Introduction of the Lactobacillus casei lactate dehydrogenase (LDH) gene into Saccharomyces cerevisiae under the control of the TPI1 promoter yielded high LDH levels in batch and chemostat cultures. LDH expression did not affect the dilution rate above which respiro-fermentative metabolism occurred (Dc) in aerobic, glucose-limited chemostats. Above Dc, the LDH-expressing strain produced both ethanol and lactate, but its overall fermentation rate was the same as in wild-type cultures. Exposure of respiring, LDH-expressing cultures to glucose excess triggered simultaneous ethanol and lactate production. However, the specific glucose consumption rate was not affected, indicating that NADH reoxidation does not control glycolytic flux under these conditions.
Formation of ethanol and acetate is a problem in biomass-directed applications of the yeast Saccharomyces cerevisiae. These low-molecular-mass byproducts are formed whenever respiring cultures are exposed to glucose excess [1–4]. In industry, this situation may occur as a result of imperfect mixing in large-scale fed-batch reactors. Formation of ethanol and acetate by aerobic S. cerevisiae cultures reflects an overcapacity of the enzymes involved in glycolysis and fermentation, relative to those involved in the respiratory breakdown of pyruvate [1–4].
In attempts to minimise byproduct formation via genetic engineering, it is of interest to identify the reaction(s) that control the rate of byproduct formation. In S. cerevisiae, glycolytic flux can be controlled at four levels [5, 6]: (i) transport of sugars across the plasma membrane, (ii) the activities of glycolytic enzymes, (iii) hydrolysis of the ATP generated in glycolysis, and/or (iv) reoxidation of the NADH formed by the glycolytic enzyme glyceraldehyde 3-phosphate dehydrogenase. Glucose transport and glycolysis involve the same enzymic reactions during respiratory and fermentative metabolism. In contrast and by definition, mechanisms for reoxidation of cytosolic NADH differ strongly depending on whether sugars are respired or fermented.
During respiratory growth, glycolytic NADH can be oxidised either by a mitochondrial external NADH dehydrogenase with a catalytic site facing the cytosol [7, 8], or via a glycerol 3-phosphate shuttle involving cytosolic and mitochondrial glycerol 3-phosphate dehydrogenases . In this way, the pyruvate formed in glycolysis is available for complete respiratory dissimilation, which requires pyruvate uptake by the mitochondria followed by its conversion into acetyl-CoA, the fuel of the TCA cycle . Conversely, during alcoholic fermentation, pyruvate decarboxylase converts pyruvate into carbon dioxide and acetaldehyde. Acetaldehyde is then used as an electron acceptor for NADH reoxidation by alcohol dehydrogenase.
The aim of the present study was to assess whether reoxidation of cytosolic NADH via the fermentative enzymes pyruvate decarboxylase and alcohol dehydrogenase controls the rate of byproduct formation by aerobic S. cerevisiae cultures. To address this question, a heterologous lactate dehydrogenase (LDH) was expressed in S. cerevisiae as an additional means of NADH regeneration. Although S. cerevisiae possesses endogenous LDHs, their expression is repressed by glucose and lactate is not a normal fermentation product [6, 10, 11]. Formation of low-molecular-mass byproducts was compared in steady-state glucose-limited chemostat cultures of the wild-type and of the LDH-expressing strain and after exposure of these cultures to excess glucose.
2Materials and methods
2.1Recombinant DNA techniques
The Lactobacillus casei LDH gene was modified by changing its GTG translation start codon, at the same time introducing an NcoI restriction enzyme site, and by introducing a PstI site at the 3′ end of the coding sequence. For this purpose, oligonucleotides were used with the following sequences (from 5′ to 3′): (1) CCATGGCAAGTATTACGGATAAGGATC and (2) CTATCACTGCAGGGTTTCGATGTC, defining nucleotides 163–1148 of the LDH sequence (accession number M76708, GenBank Sequence Database). The 985-bp DNA fragment defined by the two primers was PCR-amplified from a plasmid kindly provided by Robert Hutkins, University of Nebraska, and cloned into the EcoRV site of cloning vector pMOSBlue (Amersham Life Science), yielding plasmid pLC3. The modified gene was excised with NcoI and SalI and ligated into the NcoI/SalI-digested yeast integrative vector pYX012 (R&D System Europe Ltd, UK). The resulting plasmid pLC5 contains the modified bacterial LDH sequence under the control of the S. cerevisiae TPI1 promoter, and carries the auxotrophic marker URA3.
2.2Yeast strain and growth conditions
The S. cerevisiae strains used in this study were derived from GRF18 (MATαhis3 leu2). Strain GRF18U was obtained by 5-fluororotic acid (FOA) selection of spontaneous ura3 mutants . 2×108 cells pregrown to stationary phase on YEPD medium were plated on 2% glucose, 0.67% Difco yeast nitrogen base (YNB), 1 g l−1 FOA, 50 mg l−1 uracil, leucine and histidine. Single colonies were rescued and checked for uracil auxotrophy. The prototrophic, LDH-expressing strain GRFLC1+ was obtained starting from GRF18U. The uracil auxotrophy of this strain was first complemented by integration of pLC5 (see above) at the ura3 locus. Subsequently, sequential integration of pOXY042 and pOXY022 (containing the LEU2 and HIS3 marker genes; R&D System Europe Ltd, UK) at the leu2 and his3 loci complemented the leucine and histidine auxotrophies. The prototrophic reference strain GRFc was obtained by sequential integration of pOXY042 and pOXY022 at the leu2 and his3 loci of GRF18. Batch experiments were run at 30°C in shake-flask cultures (200 rpm) containing YNB without amino acids (6.7 g l−1) and glucose (20 g l−1).
Chemostat cultivation was performed at 30°C in a Biostat-Q system (B-Braun). A constant working volume of 800 ml was maintained via an effluent line coupled to a peristaltic pump. A dissolved oxygen concentration above 50% of air saturation was maintained by an airflow of 0.8 l min−1 and a stirrer speed of 800 rpm. The culture pH was maintained at 5.0 by automatic addition of KOH or HCl (2 mol l−1). Mineral medium was prepared as described by Verduyn et al. . The glucose concentration in reservoir media was 10 g l−1. Cultures were assumed to be in steady state when at least six volume changes had passed since the last change in growth conditions and the culture did not exhibit metabolic oscillations. For anaerobic cultivation, the air supply was replaced by nitrogen gas (flow rate, 0.8 l min−1). Furthermore, the reservoir medium, which was continuously flushed with nitrogen gas, was supplemented with 480 mg l−1 Tween-80 and 10 mg l−1 ergosterol (Sigma).
A concentrated, sterile solution of glucose was aseptically added to a steady-state, aerobic chemostat culture (D= 0.15 h−1) to give an initial glucose concentration of 50 mmol l−1. At appropriate intervals, 2–4-ml samples were withdrawn from the culture, cooled on ice and immediately analysed for OD660 and metabolite concentrations. During the pulse, the medium feed and effluent removal were continued.
Concentrations of carbon dioxide and oxygen in the exhaust gas were determined with a BM2001 gas analyser (Bioindustrie Mantovane, Italy). Calculations of specific oxygen consumption and carbon dioxide production were performed essentially as in .
2.6Determination of dry weight
Washed culture sample were filtered with 0.45-μm glass microfibre filters GF/A (Millipore) and dried for 36 h at 85°C. Parallel samples varied by less than 1%.
2.7Analysis of metabolites
Concentrations of glucose, acetate, glycerol, ethanol and lactate were determined with Boehringer enzymic kits #716251, #148261, #148270, #176290 and #139084, respectively. Pyruvate was assayed with enzymic kit no. 726-UV from Sigma.
2.8Preparation of cell extracts
Cell extracts were prepared essentially as described by Postma et al. , with the exception that cells were disrupted by five cycles of agitation with glass beads on a vortex (at 4°C) rather than by sonication. Protein contents of cell extracts were determined with Bio-Rad kit #500-0002, using bovine serum albumin as a standard.
Activities of enzymes in cell extracts were determined spectrophotometrically at 25°C immediately after preparation of the extracts. Reaction rates were proportional to the amount of cell extract used. Pyruvate decarboxylase (EC 220.127.116.11) was assayed as described by Postma et al. ; LDH (EC 18.104.22.168) was assayed in a reaction mixture containing 10 mM pyruvate, 0.128 mM NADH, 0.2 mM FDP, 37 mM acetate buffer, pH 5.6. One unit of enzyme activity is defined as the amount of enzyme catalysing the conversion of 1 μmol substrate min−1. Specific activities are given as U (mg protein)−1 and calculated based on an extinction coefficient of reduced pyridine dinucleotide cofactors of 6.3 mM−1 cm−1.
3.1Expression of a bacterial LDH gene in S. cerevisiae GRF18
The S. cerevisiae GRF18 background has previously been demonstrated to yield efficient lactate production after introduction of a mammalian LDH gene . Initial experiments on the prototrophic strain GFRLC1+ (containing the L. casei LDH expression cassette) and the isogenic wild-type strain GRFc were performed in glucose-grown shake-flask cultures.
Cell extracts of GRFLC1+ cultures contained high activities of LDH (up to 12 U (mg protein)−1 at the end of the growing phase). This specific activity corresponds to a significant fraction (0.5–2%) of the total protein content of yeast cells . LDH activity in cell extracts of the wild-type strain GRFc was <0.01 U (mg protein)−1.
In contrast to the wild-type strain, the transformed strain exhibited lactate production in the glucose-grown shake-flask cultures. At a cell density of 7×106 ml−1, wild-type cultures contained 16.9 mM ethanol. At the same cell density, the transformed strain exhibited a reduced ethanol concentration (12.3 mM) but in addition produced 6.2 mM lactate. This indicated that, at least in batch cultures, the heterologous LDH could compete efficiently for pyruvate and NADH with pyruvate decarboxylase and alcohol dehydrogenase.
3.2Growth of an LDH-expressing strain in steady-state chemostat cultures
A further physiological comparison of the GRFLC1+ strain and the reference strain GRFc was performed in aerobic, glucose-limited chemostat cultures. In such cultures, LDH levels in the transformed strain were stable for at least 150 generations (data not shown). The activity of LDH assayed in cell extracts was above 5 U (mg protein)−1 at all dilution rates tested, with a minimum at intermediate dilution rates (0.17–0.20 h−1; Fig. 1A).
The dilution rate above which aerobic ethanol production occurred (‘critical dilution rate’, Dc; ) was the same for the two strains (ca 0.19 h−1; Fig. 1A). In contrast to the wild-type, the LDH-expressing strain produced lactate in addition to ethanol at dilution rates above Dc. Lactate formation started at approximately the same dilution rate as ethanol formation (Fig. 1A).
The mixed lactate/ethanol fermentation that occurred at high dilution rates in the transformed strain indicates that the intracellular concentrations of pyruvate and NADH allowed pyruvate decarboxylase and LDH to operate simultaneously. The contribution of lactate production to the overall rate of fermentative metabolism increased with increasing dilution rate, but remained smaller than the rate of alcoholic fermentation (Fig. 1B). Especially at the highest dilution rate tested, lactate production by the transformed strain was accompanied by a corresponding decrease in the rate of alcoholic fermentation (Fig. 1A). This was also reflected by a small reduction in the specific rate of carbon dioxide production (data not shown). No differences between the two strains were observed with respect to the production of other metabolites (data not shown).
Physiological effects of LDH expression were also investigated in anaerobic glucose-limited chemostat cultures. In this case, only small amounts of lactate were produced by the transformed culture. This may be related to the high levels of pyruvate decarboxylase in the anaerobic cultures (Table 1).
Table 1. Example of anaerobic chemostat growth of GRFc and GRFLC1+
Growth of wild-type S. cerevisiae GRFc and the isogenic LDH-expressing strain GRFLC1+ in anaerobic, glucose-limited chemostat cultures (D= 0.1 h−1). qNAD+ represents the sum of qglycerol, qethanol and qlactate.
Dilution rate (h−1)
Biomass yield (g (g glucose)−1)
Residual glucose (g l−1)
qethanol (mmol g−1 h−1)
qlactate (mmol g−1 h−1)
qglycerol (mmol g−1 h−1)
qNAD+ (mmol g−1 h−1)
Lactate dehydrogenase (U (mg protein)−1)
Pyruvate decarboxylase (U (mg protein)−1)
3.3Exposure of glucose-limited cultures to glucose excess
To investigate whether the introduction of LDH as an additional means of reoxidising glycolytic NADH would affect the rate of glucose consumption after exposure of respiring cells to excess glucose, 50 mM glucose was added to aerobic glucose-limited chemostat cultures growing at D= 0.15 h−1. At this dilution rate, the transformed strain exhibited a lactate dehydrogenase activity of 8.75 U (mg protein)−1. The pyruvate decarboxylase activity was 0.85 U (mg protein)−1 in both strains.
Glucose consumption after the glucose pulse occurred at the same specific rate in the LDH-expressing culture as in the wild-type culture (5.5±0.1 mmol g−1 h−1; Fig. 2A). The metabolic response of strain GRFc was typical of wild-type S. cerevisiae strains [1–3]; rapid consumption of glucose was accompanied by the formation of ethanol (Fig. 2B). At the same time, small amounts of pyruvate, acetate and glycerol were produced (Fig. 2C–E). The LDH-expressing strain GRFLC1+ produced less ethanol than the wild-type. This was also evident from gas analysis: the specific rate of carbon dioxide production during the pulse was ca 20% lower than in the wild-type (data not shown). The transformed strain produced significant amounts of lactate (Fig. 2B).
Pyruvate and acetate production by strain GRFLC1+ continued after glucose had been exhausted (Fig. 2C,D). Control experiments in which lactate was pulsed to glucose-limited chemostat cultures indicated that these compounds are normal byproducts during consumption of lactate (data not shown).
The overall reoxidation of cytosolic NADH via fermentation can be estimated by combining ethanol, lactate and glycerol formation. This overall formation of fermentation products was similar for the two strains (Fig. 2F).
As previously reported for glucose-grown batch cultures [10, 16], expression of both mammalian and L. casei LDH in steady-state chemostat cultures of S. cerevisiae resulted in an altered distribution of metabolic fluxes at the pyruvate branchpoint. However, this redistribution of fluxes only affected the mode of fermentative metabolism, and not the relative contribution of respiration and fermentation to sugar dissimilation. It has recently been demonstrated that a 5–10-fold overexpression of pyruvate decarboxylase in S. cerevisiae led to a decrease of the dilution rate at which aerobic fermentation set in during aerobic, glucose-limited chemostat cultures , and thus affected the balance between respiration and fermentation. Apparently, the high pyruvate decarboxylase levels in the overexpressing strain competed for pyruvate with the enzymes involved in mitochondrial oxidation of pyruvate. The absence of such an effect in the LDH-expressing strain (Fig. 1) indicates that, even when expressed at high levels, LDH could not compete effectively for pyruvate and NADH with the mitochondria. This can be explained from the kinetic properties of these two enzyme systems: the Km of isolated S. cerevisiae mitochondria for pyruvate is ca 0.3 mM , whereas the Km of L. casei LDH is ca 10 mM .
Although the activity of LDH in cell extracts of the transformed strain was substantially higher than the activity of pyruvate decarboxylase, the specific rate of lactate formation was in all cases lower than the rate of ethanol formation. This result can also be explained in terms of affinity for pyruvate. The Km of pyruvate decarboxylase for pyruvate is only 1–3 mM at low cytosolic phosphate concentrations [18, 19]. This is substantially lower than the Km of L. casei LDH. In addition to the cytosolic concentration of pyruvate, also those of NADH and NAD+ will influence the distribution of pyruvate over pyruvate decarboxylase and LDH.
Also after exposure of respiring cells to excess glucose, lactate was produced by the LDH-expressing strain at the expense of ethanol formation (Fig. 2). As a result, the glycolytic flux remained essentially the same. This indicates that reoxidation of glycolytic NADH does not control the glycolytic flux after exposure of respiring cells to excess glucose. Thus, genetic engineering strategies that increase the capacity for mitochondrial oxidation of glycolytic (cytosolic) NADH will not simply lead to a further increase of the glycolytic flux, but should lead to an effective reduction of ethanol formation. The extent to which such strategies will reduce overall byproduct formation depends on the capacity for pyruvate dissimilation by mitochondria: when the capacity for respiration of glycolytic NADH exceeds that of pyruvate oxidation by the mitochondria, pyruvate and/or acetaldehyde will accumulate instead of ethanol .
Formation of glycerol, an important metabolite in the redox metabolism of S. cerevisiae, was not affected by the introduction of lactate dehydrogenase (Fig. 2E). This can be explained from the fundamentally different role in redox metabolism of alcoholic (and lactic) fermentation as compared to glycerol formation . Conversion of sugars into ethanol and carbon dioxide (or into lactate) is a redox-neutral process. In contrast to this, conversion of sugars into glycerol can be used to dispose of a surplus of NADH resulting from biomass synthesis or formation of oxidised metabolites [11, 21, 22]. Since expression of LDH affected neither the formation of oxidised metabolites (e.g. acetate and pyruvate) nor the biomass yield, effects on glycerol production were not expected.
Due to its high capacity for generation of NADH in glycolysis, S. cerevisiae is an interesting organism for introduction of heterologous NADH-linked reductases involved in the synthesis of fine chemicals. Our results indicate that the capacity for cytoplasmic NADH generation does not significantly exceed the capacity for NADH reoxidation via alcoholic fermentation and respiration. Consequently, when S. cerevisiae is used as a biocatalyst for bioconversions involving heterologous NADH-linked reductases, these will have to compete for NADH with pyruvate decarboxylase and alcohol dehydrogenase. In such processes it might be advantageous to reduce pyruvate decarboxylase levels to favour NADH reoxidation via the heterologous reductase.
We wish to thank Dr Robert Hutkins, University of Nebraska, for providing the LDH gene. Research in our groups is part of the project ‘From gene to product in yeast: a quantitative approach’, which is subsidised by the European Community (DG XII Framework IV Program on Cell Factories).