to ammonium, as it has previously been used in other works (Cruz et al., 2006;Ariz et al., 2011). Arabidopsis accession N1438 grown under 2.5 mM NH 4 + for 21 d showed three times less biomass compared with plants grown under NO 3 -, and the authors suggested ionic imbalance as a major cause of this toxicity (Helali et al., 2010). Similarly, Hoffman et al. (2007) reported a retardation of Arabidopsis Col-0 seedling growth under NH 4 + nutrition compared with NO 3 nutrition. The present study confirms an overall sensitivity of Arabidopsis to ammonium, since, out of the 47 genotypes, 44 had a ratio <1 (23 accessions showing significant differences in shoot biomass between both forms of nutrition). However, this study highlights large intraspecific variation of ammonium tolerance expressed as SB NH 4 + /NO 3 -, which varied between 0.36 and 1.18. These values are in agreement with the values registered by Ariz et al. (2011) working with seven different species and ammonium concentrations. Thus, the present study, working with a low ammonium concentration, reveals a similar degree of intraspecific Arabidopsis ammonium tolerance variability to the interspecific degree of ammonium tolerance variability. This underscores the high variability within a single species and the power of natural variation approaches for plant adaptation studies.

Ammonium accumulation affects plant growth
'Excessive' ammonium accumulation is toxic to cells. However, the concept of 'excessive' is extremely variable depending on the plant species and on the soil NH 4 + concentration.
In fact, ammonium toxicity is considered to be 'universal' even in species labelled as 'NH 4 + specialists' (Li et al., 2014). Excess ammonium causes an imbalance in, among others, pH homeostasis, ionic equilibrium, and primary metabolism (Britto and Kronzucker, 2002). Ammonium accumulation might derive from its direct uptake but also from amino acid deamination, protein degradation, and photorespiration. To prevent the cytosol from ammonium overload, plants deploy different strategies including AMT-type ammonium transporter regulation (Lanquar et al., 2009) or increasing ammonium assimilation (Setien et al., 2013). In the present work, as expected, NH 4 + -fed plants accumulated more NH 4 + and amino acids than NO 3 --fed plants and this NH 4 + accumulation was negatively correlated with Arabidopsis rosette biomass (Fig. 3A). Interestingly, this correlation was found for plants grown under both forms of nutrition, suggesting that ammonium accumulation negatively influences plant growth even under nitric nutrition. NH 4 + accumulation under low N supply might be due to a lack of proper carbohydrate supply for ammonium assimilation or to the toxicity caused by the excess NH 4 + as stated above. To the authors' knowledge, this is the first time that a correlation between plant shoot growth under NO 3 as sole N source and the accumulation of NH 4 + in leaves has been reported, which provides evidence of the extreme sensitivity of Arabidopsis to ammonium. Regarding the SB NH 4 + /NO 3 ratio, of the parameters determined, only ammonium, amino acid content, and GDHam activity from NO 3 --fed plants showed a significant correlation (Table 2). Multiple regression full and best models retained ammonium and amino acid content, which both show a strong correlation ( Supplementary Fig. S1 at JXB online), as significant factors explaining the variation in the SB NH 4 + /NO 3 ratio (Table 3). Interestingly, the NH 4 + content of NH 4 + -fed plants did not show any significant correlation with the SB NH 4 + /NO 3 ratio. Thus, the fact that NO 3 --fed plants with a higher NH 4 + content present a smaller rosette biomass (Fig. 3A) could explain the relationship between ammonium content of NO 3 --fed plants and the SB NH 4 + /NO 3 ratio (Fig. 3B). Alternatively, it can be speculated that evolutionarily a plant that under NO 3 nutrition is able to accumulate more ammonium could be genetically better adapted to an ammonium-based nutrition.

Role of NR, GS, and GDH in Arabidopsis response to ammonium
NO 3 absorbed from the nutrient solution is reduced to ammonium, whereas in NH 4 + -fed plants this step is bypassed and ammonium is directly assimilated for plant growth. As expected, NR activity was induced upon NO 3 exposure but it was not related to differential plant growth. Indeed, NR or nitrite reductase overexpression in tobacco, potato, or Arabidopsis did not increase plant biomass, thus nitrate reduction does not seem to be a limiting step for plant growth (Pathak et al;Masclaux-Daubresse et al., 2010). Ammonium assimilation in normal conditions in plants mainly occurs via the GS/ GOGAT cycle. There are two different GS isoforms. GS1 is encoded by five genes in Arabidopsis and functions primarily in assimilating ammonia during nitrogen remobilization. GS2 is encoded by a single gene in Arabidopsis and has been involved in assimilating the ammonia coming from nitrate reduction or photorespiration (Xu et al., 2012). In general, plants with higher GS activities are considered more tolerant to ammonium, and Cruz et al. (2006) showed a relationship between GS activity in the dark and ammonium tolerance. In this work, no difference in GS activity was found in almost every accession between NH 4 + -and NO 3 --fed plants ( Fig. 2A; Supplementary  Table S2 at JXB online) and there was no correlation between GS activity and shoot biomass in plants under both forms of nutrition (Tables 1, 12). A western blot analysis was performed in two accessions with contrasting ammonium tolerance (Te-0 and Gre-0), and in both cases there was a clear accumulation of the GS1 isoform in response to ammonium nutrition. Overall, total GS activity does not seem to be crucial for ammonium tolerance in Arabidopsis; however, GS1 could have an important role when ammonium is supplied as the N source. Moreover, out of the five genes encoding GS1 in Arabidopsis GS1;2 is the most highly expressed in leaves and it is induced by ammonium (Lothier et al., 2011). Indeed, an Arabidopsis mutant lacking GS1;2 expression exhibited reduced growth under a 7 d ammonium treatment compared with the wild type (Lothier et al., 2011). Similarly, a rice mutant in the GS1;1 gene was also more sensitive upon ammonium nutrition (Kusano et al., 2011). Thus, it remains to be determined whether GS1;2 and the rest of the GS isozymes are related to Arabidopsis variability under ammonium nutrition. Also, very recently root NADH-GOGAT has been suggested to play an important role in ammonium assimilation under ammonium nutrition (Konishi et al., 2014s).
GDH is able to catalyse the in vitro reversible amination of 2-oxoglutarate to glutamate. In vivo, the existence of the N assimilating capacity of GDH is controversial and in the last years evidence has been accumulating in favour of the major role of GDH deamination, for example by the use of 15 N-nuclear magnetic resonance (NMR) labelling studies showing that there was no direct incorporation of ammonia into glutamate when GS was inhibited (Labboun et al., 2009;Tercé-Laforgue et al., 2013). However, although in unstressed plants GDH ammonia assimilating capacity seems to be negligible, it appears that under stress conditions and under ammonium nutrition, GDH could incorporate NH 4 + Setien et al., 2013). In the present study, a contrasting behaviour of GDH activity was found. GDHam activity was generally induced upon NH 4 + exposure whereas GDHdeam activity was repressed (Fig. 2C, D; Supplementary Table S1 at JXB online). Moreover, in both NH 4 + -and NO 3 --fed plants ammonium and amino acid contents were positively correlated with GDHam activity, and not with GDHdeam activity (Tables 1, 12). Thus, the present data suggest that NH 4 + accumulation might be stimulating the ammonium-incorporating capacity of GDH rather than being a consequence of NH 4 + release associated with GDH glutamate deamination. Nevertheless, experiments designed to ascertain the actual GDHam activity in conditions of plant growth under an exclusive ammoniacal nutrition, such as by 15 N-NMR labelling, are necessary. GDH is traditionally accepted to form seven isoenzymes composed of α and β homo-or heterodimers. Recently, the existence in Arabidopsis of a third gene encoding a γ subunit has been shown (Fontaine et al., 2012). However, the activity of this γ isoenzyme was exclusively from root (Fontaine et al., 2012), which is in line with the hypothesis that each of the GDH subunits may have specific biological functions (Purnell et al., 2005;Tercé-Laforgue et al., 2013). In the present work, after SDS-PAGE, GDH was accumulated under ammonium nutrition (Fig. 2E). An accumulation of GDH polypeptides has already been reported in several species including wheat (Setien et al. 2013), pea (Ariz et al., 2013), and tomato (Setien et al. 2014). The overall data indicate a key role for GDH in Arabidopsis under NH 4 + nutrition.

Concluding remarks and future prospects
Overall, the results obtained in this work reveal that there exists high natural variation in A. thaliana growth as a function of the N source. This variation was partially due to the differential tissue NH 4 + and amino acid accumulation in both NO 3 --fed and NH 4 + -fed plants. Similarly, significant natural variability was detected in NH 4 + tolerance expressed as the SB NH 4 + /NO 3 ratio, and, interestingly, NH 4 + accumulation in NO 3 -fed plants was the parameter showing the highest relevance, which may indicate an evolutionary adaptation suggesting that plants that under NO 3 nutrition are able to accumulate more ammonium could be genetically better adapted to an ammonium-based nutrition. Although plant NH 4 + assimilation capacity is known to be a key aspect for ammonium tolerance, GS and GDH activity does not seem to be responsible for the variability shown in A. thaliana. However, the modulation of GDH activity as a function of the supplied N source was clearly observed, which suggests an important role for this enzyme in NH 4 + assimilation. Similarly, the observed higher content of the GS1 isoform in NH 4 + -fed plants could also contribute to NH 4 + assimilation. The quality of the root system has also been suggested partly to explain the differences in nitrogen uptake and NUE (Loudet et al., 2005). Furthermore, several works have highlighted the importance of the root in NH 4 + tolerance (Setien et al., 2013, Kojima et al., 2014. Thus, future works dealing with root metabolism will be useful to ascertain whether N assimilation in this organ could be related to the natural variability in NH 4 + tolerance in A. thaliana. Also, approaches using larger A. thaliana natural populations in combination with genome-wide association studies (Atwell et al., 2010) will no doubt be very helpful in elucidating the genetic basis underlying the Arabidopsis intraspecific variability in ammonium tolerance.

Supplementary data
Supplementary data are available at JXB online. Figure S1. Scatter plots of amino acids versus ammonium content of leaves of Arabidopsis thaliana grown under NH 4 + and NO 3 -. Table S1. Ammonium and amino acid content and enzyme activities: whole data set. Table S2. Full and Akaike's information criterion (AIC)selected best multiple regression models of Arabidopsis thaliana rosette biomass grown under NH 4 + or NO 3 nutrition.