Recent studies using transcript and metabolite profiles of wild-type and gene deletion mutants revealed that photorespiratory pathways are essential for the growth of Synechocystis sp. PCC 6803 under atmospheric conditions. Pool size changes of primary metabolites, such as glycine and glycolate, indicated a link to photorespiration.
The 13C labelling kinetics of primary metabolites were analysed in photoautotrophically grown cultures of Synechocystis sp. PCC 6803 by gas chromatography-mass spectrometry (GC-MS) to demonstrate the link with photorespiration. Cells pre-acclimated to high CO2 (5%, HC) or limited CO2 (0.035%, LC) conditions were pulse-labelled under very high (2% w/w) 13C-NaHCO3 (VHC) conditions followed by treatment with ambient 12C at HC and LC conditions, respectively. The 13C enrichment, relative changes in pool size, and 13C flux of selected metabolites were evaluated. We demonstrate two major paths of CO2 assimilation via Rubisco in Synechocystis, i.e., from 3PGA via PEP to aspartate, malate and citrate or, to a lesser extent, from 3PGA via glucose-6-phosphate to sucrose. The results reveal evidence of carbon channelling from 3PGA to the PEP pool. Furthermore, 13C labelling of glycolate was observed under conditions thought to suppress photorespiration. Using the glycolate-accumulating ΔglcD1 mutant, we demonstrate enhanced 13C partitioning into the glycolate pool under conditions favouring photorespiration and enhanced 13C partitioning into the glycine pool of the glycine-accumulating ΔgcvT mutant. Under LC conditions, the photorespiratory mutants ΔglcD1 and ΔgcvT showed enhanced activity of the additional carbon-fixing PEP carboxylase pathway.
Citation: Huege J, Goetze J, Schwarz D, Bauwe H, Hagemann M, Kopka J (2011) Modulation of the Major Paths of Carbon in Photorespiratory Mutants of Synechocystis. PLoS ONE 6(1): e16278. https://doi.org/10.1371/journal.pone.0016278
Editor: Tanya Parish, Queen Mary University of London, United Kingdom
Received: September 27, 2010; Accepted: December 9, 2010; Published: January 21, 2011
Copyright: © 2011 Huege et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The work was supported by a grant from the DFG (Deutsche Forschungsgemeinschaft) to M.H. and J.K. and by the Max Planck Society. The development of the CORRECTOR software tool was supported in part by the European META-PHOR project, FOOD-CT-2006-036220, and the GoFORSYS project (http://www.goforsys.de/) funded by the Bundesministerium für Bildung und Forschung (BMBF). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Cyanobacteria are considered the first organisms to have evolved the capacity for oxygenic photosynthesis around three billion years ago . The endosymbiotic uptake of an ancient cyanobacterial ancestor by a eukaryotic cell initiated the evolution of phototrophic algae and plants. Many of the initial cyanobacterial proteins are still detectable within the chloroplasts and nuclear genomes of current higher plants , . In both cyanobacteria and C3 plants, CO2 fixation is primarily catalysed by the enzyme ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco). The carboxylase reaction generates two molecules of 3-phosphoglycerate (3PGA) from ribulose-1,5-bisphosphate and CO2, whereas O2 competition at the reaction centre leads to the oxygenase products 3PGA and 2-phosphoglycolate (2PG) . The product 2PG is a cellular toxin that needs to be detoxified, as it inhibits Calvin-Cycle enzymes –. In plants, 2PG is scavenged by a sequence of reactions called the photorespiratory C2 pathway –, which regenerates one molecule of 3PGA for every two molecules of 2PG, at the cost of CO2 and NH4+ release.
In contrast to higher plants, 2PG metabolism was thought to exert negligible effects in cyanobacteria. Early studies only indicated the formation of glycolate from 2PG . Furthermore, the discovery of a sophisticated inorganic carbon (Ci) concentration mechanism (CCM) demonstrated the potential of cyanobacteria to increase the internal concentration of CO2 in the vicinity of Rubisco and thus to compensate for the low CO2 affinity of the cyanobacterial enzyme . As a consequence, the CCM was thought to be sufficient to suppress the oxygenase reaction and to make photorespiratory detoxification irrelevant for cyanobacterial metabolism.
Recent studies, however, demonstrated not only that the CCM is insufficient to prevent ribulose-1,5-bisphosphate oxygenation in an O2-containing atmosphere but also that there is active 2PG metabolism. The photorespiratory pathways were found to be essential for growth under atmospheric conditions , . Photorespiratory 2PG metabolism in the Synechocystis sp. strain PCC 6803 (hereafter Synechocystis) comprises three alternative paths: a plant-like C2 cycle, a bacterial glycerate pathway, and the complete decarboxylation of glyoxylate via oxalate. Mutants defective in specific enzymes within these pathways including glycolate dehydrogenase (ΔglcD1, sll0404) and glycine decarboxylase (ΔgcvT, sll0171) displayed growth retardation. In addition, intermediates of photorespiratory 2PG metabolism accumulated in agreement with the enzymatic defect; these intermediates included glycolate in ΔglcD1 and glycine in ΔgcvT. These phenomena were already apparent under elevated CO2 conditions (5%, high CO2, HC), and were enhanced under the low-CO2-availability conditions of ambient air (∼0.035%, low CO2, LC) , . The Ci affinity and the maximum rate of photosynthesis were found to be higher in cells grown under LC conditions than in cells grown under HC conditions .
Transcript and metabolite profiles of the wild-type (WT) strain compared to the above-mentioned mutants indicated that the photorespiratory mutants, when grown under HC conditions, display gene expression and metabolite patterns that are characteristic for WT cells when shifted from HC to LC conditions , . Characterisation of changes of metabolite pools opened the path towards currently unanswered questions, including: (1) Is the previously observed accumulation of intermediates, such as glycolate and glycine, indeed linked to photorespiration? (2) How is the ratio of Rubisco oxygenation to carboxylation influenced by altered Ci availability? (3) Which routes of carbon fixation are used under HC or LC conditions? To approach these questions, we applied a non-steady state 13Ci pulse and chase experimental design similar to work previously published by our lab . The appearance and dilution of 13C labelled intermediates within the primary metabolism of photoautotrophically grown WT and selected mutant cells was monitored following the suggestions made for dynamic flux analyses . With this approach, we reveal the use and modulation of the major paths of carbon assimilation in Synechocystis.
Experimental design of metabolic flux analysis
Pulse labelling of photoautotrophic Synechocystis cultures was performed by adding aliquots of a saturated solution of 13C labelled NaHCO3 to a final concentration of 2% (w/w). This procedure resulted in a very high carbon (VHC) pulse and was chosen to ensure a step change with the highest 13C enrichment possible. Moreover, the VHC conditions should suppress the oxygenase activity of Rubisco. We combined the 13Ci pulse of stably labelled bicarbonate with a chase using unlabelled CO2. Our experimental procedure generated an optimal rectangular step change during the 13Ci pulse and sufficient enrichment for short isotope dilution times of 10–60 min (Figure 1). In the following study, we focused on the metabolite pools that reached high 13C enrichment and thus allowed optimal GC-MS-based analysis. The VHC pulse was applied to cells that pre-acclimated to 5% CO2 (HC) or to 0.035% CO2 (LC) conditions, as previously reported . For the chase, a quick medium exchange was performed. The cells were subsequently incubated under continuous aeration with either HC or LC identical to the initial pre-acclimation. Our previous work on metabolic profiling revealed highly reproducible metabolic patterns after HC or LC acclimation and indicated that near-steady-state conditions could be achieved using standardised batch cultivation .
The cells were monitored by combined pool size and 13C enrichment analyses of (A) 3-phosphoglycerate (3PGA) and (B) phosphoenolpyruvate (PEP). The wild-type strain Synechocystis sp. PCC 6803 was probed after pre-acclimation to HC or LC conditions using a very high (2% w/w) 13C-NaHCO3 (VHC) pulse and a chase following media exchange and aeration with 5% CO2 (HC) or 0.035% CO2 (LC) of ambient isotope composition (cf. supplementary Methods S1). HC and LC chase conditions were the same as the respective pre-acclimation conditions. Arrows indicate initial sampling at ∼0.5 min after label exchange for pulse or chase. The pool size behaviour (top) and 13C enrichment data (bottom) are presented.
To analyse changes in carbon partitioning within the primary metabolism, we combined metabolic flux and pool size assessments as previously suggested . The metabolomic data were extended by parameters derived from mass isotopomer distribution analyses, i.e., the initial rate of 13C accumulation during the first 10 min and the maximum 13C enrichment at 20–60 min. Both parameters were calculated from the transient labelling kinetics of each of the monitored metabolite pools. Based on the analysis of 3PGA as the main entry point of the 13C label in photoautotrophic cultures of Synechocystis, we estimated the general features of the chosen experimental design. The 3PGA levels did not change after pre-acclimation to HC and LC or throughout the VHC pulse but increased clearly upon initiation of both the HC and LC chase (Figure 1A). As both chase conditions represented a shift to lower Ci availability, these observations were in agreement with the previously observed increase of the 3PGA pool 3 h after a shift from HC to LC . In parallel, the transient 13C labelling of the 3PGA pool was assessed during pulse and chase. Under our labelling conditions, the 3PGA pool reached its maximum saturated 13C enrichment 10 min after the pulse in both the HC and LC acclimated cells. LC cells showed a more rapid 13C accumulation. During the chase period with HC conditions, labelling returned below 10% within 30 min. Under LC conditions, a delayed chase response was observed. As expected, the empirical mass distribution vectors of 3PGA and PEP indicated homogenous 13C labelling. The kinetic behaviour of the mass distribution vectors (Figure S1) was generally in agreement with the modelled predictions reported earlier .
The presence of unlabelled ambient CO2 resulted in a perceptible isotope dilution compared to the 98% enrichment of the applied NaHCO3. For example, the final 13C enrichment in the 3PGA pool was on average >82% and >89% for the HC and LC conditions, respectively (Table S1). Moreover, the label exchange for the chase phase indicated a step change only for the HC condition. Under LC conditions, the label exchange was clearly delayed (Figure 1). Dilution and carryover effects at chase initiation were unavoidable in our hands, especially when implementing the LC chase conditions. Therefore, the slow return of 13C label in 3PGA could be caused by both an insufficient physical dilution of 13C label and by a physiological effect of the CCM, which is activated under LC conditions but suppressed under HC conditions . To account for these effects, we used the enrichment data of the first assimilation products, for example 3PGA, to correct the influence of ambient CO2 and any potential label carryover.
In this study, we analysed the pool size and labelling kinetics as 13C enrichment parameters of the observed metabolite pools. Each of these parameters will first be reported in context of the HC versus LC pre-acclimation for the WT strain. Subsequently, the respective phenotypes of the mutant strains will be presented. We focused on the mutants ΔglcD1 (Δsll0404), which is deficient for one of the two isoforms of glycolate dehydrogenase, and ΔgcvT (Δsll0171), deficient for the T-protein subunit of the glycine decarboxylase complex.
The entry points of the 13CO2 label: 3PGA and PEP
The two most rapidly and completely labelled metabolite pools were the 3PGA pool, as expected, and, unexpectedly, the PEP pool (Figure 1). The 13C enrichment was higher and the rate of 13C labelling was faster in the PEP pool compared to the 3PGA pool under all conditions and in all analysed mutants. The Student's t-test of the differences in 13C enrichment at 20–60 min after the pulse and in the initial 13C accumulation rate at 0.5–10.0 min showed that these differences were highly significant in both cases, p = 0.00008 and p = 0.050, respectively. These results are documented in more detail by the PEP/3PGA ratios of both measurements, which were consistently >1.0 (Table 1). The kinetic analysis of the PEP/3PGA ratio of 13C enrichment (Figure 2) demonstrated that the PEP pool was labelled faster in the pulse phase and was also de-labelled faster in the chase phases. This observation coincided with a general increase in both pool sizes upon shift from VHC during the pulse phase to HC or LC during the chase phase (Figure 1). The 3PGA pool increased by more than the PEP pool, as demonstrated by a reduction of the PEP/3PGA pool size ratio from above 1 to about 0.15 (Figure 2).
The pulse and chase conditions used in HC and LC acclimated Synechocystis sp. PCC 6803 wild-type were identical to the conditions reported in the legend of Figure 1. The initiation of the pulse (0 min) and chase (60 min) are indicated by arrows.
Cells pre-acclimated to the HC or LC conditions exhibited several characteristic differences in 3PGA and PEP labelling, but the initial rate of 13C accumulation and the maximum 13C enrichment of both metabolites remained strictly correlated during the pulse phase. Firstly, under LC conditions, the initial rates of 13C accumulation in both metabolite pools were faster by a factor of ∼1.3 as compared to HC conditions. Also the final 13C enrichment was ∼1.1-fold higher (cf. Table 1) with LC conditions. Secondly, the chase phase revealed a substantial difference. While metabolites in cells from both conditions behaved according to the assumption of a rectangular step change after the pulse, only metabolites under the HC chase conditions appeared to follow this assumption after the chase. In contrast, the LC cultures retained high 13C enrichment during the first 20 min of the chase until the onset of a delayed and slow de-labelling. This irregular behaviour probably resulted either from the physical carryover effect of 13C label from the preceding VHC pulse, as reported above, or from the intracellular CCM activity of Synechocystis, which is highly active under LC conditions  and should favour the re-assimilation of the 13CO2 generated from internal sources.
The 3PGA and PEP labelling of the glycine decarboxylase-deficient ΔgcvT mutant strain was similar to that of the WT strain. In contrast, the glycolate dehydrogenase mutant, ΔglcD1, showed large differences in 3PGA labelling but only slight changes in PEP 13C enrichment kinetics. Specifically, the rate of 13C accumulation and the final 13C enrichment in the 3PGA pool were lowest in the ΔglcD1 mutant (Table 1). This observation could be explained by the toxic effects of mutant-specific glycolate accumulation on Calvin-Benson cycle activities. These findings were in agreement with previous observations of reduced photosynthetic and growth rates for glycolate dehydrogenase-deficient mutants .
Monitoring the flux into the glucose-6-phosphate and sucrose pools
To analyse the downstream fate of the initial labelling products, 13C partitioning into the glucose, glucose-6-phosphate (G6P) and sucrose pools was monitored. The maximum 13C enrichment of the glucose pool was unchanged, i.p. 16% for LC (n = 2) and 13% for HC (n = 2). In contrast, the pools of both G6P and sucrose were rapidly labelled during the pulse and de-labelled during the chase. The maximum 13C enrichment of G6P was 82% for LC (n = 3) and 76% for HC (n = 3). The maximum 13C enrichment of sucrose was slightly lower at 68% for LC (n = 3) and 65% for HC (n = 3). Sucrose labelling was slightly delayed compared to G6P after the pulse, and sucrose de-labelling was consistently more rapid under the LC chase conditions (Figure S2). The differences between G6P and sucrose were less obvious under the HC chase conditions. Taken together these observations indicated a carbon flux from 3PGA via G6P towards sucrose synthesis when Synechocystis was exposed to excess Ci during the VHC pulse conditions. Furthermore, sucrose was mobilised when the cells were shifted back to lower carbon conditions. The sucrose flux appeared to represent one of the primary paths of photosynthetic CO2 assimilation under our conditions. The sucrose flux results (Table 2) were in agreement with earlier findings , that the acclimation of Synechocystis to the LC condition leads to a higher rate of photosynthesis and sucrose flux than acclimation to HC.
Monitoring photorespiratory flux: Glycolate, glycine and serine
In our study, glycolate was the first photorespiratory intermediate in the pathway that was amenable to 13C tracing analysis. The 2PG pool was typically at or below detection limits. As was expected of WT cells, the initial rate of 13C accumulation and the maximum 13C enrichment of the glycolate pool were low under VHC pulse conditions. Nevertheless, even under these conditions, we formally demonstrated 13C labelling of the glycolate pool with ∼0.6 atom% min−1 for HC conditions and up to ∼1.4 atom% min−1 for LC conditions. The glycolate/3PGA ratios of these measurements were ∼0.09-fold and ∼0.18-fold, respectively (Table S1A). Under the HC and LC chase conditions, however, the glycolate pool dropped below the detection limits in WT cells.
The ΔglcD1 mutant had previously been shown to exhibit a large increase in the glycolate pool, as shown by Eisenhut et al. , and therefore allowed detailed analysis of glycolate labelling. The glycolate pool size in the ΔglcD1 mutant responded to the reduced CO2 availability under HC and LC chase conditions (Figure 3). Initially, 13C rapidly partitioned into the constantly increasing glycolate pool. To compare the HC and LC chase responses, the kinetics of the glycolate/3PGA 13C enrichment ratio were analysed in the ΔglcD1 mutant. If glycolate were predominantly or exclusively generated by the oxygenase activity of Rubisco, we would expect nearly identical glycolate/3PGA ratios under both HC and LC conditions, even though the apparent chase kinetics of 3PGA 13C enrichment were different. This expectation is shown to be true in Figure 3 (bottom).
The data were acquired during the pulse and chase phases under HC and LC conditions with the glycolate-accumulating ΔglcD1 (Δsll0404) mutant. Note the strong increase in the glycolate/3PGA 13C enrichment ratio under both the HC and LC conditions (cf. Table S1). The initiation of the pulse (0 min) and chase (60 min) are indicated by arrows.
Furthermore, we attempted to quantify the flux of carbon into the glycolate and 3PGA pools under LC and HC chase conditions in the ΔglcD1 mutant. The glycolate turnover time was shorter than the 3PGA turnover time under LC conditions; specifically, it was 2.9 min (glycolate; k = 0.3337 min−1) compared to 90.9 min (3PGA; k = 0.0062 min−1). In contrast, the turnover times were in the same order of magnitude under HC conditions, i.e., 39.8 min (glycolate; k = 0.0251 min−1) and 25.1 min (3PGA; k = 0.0399 min−1). Taking the pool sizes of both metabolites into consideration, we calculated the molar rates of appearance under chase conditions, which may serve as estimates of the rates of oxygenation (glycolate) and carboxylation (3PGA). Under the LC condition, the molar rates of appearance were 4.80 * 10−6 µmol min−1 µg−1 Chl and 1.21 * 10−6 µmol min−1 µg−1 Chl for glycolate and 3PGA, respectively. The ratio of the molar rates of appearance was ∼4.0 (glycolate/3PGA). Synechocystis cells in the HC condition showed a 100-fold lower ratio (∼0.04 glycolate/3PGA). The molar rates of appearance were 4.27 * 10−7 µmol min−1 µg−1 Chl and 1.18 * 10−5 µmol min−1 µg−1 Chl for glycolate and 3PGA, respectively.
Moreover, the photorespiratory pathway was monitored with respect to the 13C enrichment in the glycine, serine, and glycerate pools. Again, ratios of 13C enrichment to 3PGA were used because this approach corrects for the differences in 3PGA labelling among the different experiments (Figure 4). The serine 13C enrichment ratios of WT cells were higher than the glycine 13C enrichment ratios throughout the VHC pulse (Figure 4, left panel). The 13C enrichment ratios of glycerate were equal to or less than the 13C enrichment ratios of serine (data not shown). These observations were consistent under all conditions and in all genotypes (Figure 4). Our observations thus indicated that the synthesis of serine from glycine was minimal under VHC conditions, while the conversion of glycerate to serine or the opposite reaction seemed to occur under HC and LC conditions.
Pulse and chase phases under HC conditions (top) and LC conditions (bottom) were used with the wild-type (left), the ΔgcvT (Δsll0171) mutant (middle) and the ΔglcD1 (Δsll0404) mutant (right) strains. Glycolate data from the ΔglcD1 mutant are included in the right panel. The initiation of the pulse (0 min) and chase (60 min) are indicated by arrows.
During the chase periods, the glycolate-accumulating ΔglcD1 mutant showed a transient increase in glycine and serine 13C enrichment ratios that was similar to the effect observed in the WT (Figure 4, right panel). The ΔgcvT mutant, which is deficient in glycine decarboxylase and is known to accumulate glycine under LC conditions , showed the expected enhanced accumulation of the 13C label in the glycine pool under LC chase conditions (Figure 4, central panel).
Monitoring PEP utilization: Aspartate, malate and citrate
Oxaloacetate, the product of PEP carboxylase activity, was not measured in our study. Instead PEP utilisation was monitored by assessing the kinetics of 13C enrichment in three metabolic oxaloacetate products: aspartate, malate, and citrate (Figure 5). To normalise the 13C labelling of these pools, we calculated ratios of the initial rate of 13C accumulation and maximum 13C enrichment with respect to PEP instead of the previously chosen 3PGA (Table S1B).
Data were obtained under the HC and a LC conditions with Synechocystis sp. PCC 6803 wild-type cultures. During the 13Ci-VHC pulse with LC acclimated cells, a transient mobilisation of an internal carbon source of ambient mass isotopomer composition was apparent (marked by the arrow within the lower panel). Carbon of ambient mass isotopomer composition appears to enter the aspartate pool first and to persist longest within the citrate pool. The initiation of the pulse (0 min) and chase (60 min) are indicated by arrows.
During the VHC pulse, 13C accumulation was highest in the citrate pool, followed by the malate and aspartate pools. Unexpectedly, the order changed over the monitored time (Figure 5). At the onset of the pulse phase, 13C enrichment was highest in the aspartate pool, followed by the malate and citrate pools. The reversal of 13C enrichment was noticeable in cells pre-acclimated to HC and became obvious under LC conditions. The LC condition also demonstrated the mobilisation of unlabelled carbon during the early phase of the VHC pulse. This internal carbon source first entered the aspartate pool (Figure 5, arrow) and appeared to persist for the longest time in the citrate pool. Under the same conditions, the ornithine analyte showed essentially the same internal carbon mobilisation effect as aspartate but with lower 13C enrichment (data not shown). Note that due to chemical conversions during the derivatisation of metabolites for GC-MS analysis, the measured ornithine analyte represents the sum of the ornithine, arginine, and citrulline pools.
Indications of additional 13C fixation through PEP carboxylase activity were found by analysing the ratios of the initial rates of 13C accumulation for the various compounds with respect to PEP (Table S1B). In WT cells, the 3PGA/PEP, citrate/PEP, and malate/PEP ratios of the initial rates of 13C accumulation were ∼0.96, ∼0.78, and ∼0.62, respectively. These ratios were not influenced by pre-acclimation of the WT cells to the HC or LC conditions. The aspartate/PEP ratio was, however, increased by 1.34 fold under the LC condition (0.65) compared to the HC condition (0.48). This LC acclimation effect was even more apparent in the two photorespiratory mutants, ΔglcD1 and ΔgcvT, with aspartate/PEP ratios approximating 1 under LC conditions in both mutants. A similar but less extreme effect in the mutants was observed for the malate/PEP ratios. These ratios increased by ∼1.3-fold in cells pre-acclimated to LC compared to those receiving HC. The citrate/PEP ratios showed the same tendency under the LC condition and were only marginally increased in the mutant strains compared to WT. Under HC, the citrate/PEP ratios were, however, reduced in both mutants compared to WT.
The major paths of carbon assimilation in Synechocystis
Our data were essentially consistent with the established fact that 3PGA is the first assimilation product of the Calvin-Benson cycle. Rapid labelling of 3PGA with 13C suggests that C3-photosynthesis is the dominant carbon fixation pathway in Synechocystis. Fixation of 13Ci via PEP carboxylase seems to be less important, but occured in LC acclimated cells (see below). Following the dilution of the initial 13C label of 3PGA into the metabolite pools that were measured by our GC-MS-based technology, we observed two major branches of carbon flow (Figure 6). The first branch leads from 3PGA to sucrose via G6P with minimal leakage of carbon into the free glucose pool. The flux of carbon from 3PGA through this branch into sucrose appears to reflect the photosynthesis rate of the respective pre-acclimated cells  because the enrichment of the sucrose pool is higher in LC cells than in HC cells (cf. Table 2). The observed conversion of 3-PGA into sucrose and probably also into glycogen shows that surplus carbon is converted into storage compounds. The second major path of carbon appears to lead to aspartate, malate, and citrate (Figure 6) via PEP and probably further into the amino acid pools via 2-oxo-glutarate. This second branch of 3PGA utilisation directly fuels the biosynthesis of primary metabolites, specifically amino acids that are generated via the incomplete tricarboxylic acid cycle of Synechocystis. In agreement with the interruption of the canonical tricarboxylic acid cycle in Synechocystis, fumarate and succinate were labelled more slowly than malate (data not shown).
Ellipses highlight the 13C labelled metabolites detected and evaluated in this study. The enzymatic steps affected in the ΔglcD and ΔgcvT mutants are marked by boxes.
We further conclude that C4 carboxylation via PEP carboxylase also occurs under our experimental conditions, specifically in LC acclimated cells. The joint action of C3 and C4 carboxylation via Rubisco and PEP carboxylase, respectively, has been revealed by 14C labelling experiments using different cyanobacterial strains, as shown by Döhler . However, the relative ratio is still a matter of discussion. The oxaloacetate product, which cannot directly be measured by our current technique, appears to be rapidly converted into aspartate and malate via the aspartate aminotransferase and malate dehydrogenase reactions. Of the potential oxaloacetate products, citrate is the most rapidly labelled, possibly because it receives part of its carbon backbone via acetyl-CoA. Acetyl-CoA should be highly labelled because it is generated from the rapidly labelled PEP pool via the final steps of the glycolytic pathway. However, as judged by the rates of 13C accumulation, the C3 carboxylating pathway from 3PGA via PEP appears to be the main route to transfer newly fixed organic carbon into the central metabolism of Synechocystis. The 13C enrichment of oxaloacetate products shows that primary 13Ci fixation via Rubisco is dominant under our experimental conditions.
One of the unexpected results was the observation that first aspartate and then malate and citrate receive 12C from internal stores during the 13Ci-VHC pulse (Figure 5). We interpret this observation as a carbon mobilisation from internal sources into the aspartate and ornithine/arginine/citrulline pools. The source could be fast cyanophycin degradation. Cyanophycin is a storage polymer in cyanobacteria that has a poly-aspartate backbone and arginine side chains. Cyanophycin accumulates mainly as a response to excess nitrogen in Synechocystis . The VHC pulse results in an environment characterised by high Ci availability and relatively low nitrogen. This should apply especially to LC cells, which were pre-acclimated to rather high C/N conditions. Cyanophycin degradation mobilises stored nitrogen and may ultimately lead to rebalancing of the intercellular C/N ratio. Such a mobilisation process should be considered in future studies that aim at dissecting the relative contribution of C3 and C4 metabolism to the biosynthesis of organic acid intermediates such as citrate.
Indications for carbon channelling from 3PGA into the PEP pool
Had Calvin and Benson ,  used Synechocystis to unravel the photosynthetic carbon cycle, PEP would have likely been misinterpreted as the first assimilation product of the carboxylation reactions. We found clear evidence that the PEP pool of Synechocystis is not only more rapidly labelled but also has a slightly but significantly higher final 13C enrichment than the 3PGA pool under all of our experimental conditions (Table 1, Figure 2). Even today, this phenomenon could be interpreted as evidence for the presence of alternative CO2 fixation mechanisms in cyanobacteria. Such pathways for CO2 fixation, e.g., the reductive tricarboxylic acid cycle or the carbamoylphosphate synthetase pathway, have previously been suggested to be operative in cyanobacteria . Both paths have been implicated for ancillary CO2 fixation. In view of the incomplete tricarboxylic acid cycle in Synechocystis and considering our data, i.e., the lower rates of 13C accumulation in the citrate and malate pools compared to the PEP pool, we consider it unlikely that under our growth conditions Synechocystis utilizes any of the above ancillary routes of CO2 fixation or the hydroxypropionate pathway for CO2 fixation, which was recently discovered in Chloroflexus aurantiacus . This view is supported by the absence of genes for such alternative C fixation pathways from the complete genome sequence of Synechocystis (CyanoBase at http://genome.kazusa.or.jp/cyanobase/Synechocystis).
Instead, we favour the interpretation that the majority of newly assimilated carbon is channelled from 3PGA to PEP, and the much smaller PEP pool is more quickly saturated with 13C. With this interpretation, we need to assume that two sub-pools of 3PGA exist in Synechocystis. The carboxysome, a cyanobacterial micro-compartment that is a component of the CCM and harbours most of the cellular Rubisco, may well be a rapidly labelled 3PGA pool from which PEP is produced. The second 3PGA pool may be cytosolic. The cytosolic 3PGA pool could originate from a second pool of Rubisco. Rubisco was recently indicated to exist outside of the carboxysome but attached to thylakoids and in close association with other Calvin-Benson cycle enzymes . Carbon should be channelled into PEP either within the carboxysomes or along with the export of 3PGA from the carboxysome into the cytosol. The cytosolic 3PGA pool may well be diluted with unlabelled carbon as a result of the turnover or mobilisation of previously assimilated soluble sugars or glycogen. Channelling into PEP may also indicate that the PEP carboxylase reaction for Synechocystis is more important than previously described . This conclusion is well supported by the rapid accumulation of assimilated carbon in pools downstream of oxaloacetate (cf. discussion above) and may explain enigmatic early reports of high accumulations of label in unexpected metabolite pools, such as the aspartate pool , .
Formal demonstration of photorespiratory flux
We detected a significant partitioning of the 13C label into the glycolate pool, primarily in the ΔglcD1 mutant, but also in WT cells under our VHC pulse conditions. Thus we show that even when Synechocystis cells are exposed to very high Ci concentrations, the oxygenation reaction of Rubisco is not completely suppressed. This finding is in agreement with the reduced growth and photosynthesis caused by glycolate accumulation in cells of the ΔglcD1 mutant grown under HC conditions .
The glycolate-accumulating ΔglcD1 mutant and the glycine-accumulating ΔgcvT mutant proved to be valuable tools to generate a detailed demonstration of photorepiratory flux (Figure 6). The ΔglcD1 mutant allowed the demonstration of enhanced carbon flux into the glycolate pool after a shift to lowered Ci during the chase phase (Figure 3). In addition, this mutant enabled an estimation of the rates of oxygenation and carboxylation under HC and LC chase conditions by quantification of the molar rates of 13C appearance in the glycolate pool compared to the 3PGA pool. An enhanced photorespiratory flux was found under LC conditions compared to HC conditions. In contrast, WT cells displayed a very low photorespiratory flux under ambient CO2 conditions, which was consistent with previous reports on the quantitatively low importance of photorespiratory 2PG metabolism .
The ΔgcvT mutant demonstrated enhanced 13C flux into the glycine pool under chase conditions. This evidence may explain the coinciding increase in glycine levels (Figure 4), which is possibly linked to the increased production of glycine and reduced glycine utilisation of this slow-growing mutant .
Mutants and WT compared under HC and LC conditions
HC and LC acclimated Synechocystis cells were labelled with 13C to assess the differential acclimation to high and limiting Ci availability. Labelling of 3PGA and sucrose confirmed previous observations on the respective photosynthesis rates, i.e., that HC acclimation resulted in lower maximal assimilation rates than LC acclimation, and that ΔglcD mutants appear to be partially deficient in photosynthesis. Moreover, HC acclimation was linked to greater channelling of assimilate into the PEP pool (Figure 2). The apparent differences between the LC and HC acclimations were most clearly assessed when comparing relative 13C enrichment after normalisation to either 3PGA or PEP in the different experiments (Figures 4 and 6). The ratios demonstrated only minor differences of 13C partitioning into the glycolate pool after a HC or LC chase in the ΔglcD1 mutant. In addition, flux through PEP carboxylase appeared to be slightly affected by the different conditions in WT cells. Only an increased aspartate/PEP-labelling ratio was observed in LC acclimated WT cells (Table S1B), which indicates PEP carboxylase activation under LC conditions. In contrast, in addition to the aspartate/PEP ratio, the mutants exhibited increased malate/PEP and citrate/PEP ratios for the initial rates of 13C accumulation, which is consistent with an enhanced flux through the PEP carboxylation reaction under LC conditions. The enhanced 13Ci fixation through PEP carboxylase in both photorespiratory mutants, and to a lesser extent in WT cells, could indicate that this alternative carboxylating enzyme is activated in LC cells to compensate for the inhibitory effects of glycolate or glycine on Rubisco and other Calvin-Benson cycle enzymes.
Materials and Methods
Strains and culture conditions
The strain Synechocystis sp. PCC 6803 was obtained from Prof. Murata (National Institute for Basic Biology, Okazaki, Japan) and served as the WT for this study. The generation and characterisation of the mutants in the Synechocystis sp. PCC 6803 photorespiratory pathway, i.e., ΔglcD1, which bears a defect in the glycolate dehydrogenase-coding gene sll0404, and ΔgcvT, which bears a defect in the gene sll0171 that encodes the T-protein subunit of the glycine decarboxylase complex, have been described elsewhere , . Cultivation was performed in BG11 medium  at pH 8.0 with an initial optical density (OD750) of ∼0.8, which is equal to ∼108 cells mL−1. Mutants were grown in the presence of 50 μg mL−1 kanamycin (Km) or 20 μg mL−1 spectinomycin (Sp). Potential contamination by heterotrophic bacteria was ruled out by spreading 0.2 mL of culture on LB plates.
Axenic cultures were grown photoautotrophically at 29°C in batch cultures using 3-cm glass vessels with 5-mm glass tubes for aeration with 5% CO2-enriched air (HC) with a bubbling flow rate set to 5 mL min−1. Cultures were continuously illuminated with warm light using an Osram L58 W32/3 at 130 μmol photons s−1 m−2. Pre-cultures were split into equal parts. One part continued to be grown under HC conditions while the second subculture was grown under Ci limiting conditions (LC conditions) by bubbling with ambient air containing 0.035% CO2. Acclimation to low Ci by pre-cultivation with ambient air was performed 24 h prior to the isotope-labelling experiments. The pH of the growth medium was stable under the chosen HC and LC cultivation conditions, which were essentially as described earlier .
Transient 13Ci isotope-labelling experiments
Pulse experiments with 13Ci were performed by transferring 40 mL of HC or LC pre-culture into a new cultivation vessel that was kept without bubbling under otherwise identical conditions. The transient isotope pulse was started under very-HC conditions (VHC), i.e. 2% (w/w) sodium hydrogen carbonate, by adding 10 mL of a 13C-bicarbonate stock solution prepared by dissolving 1 g sodium hydrogen carbonate 98 atom% 13C (Sigma-Aldrich) in 10 mL BG11. The pH of the 13C-bicarbonate stock solution was adjusted to 8.0. Subsequent chase experiments were performed after a 60 min pulse according to the pre-acclimation, either under 5% CO2 (HC) or 0.035% CO2 (LC) conditions. The BG11 medium was exchanged by centrifugation (3000g, 2 min, 22°C), careful removal of the supernatant and re-suspension in the original volume. The fresh medium was pre-adjusted to the ambient 12C/13C isotope ratio and CO2 concentration by bubbling with ambient air. Care was taken to minimise the carryover of 13Ci upon medium exchange. To enable fast sampling of the chase kinetics, a wash of the cell pellet had to be avoided. Samples were taken immediately after pulse or chase and at 1, 3, 5, 10, 15, 20 (or 30), and 60 min. Observations of pulse and chase kinetics were repeated with 2–3 independent cultivations.
In our hands a clear step change of 13C label was only possible using high bicarbonate concentrations at the cost of a change of osmolarity within the growth medium. The use of low enrichments for pulse and chase experiments, which are possible using the highly sensitive radioactive 14C detection methods, are not applicable to the less-sensitive 13C detection methods. To control for unavoidable imperfect step changes, we normalised the initial rates of 13C accumulation and the 13C enrichment in downstream metabolite pools of a pathway to the respective parameters of the first 13C assimilation products, i.e., 3PGA and PEP (cf. supplementary Methods S1).
GC-EI-TOF-MS analysis of metabolite pool sizes and mass isotopomer distributions
The previously described GC-EI-TOF-MS metabolite profiling technology for methoxyaminated and trimethylsilylated methanol/water-soluble metabolites from Synechocystis ,  was applied to assess a combination of both the metabolite pool sizes and the respective 13C mass isotopomer distributions from 13C labelled samples. Culture samples of 2–7 mL, equivalent to about 109 cells mL−1, were harvested and separated from the media by fast filtration in the light using a glass vacuum filtration device with controlled temperature and illumination. To minimize the time between sample collection and analysis, no wash was performed. This sampling method, which uses a filter disc to remove secreted metabolites and components of the growth medium, has been previously described . In addition, we performed rapid metabolic inactivation by immediately shock-freezing the cells on the filter disk in liquid N2 to obtain the quickest possible samples for tracing studies , . Recovery checks by internal standardisation using chemically synthesised stable isotope-labelled reference compounds had to be omitted so as not to interfere with the 13C tracing experiments. Instead, metabolite pools were normalised to the chlorophyll a content . The relative changes of normalised metabolite pool sizes were based on the peak intensities. In addition to the conventional metabolite profiling procedure, the sum of all observed mass isotopomers of characteristic fragments was calculated to enable pool size quantification in the presence of shifting mass isotopomer distributions. The metabolites glycolate, glycine, serine, glucose-6-phosphate (G6P), sucrose, glycerate, phosphoenolpyruvate (PEP) and 3-phosphoglycerate (3PGA) were externally calibrated using a dilution series of nine concentration points, ranging from 0.04 ng µL−1 to 166.67 ng µL−1 of a chemically defined mixture of authenticated reference compounds in equal amounts at 1.0 mg mL−1 for each compound .
Data processing and compound identification
GC-TOF-MS chromatograms were processed using TagFinder-Software . Analytes were manually identified using the TargetFinder plug-in of the TagFinder-Software and a reference library of ambient and 13C labelled mass spectra and retention indices (RI) from the Golm Metabolome Database (GMD, http://gmd.mpimp-golm.mpg.de/) , . A peak intensity matrix containing all available mass isotopomers of characteristic mass fragments that represented the primary metabolites under investigation was generated by TagFinder. This matrix was processed using the CORRECTOR software tool (http://www-en.mpimp-golm.mpg.de/03-research/researchGroups/01-dept1/Root_Metabolism/smp/CORRECTOR/index.html). Using this batch processing tool, we calculated the sum of mass isotopomer intensities and the 13C enrichments of mass fragments that had been annotated previously  using previously described methods , .
Calculations and statistical data mining
Data management, data transformation, calculations, and statistical analyses were performed using Microsoft Office Excel 2003 software, the R 2.9.1 statistical programming package, and SigmaPlot 11.0 software (Sysstat Software Inc., San Jose, CA, USA). Calculations of the molar rate of appearance (Ra) for the Rubisco reactions were performed according to Wolfe and Chinkes  using the equation Ra = k*Q, where Q represents the pool size of the respective metabolite. To obtain k, the turnover rate, the experimental data were fitted to the equations Et = Ep(1-e−kt) or Et = E0e−kt. The turnover rate (k−1) of a pool is given in min−1. Et represents the isotopic enrichment at time t, Ep the plateau enrichment and E0 the enrichment at t = 0. The sucrose flux was calculated based on the influx of the precursor G6P, as described by Roessner-Tunali and co-authors .
Error estimates and error minimisation are relevant when judging labelling results. The technical precision when determining the 13C enrichment of a metabolite pool is generally below 2% relative standard deviation (RSD) when using the GC-TOF-MS metabolite profiling method . In contrast, the technical error of a metabolite pool size determination using the same method is about one order of magnitude higher, in the range of 5–20% RSD . In addition to the technical error, the biological variability, i.e., the culture-to-culture differences of replicated Synechocystis experiments, was considered using the 3PGA measurement as a test case. The precision of 3PGA pool size determinations in replicate WT and mutant cultures had previously been determined as 25.6% RSD (n = 9), similar to the average 22.9% RSD (n = 9) of all metabolites observable by GC-TOF-MS profiling (cf. Table S1 of ). The culture-to-culture variation of the parameters was smaller. The maximum 13C enrichment, determined as atom% at 20–60 min, of the 3PGA or PEP pools had, on average, 6.3% RSD, while the initial rate of 13C accumulation, determined as the atom% min−1 by linear regression (r2 = 0.85–0.99), in either of the pools exhibited, on average, 17.1% RSD, as assessed by triplicate WT and duplicate mutant experiments (Table 1).
Motivation, detailed description and error assessment of the experimental design chosen for the comparative metabolic flux phenotyping of Synechocystis sp. PCC 6803 wild-type and photorespiratory mutants.
Head-to-tail view of ambient (red) and maximally 13C labeled (blue) mass spectra from the 3-phosphoglycerate (4TMS) and phosphoenolpyruvate (3TMS) analytes of Synechocystis. Mass fragments suitable for mass isotopomer distribution analysis are indicated by an asterisk . Inserts indicate the identifiers of the Golm Metabolome Database  and the expected retention index (RI) and experimental deviation (ΔRI%) within Synechocystis extracts. The empirical mass distribution vectors of the M-15+ fragments, i.e., C14H36O7PSi4 (M0-3 = 459-462) from the 3-phosphoglycerate (4TMS) analyte and C11H26O6PSi3 (M0-3 = 369-372) from the phosphoenolpyruvate (3TMS) analyte, are shown on the right, as determined from an experiment performed under the HC condition.
13C enrichment of the glucose, glucose-6-phosphate and sucrose pools in the wild-type strain Synechocystis sp. PCC 6803. Data from LC acclimation followed by a VHC pulse at t = 0 min and LC chase at t = 60 min or 120 min are shown. Note that glucose was detectable only under VHC pulse conditions in 2 of 3 replicate experiments.
Initial rates of 13C accumulation. Data were acquired during a 0.5-10.0 min 13Ci pulse and 13C enrichment at maximum labelling (20-60 min) in (A) the glycolate pool compared to the 3PGA or PEP pools and (B) aspartate, malate, and citrate pools compared to the PEP pool. The ratios of the initial rates of 13C accumulation and the 13C enrichment at maximum labelling were calculated from paired observations within each sample. Cultures were pre-acclimated and subjected to a chase with ambient isotope composition of either 5% CO2 (HC) or 0.035% CO2 (LC).
The authors acknowledge the long-standing support and encouragement of Prof. L. Willmitzer, Prof. M. Stitt and Prof. R. Bock as well as the discussions of our results with Dr. A.R. Fernie (Max Planck Institute of Molecular Plant Physiology, Germany). J.G. acknowledges the C++ programming advice of PD Dr. Tillmann Klamroth (University of Potsdam, Theoretical Chemistry, Germany).
Conceived and designed the experiments: JH JG DS HB MH JK. Performed the experiments: JH JG DS. Analyzed the data: JH JG DS HB MH JK. Contributed reagents/materials/analysis tools: JH JG DS HB MH JK. Wrote the paper: JH JG DS HB MH JK.
- 1. Knoll AH (2008) Cyanobacteria and earth history. In: Herrero A, Flores E, editors. The cyanobacteria: molecular biology, genomics and evolution. Norfolk: Caister Academic Press. pp. 1–19.
- 2. Martin W, Rujan T, Richly E, Hansen A, Cornelsen S, et al. (2002) Evolutionary analysis of Arabidopsis, cyanobacterial, and chloroplast genomes reveals plastid phylogeny and thousands of cyanobacterial genes in the nucleus. Proc Natl Acad Sci U S A 99: 12246–12251.
- 3. Deusch O, Landan G, Roettger M, Gruenheit N, Kowallik KV, et al. (2008) Genes of cyanobacterial origin in plant nuclear genomes point to a heterocyst-forming plastid ancestor. Mol Biol Evol 25: 748–61.
- 4. Tolbert NE (1997) The C-2 oxidative photosynthetic carbon cycle. Ann Rev Plant Physiol Plant Mol Biol 48: 1–25.
- 5. Kelly GJ, Latzko E (1977) Chloroplast phosphofructokinase: II. Partial purification, kinetic and regulatory properties. Plant Physiol 60: 295–299.
- 6. Husic DW, Husic HD, Tolbert NE (1987) The oxidative photosynthetic carbon cycle or C2 cycle. CRC Crit Rev Plant Sci 5: 45–100.
- 7. Norman EG, Colman B (1991) Purification and characterization of phosphoglycolate phosphatase from the cyanobacterium Coccochloris peniocystis. Plant Physiol 95: 693–698.
- 8. Somerville CR (2001) An early Arabidopsis demonstration. Resolving a few issues concerning photorespiration. Plant Physiol 125: 20–24.
- 9. Cegelski L, Schaefer J (2006) NMR determination of photorespiration in intact leaves using in vivo 13CO2 labeling. J Magn Reson 178: 1–10.
- 10. Bauwe H (2010) Recent developments in photorespiration research. Biochem Soc Trans 38: 677–682.
- 11. Colman B (1989) Photosynthetic carbon assimilation and the suppression of photorespiration in the cyanobacteria. Aquat Bot 34: 211–231.
- 12. Kaplan A, Reinhold L (1999) CO2 concentrating mechanisms in photosynthetic microorganisms. Ann Rev Plant Physiol Plant Mol Biol 50: 539–570.
- 13. Eisenhut M, Kahlon S, Hasse D, Ewald R, Lieman-Hurwitz J, et al. (2006) The plant-like C2 glycolate cycle and the bacterial-like glycerate pathway cooperate in phosphoglycolate metabolism in cyanobacteria. Plant Physiol 142: 333–342.
- 14. Eisenhut M, Ruth W, Haimovich M, Bauwe H, Kaplan A, et al. (2008) The photorespiratory glycolate metabolism is essential for cyanobacteria and might have been conveyed endosymbiontically to plants. Proc Natl Acad Sci U S A 105: 17199–17204.
- 15. Eisenhut M, Huege J, Schwarz D, Bauwe H, Kopka J, et al. (2008) Metabolome phenotyping of inorganic carbon limitation in cells of the wild type and photorespiratory mutants of the cyanobacterium Synechocystis sp. strain PCC 6803. Plant Physiol 148: 2109–2120.
- 16. Eisenhut M, von Wobeser EA, Jonas L, Schubert H, Ibelings BW, et al. (2007) Long-term response toward inorganic carbon limitation in wild type and glycolate turnover mutants of the cyanobacterium Synechocystis sp. strain PCC 6803. Plant Physiol 144: 1946–1959.
- 17. Huege J, Sulpice R, Gibon Y, Lisec J, Koehl K, et al. (2007) GC-EI-TOF-MS analysis of in vivo-carbon-partitioning into soluble metabolite pools of higher plants by monitoring isotope dilution after (13CO2)-labelling. Phytochem 68: 2258–2272.
- 18. Ratcliffe RG, Shachar-Hill Y (2006) Measuring multiple fluxes through plant metabolic networks. Plant J 45: 490–511.
- 19. Shastri AA, Morgan JA (2007) A transient isotopic labeling methodology for C-13 metabolic flux analysis of photoautotrophic microorganisms. Phytochem 68: 2302–2312.
- 20. Döhler G (1974) C4-Weg der Photosynthese in der Blaualge Anacystis nidulans (The C4-path of photosynthesis in the cyanobacterium Anacystis nidulans) Planta 118: 259–269.
- 21. Maheswaran M, Ziegler K, Lockau W, Hagemann M, Forchhammer K (2006) PII-regulated arginine synthesis controls accumulation of cyanophycin in Synechocystis sp. strain PCC 6803. J Bacteriol 188: 2730–2734.
- 22. Calvin M (1956) The photosynthetic carbon cycle. J Chem Soc 1895–1915.
- 23. Calvin M (1961) The path of carbon in photosynthesis. Nobel lectures, chemistry 1942-1962. Amsterdam: Elsevier Publishing Company. pp. 618–644.
- 24. Tabita FR (1988) Molecular and cellular regulation of autotrophic carbon dioxide fixation in microorganisms. Microbiol Rev 52: 155–189.
- 25. Tabita FR (2009) The hydroxypropionate pathway of CO2 fixation: fait accompli. Proc Natl Acad Sci U S A 106: 21015–21016.
- 26. Agarwal R, Ortlebb S, Sainisa JK, Melzer M (2009) Immunoelectron microscopy for locating Calvin cycle enzymes in the thylakoids of Synechocystis 6803. Molec Plant 2: 32–42.
- 27. Yang C, Hua Q, Shimizu K (2002) Metabolic flux analysis in Synechocystis using isotope distribution from C-13-labeled glucose. Met Engin 4: 202–216.
- 28. Hagemann M, Vinnemeier J, Oberpichler I, Boldt R, Bauwe H (2005) The glycine decarboxylase complex is not essential for the cyanobacterium Synechocystis sp. strain PCC 6803. Plant Biol 7: 15–22.
- 29. Rippka R, Deruelles J, Waterbury JB, Herdman M, Stanier RY (1979) Generic assignments, strain histories and properties of pure cultures of cyanobacteria. J Gen Microbiol 111: 1–16.
- 30. Erban A, Schauer N, Fernie AR, Kopka J (2007) Non-supervised construction and application of mass spectral and retention time index libraries from time-of-flight GC-MS metabolite profiles. Methods Mol Biol 358: 19–38.
- 31. Krall L, Huege J, Catchpole G, Steinhauser D, Willmitzer L (2009) Assessment of sampling strategies for gas chromatography–mass spectrometry (GC–MS) based metabolomics of cyanobacteria. J Chromatogr B 877: 2952–2960.
- 32. Schauer N, Zamir D, Fernie AR (2005) Metabolic profiling of leaves and fruit of wild species tomato: a survey of the Solanum lycopersicum complex. J Exp Bot 56: 297–307.
- 33. Luedemann A, Strassburg K, Erban A, Kopka J (2008) TagFinder for the quantitative analysis of gas chromatography - mass spectrometry (GC-MS) based metabolite profiling experiments. Bioinform 24: 732–737.
- 34. Kopka J, Schauer N, Krueger S, Birkemeyer C, Usadel B, et al. (2005) GMD@CSB.DB: the Golm Metabolome Database. Bioinform 21: 1635–1638.
- 35. Hummel J, Strehmel N, Selbig J, Walther D, Kopka J (2010) Decision tree supported substructure prediction of metabolites from GC-MS profiles. Metabolomics 6: 322–333.
- 36. Wittmann C, Heinzle E (1999) Mass spectrometry for metabolic flux analysis. Biotech Bioeng 62: 739–750.
- 37. van Winden WA, Wittmann C, Heinzle E, Heijnen J (2002) Correcting mass isotopomer distributions for naturally occurring isotopes. Biotech Bioeng 80: 477–479.
- 38. Wolfe RR, Chinkes DL (2004) Calculation of substance kinetics: single pool model. In: Isotope tracers in metabolic research: principles and practice of kinetic analysis, 2nd edition. Hoboken: John Wiley & Sons 21–50.
- 39. Roessner-Tunali U, Liu JL, Leisse A, Balbo I, Perez-Melis A, et al. (2004) Kinetics of labelling of organic and amino acids in potato tubers by gas chromatography-mass spectrometry following incubation in 13C labelled isotopes. Plant J 39: 668–679.
- 40. Allwood JW, Erban A, de Koning S, Dunn WB, Luedemann A, et al. (2009) Inter-laboratory reproducibility of fast gas chromatography–electron impact–time of flight mass spectrometry (GC–EI–TOF/MS) based plant metabolomics. Metabolomics 5: 479–496.