Research Article
8 September 2011

Comparative 13C Metabolic Flux Analysis of Pyruvate Dehydrogenase Complex-Deficient, l-Valine-Producing Corynebacterium glutamicum

ABSTRACT

l-Valine can be formed successfully using C. glutamicum strains missing an active pyruvate dehydrogenase enzyme complex (PDHC). Wild-type C. glutamicum and four PDHC-deficient strains were compared by 13C metabolic flux analysis, especially focusing on the split ratio between glycolysis and the pentose phosphate pathway (PPP). Compared to the wild type, showing a carbon flux of 69% ± 14% through the PPP, a strong increase in the PPP flux was observed in PDHC-deficient strains with a maximum of 113% ± 22%. The shift in the split ratio can be explained by an increased demand of NADPH for l-valine formation. In accordance, the introduction of the Escherichia coli transhydrogenase PntAB, catalyzing the reversible conversion of NADH to NADPH, into an l-valine-producing C. glutamicum strain caused the PPP flux to decrease to 57% ± 6%, which is below the wild-type split ratio. Hence, transhydrogenase activity offers an alternative perspective for sufficient NADPH supply, which is relevant for most amino acid production systems. Moreover, as demonstrated for l-valine, this bypass leads to a significant increase of product yield due to a concurrent reduction in carbon dioxide formation via the PPP.

INTRODUCTION

The amino acid l-valine is used as part of cosmetics and herbicides, but its main application is within the pharmaceutical industry, for example, as part of the antiviral drugs valacyclovir and lopinavir, which are applied against herpes simplex virus and HIV, respectively (9, 27). Therefore, it is aimed to replace the current production technique based on extraction from animal resources, such as chicken feathers, with a microbial cultivation process based on renewable herbal resources. Recently, several attempts were undertaken to develop an appropriate production process using the well-known amino acid-producing bacterium Corynebacterium glutamicum (28).
As in many microorganisms and plants, 1 mol l-valine is synthesized in C. glutamicum by the condensation of 2 mol pyruvate to acetolactate, which is further converted within three enzymatic reaction steps using 2 mol NADPH (Fig. 1). Based on the network of wild-type C. glutamicum ATCC 13032, the theoretically maximal yield was calculated to be 0.87 mol l-valine per mol glucose (1). Efficient l-valine formation was shown to be based on a sufficient availability of the precursor pyruvate (6), an appropriate overexpression of genes encoding the pathway enzymes (4, 19), and of an adequate supply with the cofactor NADPH (1).
Fig. 1.
Fig. 1. Selected part of the central metabolism in C. glutamicum. The l-valine pathway is marked in yellow. The enzymes inactivated or overexpressed in some of the analyzed strains are shown in green. Heterologous transhydrogenase genes (blue) are expressed in the strain C. glutamicum Δace Δpqo(pJC4ilvBNCE)(pBB1pntAB). Abbreviations: EMP, glycolysis; PPP, pentose phosphate pathway; TCA, tricarboxylic acid cycle; PTS, phosphotransferase system; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PK, pyruvate kinase; PDHC, pyruvate dehydrogenase complex; PQO, pyruvate-quinone oxidoreductase; AK, acetate kinase; PTA, phosphotransacetylase; CS, citrate synthase; PEPCx, phosphoenolpyruvate carboxylase; PEPCk, phosphoenolpyruvate carboxykinase; PCx, pyruvate carboxylase; ODx, oxaloacetate decarboxylase; ME, malic enzyme; MQO, malate-quinone oxidoreductase; MDH, malate dehydrogenase; AHAS, acetohydroxyacid synthase; AHAIR, acetohydroxyacid isomeroreductase; DHAD, dihydroxyacid dehydratase; TA, valine transaminase; THD, transhydrogenase.

Three different approaches reported to ensure sufficient pyruvate availability in C. glutamicum.

(i) Inactivation of pantothenate synthesis. Pantothenate is one building block of coenzyme A, and its formation can, e.g., be inactivated by the deletion of the panB gene, encoding the first reaction in pantothenate biosynthesis (29). Hence, by decreasing the amount of coenzyme A (CoA), the efflux into the tricarboxylic acid (TCA) cycle via the pyruvate dehydrogenase complex (PDHC) reaction can be decreased. This approach led to an intracellular accumulation of 14.2 mM pyruvate in the strain C. glutamicum ΔpanB ΔilvA with the supplementation of 0.1 μM pantothenate (2).
(ii) Inactivation of PDHC. The deletion of the aceE gene, which encodes one out of three PDHC subunits, showed a significantly higher pyruvate accumulation of 25.9 mM (6). For the growth of the cells, the complete PDHC inactivation requires supplementation with an acetyl-CoA source.
(iii) Avoiding TCA cycle efflux. In C. glutamicum, PDHC activity can be bypassed by pyruvate-quinone oxidoreductase (PQO), which is active at high intracellular pyruvate concentrations (32). Thus, the inactivation of PQO next to PDHC completely avoids the TCA cycle as the main competing pyruvate sink (5).
The l-valine formation pathway is encoded by the genes ilvBNCDE. Different genes encoding the enzymes catalyzing l-valine synthesis were analyzed in pantothenate auxotrophic strains (12, 29) and in PDHC-deficient strains, and in the latter case the overexpression of the genes ilvBNCE was shown to be most beneficial (4).
Besides precursor and pathway optimization, the availability of cofactors may strongly influence product formation (24, 30). The demand of 2 mol NADPH per mol l-valine may be met by forcing carbon flux via the pentose phosphate pathway (PPP). Interestingly, the deletion of phosphoglucoisomerase in PDHC-deficient strains showed a comparatively high l-valine yield but slow product formation and reduced strain stability during fermentations without yeast extract supplementation (1).
In this work, we applied 13C metabolic flux analysis (13C MFA) (35, 39) to the wild type and different PDHC-deficient strains, focusing especially on the flux ratio between glycolysis and PPP. The heterologous expression of a transhydrogenase was analyzed as an alternative NADPH source. Therefore, the E. coli transhydrogenase PntAB was expressed in PDHC-deficient l-valine producers. Flux analysis was performed after growth during l-valine formation to obtain a detailed quantitative understanding of the cellular metabolism in the production phase.

MATERIALS AND METHODS

Strains.

Wild-type C. glutamicum ATCC 13032 and the PDHC-deficient strains C. glutamicum ΔaceE (32), C. glutamicum ΔaceE(pJC4ilvBNCE), C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE) (5, 6), and C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE)(pBB1pntAB) (details for construction are below) were compared in this study (modified reaction steps are highlighted in Fig. 1).

Construction of a pntAB expression plasmid and measurement of transhydrogenase activity.

For the expression of the membrane-bound E. coli transhydrogenase genes pntAB in C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE), the plasmid pBB1pntAB was constructed. Therefore, a 2,985-bp fragment containing the genes pntAB with the native ribosome binding site was amplified from plasmid pEKEx2-pntAB (15) via PCR using the primers pntAB fow (5′-CATGCCTGCAGTCATCAATAAAACCG-3′) and pntAB rev (5′-GTACGCTGCAGTCTTACAGAGCTTTCAGG-3′). The resulting fragment was cut with PstI (restriction sites are underlined) and cloned into PstI-restricted plasmid pBB1 (17). The resulting plasmid, pBB1pntAB, was verified by sequencing (MWG Biotech). Plasmid pBB1pntAB expresses pntAB constitutively under the control of Ptac and is compatible with plasmid pJC4ilvBNCE.
To verify the successful expression of pntAB, transhydrogenase activities of C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE)(pBB1pntAB) and C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE) were determined. For this purpose, both strains were cultivated in CGXII medium (10) with 40 g liter−1 glucose and 5 g liter−1 acetate at 30°C as 50-ml cultures in 500-ml baffled Erlenmeyer flasks on a rotary shaker at 120 rpm. After an optical density at 600 nm (OD600) of about 8 was achieved, cells were harvested by centrifugation for 10 min at 4,500 × g, washed once with 25 ml 0.2 M Tris-HCl (pH 7.4), centrifuged again, and resuspended in 1 ml of the same buffer. The cell suspension was transferred into 2-ml screw-cap vials together with 250 mg of glass beads (diameter, 0.1 mm; Roth) and subjected to mechanical disruption three times for 30 s at speed setting 6.5 with a RiboLyser (Hybaid) at 4°C with intermittent cooling on ice for 5 min. Cell debris was removed by centrifugation for 15 min at 4,500 × g and 4°C. The resulting cell extract was subjected to ultracentrifugation for 45 min at 45,000 × g and 4°C. The sedimented membranes were resuspended in 0.5 ml 10 mM Tris-HCl (pH 8.0) and used for the measurement of transhydrogenase activity, which was performed as described previously (15). For both strains, three biological and two technical replicates were performed. The protein concentration was quantified with the bicinchoninic acid (BCA) protein assay (Pierce) with bovine serum albumin as the standard. Assays were linear over time and proportional to the protein concentration. C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE) showed no transhydrogenase activity, whereas C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE) (pBB1pntAB) showed 0.12 ± 0.03 U per mg protein, verifying the successful expression of pntAB.

Cultivation conditions.

A first preculture of 50 ml LB medium (31) in 500-ml shake flasks without baffles was inoculated with 500 μl cryoculture and incubated for 10 h at 30°C and 150 rpm. One ml of this culture was transferred to 50 ml modified CGXII minimal medium. The second preculture was incubated under the same conditions for 15 h and then used for the inoculation of the main cultures. CGXII medium according to Bartek et al. (1) contained 222 mM glucose (40 g liter−1) and 100 mM acetate (6 g liter−1) as carbon sources. Morpholinepropanesulfonic acid (MOPS) buffer (42 g liter−1) was used in the second preculture. All MFA experiments were performed in 3.6-liter vessels (Sixfors; Infors) with working volumes of 1 l. The dissolved-oxygen concentration was maintained above 30% at 30°C and pH 7.2.
In this study, we aimed to analyze the metabolic flux pattern of PDHC-deficient strains in the l-valine production phase, where these strains cannot grow any longer (3). Thus, the cells first had to be grown in the presence of acetate using unlabeled glucose to form biomass. After metabolic quasi-stationary conditions were achieved in the l-valine production phase, unlabeled glucose was exchanged by labeled glucose for flux determination. Thus, the cultivation of the PDHC-deficient strains consisted of the following four phases. (i) For the initial batch phase, 8 g glucose per liter and 6 g acetate per liter were provided. (ii) After the depletion of acetate, the batch phase was followed by an acetate feed using glacial acetic acid as the acetate source. Additional acetate (2.4 g liter−1) was fed within 4 h (0.6 g h−1) to increase cell density. (iii) Within the third cultivation phase, the remaining glucose was used for product formation without growth. (iv) A mixture of differently labeled glucose species was added directly after the depletion of the unlabeled glucose. Glucose depletion was determined by frequent sampling and the demand for oxygen. The oxygen demand, as shown by agitation, is decreased within seconds after glucose depletion. Sampling for flux determination was done in this fourth phase of cultivation between 2 and 4 h after the addition of the labeled glucose. The labeling pattern of each cultivation was analyzed in six technical replicates.
For the wild-type strain, metabolic fluxes were analyzed during growth. Hence, in this case the bioreactor medium contained exclusively 111 mM labeled glucose (20 g liter−1) as the C source, and sampling for metabolic flux analysis was done 8.5 h after inoculation during the exponential growth phase. Also for the wild type, six replicates were analyzed in parallel.
The labeling mixture contained 4% unlabeled [U-12C]glucose, 60% 1-13C-labeled glucose, and 36% uniformly labeled [U-13C]glucose for the wild type and all PDHC-deficient mutant strains. Glucose compositions were determined by conducting a priori experimental design studies as described elsewhere (23). Optimal mixtures yielding the highest information content in expected labeling measurements were calculated separately for each strain construct (see Fig. S1 in the supplemental material). However, this calculation showed very comparable results, thus the same mixture of labeled glucose could be applied in every experiment.

Analytics for intracellular metabolites.

For analyzing the 13C labeling state of intracellular metabolites, samples of 5 ml were quenched in cold methanol solution (−50°C, 60% [vol/vol]) to immediately stop cell metabolism (8). Subsequently, cells were separated from the cultivation medium by centrifugation, and intracellular metabolites were extracted by a methanol-chloroform extraction procedure (34). The cell pellet was resuspended in 1 ml ice-cold TE buffer (10 mmol Tris-HCl per liter, 1 mmol Na-EDTA per liter, pH 7.6) and 1 ml methanol solution (−20°C, 50% [vol/vol]), and an extraction volume of 2 ml chloroform (−20°C) was added. The extraction of intracellular metabolites was carried out by the overhead shaking of the homogenized solution (overhead shaker; Heidolph Instruments, Schwabach, Germany) for 2 h (−20°C). Following a centrifugation step (10,286 × g, 10 min, −20°C; Avanti30; Beckman Coulter GmbH, Krefeld, Germany), the aqueous methanol phase was separated from the chloroform phase, filtered (0.2 μm; FP30; Schleicher & Schuell, Dassel, Germany), and immediately analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS).
LC-MS/MS analysis was performed on an Agilent 1100 high-performance liquid chromatography (HPLC) system (Agilent Technologies, Waldbronn, Germany) coupled with an API 4000 (Applied Biosystems, Concord, Canada) mass spectrometer equipped with a TurboIon spray source. For the separation of metabolites from the Embden-Meyerhof-Parnas pathway (EMP), PPP, and TCA cycle, an ion pair chromatography method with tributylamine as the ion pair reagent was employed (18). The separation of amino acids was achieved by ion-exclusion chromatography (33). The injection of 10 μl of cooled samples (4°C) was realized by a temperature-controlled autosampler. Data were acquired and evaluated via Analyst software (version 1.4; Applied Biosystems, Concord, Canada). Specific mass isotopomer distributions were determined based on the selected reaction monitoring (SRM) transition pattern of each metabolite as described elsewhere (26).

Determination of extracellular rates.

Biomass formation was determined by measuring the OD546 (UV-1700; Shimadzu, Kyoto, Japan). Cell dry weight (CDW) was calculated using the experimentally determined factor of 0.33 g CDW per 1 OD unit. Glucose concentration was determined using the AccuChek sensor and the Ebio compact system (Eppendorf, Hamburg, Germany). Concentrations of extracellular organic acids were monitored using an Aminex HPX-87H column (Bio-Rad, München, Germany) eluted at 40°C with 0.1 mol H2SO4 per liter. UV absorption was detected at 215 nm. Specific rates for growth, uptake of glucose, and by-product formation were determined based on the measured process data during the 13C MFA experiments.

13C metabolic flux analysis.

13C MFA is a well-established method, and details concerning the general modeling framework of classical stationary 13C labeling experiments can be found elsewhere (23, 36-38). Briefly, to determine the unknown metabolic fluxes from the measured uptake and excretion rates as well as from the labeling data, a mathematical modeling and computational simulation procedure has to be applied. The modeling approach basically relies on the formulation of mass/isotopomer balancing, and due to the nonlinear nature of the balance equations (37), the fluxes cannot be explicitly expressed as functions of the measured labeling patterns. Thus, an iterative procedure has to be applied for flux estimation, i.e., minimizing the difference between the measured and simulated 13C labeling distributions. In particular, a hybrid optimization strategy consisting of an evolutionary algorithm complemented by a multistart gradient-based sequential quadratic programming (SQP)-like optimization routine is applied. To derive confidence regions for the unknown metabolic fluxes, a sensitivity-based statistical analysis relying on the standard linearization approach (38) was carried out. The computational workflow was performed with the software package 13CFLUX2 (www.13cflux.net).
From the estimated flux values and corresponding standard deviations, flux ratios were determined by applying the Gaussian law of error propagation.

Metabolic network models.

As a basis for the experimental design studies, metabolic models were set up containing the EMP, PPP, TCA cycle, anaplerosis, and l-valine biosynthesis reactions for C. glutamicum. The formation of extracellular by-products such as l-alanine and 2-ketoisovalerate was included. Additionally, information on inactive reaction steps was incorporated according to the genetic background of each strain. The PDHC was inactive in all strains except for wild-type C. glutamicum (see Table S1 in the supplemental material). The reaction catalyzed by the pyruvate-quinone oxidoreductase (see Table S1) was not considered for modeling the strains C. glutamicum ΔaceE Δpqo (pJC4ilvBNCE) and C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE)(pBB1pntAB).
For estimations of intracellular fluxes based on the measured isotopic stationary labeling data, a reduced network without the TCA cycle reactions was used. All applied network models, including C atom transitions, are given in Table S1 and S2 in the supplemental material. Linear relationships between metabolite demand and biomass growth are assumed for the determinatio n of effluxes from precursor metabolites into biomass synthesis in the wild-type strain (see Table S3 in the supplemental material).

RESULTS

Growth, product formation, and, most relevantly, the flux ratio between glycolysis and PPP were analyzed in wild-type C. glutamicum and several PDHC-deficient production strains. The single-deletion mutant C. glutamicum ΔaceE was compared to the same host harboring the plasmid pJC4ilvBNCE and the double deletion mutant C. glutamicum ΔaceE Δpqo also harboring this plasmid. Furthermore, the strain C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE)(pBB1pntAB) with the additional expression of an E. coli transhydrogenase was analyzed (Fig. 1).

Biomass and product formation.

The wild type showed a constant growth rate (μ) of 0.43 h−1 until glucose depletion, which was already observed in former experiments (2). In contrast, no or only negligible growth was observed for the PDHC-deficient strains, at the latest a few hours after acetate depletion (Fig. 2). A final optical density (OD) of 50 was found for the wild type. A final OD of around 30 was detected for the PDHC-deficient strains except for the strain C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE)(pBB1pntAB), which showed a final OD of 20. Here, reduced growth and a comparable late start of l-valine formation may be caused by the additional plasmid harboring the transhydrogenase gene. The delayed start of product formation, i.e., 15 h after inoculation, resulted in a reduced total amount of l-valine (150 mM for the strain C. glutamicum ΔaceE(pJC4ilvBNCE) and 125 mM for C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE)(pBB1pntAB).
Fig. 2.
Fig. 2. Cultivation profiles of analyzed C. glutamicum strains. The arrow shows when labeled glucose was added; the dotted red line shows the time of sampling. Strains were C. glutamicum ATCC 13032 (wild type) (a), C. glutamicum ΔaceE (b), C. glutamicum ΔaceE(pJC4ilvBNCE) (c), C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE) (d), and C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE)(pBB1pntAB) (e). Symbols: optical density (green diamonds), glucose (blue circles), l-valine (orange squares), l-alanine (gray hexagons), ketoisovalerate (purple triangles), acetate (light blue inverted triangles) (only PDHC-deficient strains).
In the presence of labeled glucose, constant specific rates for glucose uptake and (by-)product formation were determined (Table 1 ). The most relevant by-product was l-alanine, which was formed at a rate of 0.26 mmol gCDW−1 h−1 when cultivating the strain C. glutamicum ΔaceE. All other strains showed no or only negligible formation of l-alanine and ketoisovalerate and hence much higher rates of l-valine synthesis. A maximum specific l-valine production rate of 0.65 mmol gCDW−1 h−1 was observed for the double deletion mutant C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE) and 0.58 mmol gCDW−1 h−1 with additional transhydrogenase gene (pntAB) expression, which is comparable to that of the strain without transhydrogenase expression. In all cases carbon recovery ratios were found in a range between 100 and 120% (see Fig. S2 in the supplemental material).
Table 1.
Table 1. Experimentally determined (exp) and simulated (sim) substrate uptake and product excretion rates during fed-batch cultivation of C. glutamicum wild-type and l-valine producer strainsc
Rate measuredUptake and production excretion rates of:
ATCC 13032ΔaceΔace(pJC4ilvBNCE)ΔaceΔpqo(pJC4ilvBNCE)Δace Δpqo(pJC4ilvBNCE) (pBB1pntAB)
ExpSimExpSimExpSimExpSimExpSim
uptGLC2.06 ± 0.412.28 ± 0.270.56 ± 0.120.53 ± 0.070.70 ± 0.140.70 ± 0.090.71 ± 0.140.73 ± 0.090.43 ± 0.090.51 ± 0.03
excVAL  0.10 ± 0.020.10 ± 0.020.39 ± 0.080.39 ± 0.080.64 ± 0.130.58 ± 0.130.58 ± 0.120.46 ± 0.12
excALA  0.26 ± 0.050.26 ± 0.050.01 ± 0.000.01 ± 0.000.01 ± 0.000.01 ± 0.000.01 ± 0.000.01 ± 0.00
excKIV  0.04 ± 0.010.04 ± 0.00  0.01 ± 0.000.01 ± 0.000.02 ± 0.010.00 ± 0.00
excCO21.65 ± 0.33a1.57 ± 0.25b0.83 ± 0.170.55 ± 0.080.39 ± 0.081.10 ± 0.140.71 ± 0.141.41 ± 0.180.45 ± 0.090.75 ± 0.03
a
The experimentally determined carbon dioxide formation rates are inaccurate (too low), since the paramagnetic off-gas analyzer (Fischer-Rosemount, Haan, Germany) used predominantly determines 12CO2. Hence, these rates were omitted for flux fitting.
b
The simulated carbon dioxide formation rates applying 13C MFA represent only the sum of fluxes from carbon dioxide-producing reactions in glycolysis and the pentose phosphate pathway.
c
Specific rates for the uptake of glucose and the excretion of l-valine, l-alanine, ketoisovalerate, and CO2 are denoted by uptGLC, excVAL, excALA, excKIV, and excCO2, respectively. Relative measurement errors of 20% were applied for flux fitting. Rates are given as mmol gCDW−1 h−1.

Metabolite labeling data.

The labeling fractions of intermediates from EMP and PPP in all PDHC-deficient strains were comparable to those of the wild type (see Fig. S3 in the supplemental material). The wild type was grown directly on labeled substrate for 8.5 h (doubling time [td] = 1.61 h), and this gives clear evidence that isotopic stationarity was reached in all analyzed strains in these network parts.
The situation is quite different for the TCA cycle intermediates (e.g., 2-ketoglutarate and succinate) as well as for some free amino acids (e.g., l-glutamate). Here, the unlabeled fraction (designated m + 0) was significantly larger in all PDHC-deficient strains. Most interestingly, a gradual increase in the m + 0 fractions was found along the strain series that corresponds to an increasing limitation of TCA cycle activity by the deletion of aceE, overexpression of the l-valine pathway, and deletion of pqo. Thus, it is likely that no isotopic stationarity is reached in the TCA cycle intermediates of the PDHC-deficient strains within the experiment, and that the delay in labeling enrichment is caused mainly by an insufficient flux into acetyl-CoA as the substrate of TCA.

Quantification of intracellular fluxes.

Metabolic flux analysis was started with the same network models as used in the experimental design studies, i.e., also containing the reactions of the TCA cycle and the anaplerosis of C. glutamicum. For the estimation of intracellular fluxes, all measured extracellular rates as well as the mass isotopomer data of intracellular metabolites were used. By applying a multistart strategy, repeated flux estimations were performed with different initial values for all free fluxes.
The measured mass isotopomers of TCA cycle intermediates could be explained only by increasing the degree of freedom in the model, such as using an additional influx of unlabeled material into acetyl-CoA. On the one hand, this clearly supports the hypothesis of the isotopic nonstationarity in the TCA cycle. On the other hand, the classical 13C MFA is restricted to isotopic stationary states, and a sufficient description of these metabolites can be achieved only by applying isotopic nonstationary modeling (25), which was not the intention of this study.
Hence, the network model for flux quantification was reduced to the reactions of the EMP and PPP only. It seems to be valid to use a model without regarding TCA intermediates, as during the labeling experiment glucose was the only substrate and no gluconeogenesis is expected under the conditions tested. Running the optimization with the focused network, globally optimal solutions were found in each case, leading to good and reproducible agreements of measurements and model predictions (Table 1; also see Table S4 in the supplemental material).
Figure 3 shows the resulting flux maps for the exponential growth phase of the C. glutamicum wild type and the product formation phase without growth for the PDHC-deficient strains. Absolute flux values and corresponding standard deviations are given in Table 2 . Only 20% ± 5% of glucose uptake is transferred to l-valine in the PDHC-deficient strain without the overexpression of the genes encoding the l-valine pathway (Fig. 3b). By-products such as l-alanine or ketoisovalerate were formed from the remaining carbon, but a significant efflux into the TCA cycle was estimated as well, which is in accordance with the measured entrance of labeling material into TCA cycle intermediates (see Fig. S2 in the supplemental material). l-Valine formation increased to 56% ± 13% and 79% ± 20% of glucose uptake in the presence of the plasmid pJC4ilvBNCE (Fig. 3c and d). The additional expression of the E. coli transhydrogenase pntAB genes even resulted in a carbon flux of 92% ± 23% toward l-valine (Fig. 3e), which exceeded the theoretical maximum yield for the C. glutamicum wild type under nongrowing conditions. Clearly, the estimation of the theoretical maximum yield depends on the network structure, and hence the directly comparable upper limit for the strain possessing transhydrogenase activity is 1 mol l-valine per mol glucose (1).
Fig. 3.
Fig. 3. Quantification of intracellular fluxes in wild-type C. glutamicum and different l-valine production strains. (a to e) Estimated flux distributions. Specific values denote the molar percentage of glucose uptake. (f) Absolute flux values for NADPH-related reactions.
Table 2.
Table 2. Estimated intracellular metabolic fluxes of C. glutamicum wild-type and l-valine producer strainsa
ReactionbATCC 13032ΔaceΔace(pJC4ilvBNCE)Δace Δpqo(pJC4ilvBNCE)Δace Δpqo(pJC4ilvBNCE)(pBB1pntAB)
PGI0.63 ± 0.130.12 ± 0.030.00 ± 0.05−0.09 ± 0.060.22 ± 0.03
PFK1.54 ± 0.210.39 ± 0.050.47 ± 0.060.46 ± 0.060.42 ± 0.02
FBA1.54 ± 0.210.39 ± 0.050.47 ± 0.060.46 ± 0.060.42 ± 0.02
TPI1.54 ± 0.210.39 ± 0.050.47 ± 0.060.46 ± 0.060.42 ± 0.02
GAPDH3.45 ± 0.480.92 ± 0.131.17 ± 0.151.19 ± 0.150.93 ± 0.05
ENO2.89 ± 0.500.92 ± 0.131.17 ± 0.151.19 ± 0.150.93 ± 0.05
PKNDNDND0.46 ± 0.07ND
PGD1.57 ± 0.250.41 ± 0.070.70 ± 0.120.83 ± 0.130.29 ± 0.03
RPE0.95 ± 0.170.27 ± 0.050.47 ± 0.080.55 ± 0.080.15 ± 0.02
RPI0.63 ± 0.100.14 ± 0.020.23 ± 0.040.28 ± 0.040.10 ± 0.01
TAL0.42 ± 0.090.14 ± 0.020.23 ± 0.040.28 ± 0.040.10 ± 0.01
TKT10.53 ± 0.090.14 ± 0.020.23 ± 0.040.28 ± 0.040.10 ± 0.01
TKT20.53 ± 0.090.14 ± 0.020.23 ± 0.040.28 ± 0.040.10 ± 0.01
bsVAL1 0.14 ± 0.000.39 ± 0.080.59 ± 0.130.46 ± 0.00
bsVAL2 0.14 ± 0.000.39 ± 0.080.59 ± 0.130.46 ± 0.00
bsVAL3 0.10 ± 0.020.39 ± 0.080.58 ± 0.130.46 ± 0.12
bsALA 0.26 ± 0.050.01 ± 0.000.01 ± 0.000.01 ± 0.00
a
Fluxes are given in mmol gCDW−1 h−1. ND, not determinable, i.e., flux is not statistically identifiable.
b
PGI, phosphoglucose isomerase; PFK, phosphofructokinase; FBA, fructose bisphosphate adolase; TPi, triose phosphate isomerase; ENO, enolase; PK, pyruvate kinase; PGD, 6-phosphogluconate dehydrogenase; RPE, ribulose-5-phosphate 3-epimerase; RPI, ribose-5-phosphate isomerase; TAL, transaldolase; TKT1, transketolase I; TKT2, transketolase II; bsVAL1, bsVAL2, and bsVAL3, biosynthesis pathway l-valine; and bsALA, biosynthesis l-alanine.
In this study, the carbon flux along the PPP was of special interest, since usually significant amounts of the cofactor NADPH are generated in this pathway, which are needed for l-valine synthesis (Fig. 1). Whereas 69% ± 14% of the glucose uptake was directed through the PPP in the wild type, the split ratio gradually increased after PDHC inactivation to 78% ± 18% and further increased to 100% ± 21% and 113% ± 22% in strains with the plasmid-encoded enhancement of l-valine biosynthesis. The additional overexpression of genes encoding the transhydrogenase led to a reduced ratio of 57% ± 6% toward PPP, which is even lower than the wild-type value. Comparing the demand of NADPH for l-valine synthesis with the supply via the PPP, a good correspondence is found, except for the transhydrogenase strain (Fig. 3f). Without transhydrogenase, the flux through NADPH-forming reactions was always found to be higher than the fluxes accompanied with NADPH consumption for l-valine production.

DISCUSSION

The calculation of intracellular reaction rates presented here is based on the classical 13C MFA approach that strictly relies on the assumption of isotopic stationarity (25). As already shown in former experiments following the labeling of transients in a C. glutamicum l-lysine producer and E. coli wild type (25, 26), isotopic stationarity is rapidly achieved (i.e., within a few minutes) for intermediates of the EMP and PPP. In contrast, the time constants for TCA cycle intermediates are much higher, leading to significantly delays in labeling dynamics; isotopic stationary states are not reached until 1 h of labeling. The experimental conditions chosen here even enforce this effect, since glucose uptake during the l-valine production phase in the PDHC-deficient strains is clearly reduced compared to the exponential growth of the wild type (Table 1). Furthermore, PDHC inactivation interrupts the most important connection between glycolysis and the TCA cycle. Thus, isotopic stationarity cannot be achieved in PDHC-deficient strains in the phase of product formation during a reasonable experimental labeling duration. Besides anaplerotic reactions, no flux from pyruvate into the TCA cycle is possible at all after the deletion of both aceE and pqo. Hence, we considered only the intermediates of EMP, PPP, and l-valine biosynthesis for calculating the in vivo fluxes in the C. glutamicum strains under investigation.
The consistency of this approach also is shown in the resulting flux distributions (Fig. 3). Besides the wild type, only the strain C. glutamicum ΔaceE shows a significant efflux into the TCA cycle via anaplerosis and the pyruvate-quinone oxidoreductase reaction (encoded by pqo). This flux is strongly reduced after the overexpression of the l-valine pathway genes and is statistically negligible after pqo deletion.
The formation of amino acids is dependent on cofactor availability; in particular, the synthesis of 1 mol l-valine in C. glutamicum consumes 2 mol NADPH. Three sources of NADPH are known in C. glutamicum. It can be formed by isocitrate dehydrogenase within the TCA cycle (11). Clearly, this is disadvantageous for l-valine formation, since the pyruvate pool should be redirected mainly toward l-valine rather than into the TCA cycle. Alternatively, the malic enzyme can form NADPH within a reaction cycle, including pyruvate carboxylase and malate dehydrogenase (13). The third and most relevant source is the PPP (21), where NADPH is formed by the reactions of glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase.
Compared to that of l-valine, the formation of l-lysine with the consumption of 4 mol NADPH is even more demanding (15). Several l-lysine-producing C. glutamicum strains have been analyzed by 13C MFA, resulting in split ratios between 60 and 70% PPP (16, 20). Compared to the results shown here, the split ratio of the l-lysine producers are significantly lower. However, in a former study a flux ratio of 66% PPP during l-lysine production was found, but they also found a reduced split ratio of 36% for the wild type and just 25% for the PPP under l-glutamate-forming conditions (22). Certainly a direct comparison to our results is difficult, since all of these 13C MFA studies were conducted in either shake flasks or chemostat experiments.
A split ratio of 60% during exponential growth was determined for a pantothenate-overproducing C. glutamicum strain cultivated in a 2-liter bioreactor under fed-batch conditions applying pure stoichiometric flux analysis (7). The ratio increased to 75% during pantothenate production, which also can be explained by the NADPH-supplying function of the PPP as described here. Pantothenate synthesis branches off the l-valine pathway at the common precursor ketoisovalerate and requires 2 mol NADPH.
The EMP/PPP split ratios found in this study are, in their entirety, comparatively high, which might be due to the bioreactor cultivation and in particular the applied fed-batch approach. However, the tendencies found are substantiated by the mentioned former studies, and the estimated fluxes are statistically significant (Table 2). While the wild type showed a split ratio of 69% between the EMP and PPP without any by-product excretion, the PPP flux increased to 78% in the strain C. glutamicum ΔaceE. In addition to l-valine, this strain excreted l-alanine and ketoisovalerate (Fig. 2), the formation of which is also NADPH dependent (1 mol NADPH per mol for each). The PPP flux further increased to 100% in the strain C. glutamicum ΔaceE(pJC4ilvBNCE) and showed a maximum of 113% along the strain series in C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE). It can be concluded that this is caused mainly by the increasing formation of l-valine. Generally speaking, the PPP flux increased when an additional NADPH sink was expressed.
The situation was different for the strain C. glutamicum ΔaceE Δpqo(pJC4ilvBNCE)(pBB1pntAB). The additional expression of the E. coli pntAB genes led to transhydrogenase activity in C. glutamicum. The enzyme reversibly catalyzes the reduction of NADP with NADH (14). NADH is formed within the EMP via the glyceraldehyde 3-phosphate dehydrogenase reaction (Fig. 1). Since the analyzed cells were in the production phase under nongrowing conditions, the only NADH sinks are certain requirements for maintenance metabolism. Therefore, it can be assumed that a surplus of NADH led to a shift of the reaction equilibrium of the transhydrogenase toward NADPH formation. Consequently, NADPH formation by PPP was less relevant to meet the NADPH demand of l-valine formation, and the flux ratio was shifted back toward the EMP.
Alternatively, the inactivation of phosphoglucoisomerase (Fig. 3) has been discussed (5). l-Valine formation was significantly increased in PGI-deficient strains, but complex medium compounds had to be supplemented, and reduced growth and production rates were observed (1). Most likely, these cells produce too much NADPH due to a strongly enhanced flux via the PPP and are impaired by enforced NADPH regeneration under nongrowing conditions.
In contrast, the strain C. glutamicum ΔaceE Δpqo (pJC4ilvBNCE)(pBB1pntAB) showed a comparatively high l-valine formation rate and only minor differences in growth compared to those of the other PDHC-deficient strains. The slightly reduced growth and thus a delayed start of product formation may at least partly be explained by the metabolic burden of the second plasmid. The high l-valine yield can be explained by its increased capability to serve the high NADPH demand, since the strain is (in)directly able to use both cofactors for l-valine formation, NADH formed in glycolysis and NADPH formed in the PPP. Hence, it can be concluded that the transhydrogenase activity leads to more flexibility in adapting the EMP/PPP split ratio to the demand of NADPH for growth, maintenance, and l-valine production. At the same time, NADPH formation is decoupled from carbon dioxide formation within the PPP, which significantly improves the l-valine yield.
Finally, our results underline the high importance of cofactor supply for l-valine formation. NADPH supply can be most sufficiently ensured by the expression of the transhydrogenase genes pntAB. Moreover, the flux ratios determined by 13C MFA showed that the split ratio between EMP and PPP is strongly influenced by the NADPH demand in the investigated l-valine producer strains.

Acknowledgements

This work was financially supported by the Fachagentur Nachwachsende Rohstoffe (Agency for Renewable Resources) of the BMVEL, German Federal Ministry of Food, Agriculture and Consumer Protection (grant 04NR003/22000304), and by Evonik Degussa GmbH.
We thank Robert Gerstmeir and Andreas Karau from Evonik Degussa GmbH for fruitful cooperation and the valuable discussion of results.

Supplemental Material

File (supplemental_data_bartek_et_al_2011_r1.doc)
ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.

REFERENCES

1.
Bartek T. et al. 2010. Importance of NADPH supply for improved l-valine formation in Corynebacterium glutamicum. Biotechnol. Prog. 26:361–371.
2.
Bartek T., Makus P., Klein B., Lang S., and Oldiges M. 2008. Influence of l-isoleucine and pantothenate auxotrophy for l-valine formation in Corynebacterium glutamicum revisited by metabolome analyses. Bioprocess Biosyst. Eng. 31:217–225.
3.
Bartek T. et al. 2010. Studies on substrate utilisation in l-valine-producing Corynebacterium glutamicum strains deficient in pyruvate dehydrogenase complex. Bioprocess Biosyst. Eng. 33:873–883.
4.
Bartek T. et al. 2010. Analysing overexpression of l-valine biosynthesis genes in pyruvate-dehydrogenase-deficient Corynebacterium glutamicum. J. Ind. Microbiol. Biotechnol. 37:263–270.
5.
Blombach B., Schreiner M. E., Bartek T., Oldiges M., and Eikmanns B. J. 2008. Corynebacterium glutamicum tailored for high-yield l-valine production. Appl. Microbiol. Biotechnol. 79:471–479.
6.
Blombach B. et al. 2007. l-Valine production with pyruvate dehydrogenase complex-deficient Corynebacterium glutamicum. Appl. Environ. Microbiol. 73:2079–2084.
7.
Chassagnole C., Letisse F., Diano A., and Lindley N. D. 2002. Carbon flux analysis in a pantothenate overproducing Corynebacterium glutamicum strain. Mol. Biol. Rep. 29:129–134.
8.
de Koning W. and van Dam K. 1992. A method for the determination of changes of glycolytic metabolites in yeast on a subsecond time scale using extraction at neutral pH. Anal. Biochem. 204:118–123.
9.
Eggeling L., Pfefferle W., and Sahm H. 2001. Amino acids, p. 281–303.In Ratledge K. and Bjoern C.(ed.), Basic biotechnology. Cambridge University Press, Cambridge, United Kingdom.
10.
Eikmanns B. J., Metzger M., Reinscheid D., Kircher M., and Sahm H. 1991. Amplification of three threonine biosynthesis genes in Corynebacterium glutamicum and its influence on carbon flux in different strains. Appl. Microbiol. Biotechnol. 34:617–622.
11.
Eikmanns B. J., Rittmann D., and Sahm H. 1995. Cloning, sequence-analysis, expression, and inactivation of the Corynebacterium glutamicum icd gene encoding isocitrate dehydrogenase and biochemical characterization of the enzyme. J. Bacteriol. 177:774–782.
12.
Elisáková V. et al. 2005. Feedback-resistant acetohydroxy acid synthase increases valine production in Corynebacterium glutamicum. Appl. Environ. Microbiol. 71:207–213.
13.
Georgi T., Rittmann D., and Wendisch V. F. 2005. Lysine and glutamate production by Corynebacterium glutamicum on glucose, fructose and sucrose: roles of malic enzyme and fructose-1,6-bisphosphatase. Metabolic Eng. 7:291–301.
14.
Jackson J. B. 2003. Proton translocation by transhydrogenase. FEBS Lett. 545:18–24.
15.
Kabus A., Georgi T., Wendisch V. F., and Bott M. 2007. Expression of the Escherichia coli pntAB genes encoding a membrane-bound transhydrogenase in Corynebacterium glutamicum improves l-lysine formation. Appl. Microbiol. Biotechnol. 75:47–53.
16.
Kiefer P., Heinzle E., Zelder O., and Wittmann C. 2004. Comparative metabolic flux analysis of lysine-producing Corynebacterium glutamicum cultured on glucose, or fructose. Appl. Environ. Microbiol. 70:229–239.
17.
Krause F. S. et al. 2010. Increased glucose utilization in Corynebacterium glutamicum by use of maltose, and its application for the improvement of l-valine productivity. Appl. Environ. Microbiol. 76:370–374.
18.
Luo B., Groenke K., Takors R., Wandrey C., and Oldiges M. 2007. Simultaneous determination of multiple intracellular metabolites in glycolysis, pentose phosphate pathway and tricarboxylic acid cycle by liquid chromatography-mass spectrometry. J. Chromatogr. A 1147:153–164.
19.
Magnus J. B., Oldiges M., and Takors R. 2009. The identification of enzyme targets for the optimization of a valine producing Corynebacterium glutamicum strain using a kinetic model. Biotechnol. Prog. 25:754–762.
20.
Marx A., deGraaf A.A., Wiechert W., Eggeling L., and Sahm H. 1996. Determination of the fluxes in the central metabolism of Corynebacterium glutamicum by nuclear magnetic resonance spectroscopy combined with metabolite balancing. Biotechnol. Bioeng. 49:111–129.
21.
Marx A., Eikmanns B. J., Sahm H., de Graaf A. A., and Eggeling L. 1999. Response of the central metabolism in Corynebacterium glutamicum to the use of an NADH-dependent glutamate dehydrogenase. Metabolic Eng. 1:35–48.
22.
Marx A., Striegel K., de Graaf A. A., Sahm H., and Eggeling L. 1997. Response of the central metabolism of Corynebacterium glutamicum to different flux burdens. Biotechnol. Bioeng. 56:168–180.
23.
Mollney M., Wiechert W., Kownatzki D., and de Graaf A. A. 1999. Bidirectional reaction steps in metabolic networks. 4. Optimal design of isotopomer labeling experiments. Biotechnol. Bioeng. 66:86–103.
24.
Moritz B., Striegel K., de Graaf A. A., and Sahm H. 2002. Changes of pentose phosphate pathway flux in vivo in Corynebacterium glutamicum during leucine-limited batch cultivation as determined from intracellular metabolite concentration measurements. Metabolic Eng. 4:295–305.
25.
Noack S., Noh K., Moch M., Oldiges M., and Wiechert W. 2010. Stationary versus non-stationary (13)C-MFA: a comparison using a consistent dataset. J. Biotechnol. 154:179–190.
26.
Nöh K. et al. 2007. Metabolic flux analysis at ultra short time scale: isotopically non-stationary 13C labeling experiments. J. Biotechnol. 129:249–267.
27.
Park J. H., Lee K. H., Kim T. Y., and Lee S. Y. 2007. Metabolic engineering of Escherichia coli for the production of l-valine based on transcriptome analysis and in silico gene knockout simulation. Proc. Natl. Acad. Sci. U. S. A. 104:7797–7802.
28.
Park J. H. and Lee S. Y. 2010. Fermentative production of branched chain amino acids: a focus on metabolic engineering. Appl. Microbiol. Biotechnol. 85:491–506.
29.
Radmacher E. et al. 2002. Linking central metabolism with increased pathway flux: l-valine accumulation by Corynebacterium glutamicum. Appl. Environ. Microbiol. 68:2246–2250.
30.
Ruklisha M., Paegle L., and Denina I. 2007. l-valine biosynthesis during batch and fed-batch cultivations of Corynebacterium glutamicum: relationship between changes in bacterial growth rate and intracellular metabolism. Process Biochem. 42:634–640.
31.
Sambrook J. and Russell D. W. 2001. Molecular cloning, 3rd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
32.
Schreiner M. E. and Eikmanns B. J. 2005. Pyruvate:quinone oxidoreductase from Corynebacterium glutamicum: purification and biochemical characterization. J. Bacteriol. 187:862–871.
33.
Thiele B. et al. 2008. Analysis of amino acids without derivatization in barley extracts by LC-MS-MS. Anal. Bioanal. Chem. 391:2663–2672.
34.
Villas-Bôas S. G., Hojer-Pedersen J., Akesson M., Smedsgaard J., and Nielsen J. 2005. Global metabolite analysis of yeast: evaluation of sample preparation methods. Yeast 22:1155–1169.
35.
Wiechert W. 2001. C-13 metabolic flux analysis. Metabolic Eng. 3:195–206.
36.
Wiechert W. and deGraaf A.A. 1997. Bidirectional reaction steps in metabolic networks. 1. Modeling and simulation of carbon isotope labeling experiments. Biotechnol. Bioeng. 55:101–117.
37.
Wiechert W., Mollney M., Isermann N., Wurzel W., and de Graaf A. A. 1999. Bidirectional reaction steps in metabolic networks. 3. Explicit solution and analysis of isotopomer labeling systems. Biotechnol. Bioeng. 66:69–85.
38.
Wiechert W., Siefke C., deGraaf A.A., and Marx A. 1997. Bidirectional reaction steps in metabolic networks. 2. Flux estimation and statistical analysis. Biotechnol. Bioeng. 55:118–135.
39.
Zamboni N., Fendt S.-M., Ruhl M., and Sauer U. 2009. (13)C-based metabolic flux analysis. Nat. Protoc. 4:878–892.

Information & Contributors

Information

Published In

cover image Applied and Environmental Microbiology
Applied and Environmental Microbiology
Volume 77Number 1815 September 2011
Pages: 6644 - 6652
PubMed: 21784914

History

Received: 14 March 2011
Accepted: 13 July 2011
Published online: 8 September 2011

Permissions

Request permissions for this article.

Contributors

Authors

Tobias Bartek
Institute of Bio and Geo Sciences, IBG-1, Biotechnology, Forschungszentrum Jülich, D-52425 Jülich, Germany
Present address: Lonza Custom Manufacturing, Microbial Services, Lonza AG, CH-3930 Visp, Switzerland.
Bastian Blombach
Institute of Microbiology and Biotechnology, University of Ulm, D-89069 Ulm, Germany
Siegmund Lang
Institute for Biochemistry and Biotechnology, Technical University Braunschweig, D-38106 Braunschweig, Germany
Bernhard J. Eikmanns
Institute of Microbiology and Biotechnology, University of Ulm, D-89069 Ulm, Germany
Wolfgang Wiechert
Institute of Bio and Geo Sciences, IBG-1, Biotechnology, Forschungszentrum Jülich, D-52425 Jülich, Germany
Marco Oldiges
Institute of Bio and Geo Sciences, IBG-1, Biotechnology, Forschungszentrum Jülich, D-52425 Jülich, Germany
Katharina Nöh
Institute of Bio and Geo Sciences, IBG-1, Biotechnology, Forschungszentrum Jülich, D-52425 Jülich, Germany
Stephan Noack [email protected]
Institute of Bio and Geo Sciences, IBG-1, Biotechnology, Forschungszentrum Jülich, D-52425 Jülich, Germany

Metrics & Citations

Metrics

Note:

  • For recently published articles, the TOTAL download count will appear as zero until a new month starts.
  • There is a 3- to 4-day delay in article usage, so article usage will not appear immediately after publication.
  • Citation counts come from the Crossref Cited by service.

Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. For an editable text file, please select Medlars format which will download as a .txt file. Simply select your manager software from the list below and click Download.

View Options

Figures and Media

Figures

Media

Tables

Share

Share

Share the article link

Share with email

Email a colleague

Share on social media

American Society for Microbiology ("ASM") is committed to maintaining your confidence and trust with respect to the information we collect from you on websites owned and operated by ASM ("ASM Web Sites") and other sources. This Privacy Policy sets forth the information we collect about you, how we use this information and the choices you have about how we use such information.
FIND OUT MORE about the privacy policy