Requirement for the plastidial oxidative pentose phosphate pathway for nitrate assimilation in Arabidopsis


  • John D. Bussell,

    1. Australian Research Council Centre of Excellence in Plant Energy Biology (M316), University of Western Australia, Crawley, WA, Australia
    Search for more papers by this author
  • Olivier Keech,

    1. Australian Research Council Centre of Excellence in Plant Energy Biology (M316), University of Western Australia, Crawley, WA, Australia
    Current affiliation:
    1. UPSC, Department of Plant Physiology, Umeå University, Umeå, Sweden
    Search for more papers by this author
  • Ricarda Fenske,

    1. Metabolomics Australia, University of Western Australia, Crawley, WA, Australia
    Search for more papers by this author
  • Steven M. Smith

    Corresponding author
    • Australian Research Council Centre of Excellence in Plant Energy Biology (M316), University of Western Australia, Crawley, WA, Australia
    Search for more papers by this author

For correspondence (e-mail


Sugar metabolism and the oxidative pentose phosphate pathway (OPPP) are strongly implicated in N assimilation, although the relationship between them and the roles of the plastidial and cytosolic OPPP have not been established genetically. We studied a knock-down mutant of the plastid-localized OPPP enzyme 6-phosphogluconolactonase 3 (PGL3). pgl3-1 plants exhibited relatively greater resource allocation to roots but were smaller than the wild type. They had a lower content of amino acids and free inline image in leaves than the wild type, despite exhibiting comparable photosynthetic rates and efficiency, and normal levels of many other primary metabolites. When N-deprived plants were fed via the roots with inline image, pgl3-1 exhibited normal induction of OPPP and nitrate assimilation genes in roots, and amino acids in roots and shoots were labeled with 15N at least as rapidly as in the wild type. However, when N-replete plants were fed via the roots with sucrose, expression of specific OPPP and N assimilation genes in roots increased in the wild type but not in pgl3-1. Thus, sugar-dependent expression of N assimilation genes requires OPPP activity and the specificity of the effect of the pgl3-1 mutation on N assimilation genes establishes that it is not the result of general energy deficiency or accumulation of toxic intermediates. We conclude that expression of specific nitrate assimilation genes in the nucleus of root cells is positively regulated by a signal emanating from OPPP activity in the plastid.


The pentose phosphate pathway (PPP) is essential for metabolism as it generates reducing equivalents (i.e. NADPH) as well as ribulose-5-P and erythrose-4-P necessary for nucleotide synthesis and for production of aromatic amino acids, respectively. Often seen as an anabolic counterpart to glycolysis, the PPP can be divided into two distinct phases: one oxidative and the other non-oxidative. The oxidative PPP (OPPP) converts a six-carbon sugar phosphate, glucose-6-P, to a pentose phosphate, ribulose-5-P, which then becomes the substrate of the non-oxidative PPP (Kruger and von Schaewen, 2003). The OPPP includes two NADPH-producing steps: glucose-6-P dehydrogenase (G6PDH) and 6-phosphogluconate dehydrogenase (6PGDH). Glucose-6-P dehydrogenase produces a lactone (6-phosphoglucono-δ-lactone), which is hydrolyzed by 6-phosphogluconolactonase (6PGL) to form 6-phosphogluconate, the substrate for 6PGDH. Oxidation of 6-phosphogluconate by 6PGDH is concomitant with decarboxylation. Its role in the provision of reductant for biosynthetic reactions, including fatty acid synthesis and assimilation of inorganic nitrogen and sulfur, is particularly important in non-photosynthetic tissues such as roots and developing embryos (Leustek et al., 2000; Kruger and von Schaewen, 2003).

In plants, nitrogen uptake and its assimilation are strongly dependent on availability of photosynthetically derived carbon for energy and for carbon skeletons. It is also linked to the diurnal cycle, light intensity, water supply and the availability of other nutrients (Nunes-Nesi et al., 2010). In Arabidopsis roots, the uptake of nitrogen is generally achieved via import of inline image (nitrate) by root transporters. The subsequent assimilation of inline image is initiated by its reduction to inline image (ammonium) through the action of nitrate reductase (NR) and nitrite reductase (NiR). The final incorporation of inline image into amino acids results from the joint action of glutamate synthase (GOGAT) and glutamine synthetase (GS) (Coruzzi, 2003). The degree to which inline image is assimilated (via inline image) into amino acids in roots and then transported to other parts of the plant versus transport of inline image (or inline image) for assimilation at another location remains an open question (Nunes-Nesi et al., 2010). Regardless, this process requires bidirectional root–shoot transport of metabolites and implies root–shoot signaling to coordinate the N and C requirements of the plant.

Linkage between the OPPP and inorganic nitrogen assimilation in plastids was established by the observation that the OPPP supplies NADPH to NiR (Bowsher et al., 2007) and GOGAT (Bowsher et al., 1992; Esposito et al., 2005). At the level of transcription, OPPP genes (two of the six G6PDH genes and two of the three 6PGDH genes in Arabidopsis) are markedly induced in response to inline image treatment of seedlings starved of inline image (Wang et al., 2003). This occurs even in a NR knockout mutant, suggesting that inline image, rather than some downstream metabolite, is the inducing signal (Wang et al., 2004). The OPPP appears to be required for induction of the major root inline image uptake transporters NRT1;1 and NRT2;1 (Lejay et al., 2008). Based on these observations, it has been suggested that the OPPP might generate a signaling molecule, signal through cellular redox status, and/or be a root sensor for sucrose from photosynthesis (Lejay et al., 2008).

The Arabidopsis genome has five genes that encode 6PGL proteins. Four of these are cytosolic proteins (Kruger and von Schaewen, 2003), while PGL3 has both plastid- and peroxisome-targeting signals (Reumann et al., 2004, 2007). Knockouts of three of the four cytosolic forms (PGL1, PGL2, PGL5) have no obvious mutant phenotype (Xiong et al., 2009); the fourth, PGL4, has not been studied. In contrast, homozygous T-DNA knockouts of PGL3 are lethal (Meinke et al., 2008). In order to study the role of PGL3 and of the OPPP in nitrogen assimilation, we obtained a T-DNA insertion line in PGL3 (At5g24400). This mutant, named pgl3-1, is a knockdown that remains viable, albeit with reduced plant stature, and has previously been shown to exhibit enhanced resistance and constitutively activated pathogenesis-related gene expression during normal growth (Xiong et al., 2009). In this paper, we describe the use of pgl3-1 to seek genetic evidence for the role of plastidial OPPP function in inline image assimilation and to investigate whether sugar-stimulation of N assimilation gene expression is dependent upon plastidial OPPP activity.


Identification of the pgl3-1 mutant

Eight Arabidopsis thaliana T-DNA lines with insertions in or near At5g24400 were screened. Three of these occurred in the middle of the gene, two in the promoter sequence just upstream from the 5′-untranslated region (UTR), and three in the 3′-UTR (Figure 1a). Homozygotes could not be obtained for T-DNAs inserted between the start and stop codons, and these were assumed to be lethal (FLAG_219G10, SK_978, emb2024). Promoter-located insertion lines and two of the 3′-UTR lines grew normally and were phenotypically indistinguishable from wild-type plants. These lines were not studied further. However, as previously noted by Xiong et al. (2009), SALK_005685 was stunted in growth compared to the wild type. Quantitative (q)RT-PCR analysis revealed that shoots and roots of the SALK_005685 T-DNA homozygotes had only about 20% of the wild-type level of PGL3 transcript (Figure 1b). Soluble proteins were extracted from roots and leaves of wild-type and pgl3-1 plants and a Western blot probed with an antibody against PGL3. This revealed greatly reduced abundance of the encoded protein in both roots and leaves of the mutant (Figure 1c). The protein detected by Western blots was approximately 28 kDa, the size expected for PGL3 after cleavage of the predicted plastid-targeting peptide. The annotated position of the SALK_005685 T-DNA was just downstream of the stop codon. Sequencing across the T-DNA borders confirmed this, but also revealed that the insertion was present as an inverted repeat and resulted in deletion of 28 bp surrounding the stop codon (Figure S1). The predicted translation of the mutant protein replaced six terminal residues of the native protein with eight amino acids encoded by the T-DNA before the first stop codon was reached (Figure 1d). Xiong et al. (2009) have previously designated SALK_005685 as pgl3 and observed FLAG_219G10 to be embryo lethal. As we have used both of these lines in our experiments we accordingly distinguish them by assignment as pgl3-1 and pgl3-2.

Figure 1.

Isolation of pgl3-1 mutants. (a) Scheme of the At5g24400 (PGL3) gene and position of T-DNA insertions. The viability of T-DNA lines is indicated. Note that the SALK_005685 T-DNA (pgl3-1) is inserted as an inverted repeat. (b) Relative abundance of PGL3 transcript in roots and shoots of 7-day-old wild-type and pgl3-1 seedlings. Seedlings were grown on 0.5 ×  MS plates (without added sucrose). Values are expression relative to an internal reference gene (CACS) and represent the mean ± SE of at least three biological replicates. (c) Western blot of PGL3 protein in roots and shoots of 5-week-old hydroponically grown plants. Soluble proteins were extracted from approximately 250 mg of leaf and root tissue. Forty micrograms of leaf and 10 μg of root protein were loaded into the lanes. A c. 28 kDa band corresponding to the size of the specific, mature (pre-sequence-cleaved) PGL3 protein can be seen in wild-type (Col-0) lanes and is barely detectable in pgl3-1 lanes (arrowhead). Two independently grown samples are presented for each tissue and genotype. α-Tubulin (α-tub) was used as a loading control. M denotes the lane in which a size marker was loaded. (d) FST re-sequencing across the upstream-facing left border of the SALK_005685 T-DNA indicated that the insertion was associated with deletion of a short segment of the coding sequence. Translation of this chimeric sequence suggested that eight amino acids encoded by the T-DNA border sequence replaced six terminal residues of the native protein.

Subcellular localization and expression pattern of PGL3

6-Phosphogluconolactonase 3 is predicted to have both plastid- and peroxisome- (PTS1) targeting signals (Kruger and von Schaewen, 2003; Reumann et al., 2004, 2007). Moreover, the plastid-targeting peptide is bracketed by two possible methionine initiation codons. To examine subcellular localization of the protein, we constructed a PGL3-GFP fusion reporter by inserting the GFP coding sequence into the PGL3 coding sequence at a position corresponding to 13 amino acids upstream from the stop codon (Figure 2a). This position was chosen to preserve secondary structure predictions ( and both N- and C-terminal targeting sequences (Tian et al., 2004). Despite preserving the PTS1, we were able to localize the protein only to plastids during the transient expression assays by biolistic bombardment (Figure 2b–g).

Figure 2.

Subcellular localization of 6-phosphogluconolactonase 3 (PGL3). (a) Scheme of the GFP construct used in the localization experiments. GFP was cloned in-frame 13 amino acid residues upstream from the stop codon. Two potential start codons are indicated, as are alternative targeting signals. (b)–(g) In vivo targeting of PGL3. Biolistic bombardment was used to co-transform the PGL3-GFP plasmid with organelle markers into cultured Arabidopsis protoplasts. The small subunit of RuBisCO fused to red fluorescent protein (RFP) (SSU-RFP) was used as the plastid marker and RFP fused to the peroxisome targeting signal SRL (RFP-SRL) (Pracharoenwattana et al., 2005) as the peroxisome marker.

Publicly available microarray data were used to determine patterns of PGL3 expression. The developmental map generated by the eFP browser (Winter et al., 2007) indicated that PGL3 was expressed quite uniformly in all tissues, with slightly elevated expression in developing seeds, particularly stages 5–7 (Figure S2a). These stages correspond to those at which abortion occurs in pgl3-null mutants such as emb2024 (Meinke et al., 2008). Genevestigator (Hruz et al., 2008) data indicated that transcription of the gene was relatively high and stable throughout development (Figure S2b). Anatomically, expression was marginally higher in roots and substantially lower in anthers and pollen, but otherwise largely invariant (Figure S2c). Expression levels of PGL3 were remarkably resistant to perturbation by biotic and abiotic factors. Of the 1627 perturbations summarized by Genevestigator, only 40 (2.5%) result in a greater than 1.5-fold change (< 0.05) and of these only four were greater than two-fold (< 0.05).

Growth phenotype of pgl3-1

The vigour of pgl3-1 varied dramatically depending on the photoperiod in which it was grown. Specifically, the mutant rosette was much smaller relative to the wild type in longer light periods (Figure 3a–c). At the same age, soil-grown pgl3-1 plants grown in long days (16 h light, 8 h dark) were much smaller than the wild type. The relative size of mutant plants compared to with the wild type increased in medium (12:12 light:dark) and short (8 h light, 16 h dark) days (Figure 3a–c). Because PGL3 has both plastid- and peroxisome-targeting signals (Figure 2a), we undertook complementation of the pgl3 mutants using three different constructs corresponding to the full-length protein, and plastid- or peroxisome-specific targeted variants. Both pgl3-1 and the embryo lethal allele pgl3-2 were complemented by the full-length or plastid-specific variants, but not by the peroxisome-specific construct (Figure 3a,d). Complementation of pgl3-2 was assessed by the recovery of FLAG_219G10 T-DNA homozygotes in the progeny of transformants (Figure 3d).

Figure 3.

Phenotypic characterization and phenology of pgl3-1. (a)–(c) Appearance of typical wild-type (WT), pgl3-1 and complemented plants just prior to bolting. PGL3-full, PGL3-px, and PGL3-cp indicate pgl3-1 plants transformed with cDNAs corresponding to full-length, peroxisome-only, or plastid-only targeted constructs for complementation tests. (a) Plants grown in long-day conditions (LD, 16 h light: 8 h dark). (b) Medium days (MD, 12 h light: 12 h dark). (c) Short days (SD, 8 h light: 16 h dark). (d) Transformation of pgl3-1 and pgl3-2 (embryo lethal) by full-length, peroxisome only and plastid only targeted variants of 6-phosphogluconolactonase 3 (PGL3) showed that the plastid targeted PGL3 is sufficient to rescue both the hypomorphic allele (pgl3-1) and the embryo lethal knockout and pgl3-2. Rescue of the pgl3-1 phenotype is shown in (a). (e), (f) Plants grown in conditions depicted in (a)–(c) and also in MD hydroponics (MDHyd) were characterized for phenology, including number of days until bolting (e) and number of leaves (f). (g) Hydroponically grown MD plants were harvested at bolting for comparison of root and shoot fresh weight. Values in (e)–(g) represent the mean ± SE of at least 10 plants.

Grown under short-, medium- or long-day photoperiods, the pgl3-1 mutant was significantly slower to bolt than wild-type plants (Figure 3e). Under short-day growth conditions, the mutant had a comparable number of leaves to the wild type at bolting (Figure 3f). pgl3-1 plants grown under medium-day conditions had somewhat fewer leaves at bolting than wild-type plants, and under long-day conditions had marginally (but significantly) fewer leaves than wild-type plants. The wild type and pgl3-1 respectively maintained similar growth and phenology in a hydroponic growth system (under medium days) as they had in soil under the same photoperiod. Likewise, the phenotypic differences between the genotypes were consistent in soil or hydroponics, with the exception that for hydroponically grown plants there was no difference between Col-0 and pgl3-1 for the age of plants at bolting (Figure 3e, f). Use of the hydroponic system allowed comparison of the fresh weight of roots and shoots at bolting, and showed that in medium days, the mutant had about 40% the total fresh weight of the wild type. However, the roots contributed relatively more to total fresh weight in the mutant (Figure 3g). Thus, the ratio of root to shoot fresh weights of Col-0 plants was about 0.70 compared with 1.38 for pgl3-1.

As pgl3-1 exhibited such a drastic reduction in growth in response to an increased photoperiod, we examined photosynthetic parameters to determine whether there were any inherent differences in photosynthetic efficiency of the mutant compared to the wild type. The maximum quantum yield in the dark-adapted state (Fv/Fm) between wild type and mutant was not significantly different for intact 4-week-old hydroponically grown plants, or for individual leaves from 7-week-old plants grown in soil under short-day conditions (Figure 4a,b). The dark-adapted minimum fluorescence (F0) was not significantly different between wild type and pgl3-1, indicating that there was no difference in the size of antennae. To further investigate potential defects in the photosynthetic apparatus, dark–light/recovery induction curves were recorded. The qL (photochemical quenching, a measure of the fraction of open photosystem II reaction centers) as well as the regulated and non-regulated heat dissipation mechanisms, Y(NPQ) and Y(NO) respectively, were estimated. These parameters provide important information on the partitioning of excitation energy. qL, Y(NPQ), and Y(NO) were almost identical in leaves from wild-type and mutant plants grown hydroponically (Figure 4c,e,g). However, when similar experiments were carried out with leaves from soil-grown plants, the mutant showed a greater photochemical efficiency (qL) and a correspondingly lower Y(NPQ) than leaves from wild-type plants (Figure 4d,f,h). Although reproducible across experiments, and despite our efforts to obtain the best match between mutant and wild-type leaves, we cannot rule out the possibility that the greater photochemical efficiency in soil-grown pgl3-1 plants results from the growth differences between the two genotypes. Regardless, it can be concluded that photosynthesis in the mutant is at least as efficient as in the wild type and that the phenotypic differences are unlikely to result from an altered photosynthetic capacity. Additional experiments would clarify whether this greater photochemical efficiency in pgl3-1 plants simply results from a growth difference.

Figure 4.

Photosynthetic characterization of pgl3-1. Plant were grown hydroponically (a, c, e, g) or in soil (b, d, f, h). (a, b) Maximum quantum yield in the dark-adapted state (Fv/Fm); (c, d) qL (photochemical quenching). Regulated (e, f) and non-regulated (g, h) heat dissipation mechanisms, Y(NPQ) and Y(NO), respectively.

To further elaborate plant-scale physiological function, 4-week-old medium-day soil-grown plants were stained for starch at the beginning and end of the photoperiod. This showed that pgl3-1 was able to synthesize and metabolise starch in a similar fashion to the wild type (Figure S3). Similar results were obtained for hydroponically grown plants. Additional photoassimilates, such as fumarate, can also accumulate massively in Arabidopsis leaves during the light period (Pracharoenwattana et al., 2010). However, there were no differences between pgl3-1 and the wild type in fumarate reserves (Table S2). Thus, perturbation to starch cycling and the availability of photosynthetic reserves are also unlikely to account for the mutant's poor vigor.

pgl3-1 is altered in amino acids and free inorganic N abundance

We next utilized metabolite analysis to assess the nitrogen status of the mutant. Col-0 and pgl3-1 plants were grown hydroponically in medium days under N-replete conditions, then primary metabolites were extracted from roots and shoots. There were marked differences in the abundance of primary N-metabolites between mutant and wild-type leaves. Free inline image was decreased in rosette leaves of pgl3-1 (Figure 5a). Free inline image, however, was unaltered (Figure 5b). The total abundance of amino acids was significantly lower in mutant compared with wild-type shoots (Figure 5c). The individual amino acids that contributed most to the differences observed in leaves were Pro, Ser, Asp, Asn, Gln, and Glu (Table S2). By contrast, in roots the only significant difference between Col-0 and pgl3-1 was Ala, which was reduced in the mutant (Table S2). The amount of nitrate and total amino acids in roots was about 10% of that found in shoots, but there were no significant differences in total amino acids, inline image, or inline image between mutant and wild-type roots (Figure 5, Table S2).

Figure 5.

Analysis of N status of wild-type and pgl3-1 plants. Plants were grown under N-replete conditions (a)–(c) or subject to 7 days' N starvation (d)–(f) before being analyzed for free inline image, inline image, and amino acid composition. For each panel, results are mean ± SE (= 4) and at least four technical replicates were done for each biological replicate. Values for pgl3-1 that are significantly different (t-test < 0.05) from the corresponding wild-type value are indicated by *. Amino acid abundance in (c) and (f) is given as arbitrary units per g fresh weight (gFW) relative to an internal standard (ribitol).

Response to N-starvation was assessed by growing plants hydroponically under a medium-day photoperiod for 4 weeks then switching them to medium lacking a N source for 1 week. After this growth regime, the ratio of root:shoot fresh weights remained similar to that reported for plants grown for 5 weeks under N-replete conditions: 0.60 for Col-0 and 1.18 for pgl3-1. In N-starved plants, the amount of free inline image was substantially lower than for plants grown under N-replete conditions, although the mutant had slightly more inline image than the wild type (Figure 5d). In contrast, inline image levels were slightly higher in roots and leaves under N-starved than N-replete growth conditions, and pgl3-1 leaves had a reduced level of inline image (Figure 5e). Total amino acids were again reduced in mutant leaves (Figure 5f).

To test whether the differences in primary N metabolites between pgl3-1 and the wild type were due to different capacities for N uptake and assimilation, we utilized 15N-labeled inline image in feeding experiments. The N-starved plants were supplied with 1 mm 15inline image and the uptake and assimilation of labeled nitrate into amino acids was assayed. Despite the reduced N status of pgl3-1 compared with wild-type leaves (Figure 5), short-term assimilation of 15N was at least as efficient in the mutant as it was in the wild type (Figure 6). In roots, 15N-labeled Asp, Glu, and Gln enriched appreciably over 2 h. Other amino acids that we could reliably detect experienced only minor 15N enrichment (Ala, Leu, Ile) or remained un-enriched. Notably, there was no significant difference in 15N-enriched amino acids between pgl3-1 and wild-type roots. In shoots, there was greater enrichment in 15N-labelled amino acids in pgl3-1 than in the wild type. Thus, Asp, Glu, Gln, Asn, Ala, Gly, Ser, and Ile were all about two-fold or more 15N-enriched on a leaf fresh weight basis in pgl3-1 than in the wild type.

Figure 6.

15N-enrichment in amino acids of N-starved wild-type (Col-0) and pgl3-1 plants resupplied with inline image. (a) Roots. (b) Shoots. For each panel, results are mean ± SE (= 4) and at least four technical replicates were done for each biological replicate.

pgl3-1 exhibits a normal transcriptional response to NO3 resupply

Treatment of Arabidopsis with inline image induces expression of a range of genes including those of the OPPP, inline image transporter (NRT) and assimilation genes (Wang et al., 2003). This typical response is displayed in the wild type and pgl3-1 (Figures 7a and S4a,b). We selected classical N-inducible genes for qRT-PCR analysis and showed that 6PGDH2, G6PDH3, G6PDH2, NRT1;1, NRT2;1, NIA1, NIA2, and NiR (see Wang et al., 2003) were strongly induced in both wild-type and pgl3-1 roots 2 h after resupply of N (Figure 7a). Similar to findings of Wang et al. (2003), N assimilation genes including NADH-GOGAT, GS isoforms, and ASN2 were not greatly affected. There was very little difference between pgl3-1 and the wild type in the transcriptional response to resupply of N. In shoots, OPPP genes (G6PDH2, G6PDH3, 6PGDH2) and inline image reduction genes (NIA1, NIA2, NiR) were induced (Figure S4b), although not as markedly as in roots. NADH-GOGAT and ASN2 were also induced. As in roots, there was very little to distinguish the wild type and pgl3-1 shoot transcriptional responses to N resupply.

Figure 7.

Quantitative RT-PCR analysis of the abundance of N-responsive transcripts in roots of wild-type (Col-0) and pgl3-1 plants. (a) Plants grown for 28 days in hydroponics, starved of N for 7 days and then resupplied with 1 mm inline image. Material for assay was collected immediately prior to or 2 h after N resupply. (b) Plants were grown hydroponically under N-replete conditions for 5 weeks and then transferred for 4 h in the middle of the photoperiod to fresh media without (white columns) or with (black columns) 3% sucrose. Values are expression relative to an internal reference gene (CACS) and represent the mean ± SE of at least three biological replicates. Expression levels in pgl3-1 that are significantly different (t-test, < 0.05) from the corresponding wild-type values for the same treatment are indicated by *. Quantitative RT-PCR analysis of genes encoding these (and other) transporter and metabolic genes in roots and leaves of these plants is provided in Figures S4(a)–(d).

Transcriptional responses of pgl3-1 to sucrose supplementation

We next examined the transcriptional effect of altered C status. Plants were grown hydroponically under a 12 h light, 12 h dark photoperiod for 5 weeks then transferred for 4 h to fresh Hoagland's solution with or without a supplement of 3% (w/v) sucrose. OPPP genes (G6PDH2, G6PDH3 6PGDH2) and genes of nitrate and nitrite reduction (NIA1, NIA2, NiR) were all induced about two-fold in wild-type roots but not in pgl3-1 (Figure 7b). Cytosolic glutamine synthetase (GS1-1; At5 g37600), although not induced by sugar, was 2.5 times higher in pgl3-1 than in the wild type (Figure 7b). Collectively, these results indicate misregulation of OPPP and N reducing and assimilating genes in pgl3-1. Misregulation of sulfur transporters was also apparent in the mutant, with SULTR1;1 at least two-fold higher in the mutant and SULTR3;5 experiencing a substantially reduced expression in the mutant in response to sucrose, an effect that was not seen in the wild type (Figure S4c).

In shoots of hydroponically grown plants, there was very little difference in the transcript profile of OPPP, nutrient transporter and N assimilation genes between the mutant and wild type (Figure S4d). The main exceptions were NIA1 and NIA2, expression of which was not altered in the wild type in response to sugar, but was initially about two-fold higher in the mutant and decreased in abundance to approximately wild-type levels in response to sucrose treatment, and GS1-1 transcription which, although not altered by addition of sucrose, was uniformly about 1.5-fold higher in pgl3-1 leaves.


Plastid-targeted PGL3 is sufficient to rescue the pgl3-1 phenotype

The operation in plants of both a plastidial and a cytosolic OPPP is well established (Kruger and von Schaewen, 2003). There is now also increasing evidence of localization of the pathway to peroxisomes (Kaur and Hu, 2011), including facultative targeting of G6PDH1 to peroxisomes by an internal PTS1 and interaction with G6PDH4, dependent on key cysteine residues (Meyer et al., 2011). The PGL3 protein is potentially targeted to plastids or peroxisomes by the presence of N- and C-terminal targeting signals (Reumann et al., 2007; Xiong et al., 2009). Indeed, it is interesting to note that almost all land plant 6PGL proteins have conserved a PTS1 (or PTS1-like) tripeptide at or close to their C-termini (Figure S5). Despite this, pgl3-1 and embryo lethal T-DNA knockouts of PGL3 can be rescued by plastid-only constructs lacking the PTS1 (Figure 3) or by transformation with PGL3::PGL3-GFP or 35S::PGL3-GFP where the GFP tag masks the PTS1 (Xiong et al., 2009). Thus, localization of PGL3 to the plastid is essential, a point emphasized by mass spectrometry-based proteomic studies that have repeatedly found PGL3 in plastids (Zybailov et al., 2008; Ferro et al., 2010; Olinares et al., 2010) but not yet in peroxisomes (Reumann et al., 2007, 2009; Eubel et al., 2008).

Flux through the plastidial OPPP is essential for N assimilation and plant development

Biochemical studies have linked the plastidial OPPP to provision of reductant for N assimilation in non-photosynthetic tissues (Bowsher et al., 1992, 2007). In fact, the OPPP itself (rather than the activities of the N assimilation enzymes) is thought to limit root N assimilation (Foyer et al., 2011). Exposure of nitrate-starved roots to nitrate activates the expression of genes for nitrate uptake (NRT) and reduction (OPPP, NR, NiR) in the root (Figure 7; Wang et al., 2003). Supply of sugars from the shoot activates expression of NRT and nitrate reduction genes in the root (Figure 7; Lejay et al., 2008). The absence of strong induction of transporter genes (e.g. NRT1;1 and NRT2;1 amongst others) in wild-type roots in response to sugar treatment was somewhat surprising, because such an increase has previously been reported (Lejay et al., 2008). However, although that study used a similar hydroponic set-up to the one employed here, it additionally utilized a 40-h dark pre-treatment to repress light- or (photosynthetically fixed) sugar-inducible genes before feeding plants with sugars and other compounds. This would have strongly modified the overall C/N signaling in comparison to the present experiment.

Sugar-inducible N assimilation genes (NIA, NiR and those of the OPPP) were not induced in pgl3-1 by sucrose supplement. Thus, unlike the inline image induction of these genes, sugar-stimulated induction is dependent upon plastidial OPPP activity. The cytosolic OPPP apparently cannot substitute for loss of the pathway in the plastid, implying a plastidial signal. Collectively, our results strongly suggest that expression of N assimilation genes in Arabidopsis roots in response to sugar is dependent on operation of the plastidial OPPP. Indeed, knockdown of synthesis of GPT1, the major port of entry of G6P into non-photosynthetic plastids, resulted in embryos that had reduced accumulation of transcripts for OPPP genes, suggesting that induction of expression of these genes is dependent on flux through the pathway (Andriotis et al., 2010). It is noteworthy that in pgl3-1 the main transcriptional and metabolic effects appear to be spatially separated. That is, transcriptional differences are seen in the roots and metabolite differences in the leaves. We propose that the poor vigor of the mutant is due to reduced carbon flow through the plastidial OPPP and a consequent inability to efficiently assimilate inline image into organic molecules. The extent to which this occurs in roots and shoots is unknown and will require further investigation.

pgl3-1 plants exhibit a N-starved profile

The pgl3-1 mutant grew poorly compared with the wild type, particularly under diurnal growth conditions with longer light periods. We used a hydroponic growth system combined with a medium-day (12 h light, 12 h dark) photoperiod that reduced the difference between mutant and wild type. In this system, pgl3-1 plants displayed several symptoms typical of N starvation. First, root growth was proportionally greater in pgl3-1 than in wild type plants, a characteristic previously noted for N-starved plants (Scheible et al., 1997b). Second, for both roots and shoots of pgl3-1, transcript of cytosolic GS1 (but not plastidial GS2) was elevated compared with wild type (Figure 7B and Figure S4c). Such elevated expression of cytosolic GS transcripts has previously been associated with N starvation. For example, transcripts for cytosolic GS1, but not plastidial GS2, were increased in leaves of nitrate reductase-deficient tobacco (Scheible et al., 1997a), and at the level of protein abundance GS1 was elevated in barley grown under long-term low-inline image conditions (Møller et al., 2011). Third, the pgl3-1 phenotype is reminiscent of wild-type plants grown under mild but sustained N limitation (Tschoep et al., 2009). In that experiment, wild-type Arabidopsis plants were grown under similar conditions to the present experiment (12 h photoperiod for 35 days), but either on N-poor or N-rich peat-based soil mix. The plants grown on N-poor substrate reached less than half the shoot biomass and contained lower leaf Asp, Asn, Ala, and much lower leaf inline image compared with plants grown on high N. Although the difference in leaf inline image between pgl3-1 and wild type (Figure 5a) was not as extreme as between the previously described wild-type plants grown on low and high N, the pgl3-1 plants exhibited a similar N-deficient metabolic profile. However, unlike the comparison of low and high N (Tschoep et al., 2009), pgl3-1 did not exhibit increased serine, or decreased malate and fumarate in response to low N. This is probably because there was sufficient inline image in the medium of pgl3-1 plants to signal that they were growing under N-replete conditions, despite the apparently compromised N assimilation in planta.

Grown under N-replete or N-deficient conditions, pgl3-1 leaves had substantially lower abundance of free inline image and amino acids. However, when starved of N and then resupplied with nitrate, the pgl3-1 mutant roots were at least as efficient at assimilating resupplied inline image as those of the wild type. Surprisingly, however, in leaves 15N enrichment of amino acids occurred at about twice the rate in pgl3-1 as in Col-0. Although this result at first appears contrary to what might have been hypothesized, we attribute it to the smaller amino acid pool sizes in the pgl3-1 shoot, in which the total amount of amino acids was about half that of the wild type. Additionally, the greater root:shoot ratio of pgl3-1 might confer a greater relative uptake of 15N. Thus, there is no evidence that the short-term rate of N assimilation is appreciably lower in N-starved pgl3-1 plants than in the wild type. This indicates that the OPPP may not be limiting for N uptake under N starvation.

Severity of pgl3 mutant phenotypes

The function of 6PGL proteins is the hydrolysis of 6-phosphoglucono-δ-lactone (Miclet et al., 2001). This compound is reported to be toxic due to its highly electrophilic nature, and although it also spontaneously hydrolyzes to 6-P-gluconate, 6-phosphoglucono-δ-lactone has a significant life span (60 min) in vivo (Miclet et al., 2001). However, photosynthetic efficiency and the abundance and cycling of storage carbon are not affected in pgl3-1, suggesting that if the lactone accumulates in the mutant, it does not compromise major plastid functions. 6-phosphoglucono-δ-lactone is unlikely to accumulate in the cytosol as there are four genes encoding cytosolic 6PGL. We think it more plausible that the pgl3-1 mutant phenotype is due to reduced function.

Null mutants of PGL3 are post-zygotically lethal, with abortion occurring most commonly at the globular to small heart shape stages (Meinke et al., 2008; see Lethality for loss of genes of the PPP and OPPP has been previously noted. For example, of the three ribulose-5-phosphate 3-epimerase genes encoded in the Arabidopsis genome, knockout of the sole plastid localized form (EMB2728) is lethal (Meinke et al., 2008), as is loss of the aforementioned GPT1, (Niewiadomski et al., 2005). Although the plastidial OPPP may contribute to synthetic reactions during seed filling (Schwender et al., 2003), loss of the pathway may be lethal during embryo development because, in non-photosynthetic plastids of developing embryos, loss of OPPP-derived NADPH compromises the capacity of the embryo to tolerate production of reactive oxygen species during chlorophyll synthesis (Andriotis et al., 2010).

Unfortunately, there are few genetic studies documenting the consequences to plant growth of reduced OPPP flux. In one study, Averill et al. (1998) showed that cytosolic 6PGDH knockout mutants of Zea mays, unlike the wild type, failed to induce the OPPP in response to nitrite. Moreover, nitrite was taken up at a slower rate than in the mutants. In tobacco, levels of leaf plastidial G6PDH (P2-isoform) were altered by the introduction of overexpression or antisense gene constructs (Debnam et al., 2004). Antisense lines with reduced leaf G6PDH activity exhibited increased levels of soluble sugars and sugar phosphates in leaves and displayed an increased abundance of reduced glutathione. In similar fashion, in the pgl3-1 mutant, reduced flux through the plastidial OPPP affects cellular redox state, resulting in a greater proportion of reduced glutathione in a larger total glutathione pool (Xiong et al., 2009). Perturbed redox poise may be expected to result in altered activity of the OPPP (and consequently regulation of N assimilation) because G6PDH itself is redox-regulated (Foyer et al., 2011). Indeed, alteration of OPPP activity in cytosol or plastid potentially impacts on the operation of the pathway in the alternate compartment. For example, Wakao et al. (2008) observed that although cytosolic G6PDH activity is likely to reductant (NADPH) for seed filling, g6pdh5 g6pdh6 double knockout plants exhibiting complete loss of cytosolic G6PDH activity actually had bigger seeds with relatively higher oil content than the wild type. They proposed this to be due to an increase in carbon substrates available for synthesis of fatty acids in plastids, implying increased transport of hexose phosphates into that organelle. In the pgl3-1 mutant, the opposite might be the case, with the presumably lower flux through the plastidial OPPP leaving more G6P substrate available to the cytosolic OPPP. Indeed, it seems possible that Xul-5-P, as a product of the cytosolic OPPP, could be diverted into the plastid via the XPT transporter (Eicks et al., 2002), thus bypassing the plastidial OPPP.

In conclusion, the OPPP mutant pgl3-1 displays reduced abundance of N metabolites combined with an altered transcriptional response to sugar stimulus. We propose that whilst the flow of C through the OPPP facilitates N homeostasis when both N and new C skeletons (through photosynthesis) are plentiful, in the face of N starvation, OPPP flux is reduced. Additionally, the observation that transcriptional differences between pgl3-1 and the wild type occurred in the root while metabolic effects were seen in the rosette raises questions about the signaling mechanisms controlling root/shoot communications, in particular to sustain an efficient N assimilation and a subsequently adapted plant growth. We hypothesize that expression of nitrate assimilation genes in roots is positively regulated by a signal emanating from OPPP activity in the plastid. Further studies employing knockdown OPPP mutants to modulate the flux through the pathway will be required to provide additional insights into the regulation of N assimilation by an OPPP-dependent flow of C. Establishing whether the OPPP-derived signal originates from plastids located in shoots or in roots will also be important to help in its subsequent identification.

Experimental procedures

Plant material and growth conditions

Arabidopsis thaliana seeds of ecotype Col-0 were obtained from the Nottingham Arabidopsis Stock Centre ( T-DNA insertion lines were identified from the SIGnAL database ( SAIL_89_B08, SAIL_1239_H12 (Sessions et al., 2002), SALK_005685 (pgl3-1), SALK_117592 (Alonso et al., 2003), and emb2024 (stock number CS16134) (McElver et al., 2001) were obtained from the Arabidopsis Biological Resource Center (ABRC; FLAG_219G10 (pgl3-2) was obtained from the Versailles collection (Samson et al., 2004), GK_160H04 from the GABI-Kat collection (Rosso et al., 2003), and SK978 from the Saskatoon collection (Robinson et al., 2009). SALK_005685 and FLAG_219G10 have been described previously (Xiong et al., 2009). Primers used for genotyping are detailed in Table S1. Genotyping of SAIL_89_B08, SAIL_1239_H12, emb2024, and FLAG_219G10 used appropriate T-DNA border primers in conjunction with ‘FLAG_219G10’ genomic primers, whilst SALK_005685, SALK_117592, GK_160H04, and SK978 utilized the T-DNA vector-specific border primers with ‘SALK_005685’ genomic primers.

Plants were grown in soil, on plates, or hydroponically. All plants were grown in growth rooms with light intensity of 120 μmol m−2 sec−1 and a temperature of 22°C during the light period and 16°C for the dark period. The soil mix contained 2:1 peat:vermiculite. Plates contained half-strength MS salts and 0.5 g l−1 2-(N-morpholine)-ethanesulfonic acid (MES) adjusted to pH 5.8 with KOH. Plates were supplemented with sucrose as indicated. The hydroponics set up was as described (Gibeaut et al., 1997; Heeg et al., 2008). Briefly, rock-wool plugs were inserted into black cut-off 1.5 ml microfuge tubes. These were placed in 250 ml, 24-well floating microtube racks (Astral, I5100-43, Each box contained four to six tubes and 160 ml modified half-strength Hoagland's liquid medium [2.0 mm Ca(NO3)2·4H2O, 2.0 mm KNO3, 0.5 mm NH4NO3, 0.5 mm MgSO4·6H2O, 0.5 mm KH2PO4, 50 μm KCl, 40 μm Fe-EDTA, 25 μm H3BO3, 2.0 μm MnCl2·4H2O, 2.0 μm ZnSO4·7H2O, 0.5 μm CuSO4·5H2O, 0.5 μm (NH4)6Mo7O24·4H2O, 0.15 μm CoCl2·6H2O, 0.5 g l−1 MES, pH 5.8], sufficient to wick through the rock wool and effect germination. These nutrient conditions are referred to as ‘N-replete’ and contain in total 7 mm N (6.5 mm inline image + 0.5 mm inline image). The hydroponic medium was changed every 2–3 days ensuring that nutrients were not limiting during vegetative growth of the plants. For sugar supplement experiments, plants were moved to new media containing 0 or 3% sucrose. All hydroponics experiments were done in a 12 h:12 h light:dark photoperiod and involved plants grown for 4–5 weeks, just prior to bolting. For N starvation experiments, N sources in the media were switched for chloride salts (and NH4NO3 removed). Plants were grown under N-replete conditions for 4 weeks in 12 h:12 h light:dark, switched to media lacking N for 7 days, then resupplied with 1 mm KNO3.


The predicted protein sequence of PGL3 (At5g24400) was analysed for antigenicity using the IEDB Analysis Resource ( Residues 220–978 of the coding sequence (corresponding to amino acids 73–325, the mature PGL3 protein) were amplified and cloned into pDONR207, then into pDEST17 for expression in BL21 (DE3) pLysS cells (Clontech, The expressed antigen was purified using a His-Tag column (Ni-NTA; Qiagen, and concentrated for inoculation of a rabbit (service provided by IMVS Adelaide, Non-reactivity of the pre-bleed and reactivity of the test bleed were confirmed prior to bleed-out. Western blotting was done onto a Hybond ECL membrane using a Bio-Rad mini-transfer cell ( and detection performed using a Western Breeze system (Invitrogen, with 1/2000 dilution of the 1° antibody. α-Tubulin (Sigma T-5168, was used as a control (1/1000 dilution).

Localization of GFP

The PGL3-GFP was made by triple template PCR (Tian et al., 2004), with GFP inserted into the coding sequences of PGL3 at a position corresponding to 13 amino acid residues upstream from the stop codon. This construct was inserted by Gateway cloning into pGREEN180A, a pGREEN179 vector modified by insertion of a Gateway A cassette (Wiszniewski et al., 2009). Sequences of PGL3-TTGFP cloning primers P1, P2, P3, and P4 are given in Table S1. Cells from a long-term dark-grown Arabidopsis protoplast culture (Millar et al., 2001) were isolated 5 days after subculturing onto on 55 mm diameter Whatman filter paper by gentle vacuum via a Buchner funnel. The cells on the filter paper were placed on plates containing 1 ×  MS salts, 3% (w/v) sucrose, 0.2 m Mannitol, 0.5 μg ml−1 NAA, 50 ng ml−1 kinetin, and 0.6% agar. Plasmids were coated onto 1 μm gold particles and biolistic bombardment of the protoplast cells immobilized on the filter paper was done using a PDS-1000 particle delivery system (Bio-Rad) as described (Thirkettle-Watts et al., 2003). Cells were examined by fluorescence microscopy 24–48 h after bombardment.


Complementation of pgl3-1 and pgl3-2 was attempted with a variety of constructs. Coding sequences corresponding to the full-length protein (primers P1 and P4 from Table S1), or truncated variants containing either the plastid-targeting (primers P1 and cp_P4) or the peroxisome PTS1-targeting (primers px_P1 and P4) signals were cloned by Gateway (Invitrogen) methods into pDONR207 and then into pGREEN180A. These constructs were introduced into pgl3-1 and pgl3-2 by floral dipping (Clough and Bent, 1998). Transformants were selected on hygromycin and tested for their capacity to complement the mutants.

Quantitative RT-PCR

The RNA extraction and qRT-PCR were done as described previously (Wiszniewski et al., 2012). CACS and YLS8 were used as internal reference genes for the calculation of relative transcript abundance. The two reference genes yielded comparable relative abundances of the different transcripts so data presented all use CACS as the reference gene.

Metabolite profiling

Starch staining was done by harvesting rosettes into 80:20 ethanol:water in 50 ml tubes. These were immersed in a boiling water bath until tissues were clarified. Rosettes were stained for 10 min in ¼ dilution of Lugol (Sigma L6146), rinsed with water and photographed. Ammonium (Brautigam et al., 2007) and nitrate (Cataldo et al., 1975) were assayed using a spectrophotometer (Shimadzu UV-1800, in 1 ml reaction volumes.

Amino and organic acids were analyzed by gas chromatography (GC)-MS (Agilent 7890A gas chromatograph coupled with a 5975C Inert XL mass selective detector (MSD) with triple-axis detector; as tert-butyldimethylsilyl derivatives as described by Keech et al. (2012).

To assay incorporation of 15N into amino acids, N-starved plants (see above) were resupplied with K15NO3 (95.23% 15N). Plant tissue was frozen and ground in liquid nitrogen. Amino acids were extracted from 30 mg of ground tissue in 500 μl of extraction buffer (20:3 methanol:water) heated on a Thermomixer with shaking at 13 000 g at 20°C for 20 min. Cell debris was removed by centrifugation for 3 min at room temperature. For each sample 100 μl of the supernatant was transferred to a glass sample preparation vial. Solid phase extraction (SPE) and derivatization of amino acids was done with an EZ:faast For Free (Physiological) Amino Acid Analysis kit as described in the manual (Phenomenex KGO-7166, For GC-MS an Agilent 6890N gas chromatograph coupled with a 5975 Inert MSD was used. The GC-MS was fitted with a 10 m × 0.25 mm ZB-AAA column and instrument settings and sample injections were as described in the EZ:faast manual. Fragments in the range of 45–450 atomic mass units were analyzed after 1 min solvent delay. The GC-MS data were processed by MSD ChemStation. Amino acid derivatives were identified by comparison of retention times and mass spectra of authentic standards. The amount of each amino acid was calculated using the extracted ion current signal of a metabolite-specific ion fragment. Each peak was normalized to the Norvaline internal standard and the tissue weight. A quantitative statement was achieved by creating a five-point calibration curve made from a standard mix containing all measured amino acids (range 5–400 nm). Calibration curve samples were prepared and run in the same way as experimental samples.

Fragments used for data analysis only contained one amino group, hence the specific ion fragment(N15 labeled) = specific ion fragment(N14 labeled) + 1. Controls were used to determine the natural 14N/15N contribution.

Photosynthetic parameters

All fluorescence measurements were done using an IMAGING-PAM M-Series (Walz, Germany). A dark–light/recovery induction curve was recorded using the standard settings of the manufacturer. Actinic light was set at 186 μmol quanta m−2 sec−1, data points were recorded every 20 sec, and the duration time of experiments was set to 600 sec. Leaves/plants grown in short days (8 h light/16 h dark) were dark-adapted for 30 min prior fluorescence measurements. F0 was set at 0.8 and values for the maximum quantum yield in the dark-adapted state (Fv/Fm), the photochemical quenching efficiency (qL), the regulated (Y(NPQ)) and non-regulated (Y(NO)) non-photochemical quenching were calculated using the software ImagingWin (Version 2.32, Walz,


This work was supported by the Australian Research Council (grant numbers FF0457721 and CE0561495), the Western Australian Government's Centre of Excellence Program and the Swedish Foundation for International Cooperation in Research and Higher Education (STINT).