Assimilation of excess ammonium into amino acids and nitrogen translocation in Arabidopsis thaliana– roles of glutamate synthases and carbamoylphosphate synthetase in leaves


A. Suzuki, Unité de Nutrition Azotée des Plantes, Institut National de la Recherche Agronomique, Route de St-Cyr, 78026 Versailles cedex, France
Fax: +33 1 30 83 30 96
Tel: +33 1 30 83 30 87


This study was aimed at investigating the physiological role of ferredoxin-glutamate synthases (EC, NADH-glutamate synthase (EC and carbamoylphosphate synthetase (EC in Arabidopsis. Phenotypic analysis revealed a high level of photorespiratory ammonium, glutamine/glutamate and asparagine/aspartate in the GLU1 mutant lacking the major ferredoxin-glutamate synthase, indicating that excess photorespiratory ammonium was detoxified into amino acids for transport out of the veins. Consistent with these results, promoter analysis and in situ hybridization demonstrated that GLU1 and GLU2 were expressed in the mesophyll and phloem companion cell–sieve element complex. However, these phenotypic changes were not detected in the GLU2 mutant defective in the second ferredoxin-glutamate synthase gene. The impairment in primary ammonium assimilation in the GLT mutant under nonphotorespiratory high-CO2 conditions underlined the importance of NADH-glutamate synthase for amino acid trafficking, given that this gene only accounted for 3% of total glutamate synthase activity. The excess ammonium from either endogenous photorespiration or the exogenous medium was shifted to arginine. The promoter analysis and slight effects on overall arginine synthesis in the T-DNA insertion mutant in the single carbamoylphosphate synthetase large subunit gene indicated that carbamoylphosphate synthetase located in the chloroplasts was not limiting for ammonium assimilation into arginine. The data provided evidence that ferredoxin-glutamate synthases, NADH-glutamate synthase and carbamoylphosphate synthetase play specific physiological roles in ammonium assimilation in the mesophyll and phloem for the synthesis and transport of glutamine, glutamate, arginine, and derived amino acids.


asparagine synthetase




carbamoylphosphate synthetase (EC




ferredoxin-glutamate synthase (EC


ferredoxin-dependent nitrite reductase


glycine decarboxylase complex


glutamate dehydrogenase


glutamate synthase


glutamine synthetase (EC


NADH-glutamate synthase (EC


N-acetyl-glutamate kinase


nitrate reductase


Inorganic nitrogen in the form of nitrate and ammonium in the soil is absorbed by roots across the plasma membrane, and it is in part transported via the xylem to leaves prior to incorporation into amino acids in Arabidopsis [1]. Primary nitrogen reduction from nitrate to ammonium is catalyzed by cytosolic nitrate reductase (NR; EC, and then by plastidial ferredoxin (Fd)-dependent nitrite reductase (Fd-NiR; EC Photorespiratory glycine oxidation in the mesophyll mitochondria releases the bulk of ammonium at high rates of as much as 10–20-fold those of primary nitrate reduction in leaves [2]. Primary and photorespiratory ammonium assimilation into amino acids could take place by four distinct pathways in Arabidopsis, to meet the needs of protein synthesis, the maintenance of amino acid pool levels within the leaves, and nitrogen transport to the growing apical sinks and roots via the phloem. First, the glutamine synthetase (GS)–glutamate synthase (GOGAT) cycle is the major assimilatory pathway. Glutamine is generated from ammonium and glutamate by cytosolic GS1 and plastidial GS2 (EC Then, GOGAT transfers the glutamine amide group to the 2-position of 2-oxoglutarate to yield two molecules of glutamate, one of which is cycled to GS. The Arabidopsis nuclear genome carries multiple genes for many of the nitrogen assimilatory enzymes, and GOGAT exists as Fd-GOGAT (EC, encoded by GLU1 and GLU2, and as NADH-GOGAT (EC, encoded by GLT [3]. Second, either ammonium or a glutamine amide group is integrated into asparagine by cytosolic asparagine synthetase (AS) [ammonia-ligasing AS (EC or glutamine-hydrolyzing AS (EC] [4]. Third, carbamoylphosphate synthetase (CPSase) forms carbamoylphosphate (CP) using bicarbonate (inline image), ATP and ammonium (ammonia-ligasing CPSase; EC, or the glutamine amide group (glutamine-hydrolyzing CPSase; EC [5]. In Arabidopsis, a single copy each of carA and of carB encode the small subunit and large subunit, respectively. The small and large subunits form a single heterodimeric enzyme that supplies CP as a precursor for arginine and pyrimidine synthesis [6]. Finally, mitochondrial NADH-glutamate dehydrogenase (EC could alternatively incorporate ammonium into glutamate in response to high levels of ammonium under stress [7].

GOGATs are involved in the major synthetic pathway of glutamate from primary and photorespiratory nitrogen [8], and CPSase seems to catalyze a committed step to recover photorespiratory nitrogen in amino acid synthesis in Arabidopsis [9]. The amino acids are then translocated in the apoplasm and in the phloem via the plasma membrane-located amino acid transporters [10]. Glutamine and asparagine, and to a lesser extent arginine, glutamate, and aspartate, are transported in Arabidopsis phloem sap for use in sink cell development [11]. Therefore, we investigated whether two Fd-GOGAT isoenzymes and NADH-GOGAT play overlapping or distinct roles in nitrogen assimilation into amino acids for transport in planta using mutants deficient in GLU1, GLU2, or GLT. Despite the in silico data of the Arabidopsis databases, experiments on the in vivo function of CPSase remain largely unaddressed. Inasmuch as arginine synthesis from CP relies on the regulation of glutamate conversion to ornithine [6], we studied the impact of CPSase on overall arginine synthesis in the carB mutant. In fact, amino acid synthesis is tightly correlated with amino acid transport under the fine control of the cellular and subcellular expression of the nitrogen assimilatory genes and of the encoded enzymes [12]. Despite their primary importance, the spatial location and expression patterns have not been investigated for Fd-GOGAT isoenzymes, NADH-GOGAT and CPSase in Arabidopsis. Thus, we defined their subcellular localization and cell type-specific and tissue-specific expression patterns by promoter::GUS fusion expression in transgenic Arabidopsis, in situ mRNA hybridization, and immunohistochemical localization. The results showed that each isoenzyme of Fd-GOGAT, NADH-GOGAT and CPSase had distinct physiological relevance in the mesophyll and in the phloem for the biosynthesis and transport of amino acids under photorespiratory and nonphotorespiratory conditions.


Expression of the genes for GOGATs and CPSase

In order to understand the physiological role of GOGATs and CPSase, we first examined the expression pattern of the genes encoding these enzymes in leaves and roots from 42-day-old Arabidopsis plants. A search of the Arabidopsis genome database [13] revealed that there are two genes for Fd-GOGAT [GLU1 (AGI: At5g04140)] and GLU2 (At2g41220)], and one gene for NADH-GOGAT [GLT (At5g53460)]. GLU1 and GLU2 are composed of 33 exons coding for a protein of 165 kDa, containing a class II (purF)-type glutaminase domain and short regions for binding to FMN and iron sulfur center. GLT is composed of 20 exons encoding a large protein of 240 kDa. CPSase is encoded by two genes: carA (At3g27740) and carB (At1g29900). carA is composed of 10 exons encoding the 40 kDa small subunit. The small subunit contains a class I (trpG)-type glutaminase domain to hydrolyze glutamine to ammonia. carB is composed of three exons encoding a 120 kDa large subunit, consisting of the duplicated synthetase regions and the ATP-binding domains to synthesize CP. GLU1 was mainly expressed in the leaves, at significantly higher level than GLT and GLU2 (Fig. 1A). Although GLT and GLU2 were expressed in the leaves and in the roots, GLT mRNAs were at least seven-fold more abundant than GLU2 mRNAs (Fig. 1A). carA and carB were more highly expressed in the leaves than in the roots, and both leaves and roots contained slightly more abundant carB mRNAs (Fig. 1B). Among the cytosolic GS1 genes, higher mRNA levels were found for Gln12, Gln13 and Gln14 than for Gln11 in the leaves (Fig. 1C). The highest mRNA level was also found for Gln12 in the roots (Fig. 1C). As compared with gln12, the chloroplastic GS2 mRNAs were more abundant than gln12 mRNAs in the leaves and in the roots (about two-fold and 1.5-fold, respectively) (data not shown).

Figure 1.

 Transcript levels of the genes for GOGATs, CPSase and GSs in leaves and roots of Arabidopsis. Arabidopsis plants were grown for 42 days by hydroponic culture using 5 mm nitrate [37] in air supplemented with 3000 p.p.m. CO2. Transcript levels were determined by quantitative real-time RT-PCR. (A) GOGAT genes: GLU1, GLU2, and GLT. (B) CPSase genes: carA and carB. (C) GS1 genes: Gln11, Gln12, Gln13, and Gln14. The values are expressed as percentage ± standard error relative to the marker EF1α gene.

Characterization of the T-DNA mutants for GOGATs and CPSase

With a reverse genetic screen, individual plants with homozygous mutant alleles were identified for GLU2, GLT and carB by PCR in combination with the primers specific for the T-DNA left and right borders. The GLU2 mutant was truncated by a T-DNA insertion in intron 9 (Fig. 2A). With the use of primers downstream of the insertion site, the GLU2 mRNA level was approximately 10% of the wild-type level in leaves (Fig. 2D). The GLT mutant was characterized by a T-DNA insertion in exon 13 about 50 amino acids upstream of the FMN-binding domain (Fig. 2B). The GLT T-DNA mutant contained about 20% of the wild-type level of GLT mRNA (Fig. 2D). The carB mutant was disrupted by a T-DNA insertion in the promoter close to 600 nucleotides upstream of the initial ATG codon (Fig. 2C). The carB mutant expressed about 10% of the wild-type level of carB mRNA (Fig. 2D). To evaluate whether the decrease in the GOGAT transcripts correlates with a functional deficiency, we assayed GOGAT activities in leaves from plants grown in air or in high CO2 (3000 p.p.m.), where photorespiration is repressed. The Fd-GOGAT activity, encoded by GLU1 and GLU2, was reduced to less than 3% in the GLU1 mutant (ethylmethanesulfonate-mutagenized CS254 line) [2], whereas almost wild-type activity was recovered in the GLU2 mutant in air and in high CO2 (Table 1), indicating that GLU1 encodes the major Fd-GOGAT isoenzyme. The NADH-GOGAT activity, encoded by GLT and representing only 3% of the total GOGAT activity, was reduced to approximately one-fourth in the GLT mutant, whereas NADH-GOGAT activity was less affected in the GLU1 and GLU2 mutants, irrespective of the photorespiratory conditions (Table 1). We also assayed GS and glutamate dehydrogenase (GDH), as these enzyme activities are closely related to ammonium assimilation. The GS activity was not affected in the mutants, except for a slight reduction in the GLT mutant in high CO2 (Table 1). The GDH activity varied between 75% and 135% of the wild-type activity for glutamate synthesis and between 45% and 65% for glutamate oxidation in the three mutants (Table 1).

Figure 2.

 Schematic presentation of the gene structure with the T-DNA insertion site, and RT-PCR analysis of transcript levels in the Arabidopsis T-DNA insertion mutants. (A) GLU2 with T-DNA insertion in intron 9. (B) GLT with T-DNA insertion in exon 13. (C) carB with T-DNA insertion in the promoter. Gray triangles correspond to T-DNA, which is not to scale. Boxes indicate exons, and lines indicate 5′-flanking regions and introns. The nucleotide sequences at the gene–insertion junction are shown. The number of the first nucleotide refers to the position relative to A of the initial translation initiation ATG codon for methionine. (D) Transcripts estimated by RT-PCR for GLU1, GLU2, GLT, carB and 25S ribosomal RNA (rRNA) in the T-DNA mutants for GLU2, GLT and carB and the wild-type Arabidopsis (WT).

Table 1.   Activities of Fd-GOGAT, NADH-GOGAT, GS and GDH in the mutants and wild-type (WT) leaves of Arabidopsis under 3000 p.p.m. CO2 or atmospheric air. The enzyme assay conditions are described in Experimental procedures. GDH was assayed for NADH-dependent reductive amination (NADH-GDH) and oxidative deamination (NAD-GDH) of glutamate. The activity is expressed as μmol of glutamate formed (GOGATs), μmol of hydroxylamine formed (GS), or μmol of NADH oxidized (or of NAD reduced) (GDH) h−1·g−1 fresh weight.
Arabidopsis lines GLU1 GLU2 GLTWT
 Fd-GOGAT0.5 ± 0.126.4 ± 2.427.3 ± 2.128.7 ± 2.2
 NADH-GOGAT0.8 ± 0.10.6 ± 0.10.2 ± 0.10.8 ± 0.1
 GS115.5 ± 10.3115.0 ± 12.387.0 ± 8.1114.0 ± 10.5
 NADH-GDH28.6 ± 2.437.7 ± 4.226.8 ± 2.536.5 ± 3.2
 NAD-GDH5.5 ± 0.613.4 ± 1.57.7 ± 0.612.3 ± 1.9
 Fd-GOGAT0.5 ± 0.128.5 ± 2.329.2 ± 2.730.2 ± 3.3
 NADH-GOGAT0.8 ± 0.10.7 ± 0.10.2 ± 0.10.9 ± 0.1
 GS84.1 ± 8.976.8 ± 7.176.6 ± 6.775.0 ± 7.0
 NADH-GDH45.7 ± 5.932.6 ± 2.844.2 ± 3.938.2 ± 3.9
 NAD-GDH5.9 ± 0.713.4 ± 1.111.1 ± 1.09.9 ± 0.7

Phenotypic changes in the GOGAT and CPSase mutants

As our target was to evaluate the impact of gene function on ammonium assimilation and amino acid metabolism, we determined the levels of ammonium and free amino acids in leaves and compared them to the levels in the control wild-type lines. The GLU1 mutant accumulated a large amount of ammonium 48 h after transfer from high CO2 to air, owing to photorespiratory ammonium release (Fig. 3A). A slight accumulation of photorespiratory and nonphotorespiratory ammonium was detected in the GLT mutant (Fig. 3A). By contrast, the GLU2 and carB mutants contained a wild-type level of ammonium (Fig. 3A,B). The ammonium level of the control wild-type line of the GLU1 mutant 48 h after transfer from high CO2 to air (0.66 μmol·g−1 fresh weight) (Fig. 3A) was higher than that of the control wild-type line of the carB mutant in air (0.5 μmol·g−1 fresh weight) (Fig. 3B). This may be explained in part by the two experiments being performed independently. However, many of the reactions of nitrogen assimilation and amino acid synthesis depend on ATP, reduced Fd, and NAD(P)H, and take place in the chloroplast. Elevated CO2 causes an imbalance of energy and electron transport because of the lack of photorespiration, which dissipates excess photochemical energy and reducing equivalents [14]. This increases the number of chloroplasts and starch grains per mesophyll cell [15], and higher ammonium accumulation suggests that the control wild-type line did not completely recover the nitrogen assimilatory capacity damaged in high CO2. In high CO2, the GLU1 and GLT mutants had reduced glutamate levels and increased glutamine levels (Fig. 3C). The glutamate and glutamine levels were unaffected in the GLU2 mutant (Fig. 3C). These observations indicate that the GS/GLU1 Fd-GOGAT and GS/GLT NADH-GOGAT cycles are involved in nonphotorespiratory ammonium assimilation. In air, the highest glutamine/glutamate ratio of 13.3 was obtained for the GLU1 mutant, confirming that the GS/GLU1 Fd-GOGAT cycle is the main pathway of photorespiratory ammonium reassimilation (Fig. 3D). No impairment in glutamine to glutamate conversion was observed in the GLT and GLU2 mutants, whereas the GLT mutant accumulated asparagine (Fig. 3C,D). As CPSase supplies CP for arginine synthesis, the amino acid levels of the urea cycle were determined. Despite a tight linkage of CPSase to arginine synthesis, the carB mutant showed negligible effects on overall arginine levels. The levels of ornithine, citrulline and arginine remained low, at between 0.01 and 0.04 μmol·g−1 fresh weight (Fig. 3E). However, arginine accumulated up to 70-fold and 80-fold in the carB mutant and in the wild-type plants on 2 mm ammonium medium as compared with nitrate medium (Fig. 3E,F). The results suggest that excess ammonium was incorporated into arginine as a nitrogen storage compound. The GLU1 mutant showed a 5.8-fold increase in arginine relative to the wild-type plants in air, whereas in high CO2, arginine remained at a wild-type level, indicating that the high level of photorespiratory ammonium was in part refixed into arginine as a detoxification molecule (Fig. 3H).

Figure 3.

 Ammonium and amino acid contents in leaves of the GLU1 (ethylmethanesulfonate-mutagenized CS254 line), GLU2, GLT and carB mutants and the wild-type Arabidopsis (WT). Arabidopsis plants were grown for 42 days by hydroponic culture using 5 mm nitrate [37] in air supplemented with 3000 p.p.m. CO2, and then in air for 48 h. (A) Ammonium contents under high-CO2 conditions and in air. (B) Ammonium contents in air. (C) Glutamine (GLN), glutamate (GLU), asparagine (ASN) and aspartate (ASP) contents under high-CO2 conditions. (D) Glutamine, glutamate, asparagine and aspartate contents in air. (E) Ornithine (ORN), citrulline (CIT) and arginine (ARG) contents in leaves of Arabidopsis plants cultured with 5 mm nitrate in air. (F) Ornithine, citrulline and arginine contents in leaves of Arabidopsis plants cultured with 2 mm ammonium in air. (G) Ornithine, citrulline and arginine contents under high-CO2 conditions. (H) Ornithine, citrulline and arginine contents in air. Arabidopsis lines represent the EMS mutant for GLU1 (GLU1), T-DNA mutants for GLU2 (GLU2), GLT (GLT), and carB (carB), and wild-type Arabidopsis. The amino acid contents represent means of analysis on leaves from five independent plants.

Changes in gene expression patterns caused by exogenous ammonium

As the endogenous photorespiratory ammonium affected the levels of ammonium and amino acids in the GLU1 and GLT mutants (Fig. 3), we investigated whether expression of the ammonium assimilatory genes of GOGAT, CPSase and GS1 is modified in response to exogenous excess ammonium (10 mm), provided as a supplement to the culture medium. Both GLU1 and GLT were expressed at higher levels than GLU2 (Fig. 4A). The ammonium caused up to 4.7-fold induction of GLT mRNAs; the GLU1 mRNA was induced to a lesser extent (Fig. 4A). The level of carA mRNA was unaffected and that of carB mRNA was lowered by the ammonium treatment (Fig. 4B). The GS1 genes exhibited the contrasting patterns in response to excess ammonium: a decrease in the Gln12 mRNA and increases in the Gln11 and Gln13 mRNAs (Fig. 4C).

Figure 4.

 Regulation of transcript levels of the genes for GOGATs, CPSase and GS1 in Arabidopsis leaves in response to exogenous ammonium. Arabidopsis seedlings were grown for 12 days on Petri dishes with 5 mm nitrate, and then for 48 h in the absence or in the presence of 10 mm ammonium. Transcript levels were determined by real-time RT-PCR. (A) GOGAT genes: GLU1, GLU2, and GLT. (B) CPSase genes: carA and carB. (C) GS1 genes: Gln11, Gln12, Gln13, and Gln14. The values are expressed as percentage ± standard error relative to the marker EF1α gene.

Expression of promoter::GUS fusions

To investigate the tissue-specific expression of the genes for GOGATs and CPSase, transgenic lines expressing an N-terminal translational construct fused to a GUS reporter gene were generated. The promoter region upstream of ATG, including a partial coding sequence, was isolated by PCR from GLU1 (2385 bp) (−1931/454), GLU2 (1501 bp) (−1089/412), carA (1121 bp) (−1021/100), and carB (992 bp) (−922/70). The translational fusions to the uidA gene under the control of the gene promoter were constructed by inserting the PCR product in-frame to the 5′-end of the GUS reporter gene. In the leaf sections of the transformed Arabidopsis lines, the GLU1::GUS fusion was expressed in chloroplasts of the mesophyll (Fig. 5A). Furthermore, a high level of expression was detected in the vascular cells of minor veins (Fig. 5B). GUS activity was detected in a layer of cells composed of the companion cell–sieve element complex close to the several xylem tracheary elements (Fig. 5B). A low level of GLU2::GUS expression was found not only in the mesophyll chloroplasts, but also in the phloem of minor veins (Fig. 5C). A high level of carA::GUS expression was found in a cell layer close to the tracheary elements of the vascular bundle, together with its neighboring mesophyll cells (Fig. 5D). The expression of carB::GUS was associated with the mesophyll chloroplasts and the companion cell–sieve element complex in the phloem of minor veins (Fig. 5E). In the leaf sections from the plant transformed with empty vector, no staining was detected (Fig. 5F).

Figure 5.

 Histochemical analysis of promoter::GUS expression for GLU1, GLU2, carA and carB in Arabidopsis leaves. (A) Mesophyll section for GLU1. (B) Mesophyll and vascular section for GLU1. (C) Mesophyll and vascular section for GLU2. (D) Mesophyll and vascular section for carA. (E) Mesophyll and vascular section for carB. (F) Control mesophyll and vascular section from Arabidopsis transformed with an empty vector. bs, bundle sheath; cc, companion cell; chl, chloroplast; mc, mesophyll cell; se, sieve element; te, tracheary element. Bar: 10 μm.

In situ hybridization of the transcripts

In situ hybridization analysis was carried out to determine the tissue-specific expression pattern of GLU1, GLU2 and GLT in the leaf sections. After hybridization to the antisense RNA probe, the GLU1 mRNAs were found on the periphery of the mesophyll chloroplasts (Fig. 6A). In addition, specific staining appeared in the phloem against a pale background (Fig. 6B), consistent with the GLU1 promoter expression patterns (Fig. 5). The GLU2 mRNAs were found around the mesophyll chloroplasts (Fig. 6C). Furthermore, strong GLU2 mRNA staining was detected in the phloem adjacent to the mesophyll (Fig. 6E). The sense GLU2 mRNA probe gave no specific signal in the mesophyll or in the vascular cells (Fig. 6D). The GLT mRNAs were strongly expressed in the phloem, whereas a weak GLT mRNA signal was associated with the mesophyll (Fig. 6F), indicating that GLT was mainly expressed in the vascular cells.

Figure 6.

In situ hybridization of the transcripts of GLU1, GLU2 and GLT in Arabidopsis leaves. (A) Mesophyll section hybridized with the antisense GLU1 mRNA probe. (B) Vascular section hybridized with the antisense GLU1 mRNA probe. (C) Mesophyll section hybridized with the antisense GLU2 mRNA probe. (D) Mesophyll and vascular section hybridized with the sense GLU2 mRNA probe. (E) Vascular section hybridized with the antisense GLU2 mRNA probe. (F) Mesophyll and vascular section hybridized with the antisense GLT mRNA probe. bs, bundle sheath; cc, companion cell; chl, chloroplast; mc, mesophyll cell; pp, phloem parenchyma cell; se, sieve element; te, tracheary element. Bar: 10 μm.

Immunohistochemical localization

As the GLU1::GUS fusion and the GLU1 mRNAs were expressed both in the mesophyll cells and in the vascular cells, we examined the localization of Fd-GOGAT by the indirect immunofluorescence method, using a specific antibody against tobacco Fd-GOGAT as the primary antibody [4]. With the use of confocal laser-scanning microscopy, the Alexa 405 fluorochrome signal was detected in the mesophyll cells and in the vascular cells of minor veins bordering the mesophyll cells (Fig. 7A). With higher-magnification resolution, the specific fluorescence of Fd-GOGAT was found to be located in the mesophyll chloroplasts (Fig. 7C). The immunofluorescent signal and the corresponding transmission microscopy of the magnified vascular section showed that the specific signal was associated with the clustered oval companion cells, which flanked the sieve elements in close vicinity to the phloem parenchyma (Fig. 7E,F). With nonimmune serum as the first antibody, no signal was found in the leaf sections (Fig. 7B,D).

Figure 7.

 Immunohistochemical localization of Fd-GOGAT in Arabidopsis leaves. (A) Mesophyll and vascular section hybridized with the antibody against Fd-GOGAT as the primary antibody. (B) Control mesophyll and vascular section hybridized with nonimmune serum as the primary antibody. (C) Mesophyll section hybridized with the antibody against Fd-GOGAT as the primary antibody. (D) Control mesophyll section hybridized with nonimmune serum as the primary antibody. (E) Vascular section hybridized with the antibody against Fd-GOGAT as the primary antibody. (F) Transmission of vascular section corresponding to (E). bs, bundle sheath; cc, companion cell; chl, chloroplast; mc, mesophyll cell; phl, phloem; pp, phloem parenchyma cell; se, sieve element. Bar: 8 μm.


Recovery of excess ammonium into amino acids in the mesophyll

The expression analysis showed that the GLU1 mRNAs were mainly expressed in leaves, in which the GS1 and GS2 genes were coexpressed (Fig. 1). The GLU1 mRNAs were found around the mesophyll chloroplasts, where Fd-GOGAT protein was immunohistochemically located (Figs 5–7). GLU2, the other Fd-GOGAT gene, was also expressed in the mesophyll cells, albeit at lower levels than GLU1 (Figs 5 and 6). The high level of expression of GLU1 in comparison with that of GLU2 and the conditional lethal phenotype of the GLU1 mutant confirm that the defect in the GLU1 Fd-GOGAT cycle caused the inhibition of photosynthesis, owing to the extensive release of photorespiratory ammonium (up to 5–20 μmol·h−1·g−1 fresh weight) [2,16,17]. The high levels of glutamine and glutamate (nitrogen-rich five-carbon amino acids) and asparagine and aspartate (four-carbon amino acids) (up to 80% of the total amino acids) (Fig. 3) suggest that excess photorespiratory ammonium was detoxified, in part, in the form of amino acids for export out of parenchyma cells of the veins. The high glutamine/glutamate ratio in the GLU1 mutant (13.3) as compared with the wild type in air (1.4) (Fig. 3) reflects the inability of mitochondrial GDH to act as an alternative ammonium assimilatory pathway in the leaves, as GDH is a vascular-located enzyme [18]. As demonstrated here, the minor effects on ammonium accumulation in the GLU2 mutant in air (Fig. 3) provide evidence that the GS/GLU2 Fd-GOGAT cycle does not contribute to photorespiratory ammonium reassimilation. The low GLU2 mRNA levels in the leaves (Figs 1 and 4) suggest that GLU2 Fd-GOGAT supplies a constitutive level of glutamate to maintain a basal level of protein synthesis.

The high levels of photorespiratory ammonium in the GLU1 mutant seem to be shifted in part to the CPSase pathway, resulting in substantial accumulation of arginine (Fig. 3). Arginine synthesis involves ornithine formation from glutamate [6]. Carbamoylation of the ornithine δ-amino group with CP leads to the formation of citrulline as a precursor of arginine synthesis (see Fig. 8 for a diagram of arginine synthesis). It has been proposed that photorespiratory ammonium released by mitochondrial glycine decarboxylase complex (GDC; EC is reassimilated into glutamine by GS, and then into CP by CPSase in the mitochondria [19]. However, the subcellular compartmentation of CPSase has been unclear. We showed that the promoter from either carA or carB directed the GUS signal to the mesophyll chloroplasts (Fig. 5), indicating that photorespiratory ammonium is shuttled via glutamine to CP in the chloroplasts. Glutamine is hydrolyzed via the class I or trpG-type glutaminase of the CPSase small subunit. The carB domain of the CPSase large subunit forms the Cys-NH2 intermediate by the conserved triad (Cys293-His377-Glu379) to activate inline image-dependent ATP cleavage prior to release of CP [20]. The databases also predict importation of the large subunit (cleavage at Cys62) and small subunit (cleavage at Val33) to the chloroplast stroma [21,22]. In addition, plastid-located carbonic anhydrase 1 (At1g58180, cleavage at Ala113) and cytosolic carbonic anhydrase 2 (At5g14740) can increase the inline image supply via CO2/inline image interconversion [23,24]. Consistently, mitochondria have been shown to be unable to use ammonium, and only 0.2% of [15N]ammonium from [15N]glycine was metabolized to [15N]glutamate, at a rate of 2.64 nmol·h−1·mg−1 protein [25]. However, it has been shown that GS is localized to the mitochondria and that the mitochondria are highly capable of using glycine to convert ornithine to citrulline (up to 126 μmol·h−1·mg−1 protein) [9]. Because of a lack of bioinformatic tools to predict to what extent the large and small precursors are seemingly dual-targeted, a dual organelle location of the CPSase in the chloroplasts and mitochondria cannot not be excluded.

Figure 8.

  Proposed diagram for the role of GOGATs and CPSase in primary nitrogen assimilation, the photorespiratory nitrogen cycle, and nitrogen translocation. The organelle localizations and stoichiometries of the interconnected enzymatic reactions are not included. CH2-THF, N5,N10-methylene tetrahydrofolate; FdH, reduced ferredoxin; glycolate-P, 2-phosphoglycolate; N-acetylglutamate-5-P, N-acetyl-glutamate 5-phosphate; OH-pyruvate, hydroxypyruvate; OTC, ornithine transcarbamoylase (EC; PGA, 3-phosphoglycerate; RuBP, ribulose 1,5-bisphosphate.

Excess ammonium from either endogenous photorespiration or exogenous medium appears to be, in part, shuttled to arginine (Fig. 3). The fact that there were only slight effects of the carB mutation on overall arginine synthesis, either with excess ammonium or under standard nitrate conditions, suggests that CPSase is not the limiting enzyme for arginine biosynthesis. However, the GLU1 mutant accumulated arginine at a higher level than the wild-type plants under photorespiratory conditions (Fig. 3). It can thus be assumed that photorespiratory ammonium was shuttled to arginine under the control of N-acetyl-glutamate kinase (NAGK; EC, a key regulatory enzyme in the arginine synthetic pathway [6].

Nitrogen entry into amino acids and translocation in the vascular tissue

Under high-CO2 conditions, when photorespiration is suppressed, leaf cells depend on the importation of nitrogen via the tracheary elements for amino acid synthesis and subsequent export of the derived amino acids via phloem sieve elements for use by sink cells (Fig. 8). Cellular localization of GOGATs and CPSase in the vascular tissue has been unknown in Arabidopsis. To dissect the regulation of amino acid translocation, we determined whether GOGATs and CPSase were localized in the phloem companion cell–sieve element complex. Cis-acting regulatory elements upstream of ATG were examined in silico, using the place database [26]. The TATA or TATA-like boxes were identified for GLU1 (−61TTATTT−56 and −37TTATTT−32), GLU2 [−506TTATTT−501 and −90TTATTT−85 (−strand)], GLT (−311TATAAAT−305), carA (−277TATATAA−271 and −188TTATTT−183), and carB [−361TTATTT−356 and −143TTATTT−138 (−strand)]. Consistent with the mesophyll localization, cis-elements active in mesophyll expression were found: Mem1 motif (CACT) [27]; GLU1 (at positions −258/−255, −220/−217, and −129/−216), GLU2 (−223/−220, −218/−215, and −139/−216), GLT (−310/−307, −28/−25, and −24/−21), carA (−237/−234, −151/−148, and −115/−112), and carB (−405/−402 and −79/−76). The Mem1 sequence is supposed to direct mesophyll expression as a result of transcription repression in the vascular bundle [27]. In addition, the strong cis-elements that determine vascular patterning were identified: the BS1 motif [28] [carA (−875AGCGGG−869), −strand] and the NtBBF1 motif (ACTTTA) [GLU1 (−1180/−1175), GLU2 (−381/−376), GLT (−1499/−1494), carA (−237/−232), and carB (−526/−521)]. The NtBBF1 motif directs expression of the oncogene rolB in phloem and xylem parenchyma [29]. By in situ hybridization, the GLT mRNAs were found to be confined to the phloem companion cell–sieve element complex (Fig. 6). The GLT mutant showed strong inhibition of primary ammonium assimilation, with a high glutamine/glutamate ratio under high-CO2 conditions (Fig. 3). The impairment was caused by a lack of only 3% of the total GOGAT activity, which was restricted to the small compartment as low as 5% of mature leaf volume [30]. This underlines the importance of the companion cells for amino acid trafficking. Moreover, we found that GLU1 was coexpressed in this cellular compartment (Figs 5–7). Fd-GOGAT did not compensate for the low NADH-GOGAT activity in the leaves. This would reflect a supply of glutamate from the roots by NADH-GOGAT, which was highly expressed in the roots (Fig. 1).

The companion cell–sieve element complex is the first cell to collect the solutes and signals, which derive from the phloem transport and from the retrieval by xylem-to-phloem pathway. A high level of exogenous ammonium was found to induce the expression of GLT, Gln11, and Gln13 (Fig. 4). By contrast, the decrease in the Gln12 mRNA level suggests that the Gln12 expression is regulated by a leaf-specific mechanism, because in roots ammonium induces Gln12 expression in the vascular pericycle [31] for the synthesis and transport of glutamine to the leaves. Also, the high level of ammonium was found to induce ASN1 and ASN2 expression (data not shown). Coexpression of the genes for GS1, GOGAT and AS in the leaf phloem (Figs 5–7, and data not shown) indicates that glutamine/glutamate and asparagine/aspartate are, in part, produced in this cellular compartment. The data support the view that the GS1/NADH-GOGAT cycle in the phloem functions to enable the entry of excess ammonium and incoming primary nitrogen into glutamine and glutamate (Fig. 8). These amino acids are trafficked under the fine control of amino acid transporters [32,33]. Active uptake into yeast cells suggests that the basic amino acids arginine and lysine are transported by the specific permeases (AAP3 and AAP5) for their retrieval along the translocation pathway and accumulation [34]. Based on the finding of CPSase in the phloem, the low ability of [14C]arginine to move out of the vascular bundles can be attributed to the synthesis of arginine precursor by CPSase in the vascular tissue (Fig. 5). Consistently, leaves fed with [14C]arginine via the xylem saps show more extensive labeling of the vein than the mesophyll, and the reverse holds for [14C]glutamate and [14C]aspartate [35].

In conclusion, in response to the enhanced levels of photorespiratory ammonium and exogenously added ammonium, high levels of ammonium were converted to amino acids to allow for transport in Arabidopsis. When the GLU1 mutant was impaired in the photorespiratory nitrogen cycle, owing to the absence of the GS/GOGAT cycle, photorespiratory ammonium seemed to be shifted to arginine via glutamine and CP generated by CPSase in the chloroplasts. The strong defect of the GLT mutant with regard to the assimilation of primary ammonium when grown under high-CO2 conditions underlines the importance of amino acid synthesis and trafficking via the phloem companion cell–sieve element complex, where NADH-GOGAT was mainly located. Coexpression of the genes for Fd-GOGAT, NADH-GOGAT, GS1 and AS in the phloem is consistent with the view that the major nitrogen carriers – glutamine, glutamate, asparagine, and aspartate – are partly produced in the phloem for translocation through the vascular bundle (Fig. 8). Further experiments are required to evaluate the excess ammonium assimilation into CP associated with the formation of citrulline intermediates in arginine synthesis, and also the contribution of the phloem to nitrogen metabolism in Arabidopsis.

Experimental procedures

Isolation of homozygous T-DNA insertion lines

Seeds of T-DNA insertion mutants from a T-DNA mutagenized Arabidopsis thaliana Col-0 ecotype for GLU2 (SALK-018671), GLT (SALK-072454) and carB (SALK-034177) were obtained from the Nottingham Arabidopsis Stock Centre (Nottingham, UK). Homozygous mutant lines were isolated by PCR with the gene-specific primers and T-DNA border primer. The first PCR was carried out using the following gene-specific primers: GLU2 (SALK_087050) LP, 5′-AAACCTGCGAAACCTGAAGCC-3′; GLU2 RP, 5′-TCA CCAAGCAAACCCTCAAGC-3′; GLT (SALK_072454) LP, 5′-TCTCTGGAGGCGCATACAACC-3′; GLT RP, 5′-CCAGCGAGATGCACCAGTACC-3′; carB (SALK_ 034177) LP, 5′-GAGAAGGACATGCGGTACTAG-3′; and carB RP, 5′-AGTGAGACACGAGAGAGAGGG-3′. The reaction mixture consisted of 0.4 ng of genomic DNA isolated from rosette leaves, 10 pmol of forward primer, 10 pmol of reverse primer and 0.2 units of Taq polymerase in a total volume of 25 μL. The following program was used: presoaking at 95 °C for 3 min, and 35 cycles of 94 °C for 30 s, 55–69 °C for 1 min 30 s, and 72 °C for 1 min 30 s, with postsoaking at 72 °C for 10 min. The second PCR analysis was carried out using one of two gene-specific primers (forward or reverse) and the following LBb1 border primer: 5′-GCGTGGACCGCTTGCTGCAATT-3′. The T-DNA insertion was located, and levels of transcripts downstream of the insertion site were determined by RT-PCR. Amplified fragments were visualized by ethidium bromide staining in agarose gels, and bands were quantified by scanning with an ImageGauge imaging system (Fujifilm S.A.S., St-Quentin, France).

Plant culture

A. thaliana ecotype Col-0, T-DNA insertion mutants, the EMS-mutagenenized GLU1 mutant (strain CS254) [2,14] and the transgenic plants were grown either under hydroponic conditions in a growth chamber or on Petri dishes.

Hydroponic culture

For hydroponic culture, surface-sterilized seeds were cold-treated at 4 °C for 2 days in the dark. Seeds were sown on top of Eppendorf tubes cut at the bottom and filled with a medium consisting of a half-strength standard nutrient solution and 0.8% agar as described in [36]. Seedlings were cultured using the standard solution [37] under an 8 h light (150 μmol·photons·m−2·s−1; 21 °C)/16 h dark (17 °C) photoperiod with 80% relative humidity. After 42 days, leaves and roots were harvested at 4 h into a light period, frozen in liquid nitrogen, and kept at −70 °C prior to analysis.

Growth on Petri dishes

Surface-sterilized Arabidopsis seeds were germinated on Petri dishes using a nutrient solution containing 5 mm nitrate, 3% sucrose, and vitamins. Plants were incubated vertically for 12 days under an 8 h light (150 μmol·photons·m−2·s−1; 21 °C)/16 h dark (17 °C) photoperiod. For the ammonium induction experiments, seedlings were then transferred to the nutrient solution supplemented with 10 mm ammonium for an additional 48 h under the same regime.

Real-time RT-PCR analysis

Total RNA was extracted, and first cDNA strands were synthesized from 2 μg of RNA, using an Invitrogen RT kit (Invitrogen SARL, Cergy Pontoise, France). Real-time RT-PCR was carried out with a RealMasterMix Cybr Rox 2.5x kit according to the manufacturer’s instructions (5 PRIME; Dominique Dutscher SA, Brumath, France). Amplification was carried out with the following conditions, using 1 μL of a 1 : 10 or 1 : 20 dilution of cDNA in a total volume of 20 μL: 2 min at 95 °C, and 40 cycles of 95 °C for 19 s, 55 °C for 15 s, and 68 °C for 40 s, on an Eppendorf Realplex2 MasterCycler (Eppendorf SARL, Le Pecq, France). A melting curve was obtained to confirm the specificity of the amplification. For the genes of the multigene family, the following primer sets were designed along the nonconserved stretches of the genes. The results were expressed as percentage relative to EF1α (At5g60390) as a constitutive gene. The primers used for quantitative real-time PCR are listed in Table 2.

Table 2.   Primers used for quantitative real-time RT-PCR analysis. Amplification was carried out as described in Experimental procedures. F, forward primer; R, reverse primer.
 Primer (5′- to 3′)

Construction of GUS fusions by attB recombination reactions

Binary vectors containing the promoter sequence upstream of ATG carrying a partial coding sequence of GLU1 (2362 bp), GLU2 (1580 bp), carA (1098 bp) or carB (1071 bp) was inserted in front of GUS by site-specific recombination using Gateway vectors [38]. The 5′-flanking regions were amplified by PCR using the following gene-specific primers by introducing attB1 (5′-AA AAA GCA GGC T-3′) and attB2 (5′-A GAA AGC TGG GT-3′) recombination sites at the 5′-ends and 3′-ends, respectively: GLU1 forward, 5′-AAAACCCTAAACCCCCAATGT-3′; GLU1 reverse, 5′-GAGCATCTTTGACAACTCCATGTG-3′; GLU2 forward, 5′-TCGTGGTGGTTGATTCATTTT-3′; GLU2 reverse, 5′-TGTGTTCCATACAACCAAGTGC-3′; carA forward, 5′-CACACCAATCTTTACGAGT-3′; carA reverse, 5′-CGACAGAAACCCTAAATCCACCGC-3′; carB forward, 5′-TGTCCAGTTGCACCATTATCAAA-3′; and carB reverse, 5′-CCGGATTGGATTTGGAAGAAGCG-3′. The PCR products were cloned into pDONR207 and then in front of the N-terminus of GUS in pMDC163, according to the manufacturer’s instruction (Invitrogen SARL). Recombinant vectors containing the GUS fusion construct were transferred by electroporation into Agrobacterium tumefaciens (strain C58PMP90) [39]. A. thaliana ecotype Col-0 was transformed by dipping floral tissues into transformed Ag. tumefaciens containing 5% sucrose and 0.005% (v/v) surfactant Silvet L-77 [40]. Transformants were recovered from the seeds selected on Murashige–Skoog medium containing 30 mg·L−1 hygromicine B.

GUS histochemical analysis

In situ staining of GUS activity was carried out by incubating tissues in 50 mm sodium phosphate (pH 7.0), 0.1 mm K3[Fe(CN)6], 0.1 mm K4[Fe(CN)6] and 1.9 mm 5-bromo-4-chloro-3-indolyl β-d-glucuronic acid (X-glucuronide) at 37 °C for 2–18 h according to [41]. After removal of chlorophyll with 70% ethanol, tissues were embedded in resin using Kuzler Histo-Technique-Set 7100 (Heraeus Kuzler, Friedrichsdorf, Germany) as described in [42]. Eight-micrometer sections were cut with a microtome (Jung RM 2055; Leica Microsystems, Wetzlar, Germany). GUS staining was visualized using a Leica DMR microscope (Leica Microsystems).

In situ hybridization

Tissue inclusion

Leaf tissues were fixed in 4% (v/v) paraformaldehyde and 0.1% Triton X-100 in NaCl/Pi (10 mm sodium phosphate, pH 7.0, 130 mm NaCl). Tissues were dehydrated in a gradual series of ethanol in NaCl/Pi (10%, 30%, 50%, 70%, and 96%) and ethanol/histoclear [2 : 1, 1 : 1 and 1 : 2 (v/v)] at 4 °C. Tissues were then incubated in 100% histoclear, histoclear/paraffin (1 : 1, v/v) and paraffin at 59 °C.

Hybridization probe preparation

Total RNA was extracted using an RNA isolation kit (Qiagen, GmbH, Germany), and first cDNA strands were synthesized from 2 μg of RNA using an Omniscript RT kit (Qiagen). Sense and antisense DNA probes were amplified by PCR using the following gene-specific primers by introducing a T7 sequence (5′-TGTAATACGACTCACTATAGGGC-3′) at the 5′-ends of reverse and forward primers, respectively: GLU1 forward, 5′-ATCATTCAAGAGCAGGTTGT-3′; GLU1 reverse, 5′-GACAGTTGAAAGCAGTTATT-3′; GLU2 forward, 5′-TCAACATTTGATCGTGGTTT-3′; GLU2 reverse, 5′-AATCGAAAACCCTTTCTTAA-3′; GLT forward, 5′-GGTGGGCTGATGATGTATGGA-3′; and GLT reverse, 5′-CATCATCCGTTTTGGTGAGGA-3′. Amplified sense and antisense DNAs (400 ng each) were subjected to in vitro transcription using a transcription kit (Promega, Madison, WI, USA) in the presence of digoxigenin-UTP. DNAs were removed by DNase digestion. RNA probes were controlled by electrophoresis.

In situ hybridization

Eight-micrometer sections were prepared using a microtome and dried on glass slides (DAKO 2024; Dako, Basingstoke, UK). Samples were deparaffined in histoclear, hydrated through a gradual ethanol series (96%, 85%, 50%, and 30%, v/v), and washed in NaCl/Pi (6.5 mm Na2HPO4, 1.5 mm KH2PO4, pH 7.3, 14 mm NaCl, 2.7 mm KCl). Proteins were removed by proteinase K digestion (4 μg·mL−1 in 10 mm Tris/HCl, pH 7.5, 50 mm EDTA). Samples were treated with 0.5% (v/v) acetic anhydride in 1.3 m triethanolamine (pH 7.0). Samples were dehydrated in a gradual series of ethanol in NaCl/Pi (30%, 50%, 70%, 85%, 96%, and 100%, v/v), and prehybridized with 50% (v/v) formaldehyde, 5× SSC (1× SSC: 150 mm NaCl and 15 mm sodium citrate, pH 7.0), 100 μg·mL−1 tRNA, 50 μg·mL−1, heparin and 0.1% Tween-20. Samples were hybridized with the sense or antisense probe in situ hybridization solution (Dako). Slides were washed in 0.2× SSC, and then T2 solution (0.5% blocking reagent dissolved in T1 solution: 100 mm Tris/HCl, pH 7.5, 150 mm NaCl) (Roche Diagnostics Gmbh, Penzberg, Germany) and T3 solution (T1 solution with 1% BSA and 0.5% Triton X-100). Slides were incubated with anti-digoxigenin antibody conjugated with alkaline phosphatase (Roche Diagnostics Gmbh) in T3 solution. After washing with T3 solution, alkaline phosphatase activity was developed with 5-bromo-4-chloro-3-indolyl-phosphate (50 mg·mL−1) and Nitroblue tetrazolium (75 mg·mL−1). Slides were washed with TE and sealed with gel mount formol 1 (Microm Microtech France, Francheville, France). Fluorescence was observed using a Leica DMR microscope (Leica Microsystems).

Indirect immunofluorescence analysis

Leaf sections were fixed in 3.7% (w/v) formaldehyde dissolved in 50 mm Pipes buffer (pH 6.9), 5 mm MgSO4, and 5 mm EGTA (MTSB), and then in NaCl/Pi (6.5 mm Na2HPO4, 1.5 mm KH2PO4, pH 7.3, 14 mm NaCl, 2.7 mm KCl). Tissues were dehydrated in a graded series of ethanol (30%, 50%, 70%, 90%, and 97%), and embedded in wax. Eight-micrometer sections were cut with a microtome and mounted onto slides, and then dewaxed and rehydrated in a graded series of ethanol (97%, 90%, and 50%). Antigen unmasking was carried out in 10 mm citrate buffer (pH 6.0), and blocked with 1% (w/v) BSA in NaCl/Pi (blocking solution). Leaf sections were hybridized with the primary rabbit IgG against tobacco Fd-GOGAT, and then goat anti-(rabbit IgG) labeled with Alexa 405 (Molecular Probes, Carlsbad, CA, USA) dissolved in blocking solution. Preimmune serum was used as the control primary IgG. Immunofluorescence was observed with a laser diode (25 mW, 405 nm) using a Leica objective (HC PL APO 63×/1.20 Water Corr/0.17 Lbd. BL) and a spectral confocal laser-scanning microscope (TCS-SP2-AOBS) (Leica Microsystems). Low speed scan (200 lines per second) images (512 × 512 pixels) were generated, and Alexa 405 fluorescence was measured at a specific bandwidth (407–427 nm) after spectral adjustment to eliminate the background noise. The red autofluorescence of tissues was observed between 509 and 628 nm.

Amino acid analysis

Amino acids were extracted from 20 mg samples at 4 °C with 1 mL of 2% (w/v) sulfosalicylic acid. After centrifugation at 17 500 g for 15 min, supernatants were adjusted to pH 2.1 with LiOH. Total amino acid contents were estimated by the method of [43]. Amino acids were separated by ion exchange chromatography on a JLC-500/V amino acid analyzer (Jeol Ltd, Tokyo, Japan).

Enzyme preparation and assays

GS was extracted and assayed by measuring γ-glutamylhydroxamate produced by its synthetase reaction according to [44]. Fd-GOGAT and NADH-GOGAT were extracted and assayed as described in [12]. Glutamate formation with ferredoxin or NADH as electron donor was determined by HPLC. GDH was assayed for NADH-dependent glutamate synthetic activity and NAD-dependent glutamate oxidation activity as described in [44].

Determination of metabolites and total soluble proteins

Free ammonium contents were determined by the phenol hypochlorite assay of Berthelot [45]. Soluble protein contents were determined by the Coomassie Blue dye-binding assay (Bio-Rad Laboratories, Hercules, CA, USA).