Sixteen cytosolic glutamine synthetase genes identified in the Brassica napus L. genome are differentially regulated depending on nitrogen regimes and leaf senescence

Summary BnaGLN1 coding sequences and expression profiles in response to nitrogen availability and ageing are essentially conserved compared with A. thaliana, suggesting that the roles of GLN1 families are conserved among the Brassiceae tribe.


Introduction
Winter oilseed rape (Brassica napus L.) is the dominant oilseed crop in northern Europe, and nitrogen (N) fertilization is the main operational cost for farmers (50% of the total cost of production). When compared with other crops, oilseed rape is characterized by low nitrogen use efficiency (NUE) (Rathke et al., 2006). Despite a high N-uptake efficiency (Laine et al., 1993), only half the N originating from fertilizer application is recovered in the seeds (Schjoerring et al., 1995). Oilseed rape is characterized by early leaf shedding and unusual high N loss in senescing falling leaves. The plant can lose up to 15% of its entire N content in this way (Rossato et al., 2001). Leaf senescence generally corresponds to the mobilization of N reserves from source leaves to sink organs such as seeds . In oilseed rape, it has been shown that N can be remobilized from senescing leaves to expanding leaves at the vegetative stage (sequential senescence) as well as from senescing leaves to seeds at the reproductive stage (monocarpic senescence) (Malagoli et al., 2005).
The rate of senescence and remobilization of leaf N are related to the N nutrition status of the plant and to sourcesink relations (Masclaux et al., 2000). N remobilization towards new developing organs is largely dependent on senescence-related catabolism events and translocation of leaf N compounds. Amino acids derived from protein catabolism are exported via the phloem to growing parts of the plant; the concentration of amino acids in the phloem sap increases during leaf senescence (Herrera-Rodriguez et al., 2006Masclaux-Daubresse et al., 2006). In many species including B. napus, aspartate, glutamate, and their corresponding amides are the principal forms of amino N compounds transported in the phloem and play a key role in rendering N available for remobilization from senescing leaves (Tilsner et al., 2005). Enzymes involved in the biosynthesis and metabolism of amino acids destined for phloem loading are of special interest.
In plants, glutamine synthetase (GS; EC 6.3.1.2) is a key enzyme which catalyses an ATP-dependent conversion of glutamate to glutamine using ammonium derived from primary N uptake and various internal N recycling pathways including catabolic release of ammonium during senescence (Bernard and Habash, 2009). In a large variety of plants, induction of cytosolic glutamine synthetase (GS1) genes has been detected during leaf senescence, while chloroplastic synthetase isoenzyme (GS2) expression decreases (Masclaux et al., 2000;Guo et al., 2004;Martin et al., 2006). It has been proposed that in young photosynthetic leaves, the chloroplastic isoenzyme GS2 is mainly involved in the assimilation of ammonium provided by nitrate reduction and photorespiration through the GS/GOGAT cycle (Masclaux et al., 2001). In old senescing leaves, as chloroplasts are breaking down, glutamine to be exported would be synthesized by the newly expressed cytosolic GS1 isoforms (Masclaux-Daubresse et al., 2006).
The importance of GS1 in N management, growth rate, leaf senescence onset and severity, yield, and grain filling has been confirmed by co-location of quantitative trait loci (QTLs) and functional genomics approaches mainly performed on maize Martin et al., 2006) and rice (Tabuchi et al., 2005). In maize, Gln1.4 is up-regulated during senescence (Martin et al., 2005). The Gln1.4 knockout mutation led to a dysfunction in N export and a sharp reduction of kernel yield (Martin et al., 2006). GLN1.4 was proposed to be involved in re-assimilation of ammonium released during leaf protein degradation. In rice, mutants lacking OsGS1;1 are severely impaired in growth rate and grain filling, and glutamine levels in mutant leaf blades are reduced (Tabuchi et al., 2005). As the gene product is located in companion cells and parenchyma cells of leaf tissues, it has been proposed that OsGS1;1 is responsible for generation of glutamine for remobilization via the phloem.
To date, all studies on plant genomes have revealed multigenic families coding for several GS1 isoforms. In rice, three GLN1 genes have been identified, with seven in wheat, five in maize, and five in Arabidopsis thaliana. Transcriptomic data showed that three A. thaliana genes, AtGLN1.1, AtGLN1.2,and AtGLN1.4, are induced during leaf ageing (Guo et al., 2004). Promoter::GFP (green fluoresacent protein) fusions were used to investigate AtGLN1 gene expression in roots. AtGLN1.1 was localized at the root surface layer, whereas AtGLN1.2, AtGLN1.3, and AtGLN1.4 were expressed in root vascular tissues (Ishiyama et al., 2004). Detailed expression of AtGLN1 in leaves was only reported for AtGLN1.2 that is induced in root and leaf tissues under a high N regime and is mainly expressed in veins and mesophyll cells in older leaf tissues (Lothier et al., 2011). In veins, AtGLN1.2 protein was localized in the companion cells. The knock-out mutant phenotype led to the conclusion that AtGLN1.2 is essential for N assimilation under ample nitrate supply and for ammonium detoxification (Lothier et al., 2011). For all plant species, it is clear that not all GS1 isoforms participate equally in N management and remobilization. Regulation of expression is then a key clue towards the identification of GLN1 genes potentially involved in N remobilization.
Accumulation of GS1 and a decrease in GS2 polypeptides were observed in B. napus leaves after onset of leaf senescence (Ochs et al., 1999). Up to now, four closely related genes coding for GS1 isoenzymes, BnGSR1-1, BnGSR1-2, BnGSR2-1, and BnGSR2-2, have been identified using B. napus root-derived expressed sequence tag (EST) libraries (Ochs et al., 1999). Analysis of different tissue types has also revealed that these genes are expressed in senescing leaves (Buchanan-Wollaston and Ainsworth, 1997). Recent studies of Brassicaceae genomes show that the genome of B. napus, which is a recent allotetraloid (2n=4x=38, AACC) arising from the natural hybridization of monogenomic diploids Brassica rapa (AA) and Brassica oleracea (CC) (Nagaharu, 1935), contains additional genes coding for GS1 isoenzymes. Analysis of Brassica lineage genomes revealed that a wholegenome triplication occurred shortly after their divergence from Arabidopsis (Parkin et al., 2005). Therefore, gene families are more frequent, larger, and more complex in B. napus than in A. thaliana. Brassicaceae genome sequences are also highly conserved and many synthenic regions have been identified (Paterson et al., 2001;Parkin et al., 2005;Schranz et al., 2006), allowing the identification of 'true' orthologous genes between A. thaliana and B. napus.
In the present study, advantage is taken of the Brassicaceae genome structure and of its recent sequencing (unpublished) in order to identify all BnaGLN1 genes coding for GS1 isoenzymes. It is demonstrated that they are differentially regulated depending on tissue type, senescence, and N availability. The potential role of the BnaGLN1 genes in N remobilization during leaf senescence, the impact of whole-genome duplications and merging on the evolution of the GLN1 multigenic family in the Brassiceae tribe, and strategies based on knowledge transfer from A. thaliana to crop plants are discussed.
BnaGLN1 contigs were then enriched and/or their coding sequence completed with new cDNA sequences: clones from Genoplante oilseed rape cDNA libraries and the ADIZ-MPIZ 021 library corresponding to ESTs were sequenced when available (Supplementary Table S1 at JXB online). When the coding sequence from cDNAs was incomplete or no clone was available, specific primers were designed to clone the total or missing coding region (Supplementary Table S2). The amplified fragments were cloned into pGEM ® -T Easy plasmids (Promega) according to the recommendation of the supplier, and sequenced. Universal T6 and SP7 primers, as well as specific primers were used to sequence the clones on the positive and negative strands (Supplementary Table S3). All DNA sequencing was performed by Cogenics (Grenoble, France) and sequences were submitted to GenBank (accession numbers are given in Supplementary Table S1).

Sequence analysis
A global alignment of coding sequences from mRNA, inferred coding sequences, or newly created contigs from ESTs was generated with ClustalW (Thompson et al., 1994). Distance matrixes were computed using the Dnadist algorithm with a Kimura 2 nucleotide substitution model, and bootstrap analysis was performed with 1000 iterations. A consensus unrooted tree was then generated using the Neighbor-Joining method. All algorithms are contained in the Phylip 3.67 package available at the MOBYLE platform (http://mobyle.pasteur.fr). The NCBI Conserved Domain Database (Marchler-Bauer et al., 2011) was searched with translated B. napus and A. thaliana GLN1-coding sequences. A multiple protein sequence alignment was generated with the ClustalW algorithm.
Genetic mapping and genome or chromosome assignment Genetic mapping of BnaGLN1 genes was realized using three different B. napus double haploid (DH) populations. The Stellar×Drakkar (SD), Darmor×Samouraï (DS), and Darmor-bzh×Yudal (DY) populations consist, respectively, of 94, 134, and 445 genotype DH lines described by Lombard and Delourme (2001) and Delourme et al. (2006). Gene-specific primers were designed and selected for a presence/absence polymorphism in one of the three populations (Supplementary Table S2 at JXB online). Linkage analyses were performed as previously described by Auger et al. (2009) using MAPMAKER/EXP version 3.0b (Lander et al., 1987) and framework maps constructed in Lombard and Delourme (2001) and updated in Delourme et al. (2006). BnaGLN1 genes were assigned to a linkage group using the ASSIGN command (LOD threshold=8.0) and then placed in the most confident interval with the PLACE command (LOD threshold=2.0). Recombination frequencies were converted into centiMorgans (cM) with the Kosambi function (Kosambi, 1944).
BnaGLN1 gene assignment to A or C Brassica genomes was performed using a panel of diverse B. napus, B. oleracea, and B. rapa genotypes available in the authors' group. The panel contains genomic DNA from B. napus genotypes Darmor-bzh, Yudal, Stellar, Drakkar, Samouraï, aburamassari, Aviso, Tenor, Express, Montego; B. rapa genotypes Z1, C1.3, Chiifu; and B. oleracea genotypes HDEM and C102. This panel was PCR screened with specific but not polymorphic gene markers (Supplementary Table S2 at JXB online).
Chromosome assignment for BnaC.GLN1 genes was realized using monosomic and polysomic addition lines carrying one or several additional C chromosomes from Darmor-bzh. Lines were selected from a cross between B. napus Darmor-bzh and B. oleracea C1.3 (A.M. Chèvre and F. Eber, INRA Rennes, unpublished results). Genomic DNA from these lines was PCR screened with specific but not polymorphic markers (Supplementary Table S2 at JXB online).

Plant material and growth conditions
Brassica napus L. plants from the Darmor-bzh genotype were grown in a greenhouse at INRA Versailles, France. Seeds were sown on sand and watered with nutritive solution during 2 weeks in order to allow germination and subsequent growth of plantlets. When the first two true leaves appeared, plantlets were transferred into pots containing sand and were separated into two groups with contrasting N fertilization regimes (LN for low nitrate and HN for high ntrate, 0.4 mM and 8 mM NO 3 -, respectively) according to Albert et al. (2012). At 56 d after sowing, four plants of each nutrition regime were harvested and sampled. For each plant, all leaf ranks were collected: primary and secondary veins were separated from the rest of the leaf, described as the limb. All fresh samples were frozen immediately in liquid nitrogen and stored at -80 °C.
Brassica napus L. plants from the Express genotype were grown in field trials in 2009-2010, in Le Rheu (Brittany), France. Seeds were sown on 7 September 2009 with plant density set at 40 plants m -2 . The field trial was conducted with contrasting N fertilization regimes. Plant N status was monitored over the vegetative stage by calculating the nitrogen nutrition index (NNI) (Colnenne et al., 1998). The balance-sheet method was used as a decision tool for N fertilization, setting the potential yield for LN and HN regimes at 20 q ha -1 and 35 q ha -1 , respectively (Makowski et al., 2005). LN plants did not receive N fertilizer, while the HN plants received a total input of 110 kg N ha -1 spread at two different times (12 February and 19 March 2010). LN and HN plants were harvested, respectively, on 9 and 12 April, at the beginning of the flowering period when half the plants of the plot had their first flowers open on the main stem (F1, or 60 on the BBCH scale), and 400 degree-days later (base 0) on 17 and 20 May at the beginning of the seed filling period (G2, or 71-73 on the BBCH scale).
Plants from 0.5 m 2 per plot (~20 plants) were harvested in the early morning and sampled during the subsequent hour. For each batch, plants were ranked according to their length and developmental stage; the six median plants were selected for sampling. On the main stem, the lowest leaf starting to yellow (Old) and the highest leaf at least 5 cm long (Young) were selected, the petiole and main vein were removed, and limbs were sampled. The stems above young and old leaf insertions were selected and sampled over 2 cm and 4 cm, respectively. All fresh samples were frozen immediately in liquid nitrogen and stored at -80 °C.

Nucleic acid manipulation
PCRs were conducted in a 20 μl mix containing 2-10 ng of DNA, 0.25 mM dNTPs (Promega), 0.5 μM of each primer (Eurogentec, Angers, France), and 0.5 U of Taq DNA polymerase (Promega) in the appropriate buffer supplemented with 2.5 mM MgCl 2 . The amplification program was run on a PTC-225 thermocycler (MJ Research, Waltham, MA, USA) with the following conditions: 35 cycles of denaturation at 94 °C for 30 s (3 min for the first cycle), annealing at 55-60 °C for 30 s, and elongation at 72 °C for 1-2 min (10 min for the last cycle).
Total RNAs were extracted with the SV Total RNA Isolation System (Promega) from 70 mg fresh weight (FW) of ground frozen tissue. First, samples were homogenized in the RNA lysis buffer (400 μl) using TissueLyserII from Qiagen. Then, all cell debris was eliminated by filtering the lysate through a 'Nucleospin 96 RNA filter Plate' (Macherey-Nagel). The manufacturer's protocol was then followed. In order to remove any remaining DNA traces, 1.5 μg of RNA was treated with DNase using the Turbo DNA-free kit (Ambion) according to the manufacturer's protocol. The quality of RNA was assessed by an electrophoresis on agarose gel (1.3%, w/v), and the absence of DNA contamination in samples was confirmed by PCR amplification. First-strand cDNA was synthesized using 2 μg of total RNA with oligo(dT) 12-18 primers and Superscript III reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Two independent reverse transcription reactions were performed as technical replicates. cDNA samples were diluted 26-fold with sterile water before use.

Gene expression analysis with qPCR
Quantitative PCRs (qPCRs) were set up with the LightCycler 480 SYBR Green I Master mix (Roche Diagnostics) and 4 μl of diluted cDNA in a final volume of 12 μl. The concentration of specific forward and reverse primers was set at 0.42 μM. qPCRs were run on a Light Cycler LC480 (Roche Diagnostics) under the following conditions: an initial step at 95 °C for 10 min, then 50 cycles of 95 °C for 10 s, 60 °C for 10 s, and 72 °C for 20 s.
Primer pairs were designed for short and specific amplification of individual members of the multigenic BnaGLN1 family and five reference genes (Supplementary Table S4 at JXB online). Identification of B. napus reference genes for reverse transcription-qPCR analysis was based on EST sequence similarity to A. thaliana genes with verified expression stability over a wide range of tissues and growing conditions. Arabidopsis thaliana genes were selected from among the list established by Czechowski et al. (2005). Their respective coding sequences were used to retrieve highly similar B. napus ESTs from GenBank using the BlastN algorithm. One B. napus EST per reference gene was selected to design qPCR primers (Supplementary Table S4).
For each run, single product amplification was confirmed by melt curve analysis. PCR products from each primer pair and genotype were sequenced in a preliminary analysis. The amplification efficiency was assessed for each genotype, with each primer pair using a dilution curve method over six orders of magnitude, on a pool of cDNAs from different tissues and modalities. Selected primer pairs have efficiencies >1.8.
The results reported were obtained from four biological replicates and two reverse transcriptions as technical replicates. All samples, including reverse transcription and biological replicates, were run at the same time for each primer pair. Raw fluorescence data were collected and analysed with the R package 'qpcR' (Ritz and Spiess, 2008). The 'pcrbatch' function was used to select sigmoid models for the fluorescence curves and then allowing the determination of the intrinsic amplification efficiency (sig.eff) and threshold cycle (sig.CpD2) at the second derivative maximum (Rutledge, 2004). For each run, cDNA relative quantity (RQ) was calculated using the efficiency mean value from the two technical and the four biological repetitions (mean sig.eff), and the run-specific threshold cycle as: RQ=1/(mean sig.eff) (sig.CpD2).
The most stable reference genes were selected using the GeNorm method from Vandesompele et al. (2002) available through the 'SLqPCR' R package (Dr Matthias Kohl SIRS-Lab GmbH). The four reference genes BnaX.PTB, BnaX.SAND, BnaX.PP2A, and BnaX.UBC21 from the five tested were retained with an average M value equal to 0.33. For each cDNA sample, a normalization factor (NF) was calculated as the geometrical mean of RQ from the four selected genes, and normalized RQ (NRQ) was then calculated as NRQ=RQ/NF. The mean of both technical replicates was then calculated for each sample.

Identification of GLN1 coding sequences in EST databases of Brassica napus and its progenitors Brassica oleracea and Brassica rapa
A total of 588 B. napus ESTs, 126 B. rapa ESTs, and 36 B. oleracea ESTs highly similar to one or several of the five AtGLN1.1-AtGLN1.5 mRNA-coding sequences from A. thaliana were isolated from public and private Genoplante databases (Supplementary Data File S1 at JXB online). Sequence assembly and alignments with AtGLN1 mRNA sequences revealed different groups of transcripts that allowed ESTs to be grouped and 16 individual contigs that might correspond to different BnaGLN1 genes of B. napus to be extracted (Supplementary Data File S2). Eight contigs for B. rapa and seven contigs for B. oleracea were also isolated (Supplementary Data Files S1, S2). The BnaGLN1, BraGLN1, and BolGLN1 contigs from B. napus, B. rapa, and B. oleracea, respectively, show high levels of sequence similarity with all the five AtGLN1 genes ( Table 1). The analysis of similarity levels and a phylogenetic tree ( Fig. 1) allowed homologous sequences for each AtGLN1 mRNA to be clearly identified in B. napus, B. rapa, and B. oleracea. The phylogenetic tree reveals that the GLN1 mRNA sequences are divided into two distinct groups for monocotyledonous (wheat, maize, and rice) and dicotyledonous species. The Brassicaceae sequences are divided into five clusters, each one including one A. thaliana and one or several B. napus, B. oleracea, and B. rapa sequences (Fig. 1). Each B. napus sequence is closely related to one sequence from either B. oleracea or B. rapa progenitors, illustrating the ancestral relationship with the A and C Brassica genomes. Each B. napus sequence is also closely related to one of the five A. thaliana AtGLN1 mRNA sequences, allowing the identification of homeologous related sequences between the four species. The contigs were then named according to the AtGLN1.
x (x from 1-5) gene with the highest sequence similarity. Since for each AtGLN1.x sequence several BnaGLN1 sequences were found, contig names were also extended by a copy number (Cn): BnaGLN1.x_Cn (Tables 1, 12; Fig. 1). Similarly, the names of the B. rapa and B. oleracea contigs follow the same rules (BraGLN1.x_Cn and BolGLN1.x_Cn, respectively; Table 2).
Most of the EST assemblies have been confirmed through sequencing partial or full-length cDNA clones when available (Supplementary Table S1 at JXB online). For the few EST assemblies that cannot be confirmed in this way, the cloning of missing coding sequences (CDS) was performed by designing primers from the B. rapa and B. oleracea homologous contig sequences (see the Materials and methods). This allowed the completion of the BnaGLN1.4_C3 and BnaGLN1.4_C4 sequences. Except for the four BnaGLN1.1_C1, BnaGLN1.1_C2, BnaGLN1.2_C1, and BnaGLN1.2_C2 contigs, no other BnaGLN1 sequences have ever been described previously in the literature or reported in databases as glutamine synthetase gene products. While the BnaGLN1.1_C1, BnaGLN1.1_ C2, BnaGLN1.2_C1, and BnaGLN1.2_C2, sequences have been identified as BnGSR2-1, BnGSR2-2, BnGSR1-1, and BnGSR1-2 mRNA, respectively (Table 2; Ochs et al., 1999), contig analysis allowed the completion of the 5′ end (untranslated region and coding sequence) of the BnGSR2-2 sequence that was previously missing (Supplementary Data File S3 at JXB online).

Genetic localization of the BnaGLN1 loci on the A or C Brassica genome using PCR and gene name annotation
Phylogenetic analyses showed a strong relationship between each BnaGLN1 gene and a gene from one or other of the progenitors B. rapa and B. oleracea, suggesting a common ancestral origin on the A or C Brassica genome, respectively. The phylogenetic tree also shows that each B. napus sequence, related to one sequence from either progenitor, is also related to another B. napus sequence, itself related to the other progenitor. The two B. napus homeologous genes, the B. rapa and the B. oleracea genes, are thus defining in this way a homeology group (a, b, or c). It was found therefore that each of the AtGLN1.1, AtGLN1.2, and AtGLN1.5 genes is related to one homeology group, while the AtGLN1.3 and AtGLN1.4 genes are related to three and two groups, respectively. It has to be noted that the b group related to AtGLN1.3 is incomplete as no BolGLN1.3 expressed sequence has been identified. Both the homeology groups and the Brassica genome were used to ascribe names to the BnaGLN1 genes; thus, the genes are named Bna[A or C genome]GLN1.x [a, b, or c homeology group] according to Ostergaard and King (2008). A similar notation was used for the BraGLN1 and BolGLN1 genes (Table 2).
In order to identify the A or C genome origin, the potential localization of the BnaGLN1 genes on linkage groups Table 2. Names of contigs of ESTs, mRNA, and AtGLN1 homologous genes in Brassica napus, B. oleracea, and B. rapa The names of genes encoding each contig were assigned according to the Ostergaard and King (2008) Supplementary Fig. S1 at JXB online).
For the other members of the BnaGLN1 gene family, an attempt to assign the BnaGLN1 genes to the A or C genomes using mapping populations was unsuccessful. Therefore, a panel of various Brassicaceae genotypes ( Supplementary Fig.  S2 at JXB online) was used in order to detect the BolGLN1 and BraGLN1 orthologous genes, using specific BnaGLN1 primers (Supplementary Table S1). The number of genotypes used for each Brassica species was adjusted in order to take into account the possible allelic variations and to detect the presence/absence of polymorphism. Furthermore, additional lines carrying the full A genome and one or several B. napus C chromosomes were used in order to determine preciselg the localization of the BnaC.GLN1 genes (Auger et al., 2009) (Supplementary Fig. S3).
The results are summarized in Table 3 and detailed in Supplementary Figs S1, S2, and S3 at JXB online. In brief, with the exception of BnaC. GLN1.4.b, all the identified BnaGLN1 genes were assigned to the A or C Brassica genome, confirming the two by two relationship of homology between them, which allowed them to be named according to their ancestral genome origin as recommended by Ostergaard and King (2008).

Identification of GLN1 genes of Brassica rapa and Brassica napus genomes
The recently sequenced and annotated B. rapa genome (BRAD; Cheng et al., 2011) was used to perform BLAST searches and sequence alignments using the BraGLN1 contig sequences identified here. Analysis revealed eight annotated BraGLN1 genes (Table 2). Alignments between BraGLN1 gene sequences and contigs revealed potential splicing variants. Indeed, the CDS deduced from the BraGLN1.3_C2 contig appeared incomplete at the 5′ end. The most highly similar Bra001686 annotated gene on the A03 chromosome also appeared incomplete when compared with the AtGLN1.3 CDS, as it is missing the first exon. The BLAST search on A03 chromosome v1.1 revealed the presence of a sequence highly similar to the AtGLN1.3 first exon, 4 kb upstream of the identified Bra001686 sequence (bp 17 854 801 to 17 854 849). According to the BRAD annotation, this inserted region has been described as an long terminal repeat (LTR) transposon of 3746 bp on the minus strand. The identified Table 3. Genetic mapping of BnaGLN1 genes Mapping was performed using specific primers for each contig and different mapping populations or genotypes for B. napus, B. oleracea, and B. rapa. Linkage groups used to assign each gene to A or C genomes are presented.

BnaGLN1
Mapping population  B. rapa EST (EX089134) that allowed identification of the 5′ region of the BraGLN1.3_C2 contig starts in the transposon region and continues into the first identified exon of Bra001686 which corresponds to the second AtGLN1.3 exon. The sequencing programme performed at URGV allowed identification of BnaGLN1 sequences in the B. napus Darmorbzh genome (SEQ-POLYNAP, ANR-09-GENM-021). The BraGLN1 protein sequences were deduced from the identified BraGLN1 genes using the BRAD tool, and used to search the database of protein sequences built from B. napus genomic sequence analysis (unpublished data). From the BnaGLN1 protein sequences identified, genomic sequences were recovered (Supplementary Data File S4 at JXB online). Interestingly, 16 putative BnaGLN1 genes and two putative BnaGSL genes (coding for the GS2 isoform) were found. The 16 BnaGLN1 genomic sequences (Supplementary Data File S4) were used to analyse similarities with the B. napus contigs and to create a phylogenetic tree ( Supplementary  Fig. S4). Interestingly, each genomic sequence was closely associated with one contig sequence, suggesting that all the genes with the 16 BnaGLN1 contigs had been found. The deduced mRNA sequences (Supplementary Data File S5), obtained using FGENESH software available on the SoftBerry website (http://linux1.softberry.com/berry. phtml?topic=fgenesh&group =programs&subgroup=gfind), showed very a high similarity with the contig sequences (Table 4) and allowed the gene structures to be deduced (Fig. 2). Except for BnaGLN1.3_C5 and BnaGLN1.3_C6, similarities between mRNA and associated contigs were near 100% (Table 4). Knowing that contig sequences and mRNA sequences are obtained from different B. napus genotypes (Supplementary Data File S1), this indicates that there is almost no polymorphism between the different BnaGLN1 coding sequences regarding the various genotypes of B. napus used for genome and EST sequencing. The BnaGLN1 genes contained between seven and 12 exons. GLN1.4 and GLN1.5 genes have the same number of exons in both B. napus and Arabidopsis (Table 4). For the other GLN1 genes, exon numbers are different between Arabidopsis and B. napus, but quite close; for example, AtGLN1.3 and BnaGLN1.3 contain fewer exons than other AtGLN1 and BnaGLN1 genes.

BnaGLN1 protein sequence conservation
Protein sequences of the BnaGLN1 family deduced from the coding sequences of contigs or from the deduced mRNA are similar. The BnaGLN1 proteins share between 93% and 96.6% identity with the AtGLN1 proteins encoded by their respective orthologous genes (Table 1). Within each homeology group, the A and C BnaGLN1 proteins share 98.3-100% identity.
In all BnaGLN1 protein sequences, two conserved pfam domains specific to glutamine synthetase enzymes (pfam 03951 and pfam 00120) were identified (Fig. 3). The residues involved in the ammonium/glutamate-binding pocket (Eisenberg et al., 2000) are also strictly conserved. In contrast, the polar amino acids Q49 and S174, shown to be involved in the ammonium high affinity properties of AtGLN1.1 and AtGLN1.4 (Ishiyama et al., 2006), are not strictly conserved in all the BnaGLN1.1 and BnaGLN1.4 proteins. The polar Q49 was converted into an acidic glutamate E49 in all the BnaGLN1.1 and BnaGLN1.4 sequences, and the S174 is conserved only in the two BnaGLN1.4.b sequences but was converted into an A174 in the BnaGLN1.4.a and BnaGLN1.1.a sequences. Depending on the effect of such amino acid modifications, it might be possible that ammonium affinity properties have not been conserved within the BnaGLN1.1 and BnaGLN1.4 protein families. In contrast, the residues K49 and A174 present in the low affinity enzymes AtGLN1.2 and AtGLN1.3 are conserved in all the BnaGLN1.2 and BnaGLN1.3 protein sequences, suggesting the conservation of the low ammonium affinity properties in those two protein families (Fig. 3).

Expression of BnaGLN1 genes is modified depending on the nitrogen regime and leaf senescence
A first analysis of EST distribution between libraries and BnaGLN1 contigs led to the conclusion that BnaGLN1 genes are probably differentially expressed according to tissue and developmental stage (Supplementary Data File S1 at JXB online). The BnaGLN1 gene expression was monitored at the vegetative stage measuring transcript levels by quantitative realtime RT-PCR in samples of taproot, crown, limbs, and veins of plants grown under low or high nitrate conditions. Plants grown under low or high nitrate conditions grew 13 and 17 leaves, respectively. F v /F m and SPAD measurements on all the leaf ranks (numbered from the bottom leaf to the top leaf) were done to estimate the relative leaf senescence status of each leaf. From both SPAD and F v /F m as senescence markers, six leaves were selected from each nitrate condition presenting differential senescence levels to perform further experiments (Fig. 4). Leaves of rank 3, 5, 6, 7, 9, and 11 were harvested on plants grown under low nitrate conditions. Leaves 3, 5, 6, 9, 12, and 15 were harvested on plants grown under high nitrate conditions. To simplify the presentation of further results, the collected leaf ranks were renamed 1, 2, 3, 4, 5, and 6, respectively, with 1 designating the bottom-most and oldest collected leaf and 6 the youngest collected leaf. Leaves dissected as limbs, and primary and secondary veins were used to measure BnaGLN1.1 gene expression levels in the different Fig. 3. Alignment of Brassica napus and Arabidopsis thaliana GS1 proteins. Protein sequences were deduced from DNA coding sequences and aligned using Clustal. Stars indicate residues involved in the ammonium/glutamate-binding pocket (Eisenberg et al., 2000). Boxes indicate residues involved in ammonium affinity properties (Ishiyama et al., 2006). Arrows indicate conserved domains (1) pfam 03951 Gln-synt_N glutamine synthetase bet-Gasp domain; and (2) pfam 00120 gln-synt_C catalytic domain. Residues are coloured according to their polarity properties (neutral non-polar as black, neutral polar as green, acidic as red, and basic as blue).
tissues. In addition to BnaGLN1.1 expression, the expression of BnaGSL1 and BnaGSL2 encoding the chloroplastic GS2 isoenzymes was also monitored and used as a control for leaf senescence as it is known that genes encoding GS2 izoenzymes are down-regulated with leaf ageing in all the plant species studied so far . BnaGSL1 and BnaGSL2 expression levels confirmed the differential senescence phenotype of the chosen leaf ranks. Leaves 1, 2, and 3 can be considered as senescing leaves, 4 and 5 as mature leaves, and 6 as a young leaf according to Masclaux et al. (2000) (Fig. 4E, F).
Genes that are preferentially expressed under high or low nitrate conditions were identified. The results showed that regarding the N regime, all the members of the same gene family respond similarly, with a few exceptions from the BnaGLN1.3 family. It was observed that all the members of the BnaGLN1.1 and BnaGLN1.4 gene families were significantly induced under low nitrate conditions compared with high nitrate. This was observed in limbs, secondary veins ( Fig. 5A-F), and also in primary veins for some BnaGLN1.4 genes ( Supplementary Fig.  S5E-H at JXB online). Induction under low nitrate conditions was also clearly observed in the taproot and crown (Fig. 6A, B; E-H). In contrast, the two BnaGLN1.2 genes were significantly more expressed under high nitrate conditions in leaf limbs and veins but not in the taproot and crown (Fig. 5M, N; C, D). Finally no difference was observed in the expression of the BnaGLN1.3 and BnaGLN1.5 families in leaf limbs or veins between the high nitrate and low nitrate conditions (Fig. 5G-L). Surprisingly, all the BnaGLN1.3 genes are significantly induced under low nitrate conditions in the taproot but not in crown tissue (Fig. 6I-M). Therefore, N-dependent regulation might be different in the root and shoot.
BnaGLN1 genes also appeared to be differentially expressed depending on leaf ageing and senescence. The two BnaGLN1.1 genes and the four BnaGLN1.4 genes were significantly induced in leaf limbs and veins with ageing and during senescence independently of nitrate conditions (Fig. 5A-F; Supplementary Fig, S5B, E-H at JXB online). Cumulated expression of the BnaA.GLN1.5.a/BnaC.GLN1.5.a genes (Fig. 5L) also increased with leaf ageing and senescence in limbs and veins of plants grown under low and high nitrate conditions. In contrast, senescence triggers an opposite effect on the mRNA level of the two BnaGLN1.2 genes, especially under low nitrate nutrition (Fig. 5M, N). The effect of senescence was less evident under high nitrate conditions in limbs and veins due to the already high BnaGNL1.2 expression in mature leaves. In these leaves, expression profiles are biphasic, increasing from young to mature leaves then decreasing in senescing leaves. Profiles are more complex in the BnaGLN1.3 family since the expression of BnaC. GLN1.3.a,BnaA. GLN1.3.b, is repressed with leaf ageing in limbs and veins, while the expression of BnaA. GLN1.3.a (Fig. 5G) and cumulated BnaA.GLN1.3.c/BnaC. GLN1.3.c (Fig. 5K) is increased with ageing in limbs.
These results show that within all the BnaGLN1 families except BnaGLN1.3, all members show similar expression levels. The four BnaGLN1.4 genes are the most highly expressed in all the tissues studied. BnaGLN1.4 gene expression is four times higher than that of the BnaGLN1.1 genes and 20 times higher than that of BnaGLN1.2. The expression level of all the BnaGLN1.3 and BnaGLN1.5 genes is much lower, except that of the cumulated BnaGLN1.3c genes that reach a similar level to BnaGLN1.2. Table 5 summarizes the N and senescence effects observed on the BnaGLN1 expression levels.
BnaGLN1 genes are differentially expressed at the reproductive stage depending on plant organs or leaf ageing In order to monitor BnaGLN1 gene expression at the reproductive stage, plants were grown in field conditions under low or high N regimes. Two leaf ranks (young top leaf and old bottom leaf) and the two corresponding stem sections (also referred to hereafter as young and old) were collected at flowering and during grain filling.
Globally, effects of N limitation on BnaGLN1 expression were similar to those found at the vegetative stage, except that the magnitude of gene repression or induction was lower than that observed at the vegetative stage (Supplementary Table S5 at JXB online). Figure 7 reports the effect of senescence on the expression of the BnaGLN1 genes in leaves and stems of plants grown under a sufficient N regime. As a control of leaf senescence stages, the BnaGSL1 and BnaGSL2 genes are significantly more highly expressed in the young tissues than in older tissues (Fig. 7N, O). There was a sharp decrease in BnaGSL gene expression at the flowering stage, while at the seed filling stage the magnitude of BnaGSL repression was much lower but still significant.
As observed at the vegetative stage, the BnaGLN1.1 genes were up regulated with leaf and stem senescence, but this was only observed at the flowering stage (Fig. 7A, B). During seed filling, expression in leaves and stems was higher than during flowering, showing an effect of plant ageing. However, no difference was observed between the young and old leaves, suggesting that both types of leaves had become senescent between flowering and seed filling. The two BnaGLN1.1 genes were expressed more highly in leaf blades than in stems at both the flowering and seed filling stages. Similarly the BnaGLN1.2 genes were more expressed in leaf blades than in stems (Fig. 7C, D). The effect of senescence on BnaGLN1.2 genes was opposite to the effect observed on BnaGLN1.1 genes. BnaGLN1.2 expression decreased 2-to 3-fold in old leaf blades and old stems compared with young leaf blades and young stems, respectively. The biphasic profile obtained for BnaA.GLN1.2 at the vegetative stage was also observed at the flowering stage (data not shown). As observed with the BnaGLN1.1 genes, the effect of senescence was no more significant at seed filling.
Among the four BnaGLN1.4 genes, only BnaA.GLN1.4.a and BnaC.GLN1.4.a shared similar expression profiles ( Fig. 7E-H). They are preferentially expressed in leaf blades rather than in stems. In contrast to the vegetative stage, BnaA.GLN1.4.a and BnaC.GLN1.4.a tend to be repressed by senescence in leaf blades but induced by senescence in stems. This trend was especially significant at the flowering stage. In contrast, BnaA.GLN1.4.b was induced by senescence in leaf blades and stems at the flowering stage but repressed during seed filling (Fig. 7G). Finally, BnaC.GLN1.4.b expression was higher in leaf blades than in stems and was repressed by senescence at the flowering stage, similarly to the two BnaA. GLN1.4.a and BnaC.GLN1.4.a genes (Fig. 7H). In contrast to the vegetative stage, the members of the BnaGLN1.4 family have developed specificities and are differentially expressed at the flowering and seed filling stages. It is likely that they have different roles and influences on N metabolism after flowering.
Among the BnaGLN1.3 members, similar profiles were observed for BnaA. GLN1.3.a, BnaA.GLN1.3.b, and  BnaC.GLN1.3.b (Fig. 7I, K, L). These three genes are Fig. 5. Expression of BnaGLN1 genes is modified depending on nitrate availability and leaf ageing. The relative expression level of BnaGLN1 genes was monitored in limbs and secondary veins of six leaf ranks harvested on vegetative plants grown under low (white bars) or high (black bars) nitrate conditions. Leaf ranks represented as number 1 (bottom and older leaf) to 6 (top and younger leaf) showed differential senescence symptoms. Mean and standard deviation of four plant repeats are shown.
down-regulated in old leaves and stems compared with young leaves at the flowering stage. However, their expression increased sharply in old limbs at the seed filling stage. The other BnaC.GLN1.3.a and BnaA.GLN1.3.c/ BnaC.GLN1.3.c expression profiles did not show any modification associated with leaf or stem senescence (Fig. 7J, M). All the BnaGLN1.3 genes appeared to be more highly expressed in stems than in leaves, especially at the flowering stage.
It was not possible to measure BnaA.GLN1.5.c/BnaC. GLN1.5.c gene expression, possibly due to the very low expression level in vegetative tissues that cannot be accurately measured in field-grown plants.
Results obtained at the flowering and seed filling stages confirm results from the vegetative stage. BnaGLN1 genes are generally similarly regulated according to their orthology group, although exceptions were observed particularly at the seed filling stage, such as with BnaA.GLN1.4.b and BnaC. GLN1.3.a (Fig. 7G, J).

Discussion
Glutamine synthetase is a key enzyme of N metabolism involved in ammonium assimilation and remobilization. Recent studies highlight the important role of GS1 cytosolic isoenzymes for N management linked to yield establishment and seed filling in monocotyledonous crops (Tabuchi et al., 2005;Martin et al., 2006;Bernard and Habash, 2009;Swarbreck et al., 2011). The GS1-coding genes are therefore good candidates for improving yield traits and grain quality . The complexity of studying glutamine synthetases arises from the fact that two isoenzymes exist, one in the chloroplast and the other in the cytosol, and that several isoforms exist for the cytosolic enzyme. The numerous isoforms are encoded by a multigenic family, and the five GLN1 genes in A. thaliana are likely to present different roles depending on plant organs and nitrate availability in the soil (Lothier et al., 2011). Similarly the five maize GLN1 genes do not participate equally in N management at the whole-plant level (Martin et al., 2006;Hirel et al., 2007). The aim of this study was to identify the whole BnaGLN1 gene family and to characterize the expression of the different genes depending on nitrate availability as well as depending on ageing and leaf and stem tissue senescence.
Using the sequences obtained from EST libraries and genome sequencing, a total of 16 genes belonging to the BnaGLN1 family, eight genes from each of the A and C genomes, were found. In accordance with the history of B. napus genome formation (Nagaharu, 1935), it was found that each BnaGLN1 gene is closely related to a BraA.GLN1 or BolC.GLN1 gene depending on its A or C genome location. Therefore, it can be stated with confidence that all the GLN1.1 genes of B. napus have been described in this report. Sequence analyses also showed that B. napus coding sequences are highly conserved between the A and C genomes and also between B. napus genotypes. The level of sequence divergence observed in the BnaGLN1 family is between 0.9% and 2.9% SNPs (single nucleotide polymorphisms) in CDS, which is less than the preliminary observation showing sequence divergences of ~3-5% SNPs in CDS from SLR1 (Inaba and Nishio, 2002).
It is well known that B. napus shows a high degree of collinearity to its diploid progenitors B. rapa and B. oleracea (Rana et al., 2004). Many studies have investigated the segmental structure of the Brassica genomes and led to the identification and genetic mapping of syntenics blocks between A. thaliana and the Brassica genomes (Parkin et al., 2005;Schranz et al., 2006;. The number of potential BnaGLN1 genes and their localization on linkage groups can then be predicted depending on the number of times the blocks are replicated and on their localization on each B. napus linkage group. According to the whole-genome triplication event that occurred in Brassicaceae genome species after divergence from Arabidopsis (Lysak et al., 2005;Parkin et al., 2005), and the recent hybridization between B. rapa and B. oleracea leading to the appearance of B. napus, each AtGLN1 gene could have been found in triplicate in each A and C genome from B. napus to form three pairs of homeologous genes. From EST and genome sequence analyses, it is revealed here that only one homeology group exists for AtGLN1.1, AtGLN1.2, and AtGLN1.5 and two groups for AtGLN1.4. AtGLN1.3 is the only GLN1 gene for which the six BnaGLN1 orthologous genes were retained in the B. napus genome. This illustrates the massive gene loss that occurred in the Brassica lineage after the whole-genome triplication event. Indeed, using A. thaliana as an outgroup, Town et al. (2006) found that 35% of genes inferred to be present when genome triplication occurred in the Brassica lineage have been lost in B. oleracea. Similarly, whole-genome analysis of B. rapa revealed a high rate of gene loss, from 30% to 64% depending on the degree of fractionation of the region considered (X. . BnaGLN1 families are a good example of this. With the exception of BnaGLN1.4.b genes, the present study points out that pairs of BnaGLN1 homeologous genes share very similar transcription profiles. Furthermore, within one orthology group, when several groups of homeologues were retained, paralogous genes conserved similar expression profiles (BnaGLN1.3 and BnaGLN1.4). This suggests that coding but also regulatory sequences were essentially conserved after genome merging of B. napus progenitors, but also after the whole-genome duplication (WGD) and diploidization events that occurred in the Brassica lineage after the divergence from the Arabidopsis genus. WGD is generally thought to provide raw material for gene neo-and subfunctionalization, extending resilience to deleterious mutations, increasing the net speciation rate and species richness (Soltis et al., 2009), as well as providing the adaptive advantage for colonizing harsh and unstable environments (Franzke et al., 2011). On the other hand, maintenance of redundancy can confer robustness against mutations (De Smet and Van de Peer, 2012) and/or a selective advantage in increasing the abundance of encoded proteins (Bekaert et al., 2011). As GS1 is an essential enzyme of primary N metabolism, linked to central carbon metabolism via the GS/GOGAT cycle that might also play a role in the adaptation of plant to nutrient deficiency and pathogen attack (Brauc et al., 2011;Seifi et al., 2013), maintenance of multicopy of GLN1 could confer robustness against mutations. Nevertheless, GLN1 expression profiles have not been exhaustively investigated and there might be particular environmental or developmental conditions allowing the differentiation of expression profiles between homeologous and/or paralogous genes. Partially overlapping profiles could provide robustness against mutations, but also adaptive advantages for colonizing harsh and unstable environments.
Allopolyploidization involves the merger of two different, and often divergent, genomes whose reconciliation in a common nucleus often leads to myriad changes, including unequal expression of the two merging genomes. Biased expression among homeologues has been found in cotton and wheat (Pumphrey et al., 2009;Rapp et al., 2009). Previous studies have suggested bias toward the B. rapa A genome in the transcriptional expression of rRNA genes (Chen and Pikaard, 1997). In the present study it is shown that most of the BnaGLN1 homeologous pairs display similar expression levels in the various tissues studied. Differences in mRNA contents observed between homeologous pairs of A and C genome origins were generally very small. No systematic bias towards the B. oleracea C parental genome or B. rapa A genome can be identified in this study. Bias towards the B. oleracea C parental genome was identified for BnaGSL (at flowering and seed filling stages), BnaGLN1.3.a (especially at the vegetative stage), and BnaGLN1.4.a homeologues, and in favour of the B. rapa parental genome for BnaGLN1.1.a and BnaGLN1.2.a. These results are in agreement with the recent finding that nearly 7% of the potentially identified homeologous genes expressed in a leaf extract are displaying a differential expression level in favour of the A or C parental genome for 1/3 and 2/3 of the pairs, respectively, and that genes involved in metabolic processees tend to be over-represented (Higgins et al., 2012).
Beside their intrinsic expression levels, it was found that the BnaGLN1 genes are similarly regulated depending on their orthology group and that they are differentially regulated between groups. Overall it was found that specificities of expression are conserved between BnaGLN1 genes and their respective AtGLN1 orthologues, raising the hypothesis of conserved physiological functions.
In Arabidopsis, several studies have shown that AtGLN1.1 is highly induced in leaves during senescence (Guo et al., 2004) and is up-regulated when exogenous N sources are limiting (Lothier et al., 2011). Up-regulation of AtGLN1.1 under nitrate starvation is in good agreement with the kinetic properties described by Ishiyama et al. (2004) that suggested that the high affinity of AtGLN1.1 for ammonium is correlated with a role for the enzyme under low N conditions. The induction of the expression of the two BnaGLN1.1 genes in older stems and leaves is conserved, as already shown by Buchanan-Wollaston and Ainsworth (1997) and Ochs et al. (1999). It was also found that the two BnaGLN1.1 genes are overexpressed under low nitrate conditions especially at the vegetative stage. As the amino acid residues known to be involved in the high affinity of AtGLN1 towards ammonium are not conserved in any of the BnaGLN1 proteins, no information about the potential kinetic properties can be extrapolated from the protein sequence. Regarding expression profiles, it is at least suspected that BnaGLN1.1 and AtGLN1.1 proteins might have similar roles In a previous study, it was found that AtGLN1.2 is slightly induced by leaf ageing and that the expression profile is biphasic, with an increase from young to mature leaves and then a decrease in strongly senescing leaves (Diaz et al., 2008;Lothier et al., 2011). In addition, it was found that AtGLN1.2 is mainly expressed in roots and leaves under a high N regime (Guo et al., 2004;Ishiyama et al., 2004;Lothier et al., 2011). A detailed functional analysis led to the conclusion that AtGLN1.2 was involved in primary ammonium assimilation under high N regimes (Ishiyama et al., 2004;Lothier et al., 2011). BnaGLN1.2 genes (also named BnGSR2.1 and BnGSR2.2) are also more highly expressed in roots than in shoots (Ochs et al., 1999). It is shown here that the two BnaGLN1.2 genes are more expressed in young leaves than in old leaves and are overexpressed under high N regimes, suggesting a similar role to AtGLN1.2 in primary ammonium assimilation. Interestingly, a BraA.GLN1.2 gene, named BcGS1 (Sun et al., 2010), was found also to be expressed in root and induced under high N regimes, suggesting the conservation of the regulation in the Brassiceae tribe.
Similarly to AtGLN1.3, the BnaGLN1.3 genes are not induced in older leaves at the vegetative stage and do not respond to differential N regimes (Guo et al., 2004;Lothier et al., 2011). The hypothesis about the physiological roles of AtGLN1.3 in N export via phloem tissues in roots is supported by the high capacity of the enzyme for glutamine synthesis and by the location of AtGLN1.3 expression in the root vasculature (Ishiyama et al., 2004). It has to be noted that BnaGLN1.3 genes are the only BnaGLN1 genes preferentially expressed in stem tissues compared with leaf blades. As stems are richer in vascular tissues than leaf blades, the higher expression of BnaGLN1.3 in stems might be related to a potential vascular localization that remains to be explored.
AtGLN1.4 is one of the markers used for leaf senescence (Guo et al., 2004;Wingler et al., 2009). AtGLN1.4 is induced by N limitation or starvation in both the root and shoot (Ishiyama et al., 2004;Lothier et al., 2011). AtGLN1.4 protein exhibits high affinity towards ammonium and is expressed in the pericycle cells of roots (Ishiyama et al., 2004). Evidence for BnaGLN1.4a root expression was found in EST libraries. However, induction of gene expression in leaves in response to ageing and low N regime is not well conserved among the four BnaGLN1.4 orthologues in regards to developmental stages. The BnaGLN1.4 genes are significantly overexpressed in senescing leaves and under low nitrate conditions at the vegetative stage. However, this trend is not conserved at the flowering and seed filling stages. BnaA. GLN1.4.a,BnaC.GLN1.4.a,and BnaC.GLN1.4.b are clearly and significantly less highly expressed in old than in young leaves at these stages. Furthermore, as the residues conferring the high affinity property are partially conserved in BnaGLN1.4 proteins, it might be suspected that BnaGLN1.4 genes have a role when N resources are low. Surprisingly, in contrast to Arabidopsis in which AtGLN1.4 expression is one of the lowest, BnaGLN1.4 expression levels are the highest found among all the BnaGLN1 genes.
In Arabidopsis, AtGLN1.5 expression is known as the lowest of the AtGLN1 gene family. Expression was mainly found in seeds (Lothier et al., 2011), and very little is known about AtGLN1.5. BnaGLN1.5 gene expression is also very low, and it was not possible to measure it in leaf and stem tissues at flowering and vegetative stages. Similarly to AtGLN1.5, BnaGLN1.5 ESTs were found in reproductive tissues and seeds, thus suggesting specific roles during seed maturation.

Conclusion
A total of 16 B. napus GLN1 genes were identified, among which 12 have never been described. The total number of BnaGLN1 genes, their phylogenetic relationships, and genetic location are in agreement with the evolutionary history of Brassica species. Some specificities of expression seemed to be conserved among the Brassiceae tribe and especially between A. thaliana and B. napus. Regulations arising from plants interactions with their environment (such as N resources), final architecture, and therefore sink-source relations in planta, seem to be globally conserved when compared with data available from the Arabidopsis model. Considering the architectural, size, and lifespan differences between A. thaliana and B. napus, it is not surprising to find some differences in gene expression profiles. Also, due to the higher number of GLN1 genes conserved in the B. napus genome, it seems correct to find some specificities in the expression of each BnaGLN1 in contrast to genes involved in flavonoid biosynthesis that display highly conserved expression profiles between A. thaliana and B. napus during seed development and are highly dependent on tissue differentiation (Auger et al., 2009). A more detailed localization of BnaGLN1 gene expression would refine the hypothesis concerning their physiological role. Indeed, the present expression study relied on leaf blade and stem samples consisting of different tissues with contrasting physiological roles, in particular parenchyma and vascular tissues. Such a localization study could be advantageously performed in Arabidopsis and B. napus to provide a new basis for comparison of the evolution of this gene family.

Supplementary data
Supplementary data are available at JXB online Figure S1. BnaGLN1 gene localization on the A or C genome using a panel of various Brassica genotypes. Figure S2. BnaGLN1 gene localization on LGs using mono-and polysomic additional lines. Figure S3. Positions of BnaGLN1 genes on the B. napus genetic map. Figure S4. Alignments of the genomic sequence, deduced mRNA sequence, and contig of EST sequences are reported for each BnaGLN1 gene. Figure S5. Expression of BnaGLN1 genes is modified depending on nitrate availability and ageing in primary veins of vegetative B. napus plants. Table S1. Clones from Genoplante and ADIS-MPIZ oilseed rape cDNA libraries used for BnaGLN1 mRNA sequencing. Table S2. Primers used for cloning and genetic mapping of the BnaGLN1 gene. Table S3. Specific primers used for sequencing BnaGLN1 cDNAs. Table S4. qPCR primers used for BnaGLN1 gene expression analysis. Table S5. Induction of BnaGLN1 gene expression under low N fertilization in field-grown plants.
Data File S1. List and description of EST sequences belonging to each B. napus, B. oleracea. and B. rapa GLN1 contig, and tables of the distribution of ESTs between libraries according to their BnaGLN1 contigs.
Data File S2. BnaGLN1 contig sequences in fasta format. Data File S3. Global multiple alignment of nucleotide sequences in the study.
Data File S4. Genomic sequences of the BnaGLN1 genes. Data File S5. Deduced BnaGLN1 mRNA sequences.