H. Santos, Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa, Rua da Quinta Grande, 6, Apt 127, 2780-156 Oeiras, Portugal. Fax: + 351 21 4428766, Tel.: + 351 21 4469828, E-mail: firstname.lastname@example.org
The metabolism of glucose by nongrowing cells of Lactococcus lactis strain FI7851, constructed from the wild-type L. lactis strain MG1363 by disruption of the lactate dehydrogenase (ldh) gene [Gasson, M.J., Benson, K., Swindel, S. & Griffin, H. (1996) Lait76, 33–40] was studied in a noninvasive manner by 13C-NMR. The kinetics of the build-up and consumption of the pools of intracellular intermediates mannitol 1-phosphate, fructose 1,6-bisphosphate, 3-phosphoglycerate, and phosphoenolpyruvate as well as the utilization of [1-13C]glucose and formation of products (lactate, acetate, mannitol, ethanol, acetoin, 2,3-butanediol) were monitored in vivo with a time resolution of 30 s. The metabolism of glucose by the parental wild-type strain was also examined for comparison. A clear shift from typical homolactic fermentation (parental strain) to a mixed acid fermentation (lactate dehdydrogenase deficient; LDHd strain) was observed. Furthermore, high levels of mannitol were transiently produced and metabolized once glucose was depleted. Mannitol 1-phosphate accumulated intracellularly up to 76 mm concentration. Mannitol was formed from fructose 6-phosphate by the combined action of mannitol-1-phosphate dehydrogenase and phosphatase. The results show that the formation of mannitol 1-phosphate by the LDHd strain during glucose catabolism is a consequence of impairment in NADH oxidation caused by a highly reduced LDH activity, the transient production of mannitol 1-phosphate serving as a regeneration pathway for NAD+ regeneration. Oxygen availability caused a drastic change in the pattern of intermediates and end-products, reinforcing the key-role of the fulfilment of the redox balance. The flux control coefficients for the step catalysed by mannitol-1-phosphate dehydrogenase were calculated and the implications in the design of metabolic engineering strategies are discussed.
Lactic acid bacteria play a central role in the dairy industry because of their widespread use as starter cultures in milk fermentation. The relatively simple metabolic strategy of these organisms that utilize sugars primarily to generate energy and not for growth, makes them an attractive model system for studies aiming at the construction of improved strains by metabolic engineering. However, a deep understanding of the metabolic network as well as of the interdependence relationships among the different steps are essential to establish a rational and systematic methodology that will enable successful manipulation of carbohydrate metabolism.
Recent work aiming at the metabolic characterization of lactic acid bacteria revealed several strategies for regeneration of NAD+ during the metabolism of carbohydrates. For instance, Oenococcus oeni produces erythritol to consume the reduced coenzymes formed in the glycolytic pathway ; a strain of Lactobacillus plantarum deficient in both l- and d-lactate dehydrogenases produces mannitol (Man-ol) as an end-product of glucose catabolism , and disruption of the ldh gene in Lactococcus lactis strain NZ2007 leads to the conversion of acetate into ethanol as a rescue pathway for NAD+ regeneration . The importance of fulfilling the redox balance is also apparent from the metabolic behaviour of a L. lactis strain carrying a construct overproducing NADH oxidase; in such strain, the formation of lactate is drastically reduced, and a shift towards acetoin production is observed . The view that the NAD+/NADH ratio is a key factor controlling the glycolytic flux [5–8] was recently rationalized by the construction of a mechanistic model of glycolysis in L. lactis using experimental data collected by in vivo NMR. This model predicts that fructose 1,6-bisphosphate [Fru(1,6)P2] accumulation is largely driven by NAD+ limitation whose regeneration is dependent on downstream pyruvate reduction . Therefore, metabolic engineering strategies aimed at channelling the carbon flux from carbohydrates towards industrially desirable products (e.g. diacetyl) have to take into account the tight regulation of carbon flux by the NAD+/NADH ratio.
Although extensive knowledge on the physiology and genetics of L. lactis has become available in recent years, efficient strain manipulation is not yet a straightforward process. Reliable physiological information on carbon fluxes and intracellular metabolite concentrations, in different mutants bearing modifications in the activity of key-glycolytic enzymes, is essential as input data for the construction of a useful mathematical model for sugar metabolism in L. lactis. NMR analysis coupled to isotopic labelling has proved to be a powerful analytical tool for characterization of metabolism in several organisms, allowing the monitoring of the evolution of metabolite pools in a noninvasive manner [7–14]. In this work, we used in vivo13C-NMR to investigate glucose metabolism by nongrowing cells of a L. lactis strain in which the lactate dehydrogenase gene was disrupted with the purpose of increasing the pyruvate pool and thereby shifting the carbon flux toward the production of diacetyl . Instead of the expected enhancement in the production of the flavour compound diacetyl, Fru6P was channelled to produce high levels of Man-ol-1-P and Man-ol, as part of a strategy to compensate for the severe reduction in its ability to regenerate NAD+. The kinetics of the build-up and consumption of the intracellular pools of metabolic intermediates (Man-ol, Man-ol-1-P, Fru(1,6)P2, 3-phosphoglycerate and phosphoenolpyruvate) was characterized directly in living cells, without interfering with the biochemical functions.
Materials and methods
Organisms and growth conditions
L. lactis strains FI7851 (lactate dehydrogenase deficient; LDHd)  and MG1363 (parental strain) were grown as previously described by Neves et al. . For growth of the LDHd strain the medium was supplemented with erythromycin (5 µg·mL−1). To assess the stability of the LDHd mutation, the strain was transferred several times in chemically defined medium and the production of lactate and the activity of l-LDH were monitored after each transfer.
Cells were harvested in mid-logarithmic growth phase (D600 = 2.2), centrifuged, washed twice and suspended to a protein concentration of 13 mg·mL−1 in 50 mm KPi or Mes buffer, pH 6.5, for 13C- or 31P-NMR experiments, respectively. The experiments were performed at pH 6.5 and 30 °C as described previously . When no changes in the resonances due to end-products and intracellular metabolites were observed, the experiment was stopped and an aliquot of the cell suspension was passed through a French press (twice at 120 MPa); the resulting crude cell extract was incubated at 80 °C for 15 min in a sealed container, cooled on ice, the cell debris was removed by centrifugation and the supernatant solution (denoted hereafter as total cell extract) was stored at −20 °C. For protein quantification, cells were disrupted by treatment with 1 m NaOH and incubation at 80 °C for 2 min prior to determination by the method of Bradford .
Quantification of extracellular Man-ol and determination of the ratio between the isotopic enrichment of C1 and C6 in Man-ol-1-P
Intracellular and extracellular pools of Man-ol could not be distinguished in the NMR experiments described above. Therefore, extracellular Man-ol was quantified in a separate experiment performed with the set-up used for NMR measurements: following glucose addition, 1-mL samples were withdrawn at 2-min intervals. The samples were centrifuged immediately and the supernatants were frozen in liquid nitrogen. The concentration of Man-ol in the supernatant solutions was determined by 13C-NMR using [13C]methanol (49.3 p.p.m.) as concentration standard. The ratio between the isotopic enrichments on C1 and C6 of Man-ol-1-P was determined from 13C-NMR analysis of perchloric acid extracts derived from samples taken at times 14 and 16 min after addition of [1-13C]glucose.
Quantification of products
Lactate, acetoin, acetate, 2,3-butanediol, ethanol and formate were quantified in total cell extracts by 1H-NMR . The concentration of minor products (e.g. pyruvate) and metabolic intermediates that remained inside the cells (phosphoenolpyruvate, 3-phosphoglycerate, aspartate, succinate, and alanine) was determined from the analysis of 13C spectra of the total cell extracts, acquired using a pulse flip angle of 60° and a recycle delay of 60.5 s. It was verified that these conditions allowed full relaxation of all the relevant resonances. The concentration of labelled lactate determined by 1H-NMR was used as a standard to calculate the concentration of the other metabolites in the sample. The total amount of lactate was confirmed by enzymatic methods.
Quantification of intracellular metabolites in living cells by 13C-NMR
Due to the fast pulsing conditions used for acquiring in vivo13C-spectra, a direct correlation between concentrations and peak intensities could not be established. The correction factors that allowed the conversion of peak-areas to intracellular concentration were determined using different strategies. The correction factors for C1 and C6 of Fru(1,6)P2 (0.73 ± 0.04) were determined by slowing down the cell metabolic rate by suspending the cells in 50 mm potassium phosphate buffer prepared in 50% 2H2O. Under these conditions, the intensity of the resonances due to Fru(1,6)P2 and that of extracellular lactate remained constant for a long period of time, enabling the acquisition of fully and partially relaxed spectra, and the determination of correction factors to take into account saturation and nuclear Overhauser effects. This procedure was not applicable for the transient metabolites Man-ol and Man-ol-1-P, as the production of these compounds was inhibited in 50% 2H2O. Therefore, the correction factors for the resonances C1 and C6 of Man-ol and Man-ol-1-P (0.65 ± 0.03) were obtained from spectra of a total cell extract to which pure Man-ol and Man-ol-1-P were added to a final concentration of 50 mm. The correction factors for the resonances due to C3 of 3-phosphoglycerate and phosphoenolpyruvate (0.71 ± 0.04) were determined directly in a cell suspension of L. lactis after the metabolism of [1-13C]glucose; following glucose depletion, the intracellular pools of these metabolites reached a constant level. The quantitative kinetic data for intracellular metabolites were calculated from the areas of the relevant resonances, by applying the correction factors and comparing with the intensity of the lactate resonance in the last spectrum of the sequence. The 13C-NMR spectra used for calculating the correction factors were acquired with a 60° flip angle and a recycle delay of 1.5 s (saturating conditions) or 60.5 s (fully relaxing conditions).
Metabolite concentrations were calculated using a value of 2.9 µL·mg−1 of protein for the intracellular volume . The concentration limit for detection of intracellular metabolites under the conditions used to acquire spectra (30-s acquisition time) of living cells was ≈ 3 mm.
Preparation of perchloric acid extracts for the identification of Man-ol and Man-ol-1-P in the LDHd strain
Cell suspensions were prepared as described above, placed in the mini-bioreactor, and 20 mm[1-13C]glucose or [U-13C] glucose was added. Thirteen minutes after glucose addition, a sample was taken and intracellular metabolites were extracted by mixing immediately a perchloric acid/EDTA solution as described by Neves et al. . A sample of the extract was directly analysed by 13C-NMR spectroscopy; additionally, to facilitate the identification of labelled compounds, the extract components were separated by anion-exchange chromatography. The extract was loaded onto a QAE-Sephadex column equilibrated with 5 mm ammonium bicarbonate buffer (NH4HCO3), pH 8.0. The elution was performed with a linear gradient of 5 mm to 1 m NH4HCO3. Aliquots of each fraction were analysed by the method of Dubois et al. . Two fractions giving a positive reaction (eluted at 5 mm and 0.9–1 m NH4HCO3) were pooled separately, lyophilized, and analysed by 1H-NMR.
13C- or 31P-NMR spectra were acquired on a Bruker DRX500 spectrometer. All NMR experiments of living cells were run at 30 °C with a quadruple-nucleus probe head . Each type of NMR experiment was repeated at least twice and the results were highly reproducible. The 13C spectra of the perchloric acid extracts used for the identification of Man-ol-1-P and Man-ol were run with a 5-mm diameter selective probe head. 13C-Homonuclear correlation shift spectroscopy (13C COSY) were performed by collecting 4096 (t2) × 512 (t1). Carbon and phosphorus chemical shifts are referenced to the resonances of external methanol or external 85% H3PO4 designated at 49.3 p.p.m. and 0.0 p.p.m., respectively.
Calculation of metabolic fluxes and control coefficients
The control coefficients were calculated by the direct method described by Canela et al. . Given the large set of experimental data, the stoichiometric matrices were slightly over-determined (number of reactions distinguished = number of metabolites measured − 1), and, consequently, the metabolic fluxes were calculated by minimization of square deviations. matlab 5.3™ (MathWorks, Inc.) was used as a programming language due to the convenience offered by this environment for matricial calculus.
Enzyme activity measurements
Freshly prepared cell extracts were obtained by mechanical disruption in a French press of cell suspensions, prepared as described for the NMR experiments. Enzymes were assayed at 30 °C. The protein concentration was determined by the method of Bradford . LDH, pyruvate kinase, and glyceraldehyde 3-phosphate dehydrogenase were assayed as described by Garrigues et al. . 6-Phosphofructokinase activity was monitored by the method of Fordyce et al. , and NADH oxidase activity was determined as described by Lopez de Felipe et al. . Fructose bisphosphatase (EC 188.8.131.52) was assayed by the method described by Babul and Guixé. The forward reaction of Man-ol-1-P dehydrogenase was assayed by the method described by Lee et al. , while the backward activity of Man-ol-1-P dehydrogenase was measured using 0.3 mm NADH and the reaction was initiated by the addition of 3 mm Fru6P. For the determination of Man-ol-1-phosphatase (EC 184.108.40.206) cells were suspended in 25 mm Hepes/KOH buffer, pH 7.2, containing 5 mm MgCl2, 5 mm 2-mercaptoethanol, 1 mm EDTA, and a cocktail of protease inhibitors (2 µg·µL−1 leuphosphoenolpyruvatetin, 2 µg·µL−1 antipain, and 0.5 mm phenylmethanesulfonyl fluoride). Phosphatase activity was determined as described by Hult and Gatenbeck , and phosphate was measured by the method described by Ames .
[1-13C]Glucose (99% enrichment) and [U-13C]glucose (99%) were obtained from Campro and ISOTEC, respectively. [13C]Methanol (99%) was supplied by Sigma Co. Formic acid (sodium salt) was purchased from Merck. QAE-Sephadex A25 was obtained from Pharmacia Fine Chemicals. All other chemicals were reagent grade.
Identification of Man-ol-1-P and Man-ol as intermediates in glucose metabolism
Figure 1 shows a selection of 13C spectra acquired during the metabolism of 20 mm[1-13C]glucose by a cell suspension of the LDHd strain. Resonances due to lactate (at 20.4 p.p.m.), acetoin (18.8 and 25.5 p.p.m.), ethanol (17.3 p.p.m.) and 2,3-butanediol (17.2 p.p.m.) were detected. Furthermore, resonances due to intracellular metabolites previously assigned as the β-furanose form of [1-13C]Fru(1,6)P2 (66.4 p.p.m.) and the β-furanose form of [6-13C]Fru(1,6)P2 (65.1 p.p.m.) were apparent [7,13,25,26]. The resonances due to the α-furanose form were too weak to allow a reliable quantification. Therefore, the total pool of Fru(1,6)P2 was evaluated from the intensity of the resonances of the β-form and using the ratio of 82 : 18 reported for the β:α ratio at 30 °C . Two unknown resonances at 63.8 and 66.1 p.p.m. were clearly seen, reached intensity maxima approx. at 13 min and 17 min, respectively, and decreased thereafter to undetectable levels. We resorted to 13C-NMR analysis of perchloric acid extracts for identification of the unknown metabolite(s): a doublet centred at 66.10 p.p.m. and two single resonances at 63.92 and 63.87 p.p.m. were detected (data not shown). After passage through an anion-exchange column, the resonance at 63.87 p.p.m. was detected in the fraction that contained noncharged compounds, and the resonances at 66.10 and 63.92 p.p.m. were detected in the spectrum of the fraction containing negatively charged components. To facilitate the assignment, uniformly labelled glucose was supplied to a cell suspension and the resulting cell extract was treated in the same way. The 13C-NMR spectrum of the charged components showed six equal-intensity resonances at 71.62, 70.82, 69.97, 69.18, 66.10 and 63.92 p.p.m. with a coupling pattern revealed by 13C COSY. These data led us to suspect that the unknown compound was Man-ol-1-P, subsequently confirmed by addition of the pure compound to the sample. Therefore, resonances at 66.10 and 63.92 were assigned to C1 and C6 of Man-ol-1-P, respectively. A resonance at 63.87 in the noncharged fraction was assigned to C1 and/or C6 of Man-ol by spiking the sample with pure Man-ol. The assignment of Man-ol-1-P was also confirmed by 31P-NMR of the cell suspension; in the spectra obtained after glucose addition under anaerobic conditions, a strong resonance was detected at 3.9 p.p.m. and later firmly identified as Man-ol-1-P by spiking the respective perchloric acid extract with the pure compound.
Characterization of glucose metabolism by LDHd and parental strains under anaerobic conditions
The kinetics of glucose consumption and product formation as well as the evolution of intracellular metabolites in the LDHd and the parental strains under anaerobiosis are illustrated in Fig. 2. The end-products from [1-13C]glucose (20 mm) were acetoin (1.6 ± 0.2 mm), 2,3-butanediol (5.8 ± 0.3 mm), ethanol (8.4 ± 0.5 mm), lactate (9.9 ± 1.0 mm), acetate (2.6 ± 0.3 mm) and formate (12.8 ± 0.5 mm). Formate was not monitored in vivo because it does not become labelled from [1-13C]glucose. A transient accumulation of acetoin, which was subsequently converted to 2,3-butanediol, was observed (Fig. 2A). Under the same conditions, the parental strain produced lactate (37.0 ± 1.0 mm) and acetate (1.1 ± 0.2 mm) (Fig. 2B). The glucose consumption rate of the LDHd strain (0.11 µmol·min−1·mg of protein−1) was fivefold lower than that of the parental strain (0.57 µmol·min−1·mg of protein−1). The kinetics of the build-up and consumption of glycolytic intermediates by the LDHd strain (Fig. 2C) showed that Man-ol and Man-ol-1-P were the major metabolites detected during glucose catabolism. Glucose was exhausted within 14 min; Man-ol accumulated steeply while glucose was available, reaching a maximum concentration of 88 ± 4 mm. The Man-ol-1-P pool was maintained at ≈ 20 mm while glucose was available, but a steep increase, concomitant with a drastic reduction in the concentration of Man-ol, was observed at the onset of glucose depletion. Therefore, we concluded that the accumulation of Man-ol-1-P (maximum concentration 76 ± 2 mm) was associated primarily with Man-ol utilization, denoting a bottleneck at the level of Man-ol-1-P dehydrogenase. Extracellular Man-ol was only one-quarter of the total amount of Man-ol produced when the Man-ol-1-P pool was approximately constant (between 4 and 13 min), but once the Man-ol-1-P pool started to increase, the external Man-ol concentration decreased quickly to undetectable levels. The values for total Man-ol concentration (determined in vivo by NMR) and for extracellular Man-ol (determined separately in the supernatant of the cell suspension) were calculated as though this polyol was exclusively intracellular to make possible a direct comparison (Fig. 2). The ratio between the isotopic enrichment in C1 and C6 of Man-ol-1-P, at times 14 and 16 min was 0.31 and 0.52, respectively. Moreover, the rate of Fru(1,6)P2 accumulation was lower in the LDHd strain than in the parental strain, reaching a maximal concentration of 40 ± 2 mm. A secondary maximum in the concentration of Fru(1,6)P2 (21 ± 2 mm) was detected after Man-ol depletion. Finally, as the pools of Man-ol-1-P and Fru(1,6)P2 decreased, the concentrations of 3-phosphoglycerate and phosphoenolpyruvate increased steadily, levelling off at 42 ± 1 mm and 15 ± 2 mm, respectively. Considerable amounts of intracellular aspartate (17.4 ± 0.3 mm), pyruvate (9.0 ± 2 mm), and alanine (8.6 ± 0.6 mm) were also determined as well as minor metabolites, such as succinate (2.9 ± 0.5 mm), and glycerol (2.7 ± 0.5 mm).
The pattern of intracellular metabolites was much simpler in the parental strain. The Fru(1,6)P2 pool increased rapidly upon glucose addition (Fig. 2D), reaching a maximum of 50 ± 1 mm, and started to decline upon glucose exhaustion, although it remained at detectable levels for more than the subsequent 10 min. Concomitantly, the resonances due to 3-phosphoglycerate and to phosphoenolpyruvate began to increase, levelling off at 15 ± 2 and 5 ± 2 mm, respectively, as Fru(1,6)P2 became undetectable. The intracellular concentrations of 3-phosphoglycerate and phosphoenolpyruvate were about threefold lower than those detected in the LDHd strain. The ratio of the two isotopomers of Fru(1,6)P2 (β[1-13C]Fru(1,6)P2/β[6-13C]Fru(1,6)P2], determined at the maximum of the Fru(1,6)P2 pool, was 1.0 in the LDHd strain (full scrambling of label) and 1.7 in the parental strain (Fig. 3). The carbon and redox balances for the LDHd were 97 and 82%, and for the parental strain were 99 and 93%, respectively, taking into consideration both extracellular and intracellular metabolites.
Effect of oxygen on glucose metabolism by the LDHd strain
Under aerobic conditions, glucose was consumed at a rate of 0.28 µmol·min−1·mg of protein−1, and the major extracellular products were acetoin (9.5 ± 1 mm), acetate (5.9 ± 0.5 mm), lactate (5.5 ± 0.5 mm), and 2,3-butanediol (2.5 ± 0.5 mm) (Fig. 4); a fivefold increase in the acetoin production was observed when compared to that found under anaerobiosis. Only a small amount of ethanol (0.5 ± 0.1 mm), and diacetyl (0.12 mm) were produced; the synthesis of formate was completely abolished as expected from the inhibitory effect of oxygen on pyruvate–formate lyase. Fru(1,6)P2 and Man-ol increased to a maximum concentration of 35 ± 1 mm and 14 ± 1 mm, respectively, and remained constant while glucose was available; Man-ol-1-P was not detected. Concomitantly with glucose exhaustion, the Fru(1,6)P2 and Man-ol pools dropped steeply to undetectable levels, while intracellular 3-phosphoglycerate and phosphoenolpyruvate increased rapidly to concentrations of 35 ± 2 mm and 15 ± 2 mm, respectively. Aspartate and pyruvate accumulated to intracellular concentrations of 15 ± 3 mm and 6.5 ± 1 mm, respectively. Other minor intracellular metabolites detected were alanine (1.7 ± 0.2 mm), succinate (2.6 ± 0.2 mm), and malate (2.6 ± 0.2 mm). A fivefold decrease in the alanine concentration was observed when compared to the amount produced under anaerobiosis. A carbon recovery of 97% was determined.
Enzyme activities involved in glucose and Man-ol metabolism
The activities of various enzymes in L. lactis LDHd and parental strains grown on glucose are summarized in Table 1. 6-Phosphofructokinase, fructose-bisphosphatase, pyruvate kinase, NADH oxidase, and glyceraldehyde-3-phosphate dehydrogenase activities were similar in both strains studied. A residual LDH activity (60-fold lower than the parental level) was detected in extracts derived from the LDHd strain. It was verified that this LDH activity remained constant after several steps of transference of the strain into fresh medium, and accounted for the amount of lactate formed. We also verified that only the l-isomer was produced, as the concentration of lactate determined with the enzymatic kit (which is specific for the l-form) agreed with the determination by NMR, which detects the total lactate.
Table 1. Enzyme activities (U·mg of protein−1) determined in crude cell extracts of the parental and LDHd strains grown on glucose. Activities are expressed in units per milligram of protein; all the determinations were made at least in triplicate in two extracts obtained from independent cultures.
Glyceraldehyde 3-P dehydrogenase
Man-ol-1-P dehydrogenase (forward reaction)
Man-ol-1-P dehydrogenase (backward reaction)
4.5 × 10−3
12 × 10−3
14 × 10−3
Extracts of the LDHd strain catalysed the forward and backward reactions of Man-ol-1-P dehydrogenase with similar rates, whereas no significant activity was found in extracts of the parental strain. The observed conversion of Man-ol-1-P to Man-ol in vivo required the presence of a phosphatase, which could be measured in cell extracts derived from the LDHd strain, but was absent in the parental strain. ATP-dependent Man-ol kinase activity was not found in any of the strains examined in this study.
Metabolic control coefficients for Man-ol-1-P dehydrogenase
To further describe the context of Man-ol accumulation, the metabolic control coefficients were determined under aerobic and anaerobic conditions (Table 2). As the control coefficients are bound to change as the metabolism progresses, the coefficients listed refer to the mid-point glucose consumption (10 mm). These coefficients are global properties of the metabolic dynamics assessing the association between changes in the individual metabolic fluxes. Because not all metabolites accumulate above the detection limit, the metabolic steps are grouped such that the substrate and the product are measured (Fig. 2). Therefore, the control coefficients apply for each reaction ensemble considered. For example, for aerobic conditions, the conversion of glucose to Man-ol was treated as a single metabolic step because Man-ol-1-P did not accumulate above the detection limit under aerobiosis. Accordingly, under anaerobic conditions, the control coefficients for the two steps leading to the formation of Man-ol-1-P and Man-ol were calculated (Table 2).
Table 2. .Control coefficients for the conversion of glucose to Man-ol (aerobic conditions) or glucose to Man-ol-1-P and Man-ol-1-P to Man-ol (anaerobic conditions). The coefficients were estimated for a glucose concentration of 10 mm (t = 3.25 min and t = 5.25 min, for aerobic and anaerobic conditions, respectively). NA, not applicable.
Glucose → Man-ol
Glucose → Man-ol-1-P
Man-ol-1-P → Man-ol
Glucose → Fru(1,6)P2
Glucose → Man-ol-1-P
Man-ol-1-P → Man-ol
Glucose → Man-ol
Fru(1,6)P2 → 3-phosphoglycerate
3-Phosphoglycerate → pyruvate
Pyruvate → lactate
Pyruvate → acetate
Pyruvate → ethanol
Pyruvate → acetoin
Acetoin → 2,3-butanediol
In this work, we took advantage of the noninvasive characteristics and analytical power of NMR to obtain reliable metabolic data during glucose utilization by a L. lactis strain deficient in LDH. Intracellular (and extracellular) metabolite levels were determined in real time and under controlled conditions of pH, temperature and gas atmosphere. The experimental set-up allowed us to obtain a detailed picture of the time courses for the build-up and decline of the intracellular pools of major metabolites following a pulse of glucose. In addition to the glycolytic intermediates, Fru(1,6)P2, 3-phosphoglycerate, and phosphoenolpyruvate that were detected in the parental strain, the genetically modified strain accumulated transiently high amounts of Man-ol-1-P and Man-ol, under anaerobic conditions. The different patterns of intracellular metabolites clearly reflected the distinct enzyme activities and metabolic features of the two strains examined. While Fru(1,6)P2 accumulated in the parental strain when glucose was available, the LDHd strain accumulated primarily Man-ol, in addition to lower levels of Fru(1,6)P2, denoting that an alternative strategy was followed for NAD+ regeneration. The pressure to regenerate NAD+, aggravated by the deficiency in the main activity responsible for this regeneration (LDH), was in part relieved by resorting to the formation of Man-ol, which leads to NADH oxidation (Fig. 5). The level of the intermediate Man-ol-1-P was considerably lower than that of Man-ol, indicating that the step catalysed by the phosphatase was not rate limiting. In contrast, upon glucose exhaustion, Man-ol was utilized at a high rate, and Man-ol-1-P, rather than Fru(1,6)P2, accumulated due to lack of NAD+ for the oxidation of Man-ol-1-P to Fru6P. The NAD+ limitation, that hampers the oxidation of glyceraldehyde 3-phosphate, has been previously invoked to explain the high levels of Fru(1,6)P2 found in a wild-type strain of L. lactis. In the LDHd strain, it is interesting to note the shift of the redox bottleneck from the glyceraldehyde-3-phosphate dehydrogenase-catalysed step to the reaction catalysed by Man-ol-1-P dehydrogenase, when Man-ol was utilized following glucose depletion (Fig. 2).
Our results show that the transient production of Man-ol served as a regeneration pathway for NAD+. This interpretation was further supported by the shift observed in the pattern of end-products and intermediate metabolites in the presence of oxygen. Under these conditions, the contribution of the Man-ol biosynthetic pathway was strongly reduced, because the pressure to regenerate NAD+ was relieved by the activity of NADH oxidase. Accordingly, Man-ol-1-P was not detected and the Man-ol concentration was reduced sixfold. Moreover, the levels of other compounds, such as ethanol, 2,3-butanediol, lactate and alanine, whose formation is associated with NADH oxidation, were considerably decreased, whereas acetoin and acetate production was enhanced.
We verified that the production of Man-ol-1-P and Man-ol was not an artefact resulting from the use of nongrowing cells and high cell densities during the in vivo NMR experiments, as intracellular Man-ol-1-P was also detected in growing cells of the LDHd strain (≈ 15 mm during the late exponential growth phase). To our knowledge, this is the first report of the production of Man-ol and Man-ol-1-P by L. lactis, although these products have been earlier observed in other bacteria, such as Staphylococcus aureus, Streptococcus mutans, and Escherichia coli[28–32]. Recent work with a different construct of LDHd showed an alternative strategy for NAD+ regeneration in which acetate is reduced to ethanol . Interestingly, the LDHd strain studied in this work did not produce Man-ol as end-product of glucose catabolism. Instead, after glucose exhaustion, Man-ol was metabolized. Therefore, Man-ol had to be taken up by the cell in a process that is most likely mediated by the phosphoenolpyruvate:phosphotransferase system; Man-ol-specific phosphotransferase systems have been described for other bacteria [22,33,34]. Furthermore, a gene showing sequence homology with mtlA of Bacillus subtilis, encoding enzyme II of the Man-ol-specific transport system, was found in the genome of L. lactis IL1403 . Cell extracts of the LDHd strain could not catalyse ATP-dependent phosphorylation of Man-ol and, to the best of our knowledge, Man-ol kinase has not been identified in bacteria [32,36].
The labelling pattern of Fru(1,6)P2 (both the isotopomers [1-13C]Fru(1,6)P2 and [6-13C]Fru(1,6)P2 were detected) is consistent with scrambling of the 13C label at the level of trioses phosphate and backflux through aldolase, as described previously for L. lactis MG5267 . The C1/C6 ratio is lower for the LDHd strain than the parental strain, indicating more extensive scrambling occurring in the manipulated strain, as a result of the lower net carbon flux through the aldolase catalysed step: the rate of glucose metabolism was ≈ fivefold lower in the LDHd strain. These results provide further support for the view that the net flux of Fru(1,6)P2 cleavage is largely controlled at the level of glyceraldehyde-3-phosphate dehydrogenase by the availability of NAD+, a cosubstrate for the enzyme [5,7]. However, it is interesting to note that in the LDHd strain, the maximum level of the Fru(1,6)P2 pool remained high (about 35 mm) even when an alternative pathway for NAD+ was provided by the presence of oxygen (Figs 2 and 4).
Man-ol-1-P dehydrogenase, the enzyme catalysing the reversible conversion of Fru(1,6)P2 to Man-ol-1-P, was not detected in extracts of the parental strain, but was induced in the LDHd strain. In L. lactis, phosphofructokinase, pyruvate kinase, and LDH are encoded by a multicistronic operon , but, in this case, ldh inactivation did not affect the activity of phosphofructokinase or pyruvate kinase.
Disruption of the ldh gene induced a clear metabolic shift in the end-products of glucose metabolism, from a homolactic type of fermentation in the parental strain to a mixed acid fermentation. However, this disruption did not result in the desired enhancement in the pool of α-acetolactate, and in the production of diacetyl, although the flux through α-acetolactate synthase was considerably increased (from ≈ 0 in the parental strain to 0.44 in the LDHd strain). The concept that a global understanding of metabolism is mandatory to achieve an effective redirection of microbial metabolism is thus strengthened [38,39]. In fact, modulation of cellular activity either by increasing the level of expression, or deleting a single gene has proved to be an inefficient process in several cases because the flux control coefficients of individual steps are, in general, low [39,40]. For instance, overexpressing the α-acetolactate synthase gene in L. lactis MG5267 did not result in an increase of carbon flux to acetoin/diacetyl  (A. R. Neves, A. Ramos, M. Kleerebezem, J. Hugenholtz, W. M. de Vos & H. Santos, unpublished results). Complementary metabolic manipulations are needed in order to redirect metabolism efficiently at the pyruvate branching point. A useful clue was provided by our results obtained under aerobic conditions, which led to a further increase of 50% in the flux through α-acetolactate synthase. Therefore, overexpression of NADH oxidase activity combined with deletion of the α-acetolactate decarboxylase gene [41,42] is probably a correct strategy towards optimal channelling of carbon to diacetyl.
The demonstrated ability of our LDHd strain to produce Man-ol may have considerable practical interest, given the well-known applications of this compound as a sweetening and stabilizing agent in the food industry. In this respect, one of the most outstanding features of the profile of control coefficients is the very high control of the step catalysed by LDH on the reactions leading to Man-ol formation. This finding leads to the proposal that the performed disruption of the ldh gene could be a crucial step in a strategy to construct a Man-ol-overproducing strain.
This work was supported by the BIOTECH Program, contract BIO4CT-96-0498 of the Commission of the European Communities and by Fundação para a Ciência e Tecnologia (FCT), Portugal, contracts PRAXIS/PCNA/P/BIO/39/96 and PRAXIS/P/BIA/11072/1998. A. R. N. and A. R. acknowledge FCT for the award of research fellowships.