A new anaplerotic respiratory pathway involving lysine biosynthesis in isocitrate dehydrogenase-deficient Arabidopsis mutants


Author for correspondence:

Guillaume G. B. Tcherkez

Tel: +33 1 69 15 33 79

Email: guillaume.tcherkez@u-psud.fr


  • The cornerstone of carbon (C) and nitrogen (N) metabolic interactions – respiration – is presently not well understood in plant cells: the source of the key intermediate 2-oxoglutarate (2OG), to which reduced N is combined to yield glutamate and glutamine, remains somewhat unclear.
  • We took advantage of combined mutations of NAD- and NADP-dependent isocitrate dehydrogenase activity and investigated the associated metabolic effects in Arabidopsis leaves (the major site of N assimilation in this genus), using metabolomics and 13C-labelling techniques.
  • We show that a substantial reduction in leaf isocitrate dehydrogenase activity did not lead to changes in the respiration efflux rate but respiratory metabolism was reorchestrated: 2OG production was supplemented by a metabolic bypass involving both lysine synthesis and degradation.
  • Although the recycling of lysine has long been considered important in sustaining respiration, we show here that lysine neosynthesis itself participates in an alternative respiratory pathway. Lys metabolism thus contributes to explaining the metabolic flexibility of plant leaves and the effect (or the lack thereof) of respiratory mutations.


Respiration couples the production of ATP, reducing equivalents and carbon (C) skeletons to the release of large amounts of CO2 into the atmosphere: between 30% and 80% of daily photosynthetic carbon gain is released into the atmosphere by plant respiration. However, the intrinsic mechanisms of leaf respiratory metabolism responsible for CO2 production are somewhat uncertain. There is indeed a critical gap in our knowledge about metabolic pathways involved in the biosynthesis of the key respiratory intermediate 2-oxoglutarate (2OG), the precursor of amino acids glutamate (Glu) and glutamine (Gln). Despite the considerable literature on plant respiration, the provision of C skeleton acceptors through respiratory metabolism to sustain N (and sulphur) assimilation remains somewhat unclear for at least three reasons. First, during N reduction and assimilation in the light, leaf respiration (CO2 evolution) has been shown to be inhibited due to the lower activity of several enzymes of the tricarboxylic acid pathway (TCAP; Hanning & Heldt, 1993; Gessler et al., 2009). Second, despite the very clear utilization of C reserves to sustain Glu synthesis in the light (Tcherkez et al., 2009; Gauthier et al., 2010), the way by which such reserves (fumarate, malate or citrate) are converted to 2OG are not well defined and, furthermore, reserve pools may be insufficient to feed total diurnal Glu synthesis (Stitt et al., 2002). Third, Arabidopsis mutants impaired in enzymatic steps responsible for respiratory 2OG production in leaves (NAD- and NADP-dependent isocitrate dehydrogenase, abbreviated IDH and ICDH, respectively) have no obvious phenotype, suggesting that isocitrate dehydrogenase activity per se exerts little metabolic control on respiration (CO2 efflux) and N assimilation (N influx). Knock-down mutations of cytosolic ICDH lead to little metabolic effect (the majority of metabolic pools are affected by < 1.5-fold, except for glutathione and Cys; Mhamdi et al., 2010) and similarly, knock-down mutations of IDH caused variable and poorly significant changes in metabolic pools, although several TCAP intermediates accumulated under heterotrophic liquid culture conditions (Lemaitre et al., 2007). In IDH antisense tomato (Solanum lycopersicum) lines, the 2OG-to-Glu ratio is increased and little effect is seen in organic and amino acid content, despite a slightly lower labelling in TCAP intermediates upon 13C-pyruvate feeding (Sienkiewicz-Porzucek et al., 2008). From consideration of respiration rates and IDH activity, the calculation of control coefficients further indicates a very small value for IDH (Araujo et al., 2011).

The lack of clear metabolic consequences of alterations in 2OG production suggests that isocitrate dehydrogenase activity is not limiting for N assimilation and ATP production and/or that alternative pathways are possible. Several respiratory bypasses have been described, including a futile cycle with phosphoenolpyruvate carboxylase (which produces oxaloacetate then reduced to malate) and the malic enzyme (which converts malate back to pyruvate), the γ-aminobutyrate shunt or the glyoxylic pathway (Sweetlove et al., 2010). However, these pathways may operate under certain circumstances in leaves (such as hypoxia or senescence) and, quite critically, none of them provides 2OG. It is then probable that the involvement of several I(C)DH enzymes/isoforms compensates for each of the individual mutations described above. For example, in yeast (Saccharomyces cerevisiae), auxotrophy for Glu is observed only in strains with double (idh2 idp1) or triple (idh2 idp1 idp2) disruptions (in yeast, mitochondrial and cytosolic NADP-specific isocitrate dehydrogenase isozymes are denoted as idp1 and idp2, respectively) under glucose nutrition (Zhao & McAlister-Henn, 1996). The balance between cytosolic and mitochondrial ICDH and mitochondrial IDH activities may further explain the modest effects of individual mutations – provided 2OG exchanges between cell compartments via transporters (Picault et al., 2002). However, the I(C)DH proteins affected by these mutations are responsible for most of leaf I(C)DH activity.

Thus, presently, there are arguments for and against an isocitrate dehydrogenase origin of 2OG for N assimilation and Glu synthesis (Galvez et al., 1999). In addition to being sensitive to environmental conditions and, perhaps, not limiting for the TCAP, 2OG provision by I(C)DH might be supplemented by other metabolic pathways. As an aid in clarifying 2OG metabolism, we investigated the primary C metabolism of Arabidopsis mutants considerably affected in 2OG synthesis (I(C)DH activity); the objective of this study was to identify and characterize the involvement of potential respiratory bypasses when I(C)DH activity is altered.

Five genes encode different IDH sub-units (Lemaitre & Hodges, 2006) and three genes encode the four ICDH isoforms. Amongst these AtIDHV and AtICDH2 encode a mitochondrial IDH catalytic subunit and the cytosolic ICDH, respectively, that are associated with ≈90% of leaf I(C)DH activity (see Lemaitre et al., 2007 and Mhamdi et al., 2010, respectively). We took advantage of the previously characterized single mutants idhV and icdh2 and generated ‘hemihomozygous’ lines (homozygous for one mutation and heterozygous for the other) idhV −/− icdh2 +/− (abbreviated idh ICDH) and idhV +/− icdh2 –/– (abbreviated IDH icdh), the double mutant being lethal. Gas exchange showed that despite a reduced I(C)DH activity, there was no significant change in leaf respiration rate both in the dark and in the light. Metabolomic analyses indicated (light-sensitive) alterations of primary C metabolites and the Lys degradation intermediate α-aminoadipate, mostly in hemihomozygous mutants. 13C-labelling further provided evidence that effective Lys biosynthesis and conversion to Glu (inferred from labelling in Pro) were enhanced in hemihomozygous mutants, demonstrating that Lys metabolism plays the role of a respiratory bypass.

Materials and Methods

Plant lines

Both idhV (SAIL_806A06) and icdh2 (SALK_056247) Arabidopsis thaliana (L.) Heynh lines are mutants containing a T-DNA insertion in AtIDHV (At5 g03290) and AtICDH2 (At1 g65930), respectively. Both have already been characterized previously and have been shown to be specifically affected in the targeted genes (Lemaitre et al., 2007; Mhamdi et al., 2010). Both lines are knockout mutants for each gene, but with a residual I(C)DH activity (Fig. 1a). Hemihomozygous lines were generated by crossing single mutants (homozygous parents). Each genotype was identified using PCR amplification with gene-specific primers. After having generated the F1 progeny, hemihomozygous F2 plants were used for analysis.

Figure 1.

Respiration and gas exchange. In vitro NAD-dependent (IDH, closed bars) and NADP-dependent (ICDH, open bars) isocitrate dehydrogenase activity in Arabidopsis rosette leaves (a, in pmol mg−1 soluble proteins min−1), leaf respiration (CO2 evolution) in darkness (b) and in the light (c), leaf net photosynthesis (d) and RuBP carboxylation (open bars) and oxygenation (closed bars) rates (e, calculated from net photosynthesis, Rday, Γ* and Ci – see the 'Materials and Methods' section for further details). In (a), letters stands for statistical categories of total I(C)DH activity (< 0.05). In (b–e), there is no significant difference between genotypes. Error bars, ± SE.

Plant growth and sampling

After sowing on potting mix, plantlets were transplanted to individual pots and grown in a controlled environment (growth chamber) under 8 h : 16 h light : dark (short days) at an irradiance of ≈150 μmol m−2 s−1, 20 : 18°C day : night temperature, 65% humidity and nutrient solution (1 g l−1 PP14-12-32, (Plant-Prod, Puteaux, France) supplemented with 20 μl l−1 fertoligo L (Fertil, Boulogne-Billancourt, France)) twice a week. Plants were sampled 50 d after germination; samples were immediately frozen in liquid N and stored at −80°C.

Gas exchange

Photosynthesis and respiration rates were measured with the gas exchange open system Li-Cor 6400 xt using the 2-cm2 fluorescence chamber (Li-Cor, Austin, TX, USA), under a controlled humidity of 80% fixed with a dew-point generator (Li-Cor 610). Net photosynthesis (A) was measured in typical conditions (380 μmol mol−1 CO2, 21% O2, 22°C, 250 μmol m−2 s−1 PAR, 10% blue). Dark respiration was measured in the same gaseous conditions after 30 min dark acclimation. Day respiration (Rday, nonphotorespiratory CO2 evolution in the light) was obtained using the method of Laisk with three different light intensities (100, 175, 250 μmol m−2 s−1 PAR), which also provided the CO2 compensation point in the absence of day respiration (Γ*). There was no difference in Γ* between plant lines, at ≈48 μmol mol−1. ci-based RuBP carboxylation (vc) and oxygenation (vo) rates were calculated (after Von Caemmerer & Farquhar, 1981) using: vc = (A + Rd)/(1–Γ*/ci) and vo = 2vcΓ*/ci (ci, leaf internal (intracellular) CO2 mole fraction).


Analyses of GC-MS metabolomics were carried out on methanol extracts chemically derived with methoxyamine and MSTFA in pyridine, with a Pegasus III GC-TOF-MS system (Leco, Garges-les-Gonesse, France) as previously described (Bathellier et al., 2009).

13C-labelling and LC-MS measurements

Labelling with 13C-3-glucose and 13C-2-Lys was carried out on leaf disks incubated in Petri dishes in the dark with 15 mM solutions. Precautions were taken so that leaf disks floated on the solution and did not sink and thus were not subjected to hypoxia. 13C-3-glucose (99% 13C in C-3) and 13C-2-Lys (85% 13C in C-2) were sourced from Sigma. After 3 h incubation, leaf disks were frozen in liquid N and stored at −80°C for analysis. Approximately 200-mg leaf disks were ground in liquid N and the powder was extracted with 1.5 ml of 50 mM phosphate buffer (pH 7.2) containing 1% (w/v) ascorbic acid and 0.5% (w/v) dithiothreitol. Samples were heated at 100°C for 10 min and flash-cooled on ice. After 15 min of centrifugation at 14 000 g, the supernatant was filtrated through a membrane (0.2-μm pore size). The filtrate was then transferred into a glass vial containing 10 μl AraC (cytosine β-d-arabinofuranoside, 0.1 mM) as an internal standard. Samples were then analysed with a UPLC-MS system as described in Tcherkez et al. (2012ab). Briefly, they were injected into the UPLC Acquity equipped with the column UPLC-HSS T3 (2.1 × 100 mm, 1.8 μm; Waters, Guyancourt, France) under a formic acid (0.1%) gradient in water/formic acid (0.1%) in acetonitrile, coupled to electrospray mass spectrometry MicrOTOF II (Bruker Daltonics, Wissembourg, France) with N2 (spray gas) upon negative or positive ionization. Amino acids studied in this paper were identified and measured with retention times and calibration curves obtained with pure standards (Sigma). The 13C-abundance was calculated from m + 1 and m (m/z) mass signals (Sm+1 and Sm, respectively) using the following relationship numerically solved for xc (fraction of 13C):

display math

(nj, number of atoms of element j (C, N, H, O); xj, mole fraction (percentage ÷ 100) of the heavy isotope (13C, 15N, 2H, 17O) of element j). Known xj values (natural abundance) were used for elements other than C (0.00366, 0.00015, 0.00038 for 15N, 2H and 17O, respectively). The total 13C-amount in amino acids (in moles 13C per g FW) was then calculated using the absolute leaf content (C) determined by HPLC (Godel et al., 1984): 13C-amount = xc × C.

Calculations and statistics

Metabolomic hierarchical clustering (cosine correlation) and ANOVA were carried out with the software MeV (Eisen et al., 1998). Connectivity graphs between metabolites were carried out from the correlation matrix using the software Pajek (Batagelj & Mrvar, 2004) with the Kamada–Kawai algorithm. In the present paper, differences were examined with heteroscedastic Student–Welsh tests and labelled as statistically significant when < 0.05 or < 0.08 as indicated in the figures.


Mutant generation and I(C)DH activity

Already characterized idh and icdh simple mutants were crossed and the F1 generation was propagated to yield homozygous double mutants. Unfortunately, the double mutation was lethal at the globular embryo stage (Supporting Information Fig. S1). In vitro NAD- and NADP-dependent isocitrate dehydrogenase activity was measured in leaf extracts (Fig. 1a). The response of I(C)DH activity to the allelic composition appeared to be quantitative. That is, heterozygous mutants had an intermediate activity between the wild-type and homozygous single mutants. Total I(C)DH activity was the lowest in hemihomozygous plant lines.

Photosynthesis and respiration

The respiration rate in darkness and in light was similar in all genotypes, with no significant difference, including the hemihomozygous plants (Fig. 1b,c). Net photosynthesis measured with gas exchange under typical conditions (250 μmol m−2 s−1 PAR, 380 μmol mol−1 CO2, 21% O2) was similar in all genotypes (no significant difference; Fig. 1d). RuBP carboxylation (vc) and oxygenation (vo) rates were calculated from net photosynthesis, internal CO2 concentration and the CO2 compensation point in the absence of day respiration (Γ*). Clearly, there was no difference between genotypes, showing that photorespiration was unchanged in mutants (similar vo, Fig. 1e). The lack of difference in respiration and photosynthetic parameters was independent of the units used because both leaf specific area and chlorophyll content were unchanged.

Metabolomic patterns

Analyses of GC-MS metabolomic were conducted on rosette leaves of the five different plant lines, sampled either in the light (3 h light acclimation) or in the dark (1 h dark acclimation). The metabolomics analyses allowed the detection and quantification of 81 metabolites, among which 19 were significantly different between genotypes using a two-way ANOVA (the two factors used were day: night condition and genotype; Fig. 2a). As expected, 2-hydroxyglutarate (2HG) and 2OG were significantly lower in all mutant lines and 2OG was even not detected in hemihomozygous plants (Fig. 2b). Amongst significant metabolites, most of them were associated with sugar metabolism (e.g. galactose, arabinose, fructose, glucose, galactonolactone), suggesting that sucrose and/or glucose interconversion decreased in mutants (Fig. 2a). Several amino acids were also significantly altered in mutant genotypes, such as aromatics (Phe, Tyr, Trp). The Glu content was not significantly altered by mutations although the day: night pattern disappeared in the ICDH idh line (Fig. 2c). The key respiratory tricarboxylic acid, citrate, did not show large variations between genotypes but was found to be significantly less abundant in the idh mutant in the dark (Fig. 2d). α-aminoadipic acid (AAA), an intermediate of Lys degradation, appeared to be a significant metabolite (Fig. 2a) and was indeed less abundant in hemihomozygous lines, mostly in the dark (Fig. 2e). Lys was slightly but insignificantly less abundant in hemihomozygous lines both in the light and in the dark (data not shown). The decrease in both Lys and AAA contents likely reflect an enhanced consumption of Lys-related metabolites by leaf metabolism.

Figure 2.

Metabolomic array and metabolite content. (a) Heat map showing metabolites (quantified by GC-MS) that differ significantly (< 0.05, two-way ANOVA) between Arabidopsis plant lines: wild-type (wt), idh and icdh simple mutants and icdh IDH and ICDH idh partial double mutants (denoted as sesq-ICDH, sesq-IDH, respectively, in column labels). The 16 left and 16 right columns correspond to light and darkness, respectively. Each column corresponds to the average of three leaf disks. (b–i) Metabolite content and metabolic ratios in the dark (closed bars) and in the light (open bars). Asterisks: values significantly different from the wild-type (< 0.05). ND, not detected. Error bars, ± SE.

There were also very clear differences in metabolic ratios (Fig. 2f–i). Larger values of the pyruvate-to-fumarate ratio in ICDH idh hemihomozygous plants in the light indicated that pyruvate was likely less committed to the TCAP and consumed by other pathways (Fig. 2f). Despite quite large variations in the relative Gly content in the ICDH idh plants, both hemihomozygous genotypes had a larger Gly-to-Ser ratio (significantly different in icdh IDH plants; Fig. 2g). Because an alteration of photorespiratory Gly metabolism is unlikely (Fig. 1e), this effect indicates changes in cytoplasmic Ser metabolism from 3-phosphoglycerate. All mutant genotypes had relatively less ascorbate synthesized from galactonolactone in the dark (Fig. 2h) and a lower commitment from glycolysis to reductive production of lactate in the light (Fig. 2i), suggesting that there was less reductive power and an oxidative stress.

The whole metabolomic dataset was replotted as connectivity graphs showing the variational proximity between all metabolites (Figs S2–S7): pair-wise cosine correlation between metabolite is used to generate a web in which the distance represents variational dissimilarity (i.e. the closer the distance, the higher the correlation). This representation allows us to see multiple correlations, which are not apparent in classical hierarchical clustering (such as in Fig. 2a). In all genotypes, AAA was relatively close to citrate and/or 2OG, suggesting a metabolic relationship between them; Gln and Glu were clearly much closer to AAA in hemihomozygous genotypes compared to others. Overall, this analysis shows that Gln, Glu, 2HG and AAA were much more closely correlated in hemihomozygous genotypes.


The effective metabolic commitment of catabolic substrates to downstream metabolites was investigated using either 13C-3-glucose or 13C-2-Lys labelling in the dark (Fig. 3). Specific positional labelling allowed us to discriminate between pathways (see Fig. 4 for a metabolic scheme) for two reasons. (1) The 13C-atom in glucose (C-3) is lost in the form of CO2 during acetyl-CoA synthesis and thus it cannot enter directly the TCAP by citrate synthase; however, phosphoenolpyruvate carboxylase (PEPC) activity may abstract 13C from phosphoenolpyruvate and thus 13C-enrich Asp, the precursor of Lys. (2) The 13C-atom in Lys (C-2) is lost by the action of 2OG dehydrogenase and thus cannot be redistributed via the TCAP. The 13C-amount in key amino acids measured using LC-MS accurate mass determination is shown in Fig. 3. There was a significant 13C-enrichment in Lys upon 13C-glucose feeding in hemihomozygous genotypes (Fig. 3a), highlighting the de novo synthesis of Lys. Similarly, Pro was 13C-enriched especially in the ICDH idh genotype (Fig. 3b). Because 13C-atoms cannot have originated from acetyl-CoA, 13C-Pro originated from both PEPC activity (Fig. 4, light grey circles) and 13C-Lys recycling. Phe was slightly labelled (13C-amount above natural abundance, Fig. 3c) in single mutants (but not in hemihomozygous lines), indicating that the 13C-enrichment in phosphoenolpyruvate, the metabolic precursor of Phe.

Figure 3.

13C-labelling in key amino acids. 13C content (in pmol 13C mg−1 FW) in Arabidopsis partitioned to Lys, Pro, Phe and Val upon 13C-labelling with 13C-3-glucose (a–c) or 13C-2-Lys (d, e). Asterisks (*) and double S (§): 13C content significantly different from the wild-type with < 0.05 and 0.08, respectively. See Fig. 4 for the labelling processes involved. The dashed line stands for the theoretical average 13C content that would be expected if the amino acid of interest were at natural abundance (1.1% 13C). Error bars, ± SE.

Figure 4.

Simplified scheme showing the metabolism of the 13C-label. The fate of the 13C-atom is indicated with dark grey circles (13C-3-glucose labelling) or a black square (13C-2-Lys labelling). The fast 13C-redistribution after 13C-3-glucose labelling due to malate/fumarate equilibration (asterisk) is indicated with light grey circles and the subsequent 13C-redistribution via the Asp/Lys pathway is indicated with ‘L's. When several reactions are involved, two successive arrows are shown. The reaction impaired in mutants investigated here is emphasized with a thick arrow. In this figure, the redistribution of 13C from 13C-3-glucose via the pentose phosphate pathway is not shown.

Upon 13C-Lys labelling, both Val and Phe remained at nat-ural 13C-abundance, as expected (no predicted labelling of phosphoenolpyruvate; Fig. 3d,f). By contrast, Pro was significantly 13C-enriched in hemihomozygous ICDH idh genotypes, demonstrating that Lys was recycled to 2OG and then to Pro (Fig. 4, black squares). The 13C-amount in Glu and Gln did not differ between genotypes, due to their substantial content in leaves (at least ten times higher than Pro, Val or Lys), that diluted the isotopic signal considerably. Ile was not 13C-enriched (and remained at natural abundance) upon 13C-Lys feeding. Upon 13C-glucose labeling, Ile was not 13C-enriched in genotypes other than the wild-type and the icdh mutant (Fig. 3c, inset). This suggests that in both idh and hemihomozygous genotypes, the competition between Lys and Ile/Thr biosynthetic pathways for the same precursor (Asp) favoured Lys. The 13C-enrichment in AAA was not measured due to the low content and poor ionization of this metabolite and thus the impossibility of computing the 13C-signal from LC-MS data.


Respiratory control by I(C)DH activity

Despite the clear drop in leaf I(C)DH activity, the rate of respiratory CO2 evolution was unchanged in all Arabidopsis mutant lines investigated here (Fig. 1), showing little control exerted by I(C)DH on the respiratory CO2 efflux. Our results nevertheless show that the metabolic content and the 13C-enrichment patterns were different between the two hemihomozygous lines, with a more pronounced metabolic phenotype in idh ICDH (Figs 2, 3). This suggests that IDH was more intrinsically linked to TCAP metabolism. Many respiratory mutations have been found to have a modest or no effect on the respiration rate (for a review, see Tcherkez et al., 2012a,b) and isocitrate dehydrogenase has been suggested to be of minor importance in determining the rate of leaf CO2 evolution (Araujo et al., 2011). Such a resilience of the respiratory flux with respect to enzymatic alterations likely comes from TCAP flexibility; that is, reversibility of several reactions (including ICDH catalysis), enzymatic redundancy and respiratory bypasses. In addition, TCAP is believed to integrate stored organic acids, thereby attenuating its dependence upon de novo synthesis of respiratory intermediates. This means that other pathways that can simultaneously synthesize 2OG and evolve CO2 have probably occurred in mutants and compensated for the decreased I(C)DH activity: this would explain the lack of effect on both the respiratory CO2 efflux and plant growth and development; Lys synthesis and degradation does evolve CO2 and Lys degradation can produce 2OG (Fig. 4 and see later). Nevertheless, in the progeny of hemihomozygous plants, the double mutation (homozygous mutants) appeared to be lethal during embryo development (Fig. S1) suggesting that the metabolic compensation is insufficient when the reduction of I(C)DH activity is considerable and/or that the demand in 2OG is much larger in developing embryos.

Pleiotropic effects of respiratory mutations

Despite the absence of morpho-anatomical and developmental consequences, we found a clear metabolic phenotype in our hemihomozygous lines. Amongst metabolic changes, metabolites involved in redox homeostasis and Gly/Ser metabolism were the most affected (Fig. 2). However, no change in the photorespiration rate was observed (Fig. 1), clearly showing here that the Gly-to-Ser ratio was divorced from RuBP oxygenation. It is likely that the decrease in I(C)DH activity stimulated the cytoplasmic (nonphotorespiratory) biosynthesis of Ser. This effect may naturally occur upon a redox disequilibrium, because the conversion of Ser into Gly requires tetrahydrofolate (THF), thereby converted to methylene-THF. The latter is in equilibrium with its reduced form, methyl-THF, and its oxidized form, methenyl-THF. In other words, a decrease of the cellular NAD(P)H/NAD(P)+ ratio caused by the altered I(C)DH activity may have displaced these equilibria towards methenyl-THF (and formyl-THF), thus stimulating Gly conversion from Ser to alleviate the depletion in reduced THF derivatives. The predominance of redox-based mechanisms would agree with the increase in ascorbate biosynthetic commitment suggested by the relative decrease in galactonolactone (ascorbate precursor; Fig. 2). It is thus likely that many of the effects visible here are associated with a lower reductive poise in the mitochondrial and cytoplasmic compartment, which, in turn, stems from the altered production of NAD(P)H by I(C)DH activity. Such a redox effect of the icdh mutation remains equivocal, however. On the one hand, icdh2 mutants have no significant changes in redox ratios like GSH/GSSG (but an enhancement of oxidative signalling) in a Col0 wild-type background. On the other, there is a strong oxidative effect in a catalase-2 background (Mhamdi et al., 2010). In addition, in idhV mutants, the NADH/NAD ratio is higher (with a lower total NAD pool) but the NADPH/NADP ratio is lower than in the wild-type (with no change of the total NADP pool; Lemaitre et al., 2007). Antisense IDH tomato lines have lower total NAD(P) pools and NAD(P)H/NAD(P) ratios and a marked increase in the transcription of redox-related genes (Sienkiewicz-Porzucek et al., 2008). In the well-characterized respiratory mutant CMSII of wild tobacco (Nicotiana sylvestris), the increase in the CO2 evolution rate and flux through the TCAP has been found to be associated with a slightly larger NADH/NAD ratio and an enhanced resistance to oxidative stress and pathogens (Dutilleul et al., 2005). Several pleiotropic effects of respiratory mutations (including on photosynthesis) may occur and have been reviewed elsewhere (Tcherkez et al., 2012a,b). Here, many of the significant effects caused by the i(c)dh mutations are also probably pleiotropic and related to mechanisms overriding changes in cellular NAD(P)H rather than a drastic limitation in 2OG synthesis.

Lys metabolism as a respiratory bypass

The degradation of proteins and amino acids to sustain leaf respiration has been shown to occur, mostly during senescence or stressful conditions (Moller & Kristensen, 2004). However, it has also been suggested that protein degradation feeds respiratory metabolism under ‘ordinary’ conditions (Bouma et al., 1994; Lehmeier et al., 2008). In particular, Lys has been proposed to be recycled back to 2HG, which contributes to sustaining the mitochondrial respiratory chain with electrons (2HG oxidized to 2OG) under specific circumstances (Ishizaki et al., 2005; Araujo et al., 2010, 2011). It should be noted that in plants, Lys biosynthesis occurs in the chloroplast and 2HG evolved by Lys degradation should be imported into the mitochondria to be oxidized to 2OG (Araujo et al., 2010). Lys catabolism is also believed to be essential for sustaining Glu synthesis via acetyl-CoA in animals (brain tissue) and via 2HG in senescing plant leaves (for a review, see Galili, 2002; Kirma et al., 2012). Here, we show that both Lys synthesis and its degradation are enhanced in mutants and did not involve acetyl-CoA as an intermediate (Fig. 3): 13C-labelled Glc caused a 13C-enrichment in Lys in hemihomozygous lines and 13C-labelled Lys caused a 13C-enrichment in Pro in ICDH idh lines. Should acetyl-CoA be the major product of AAA degradation, the 13C-enrichment in Pro after 13C-3-Glc labelling would have been negligible. Similarly, the redistribution of the 13C-atom by the pentose phosphate pathway is unlikely because Phe was not labelled in hemihomozygous lines and was even less 13C-enriched than in the control. It should also be noted that under our conditions (positional labelling), the 13C-atom inherited from 13C-3-Glc was lost as CO2 along the classical respiratory pathway, and thus other pathways –such as, here, Lys metabolism– must be considered to explain the 13C-enrichment in Pro (Fig. 4, dark grey circles).

The Lys pathway (including Lys biosynthesis) thus appeared to be a respiratory bypass sustaining 2OG production (allocated to Glu and Pro) and compensating for the lower 2OG synthesis. Indeed, Lys catabolism has been proved able to evolve 2OG as a terminal product (Araujo et al., 2010; Engqvist et al., 2011; Fig. 4). The enhancement of the Lys pathway is probably detrimental to Thr and Ile synthesis, which starts from the same precursor aspartate-semialdehyde, because Thr and Ile contents were somewhat (but insignificantly) lower in hemihomozygous lines ( 0.1) and Ile was not 13C-enriched in hemihomozygous lines. In addition to developmental higher respiratory requirements, the embryo lethality found here in double mutants may have originated from the larger Lys demand competing with respiratory Lys consumption. We recognize that the genetic origin of the increased flux through the Lys pathway has not been investigated here. It may consist in the stimulation of transcription of genes encoding enzymes of the Lys pathway and/or post-transcriptional events. Lys synthesis and degradation is believed to be controlled mainly by feedback inhibition of dihydrodipicolinate synthase by Lys (Griffin et al., 2012) and expression of Lys-oxoglutarate reductase/saccharopine dehydrogenase, respectively (Galili, 2002; Stepansky et al., 2005). Here, Lys concentrations are not significantly different between plant lines and thus an increase of enzyme contents appears more plausible. The resulting metabolic flux through the Lys pathway is probably not large, though. In fact, in the hemihomozygous line idh ICDH, the 13C-amount is increased two-fold only upon labelling (Fig. 3). Lys is not a very abundant amino acid and thus its biosynthetic rate is certainly < 0.05 μmol m−2 s−1, which is the average value of total leaf N assimilation (Tcherkez & Hodges, 2008). In other words, it is plausible that in hemihomozygous lines, the Lys pathway and other respiratory bypasses occur concomitantly.

The Lys pathway is comparable but less efficient than usual respiratory metabolism. In fact, the whole pathway may be summarized as follows: inline image. First, it evolves CO2 during both Lys synthesis and degradation and consumes bicarbonate, giving a net CO2 evolution rate of 1 mol CO2 mol−1 Lys, equal to I(C)DH-catalyzed activity. Secondly, both Lys synthesis and degradation involve NH2-group exchanges, which are perfectly balanced; that is, the net output of the Lys pathway is the production of 1 mol 2OG mol−1 Lys. Thirdly, the Lys pathway considered from 2 PEP to 2OG, has no net ATP production and yields one reductive equivalent; the ‘ordinary’ pathway from 2 PEP to 2OG (with PEPC-catalyzed oxaloacetate production and I(C)DH activity) yields one ATP and two reductive equivalents. There is some uncertainty about the final ATP budget because different NAD(P)H dehydrogenases (complex I, external dehydrogenases) and the alternative oxidase may be involved, but overall it is likely that the Lys pathway is energetically less favourable.

Rationale and perspectives

Taken as a whole, the influence of the restriction in I(C)DH activity appears to be modest. In addition to being feebly sensitive to changes in that step (low control coefficient), leaf respiration seems to be further sustained by the Lys bypass. These results help to explain why metabolic (respiratory) mutations are associated with negligible phenotypic effects. However, two critical issues remain: Why is Lys metabolism a preferential, alternative respiratory pathway? Is the catabolism of other amino acids also involved? Significant effects are seen on aromatic amino acids, Asn and Met in hemihomozygous lines (Fig. 2). While alterations in Asn and Met content are not surprising (both derived from Asp, as Lys is), the effect on aromatic amino acids might have originated from the more intense PEPC activity, which abstracts the precursor (phosphoenolpyruvate) of shikimate synthesis to sustain Asp production. The reasons that might explain why the Lys pathway is privileged over other catabolic routes are the following: (1) the PEPC flux is always relatively large in leaves (≈5% of photosynthesis; Raven & Farquhar, 1990), thus providing considerable amounts of Asp; (2) amongst Asp-derived amino acids, Lys catabolism evolves 2OG; and (3) Lys synthesis does not require reduced one-carbon units (contrary to Thr and Ile that also derive from Asp), which are likely to be limited (see the section ‘Pleiotropic effects of respiratory mutations’). Nevertheless, further examination of in vivo respiratory fluxomics is needed to provide a better picture of integrated catabolic pathways in plant leaves. Therefore, future high-resolution 13C-labelling dynamics will be highly informative and likely to emphasize the involvement of catabolic recycling in feeding leaf respiration.


The authors acknowledge the support of the French Ministry of Higher Education and Research and the Research National Agency for their financial support through a PhD grant and a research project funding (under contract no. 08-330055), respectively.