Nitrogen is a crucial nutrient for plant growth and development. Arginine is considered to be an important amino acid for nitrogen transport and storage, playing a crucial role during plant seedling development. However, little is known about the role of arginine in nitrogen remobilization at the reproductive stage. We isolated a rice mutant nglf-1 with reduced plant height, small panicle and grain size, and low seed-setting rate (10% in nglf-1 compared to 93% in wild-type). Map-based cloning revealed that the mutant phenotype was caused by loss of function of a gene (OsARG) encoding an arginine hydrolysis enzyme, which is consistent with arginine accumulation in the mutant. The phenotype was partially corrected supplying exogenous nitrogen, and fully corrected by expression of a wild-type OsARG transgene. Over-expression of OsARG in rice (cv. Kitaake) increased grain number per plant under nitrogen-limited conditions. OsARG was ubiquitously expressed in various organs, but most strongly in developing panicles. The OsARG protein was localized in the mitochondria, consistent with other arginases. Our results suggest that the arginase encoded by OsARG, a key enzyme in Arg catabolism, plays a critical role during panicle development, especially under conditions of insufficient exogenous nitrogen. OsARG is a potential target for crop improvement.
Nitrogen is one of the most expensive nutrients to supply, and commercial fertilizers represent a major cost in plant production. Reducing fertilizer input and breeding plants with better nitrogen use efficiency is one of the main goals of research on plant breeding. The use of nitrogen by plants involves several steps, including uptake, assimilation, and remobilization during plant aging. In Arabidopsis and oilseed rape, nitrogen may be remobilized via the phloem from senescing leaves to expanding leaves at the vegetative stage, as well as from senescing leaves to seeds at the reproductive stage, in the form of nitrogen-rich amino acids (Malagoli et al., 2005; Hirel et al., 2007; Diaz et al., 2008; Lemaître et al., 2008). Diaz et al. (2005) found that individual amino acids, including γ-aminobutyrate, leucine, isoleucine, tyrosine and arginine (Arg), accumulate during leaf aging, particularly in the leaves of early-senescing lines, although there was a steady increase in nitrogen-rich Arg in all genotypes investigated. Until now, little information has been available regarding the role of Arg produced during the process of remobilization to generative organs, especially the developing rice panicle.
Arginine is an important amino acid for the transport and storage of nitrogen in plants (Van Etten et al., 1963). As one of the most functionally diverse amino acids in living cells, Arg is a precursor for the biosynthesis of polyamines, agmatine and proline, as well as the cell-signaling molecules glutamate, γ-aminobutyric acid and nitric oxide (NO) (Morris, 2002). Arginine decarboxylase, arginase and nitric oxide synthase (NOS) are key enzymes in Arg metabolism, and their relative activities control the metabolic fates of Arg (González de Mejía et al., 2003; Hao et al., 2005; Jubault et al., 2008). Plants utilize Arg for the production of polyamines (Zonia et al., 1995; Chen et al., 2004) and possibly NO, although the latter assertion is still not confirmed (Guo et al., 2003; Corpas et al., 2006; Crawford, 2006; Moreau et al., 2008). NO and polyamines are involved in plant defense, adaptation and developmental pathways (Chen et al., 2004). Arginase-negative mutants of Arabidopsis provide some evidence for Arg-derived NO. Mutation of either ARGAH1 or ARGAH2 results in increased formation of lateral and adventitious seedling roots, and Arg supplementation resulted in increased NO accumulation (Flores et al., 2008).
Arginase also hydrolyzes Arg to urea and ornithine (Orn), and transfers Arg nitrogen into urea, which is recycled by urease-catalyzed hydrolysis to ammonia (Jenkinson et al., 1996). The coordinated action of arginase and urease is thought to recycle urea nitrogen to meet the metabolic demands of developing seedlings (Zonia et al., 1995). Nitrogen mobilization during seedling development correlates with large increases in arginase gene expression as reported in several species, including soybean, broad bean, pumpkin and loblolly pine (Polacco and Holland, 1993; Todd and Gifford, 2002). To date, the physiological role of Arg catabolism in cereal crops has not been fully clarified. In the case of rice, seedling arginase activity increases, especially at the onset of root and 2 day shoot emergence during germination, corresponding to the increased turnover of Arg, although no further evidence of urea accumulation has been reported (Cao et al., 2010). In this paper, we characterize a rice mutant that is defective in arginase, and highlight the importance of Arg-derived nitrogen in rice grain development and yield formation.
Aberrant panicle phenotype in the nglf-1 mutant
The rice narrow-grain and low-fertility mutant nglf-1 was derived from an anther culture of autotetraploid indica/japonica hybrid H3774 (H2088 × H891) (Cheng et al., 2005; Qin et al., 2005). During the vegetative growth stage, the mutant and its wild-type (WT) were indistinguishable. At the ripening stage, the apparent characteristics of the mutant included a small and erect panicle accompanied by low fertility (10% in nglf-1 compared to 93% in WT) and lower kernel weight under growth conditions in Beijing in summer (39° 54′ N, temperate climate) or Sanya, Hainan Province in winter (18° 16′N, tropical climate). The abnormal phenotype was more severe in plants without exogenous nitrogen application, and included lack of seed set on the panicle. Compared to WT, the mutant displayed reduced plant height but an increase in the number of small, non-seed-setting tillers (Figure 1 and Table 1), which resulted in dry weight differences in stem, leaf sheath and panicle. However, no significant differences in total dry weight were found between the mutant and WT, which was further described in the following.
Table 1. Statistical analysis of agronomic traits
Values are means ± SE. Asterisks indicate statistically significant differences compared with wild-type.
To map the gene, we crossed nglf-1 with IRAT129, a japonica variety, to construct a genetically segregating population. Within the F2 population, nglf-1 exhibited normal Mendelian segregation (1632 normal:560 nglf-1; χ2 = 0.322; P > 0.05), indicating the involvement of a single gene mutation. Bulked segregant analysis with SSR markers covering the whole rice genome found that the RM335 marker on the short arm of chromosome 4 was linked to the nglf-1 locus, which was subsequently delimited using SSR primers AL36-1 and AL1-1, and further mapped to a 72 kb interval, delimited by the two markers In10 and AL1-1, using 510 F2 mutant plants (Figure 2a). Based on the Rice Genome Annotation Project (http://rice.plantbiology.msu.edu/), the mapped region contains 13 recognizable open reading frames (ORFs). Sequence comparison revealed a single nucleotide substitution resulting in a stop codon (TGA) in ORF8 of nglf-1. ORF8, gene LOC_Os04g01590, encodes a protein homologous to arginase and consists of six exons, with a total size of 1382 bp, including a 1023 bp coding region, a 126 bp 5′ UTR, and a 233 bp 3′ UTR. Thus the substitution occurred in the arginase domain of nglf-1, resulting in premature termination of the protein (Figure 2b), and thus ORF8 was considered a candidate for putative gene Nglf.
To verify that the mutation in LOC_Os04g01590 is responsible for the nglf-1 phenotype, we introduced a genomic DNA fragment of 5233 bp containing the entire LOC_Os04g01590 coding region, 937 bp upstream promoter sequence and 426 bp downstream sequence cloned from WT, into nglf-1. Of 26 regenerated plants (T0), 20 plants were confirmed as positive by PCR, exhibiting the same phenotype as WT. We grew five plants to the T1 generation, and found that all PCR-positive plants performed similarly to WT (Figure2c–e). We also searched the publicly available rice T-DNA database (http://signal.salk.edu/cgi-bin/RiceGE/) and found a line in which Nglf was tagged by a T-DNA (3A-10962). This allelic mutant had a panicle-defective phenotype similar to nglf-1, and was named nglf-2 (Figure S1a–c). Further analysis indicated that the insertion was in the first exon of Nglf (Figure S1 d,e), and that such an insertion completely disrupted transcription of Nglf, as we were unable to detect any transcripts of Nglf in nglf-2 by RT-PCR (Figure S1f). Together, these results confirm that LOC_Os04g01590 is Nglf, and it was renamed OsARG.
Plant arginases are encoded by a highly conserved gene
In the rice genome, OsARG is the only gene encoding arginase. A BLAST search with translated OsARG revealed that most plant arginases are encoded by one or two genes. OsARG has high identity with proteins from other organisms, including Arabidopsis thaliana (Flores et al., 2008), Solanum lycopersicum (Chen et al., 2004), Pinus taeda (Todd et al., 2001) and Glycine max (Goldraij and Polacco, 1999) (Figure S2). OsARG also has high identity with arginases from five genera of monocotyledonous plants, including the arginases in Zea mays, Sorghum bicolor, Brachypodium distachyon, Triticum aestivum and Hordeum vulgare, all of which are members of the same clade, distinguishing them from all dicotyledonous species analyzed. The shared identity may indicate a different role in monocotyledonous plants than in dicotyledonous species (Figure S3).
The OsARG protein comprises 340 residues with a predicted molecular mass of 36.96 kDa and pI of 5.90. Sequence alignment indicated a conserved arginase domain, and all plant arginases contain two His and four Asp residues that bind the Mn2+ co-factor (Chen et al., 2004). A predicted mitochondrial targeting peptide is located at the N-terminal end, and this region showed the greatest diversity among various plant arginases (Figure S2).
Expression and intracellular localization of OsARG
We examined the pattern of OsARG expression in various organs using quantitative real-time PCR. OsARG was expressed ubiquitously in all organs, including the root, stem, leaf blade, leaf sheath and panicle. Compared to WT, the expression in nglf-1 was reduced (Figure 3a). For a detailed evaluation of OsARG expression, we generated transgenic plants expressing the GUS reporter gene driven by the native promoter of OsARG. The GUS signal was found in the root, stem, leaf and panicle, with the strongest signal in the spikelets, consistent with the RT-PCR results (Figure 3c).
According to the prediction using online software SignalP 3.0 (http://www.cbs.dtu.dk/services/), a mitochondrial signal peptide is present at the 5′ end of the OsARG protein. The intracellular localization of OsARG was tested by expressing an OsARG-GFP fusion construct in rice (Oryza sativa L., var. Nipponbare). The GFP signal was visualized within the protoplast cells of the transgenic plant, and co-localized with the mitochondria-specific stain MitoTracker Red. The co-localization supports the prediction that OsARG is a mitochondrial protein (Figure 3d).
Arginine accumulated in the panicle due to abolished arginase activity in nglf-1
By detecting the amount of urea produced by adding exogenous Arg to protein extracted from various plant organs, we found that the OsARG activity in WT correlated with expression of its encoding gene, i.e. higher gene expression and protein activity were found in the panicle. This high OsARG activity was sustained until 28 days after anthesis. In contrast to WT, arginase activity in nglf-1 is very low in all organs, including the root, stem, leaf and panicle (Figure 3b).
We subsequently monitored free amino acids in the nglf-1 panicle in order to investigate whether levels of Arg and related amino acids (namely ornithine, Glu and Pro) were altered. Compared to WT, free Arg levels increased 9.7-fold, Glu levels increased almost threefold and Pro /levels were reduced by half, with a lesser reduction in Trp levels in the nglf-1 panicle (Table 2). However, no alteration of the levels of free amino acids was found in the stem or leaves, except for Asp and Lys (Table S1).
Table 2. Concentration and proportion of amino acids in the panicle of WT and nglf-1 plants
Values are means ± SD (μmol g DW−1). Three panicles 1 day after flowering from independent plants were used for extraction of free amino acids. Each amino acid proportions are given in parentheses. pSer, PhosphoSerine; β-Ala, β-alanine; β-Aiba, β-aminoisobutyric acid; γ-Aba, γ-aminobutyric acid. Asterisks indicate statistically significant differences compared with wild-type.
nglf-1 is a nitrogen utilization-related mutant, and its phenotype is partially recovered by exogenous nitrogen
Arginase is a key enzyme in arginine catabolism, and arginine nitrogen has been suggested to be recycled by the coordinated action of arginase and urease to meet the metabolic demands of developing organs. We determined total nitrogen, nitrogen per unit weight, and dry weight of various tissues at the ripening stage (Figure 4). Although no significant differences in total dry weight were found, a significant difference in total nitrogen was found between WT and nglf-1, mainly in stems, sheaths and panicles. Differences in nitrogen per unit weight were found in the stems and sheaths of WT and nglf-1. The difference in total nitrogen in the panicle was probably caused by the reduction in panicle biomass. Excess nitrogen remained in the stem, probably due to the low nitrogen utilization efficiency of developing panicles with regard to the Arg of remobilized nitrogen, which resulted in extra small, non-seed-setting tillers.
In our field experiments, we found when exogenous nitrogen was applied, the defect in the nglf-1 phenotype was alleviated. To verify the association of the nglf-1 phenotype with exogenous nitrogen application, exogenous nitrogen was supplied in the form of urea (0, 0.1, 0.2 and 0.4 g N/kg soil) to pot-cultivated nglf-1 plants. All agronomic traits surveyed in nglf-1 exhibited partial recovery, accompanied by an increase in the amount of supplied nitrogen. In the absence of exogenously applied nitrogen fertilizer, the WT plants grew normally until maturity, whereas nglf-1 almost lost the ability for normal panicle development and lacked branch and filled-grain formation. At the other three nitrogen concentrations, all of the traits surveyed in nglf-1 gradually recovered, especially the increase in seed-setting from 11% at 0.1 g N per kg soil to 62% at 0.4 g N per kg soil, showing recovery with exogenous urea supplementation (Figure 5).
Increased filled grain number in transgenic plants corresponds with increased OsARG expression
To explore the impact of OsARG expression on yield traits, OsARG was over-expressed in var. Kitaake. Three T2 lines over-expressing OsARG (OE1, OE2 and OE3) and their control transformed empty vector (CK) were selected for further analysis (Figure 6a). The plants were grown in the field under sub-optimal nitrogen feeding conditions. Compared to the control, arginase activity was increased approximately 1.4, 3.3 and 2.9-fold in the OE transgenic lines (Figure 6b). Among the agronomic traits surveyed, seed-setting rate, grain yield per plant and filled grain number per plant increased significantly for all transgenic plants over-expressing OsARG compared to CK (Figure 6c–e). The correlation coefficient calculated for the grain yield per plant versus filled grain number per plant was R2 = 0.9899 (P >0.01), indicating that the increase in filled grain number mostly contributed to the yield increase, despite no significant differences in the shoot dry weight per plant and thousand-kernel weight (Figure 6f–h).
OsARG expression is independent of Arg catabolism-related genes in nglf-1
Nitrogen catabolism in the Arg catabolism pathway involves at least three key enzymes, namely arginase (l-Arg amidohydrolase, EC 188.8.131.52), urease (urea amidohydrolase, EC 184.108.40.206) and glutamine synthetase (GS; EC 220.127.116.11) (Gerendás et al., 1998; Sirko and Brodzik, 2000). Urease is a nickel metalloenzyme that requires three urease accessory proteins for activation in Arabidopsis (AtUreD, AtUreF and AtUreG) (Witte et al., 2005). All these proteins have a corresponding homolog (Osurease, OsureD, OsureF and OsureG) in rice, with the exception of three homologous cytosolic glutamine synthetase encoding genes: OsGS1;1, OsGS1;2 and OsGS1;3 (Sakamoto et al., 1989; Tabuchi et al., 2005). All seven homologous rice genes displayed similar levels of expression between WT and mutant, indicating no profound effect on Arg catabolism-related genes in nglf-1 (Figure 7).
A small family of arginase-encoding genes that play a key role in providing nitrogen to developing seedlings has been identified in many plant species, such as loblolly pine (Todd et al., 2001), soybean (Goldraij and Polacco, 1999) and tomato (Chen et al., 2004). In Arabidopsis, such genes participate in NO accumulation (Flores et al., 2008). In this study, we identified OsARG from a rice nglf-1 mutant involved in panicle and grain development. The nglf-1 mutant is morphologically similar to the knockout mutant of OsGS1;1, which is an important gene for normal growth and grain filling in rice (Tabuchi et al., 2005). In nglf-1, a nucleotide substitution resulted in premature termination of the putative protein, accompanied by reduced expression of OsARG, and abolished arginase activity, resulting in the levels of its substrate, Arg, increasing nearly 9.6-fold, indicating blockage of the Arg catabolism pathway. Comparative analysis of the nitrogen amount in nglf--1 and WT revealed accumulation of nitrogen in stems and sheaths, and its phenotype was partially recovered by applying exogenous nitrogen. All results indicated that nglf-1 is a nitrogen remobilization-related mutant.
Arginine is one of the main amino acids in nitrogen recycling and remobilization in rice
With regard to the management of nitrogen, young developing roots and leaves of most plant species behave as sink organs for the assimilation of inorganic nitrogen and the synthesis of amino acids. These amino acids are used for the synthesis of enzymes and proteins that are mainly involved in building the plant architecture and various components of the photosynthetic machinery. Nitrogen recycling may occur before flowering, during the reproductive stage, when shoots and/or roots start to act as nitrogen sources by providing amino acids released from protein hydrolysis, which are subsequently exported to reproductive and storage organs via phloem sap translocation (Lattanzi et al., 2005; Masclaux-Daubresse et al., 2010). Wilkinson and Douglas (2003) determined the amino acid composition of phloem exudates from 16 plant species, and found that all 15 dicotyledonous plant species, including Arabidopsis thaliana, had a low proportion of Arg (< 1.6% of total amino acids) in their exudates.
In the case of rice, remobilized nitrogen from the vegetative organs accounts for 70-90% of the total panicle nitrogen (Mae, 1997; Tabuchi et al., 2007). The Arg concentration in leaves, roots and seeds varied from 0.4% to 3.4%, and abnormally high Arg levels ranging from 7.5 to 11.6% of total free amino acids, Arg which originate from the recycling of nitrogen, have been found in phloem sap (Cagampang et al., 1971; Hayashi and Chino, 1990; Sogawa et al., 2002; Takahashi et al., 2006). The difference in the Arg concentration of storage organs versus phloem sap in the reproductive stage is probably due to the activity of OsARG. The Arg concentration detected in the panicles of nglf-1 (11.2%) was consistent with the concentration in the phloem sap, indicating that OsARG is indispensible in the process of nitrogen recycling. Therefore, abolishing OsARG expression would probably block Arg catabolism, resulting in nitrogen shortage and the aberrant phenotype in the nglf-1 panicle.
Arg catabolism plays a crucial role in rice panicle development in the presence of low exogenous nitrogen supply
Arg catabolism is well known for providing a significant portion of nitrogen during seedling development. Arginase, urease and its co-enzymes, and glutamine synthetases all play important roles in this pathway. Until now, only OsGS1;1 was found to exert an effect on panicle development in rice (Tabuchi et al., 2005). Serious phenotype defects have been found for panicle traits in the nglf-1 mutant due to abolishment of OsARG activity. In addition, the response of the nglf-1 mutant to exogenous nitrogen application indicated that, at low levels of nitrogen supplementation, nglf-1 plants show notable differences in seed-setting rate compared to the WT plants, and the difference was alleviated by increasing the exogenous nitrogen supplementation. The most likely explanation why exogenous nitrogen partially restores function in the nglf-1 mutant is the presence of another, unknown nitrogen supplementation pathway. Under low exogenous nitrogen conditions, the panicle mainly utilized the nitrogen hydrolyzed from Arg catabolism; when exogenous nitrogen supplementation was increased, the pattern of nitrogen utilization in the panicles shifted to depend much more on exogenous nitrogen. The pathway that is used depends on the exogenous nutrition concentration, as described for nitrogen by Howarth et al. (2008). Whether this pathway resembles the nutrition utilization pattern in panicle development needs be evaluated in future experiments.
OsARG is a potential gene for improving rice nitrogen use efficiency
Improving nitrogen use efficiency is an important objective of breeding programs. In maize over-expressing the Gln1-3 gene, an increase in kernel number was observed under either high-nitrogen or low-nitrogen growth conditions (Martin et al., 2006). A recent investigation of rice transformed with the cytosolic glutamine synthetase (GS1) gene indicated that, due to enhanced nitrogen use efficiency, lines over-expressing GS1 exhibited a 25–35% higher spikelet yield (Brauer et al., 2011). Our data indicate the possibility of increasing OsARG expression, highlighting the potential use of manipulation of OsARG expression to improve rice yield under sub-optimal nitrogen conditions. However, extensive field experiments are needed.
In summary, we have identified an important gene, OsARG, that not only functions in nitrogen catabolism, but also participates directly in the control of grain yield in rice. Further efforts will be directed to understanding how OsARG and its potential partners are coordinated to regulate nitrogen metabolism in rice plants.
Plant materials and growth conditions
The rice mutant nglf-1 was isolated from the self-cross of a diploid plant derived from an anther culture of autotetraploid rice (Qin et al., 2005). The mutant was crossed with var. IRAT129 to construct an F2 mapping population. A set of 2017 mutant F2 progeny was selected at the panicle emergence stage for use in fine genetic mapping. Grains from IRAT129, the mutant and its wild-type were maintained in the Chinese National Key Facility for Crop Gene Resources and Genetic Improvement. The rice mutant nglf-2 was obtained from T-DNA tagging lines in the background of japonica var. Dongjin (An et al., 2003; Ryu et al., 2004). All plants were grown on the experimental farms of the Institute of Crop Science in Beijing (China) or Sanya (Hainan Province, China) during the natural growing season.
For the pot experiments, plants were grown in plastic pots containing 9 kg soil mixed with 1.72 g KH2PO4 and 1.91 g KCl. The WT and mutant were grown in completely randomized blocks with four replications, each treatment consisting of 16 plants in four separate pots. The N, P and K fertilizers were in the form of CO(NH2)2, KH2PO4 and KCl, respectively, with fertilizer distribution of 50% at basal, 30% at tillering, and 20% at heading. A total of four treatments including 0, 0.1, 0.2 and 0.4 g N/kg soil were applied.
Fine mapping and isolation of OsARG
Genomic DNA from 510 F2 plants exhibiting the mutant phenotype was subjected to linkage analysis. An initial screen for molecular markers linked to the nglf-1 locus was performed using genetic markers from available rice databases, including Gramene (http://www.gramene.org) and the Rice Genome Research Program (http://rgp.dna.affrc.go.jp/publicdata/caps/index.html). The markers used for fine mapping were produced based on comparison of genomic sequences from indica var. 9311 and japonica var. Nipponbare. The primer sequences are given in Table S2.
For functional complementation, a 5233 bp genomic DNA fragment containing the entire OsARG coding region and its immediate upstream and downstream sequence was amplified from WT genomic DNA using primers 5′-CCGGAATTCTAGGAATAGCCTCCATGTCATCG-3′ and 5′-CGGGGTACCACCACTCCACTTGAACCTCGTAA-3′. The resulting fragment was digested using EcoRI and KpnI, and inserted into the pCAMBIA1305 vector (Cambia, http://www.cambia.org/) to generate transformation plasmid pARG, which was then introduced into Agrobacterium tumefaciens strain EHA105 by electroporation and transformed into the mutant nglf-1 via complementation or into var. Kitaake to over-express OsARG, as described by Hiei et al. (1994).
Amino acid sequences homologous to OsARG were selected by BLAST search (http://blast.ncbi.nlm.nih.gov/). Multiple sequence alignments were performed using BioEdit version 7.0 (Hall, 1999). A neighbor-joining tree (Saitou and Nei, 1987) was built using MEGA version 4.0 (Tamura et al., 2007) by adopting Poisson correction distance with a bootstrap replicate number of 1000.
Total RNA was extracted using an RNAprep Pure plant kit (Tiangen, http://www.tiangen.com/). Total RNA (2 μg) was used to synthesize cDNA using a reverse transcription kit (TaKaRa, http://www.takara.com.cn/). RT-PCR was performed using the primer pairs 5′-CTTCACCTTGACGCACATC-3′ and 5′-GCGCATCTCATACTGTTCC-3′ for OsARG, and 5′-TGGAACTGGTATGGTCAAGGC-3′ and 5′-AGTCTCATGGATACCCGCAG-3′ for ACTIN, which was used as an internal control for normalization of each RNA sample.
Quantitative RT-PCR was performed using a SYBR premix Ex Taq™ kit (TaKaRa) according to the manufacturer's instructions, on a 7900HT real-time PCR system (Applied Biosystems, http://www.appliedbiosystems.com.cn/). Before comparison, relative expression levels were normalized to that of an internal control, UBIQUITIN (Os03g0234200). The primers used to specifically detect the genes are listed in Table S3.
For the promoter–GUS assay, a 2801 bp genomic fragment upstream of the OsARG putative translation start codon was amplified using primers 5′-CCGGAATTCCGTCGCTTTGGTCCCTGCCG-3′ and 5′-GGAAGATCTTATCCCCCCTTCGCTCCGCT-3′. After sequencing confirmation, the DNA fragment was cloned into the EcoRI and BglII sites of the vector pCAMBIA1381Z. The construct pOsARG–GUS was transformed into var. Kitaake by Agrobacterium tumefaciens-mediated transformation. Histochemical GUS staining was performed as described by Scarpella et al. (2003).
To determine its subcellular localization, a 1020 bp cDNA fragment of OsARG was amplified using primers 5′-TAGCTCTAGAATGGGCGGCGTGGCGGCG-3′ and 5′-CCATGGATCCCTTGGAGATCTTGGCTGTGA-3′. The fragment was inserted into the pCAMBIA1305-GFP vector, and introduced into var. Kitaake by Agrobacterium tumefaciens-mediated transformation. Mitochondria from protoplasts isolated from young transformant roots were labeled with MitoTracker Red, a mitochondrion-specific dye (Invitrogen, http://www.invitrogen.com/), as described by Han et al. (2006) and Woo et al. (2007). The images were captured using a Nikon Eclipse TE2000-U microscope (Nikon).
Arginase activity assay
Various tissues from WT and nglf-1 plants were frozen in liquid nitrogen and ground in a mortar to a fine powder. Each sample was extracted using 1.0 ml ice-cold 10 mM Tris/HCl (pH 9.0) containing 1 mM MnCl2, and the homogenate was centrifuged at approximately 14 000 g for 20 min (4°C) in a 1.5 ml microfuge tube. The supernatant was transferred into a new 1.5 ml microfuge tube. Arginase activity was assayed as described by Flores et al. (2008).
Determination of free amino acids
Free amino acid analyses were performed in triplicate for WT and nglf-1. The stem, leaf blade and panicle were harvested 1 day after flowering, ground to powder in liquid nitrogen, and then homogenized in 10 volumes of 8% 5-sulfosalicylic acid dihydrate (10 μl for 1 mg sample). The homogenate was centrifuged at 10000 g for 10 minutes, and the supernatant fraction was filtered through a syringe filter (PALL, http://www.pall.com/). The resulting filtrate was used for analysis of amino acids. Levels of amino acids were determined using a S-433D automatic amino acid analyzer (Sykam, http://www.sykam.com/) according to the manufacturer's instructions.
Total nitrogen analysis
For total nitrogen analysis, 0.1 g of oven-dried and ground tissue was weighed into 100 ml Kjeldahl digestion flasks. Concentrated H2SO4 (5 ml) was added to each flask and gently heated. When frothing ceased, the heat was increased to 280°C, and 30% v/v H2O2 was added to the flask intermittently until the digest cleared. After complete digestion, the flask was allowed to cool, and water was slowly added to make up the volume to 100 ml. A 5 ml sample was then analyzed for total nitrogen by Kjeldahl determination (KDY-9830, RuiBangXingYe, http://www.rbxycn.com/) as described by Fan et al. (2007).
We thank Chentao Lin (Department of Molecular, Cell and Developmental Biology, University of California, Los Angeles, CA), Jianfeng Ma (Institute of Plant Science and Resources, Okayama University, Japan) and Xianchun Xia (Institute of Crop Science, Chinese Academy of Agricultural Sciences, Beijing, China) for critical reading and comments on the manuscript. We also thank Joe C. Polacco and Christopher D. Todd (Biochemistry Department, University of Missouri, Columbia, MO) for technical assistance with the arginase activity assay, and Quan Zhang (School of Life Sciences, Peking University, Beijing, China) for help with subcellular localization. This work was supported by grants from the National 863 High-tech R&D Program of China (2012AA10A301), the National Special Project (2011ZX08009-003-003), and the National Natural Science Foundation of China (31771316).