Homofermentative Lactate Production Cannot Sustain Anaerobic Growth of Engineered Saccharomyces cerevisiae: Possible Consequence of Energy-Dependent Lactate Export

Abstract

Due to a growing market for the biodegradable and renewable polymer polylactic acid, the world demand for lactic acid is rapidly increasing. The tolerance of yeasts to low pH can benefit the process economy of lactic acid production by minimizing the need for neutralizing agents. Saccharomyces cerevisiae (CEN.PK background) was engineered to a homofermentative lactate-producing yeast via deletion of the three genes encoding pyruvate decarboxylase and the introduction of a heterologous lactate dehydrogenase (EC 1.1.1.27). Like all pyruvate decarboxylase-negative S. cerevisiae strains, the engineered strain required small amounts of acetate for the synthesis of cytosolic acetyl-coenzyme A. Exposure of aerobic glucose-limited chemostat cultures to excess glucose resulted in the immediate appearance of lactate as the major fermentation product. Ethanol formation was absent. However, the engineered strain could not grow anaerobically, and lactate production was strongly stimulated by oxygen. In addition, under all conditions examined, lactate production by the engineered strain was slower than alcoholic fermentation by the wild type. Despite the equivalence of alcoholic fermentation and lactate fermentation with respect to redox balance and ATP generation, studies on oxygen-limited chemostat cultures showed that lactate production does not contribute to the ATP economy of the engineered yeast. This absence of net ATP production is probably due to a metabolic energy requirement (directly or indirectly in the form of ATP) for lactate export.

Traditional uses of l(+)-lactic acid, nowadays predominantly produced via microbial fermentation, comprise applications in food, cosmetics, and pharmaceuticals (6, 26). The worldwide production of lactic acid (presently an estimated 220,000 metric tons per year) is rapidly increasing, mainly as a result of the growing market for polylactic acid (7, 15). It is expected that this biodegradable polymer, produced from renewable resources, will replace various petrochemistry-based polymers in applications ranging from packaging to fibers (35).

Lactic acid is presently mainly produced with lactic acid bacteria, such as various Lactobacillus species (6, 26). Due to the pH sensitivity of these organisms, industrial lactic acid production requires large amounts of CaCO₃ or other neutralizing agents. This complicates downstream processing and yields large amounts of gypsum as a by-product (6). Development of alternative production organisms that are more tolerant to low pH may strongly decrease the requirement for neutralizing agents and lower the cost of downstream processing.

Yeasts, including Saccharomyces cerevisiae, are known for their ability to grow at low pH. In addition, S. cerevisiae can grow anaerobically on glucose in the presence of the essential anaerobic growth factors nicotinic acid, oleic acid, and ergosterol (2, 3, 18, 28). These attributes, combined with its genetic accessibility (22), make S. cerevisiae an interesting microorganism for the production of lactic acid (17, 31). Expression in S. cerevisiae of an NAD⁺-dependent lactate dehydrogenase (LDH; EC 1.1.1.27) from either mammalian, bacterial, or fungal origin results in simultaneous formation of ethanol and lactate (1, 17, 31, 34). To minimize ethanol formation, LDH has been expressed in S. cerevisiae strains with strongly reduced pyruvate decarboxylase or alcohol dehydrogenase activities (34). Although this indeed resulted in reduced ethanol formation, the lactate productivity or yield was not drastically increased. Complete elimination of alcoholic fermentation in S. cerevisiae can be achieved by deletion of all three structural genes encoding pyruvate decarboxylase (PDC1, PDC5, and PDC6) (23). Construction of a pyruvate decarboxylase-negative S. cerevisiae strain expressing bovine lactate dehydrogenase has been reported previously (31), but its physiology has not been studied in detail.

The conversion of glucose to either ethanol or lactic acid is equivalent both in terms of redox balance and ATP yield (Fig. 1). Nevertheless, a Kluyveromyces lactis strain engineered for homolactic fermentation required oxygen for efficient lactate production (29). In our own research, we encountered a similar phenomenon in engineered homolactic S. cerevisiae strains (unpublished data).

FIG. 1.

Schematic comparison of ethanol and lactic acid fermentation. Glucose uptake by S. cerevisiae occurs via facilitated diffusion. In both situations the conversion of glucose by glycolysis results in the formation of one ATP per pyruvate produced. The excess NADH formed by glycolysis is either regenerated by pyruvate decarboxylase and alcohol dehydrogenase (alcoholic fermentation) or LDH (lactate fermentation). Ethanol is known to rapidly diffuse through the cell membrane of S. cerevisiae. The mechanism(s) of lactic acid export in S. cerevisiae is, as yet, unknown, but it may involve ATP hydrolysis.

This study aims to resolve the role of oxygen in metabolically engineered homofermentative lactate-producing S. cerevisiae. To this end we quantitatively analyzed growth, metabolite production, and energetics of a homofermentative S. cerevisiae strain (RWB850-2) in batch and continuous cultures under various defined aeration regimes.

MATERIALS AND METHODS

Strains and maintenance.

All S. cerevisiae strains used in this study (Table 1) were derived from the isogenic CEN.PK family (36). Stock cultures were prepared from shake-flask cultures (100 ml of synthetic medium in 500-ml flasks) by addition of 20% (vol/vol) glycerol and storage of 2-ml aliquots in sterile vials at −80°C.

TABLE 1.

S. cerevisiae strains used in this study

Strain	Genotype
CEN.PK182	MATapdc1Δ::loxP pdc5Δ::loxP pdc6Δ::loxP
CEN.PK111-61A	MATα ura3-52 leu2-112 his3-Δ1
RWB837	MATapdc1Δ::loxP pdc5Δ::loxP pdc6Δ::loxP ura3-52
RWB850-2	MATapdc1Δ::loxP pdc5Δ::loxP pdc6Δ::loxP ura3-52 + YEpLDH#1

Strain and plasmid construction.

RWB837 was obtained from a cross between CEN.PK182 and CEN.PK111-61A (constructed by P. Kötter, Frankfurt, Germany, and obtained from A. E. Staley Manufacturing Company) (Table 1). The resulting diploid was sporulated, and the ascites were heated for 15 min at 56°C. This spore mix was then plated on yeast extract peptone medium with 0.2% acetate as the carbon source. The resulting colonies were tested for growth on yeast extract peptone medium with glucose or ethanol. Colonies that could not grow on glucose were subsequently checked by PCR for the presence of a disrupted PDC6 gene and for the mating type. Determination of the auxotrophic markers present, in this case ura3-52, then gave RWB837. The plasmid YEpLDH#1 was constructed by cloning an NheI fragment, containing the TPI1 promoter and Lactobacillus casei LDH gene, from plasmid pLC5 (9) into YEplac195 (21) cut with XbaI. RWB837 was then transformed with YEpLDH#1, resulting in RWB850. Transformant number 2 (RWB850-2) was selected for this work. (RWB850-2 is intellectual property of the A. E. Staley Manufacturing Company.)

Media.

The synthetic medium contained (per liter of demineralized water) 5 g of (NH₄)₂SO₄, 3 g of KH₂PO₄, 0.5 g of MgSO₄ · 7 H₂O, 0.15 ml of silicon antifoam (British Drug House), and trace elements at concentrations given by Verduyn et al. (40). After heat sterilization of the medium for 20 min at 120°C a filter-sterilized vitamin solution, prepared according to Verduyn et al. (40), was added. Synthetic media for precultures and start-up batch cultivation contained 1.5% (vol/vol) ethanol as the sole carbon source. For chemostat cultivation, glucose (10.0 g liter⁻¹) was added separately after heat sterilization at 110°C. In addition, without prior sterilization acetic acid was added to the autoclaved medium to a final concentration of 19 mM (10% of total substrate carbon) to rescue the C₂ requirement of pyruvate decarboxylase-negative S. cerevisiae (19).

Aeration conditions.

The dissolved oxygen concentration was measured with an oxygen electrode (Mettler Toledo, Greifensee, Switzerland). A stirrer speed of 800 rpm was used under all conditions. For aerobic conditions, an airflow of 0.5 liters min⁻¹ was applied to keep the dissolved oxygen concentration above 60% of air saturation. For fully anaerobic conditions the fermentors were sparged with 0.5 liters of pure nitrogen (Hoek-Loos, Schiedam, The Netherlands) min⁻¹. To prevent diffusion of oxygen, the fermentors were equipped with Norprene tubing (Saint-Gobain Performance Plastics, Charny, France) and Viton O-rings (Eriks, Alkmaar, The Netherlands). Oxygen limitation was obtained by supplying a mixture of air and nitrogen to the fermentors (14). The desired percentage of air, supplied via computer-directed mass-flow controllers, was topped up with technical-grade nitrogen to a fixed total flow rate of 0.5 liters min⁻¹. In contrast to the experimental setup for strictly anaerobic chemostat cultivation (41), the setup for the oxygen gradient experiments allowed for entry of a small amount of oxygen (about 24 μmol h⁻¹) via the medium reservoirs.

Fermentor cultivation.

All fermentor cultivations were performed at 30°C in 2-liter fermentors (Applikon, Schiedam, The Netherlands) with a working volume of 1 liter. In all experiments the pH was controlled at 6.0 via automated addition of 4 M NaOH (Applikon ADI 1030 biocontroller). At this pH the largest fraction of acetic (pK = 4.74 [11]) and lactic acid (pK = 3.87; the most common handbook value) is in the dissociated form, thereby minimizing uncoupling by these weak acids (40). For chemostat cultivation, the addition of medium was controlled by a peristaltic pump. The working volume of the cultures was kept constant by means of an electrical level sensor. Chemostat cultures were assumed to be in steady state when, after at least five volume changes, the culture dry weight, specific carbon dioxide production rate, and oxygen consumption rate changed by less than 2% during 24 h. Spontaneous synchronization of the cell cycle, a phenomenon sometimes encountered with chemostat cultures of wild-type S. cerevisiae (25), was not observed with the engineered S. cerevisiae strain.

Analytical procedures.

Dry weight determination, substrate-, and metabolite analysis, in- and off-gas analysis, preparation of cell extract, enzyme assays, and protein determination were performed as described previously (38). Lactate dehydrogenase was assayed at 30°C with a reaction mixture of 2 mM MnCl₂, 1 mM fructose-1,6-diphosphate, and 0.15 mM NADH in 0.1 M imidazole at pH 6.5. The reaction was started by addition of potassium pyruvate (10 mM).

Calculation of physiological parameters during defined aeration gradients.

Defined, continuously decreasing oxygen feeds were applied to carbon-limited chemostat cultures by supplying a mixture of nitrogen and air to the cultures. During these 200-h experiments the physiological parameters were continuously changing. Therefore, calculations similar to those used by Costenoble et al. (14) were used. The specific growth rate throughout this experiment was calculated from a mass balance. With constant volume, absence of biomass in the feed, and a known dilution rate (D), the specific growth rate (μ) can be calculated according to the following equation: μ = D + (dC_x/dt) · C_x⁻¹, with C_x being the culture dry weight. Between 60 and 140 h, dC_x/dt was easily determined by linear regression (see Fig. 4A).

FIG. 4.

During defined oxygen feed gradients of S. cerevisiae RWB850-2, an increasing fraction of the glucose is fermented to lactic acid. The relative contributions of air and nitrogen to the inlet gas mixture are depicted below the graphs. Results shown are from one representative oxygen feed gradient of a set of four independent replicate experiments, results of which differed by less than 10%. (A) The measured parameters. Symbols: primary y axis, culture dry weight (▪); secondary y axis, lactate (•), glucose (○), pyruvate (□), and acetate (▴) concentration. (B) The calculated physiological parameters. Symbols: primary y axis, biomass yield on glucose (▪); secondary y axis, lactate production rate (•), glucose consumption rate (○), oxygen consumption rate (▴), and carbon dioxide production rate (▵).

Expressions for the specific substrate consumption and product formation rates were derived analogously. The specific substrate consumption rate (q_s) can be calculated from the following equation: q_s = [D · (C_s,0 − C_s) − (dC_s/dt)] · C_x⁻¹, with C_s,0 representing the reservoir substrate concentration and C_s representing the residual substrate concentration in the culture. To calculate q_p (specific product formation rate), the equation q_p = [(dC_p/dt) + D · C_p] · C_x⁻¹ was used. In this equation, C_p represents the product concentration in the culture supernatant. Because the turnover in the gas phase is much higher than that in the liquid phase, the specific oxygen consumption and carbon dioxide production rates in these transient-state cultures were calculated according to the standard procedures for steady-state cultures (39). The biomass yield on substrate was calculated by dividing the specific growth rate by the specific substrate consumption rate. Accordingly, the biomass yield on oxygen was calculated by dividing the specific growth rate by the specific oxygen consumption rate.

RESULTS

Aerobic glucose-limited chemostat cultivation.

In all experiments described, carbon-limited chemostat cultures grown at a dilution rate of 0.10 h⁻¹ were used to obtain well-defined and reproducible starting conditions. It was therefore important to know the physiological behavior of our engineered strain under these conditions. Similar to other pyruvate decarboxylase-negative strains (19), the lactate dehydrogenase-expressing, pyruvate-decarboxylase-negative S. cerevisiae strain (RWB850-2) required the addition of a C₂ compound (ethanol or acetate) to the growth media. In this study, acetate was used to fulfill the C₂ compound requirement. The data described below were obtained from six independent replicate experiments, with an average carbon recovery of 95.3% ± 1.9%. The in vitro-measured lactate dehydrogenase activity in these cultures was 25.3 ± 1.2 μmol min⁻¹ mg of protein⁻¹.

Under aerobic carbon-limited conditions, the metabolism of RWB850-2 was fully respiratory, as indicated by the high biomass yield on carbon (13.8 ± 0.1 g of biomass C mol⁻¹) and the absence of fermentation products. The specific glucose consumption rate was 1.09 ± 0.01 mmol g of biomass⁻¹ h⁻¹, and acetate was consumed at a rate of 0.37 ± 0.01 mmol g of biomass⁻¹ h⁻¹. The specific oxygen consumption rate (2.78 ± 0.11 mmol g of biomass⁻¹ h⁻¹) and the specific carbon-dioxide production rate (2.91 ± 0.14 mmol g of biomass⁻¹ h⁻¹) resulted in a respiratory quotient close to unity (1.04 ± 0.01 mol of CO₂ produced per mol of O₂ consumed). The biomass yield on oxygen was 36.0 ± 1.5 g of biomass⁻¹ mol of oxygen⁻¹. These values were similar to those obtained with the isogenic wild type grown under the same conditions (data not shown).

Aerobic versus anaerobic lactate formation.

To assess the effect of oxygen on lactate production by S. cerevisiae RWB850-2, cells pregrown in an aerobic, glucose-limited chemostat culture were exposed to excess glucose under alternating anaerobic and aerobic conditions. After stopping the medium pumps and the air supply of the chemostat culture, it was sparged with nitrogen to achieve and maintain anaerobicity. Immediately after anaerobicity was established, a glucose pulse was administered to the culture (Fig. 2). During the first 2 (anaerobic) h, lactate was produced at a rate of 0.8 ± 0.0 mmol g of biomass⁻¹ h⁻¹ (Fig. 2). After these 2 h, normal aeration was resumed and the dissolved oxygen concentration increased to above 80% of air saturation. As a result, the specific lactate production rate increased by 2.5-fold (to 2.0 ± 0.0 mmol g of biomass⁻¹ h⁻¹) compared to that of the first (anaerobic) phase of the experiment. After 2 h of aerobic incubation, aeration was stopped again and anaerobicity was reestablished. In this third phase of the experiment the specific lactate production rate (0.8 ± 0.0 mmol g of biomass⁻¹ h⁻¹) was the same as that during the first anaerobic phase.

FIG. 2.

Lactate formation upon exposure of an aerobic glucose-limited chemostat culture of S. cerevisiae RWB850-2 to excess glucose. During this experiment the conditions were changed from anaerobic to aerobic and then back to anaerobic. The dashed lines indicate a change in the aeration regime. The open and closed circles indicate glucose concentration and lactate concentration, respectively. The results depicted are the averages of independent duplicate experiments. The experimental variation was below 2% for all measurements.

During this 6-h experiment, no increase in optical density at 660 nm was observed, indicating that there was no biomass growth. Apart from lactate (40 mM), glycerol and pyruvate were formed (the latter only during the aerobic phase), each to a maximum concentration of 1 mM. In the aerobic phase, part of the glucose was converted to carbon dioxide via respiratory metabolism (data not shown).

Long-term exposure to excess glucose under anaerobic conditions.

The glucose pulse experiment described above demonstrated that, although at a low rate, anaerobic lactate production is possible for at least 2 h. To further study anaerobic lactate production by S. cerevisiae RWB850-2, cells pregrown in aerobic, glucose-limited chemostat cultures were incubated for 96 h under anaerobic conditions in the presence of excess glucose. A few hours after the beginning of the experiment the specific lactate production rate was 0.63 ± 0.01 mmol g of biomass⁻¹ h⁻¹ (Fig. 3), in good agreement with the short-term experiment discussed above (Fig. 2). This rate decreased to 0.34 ± 0.00 mmol g of biomass⁻¹ h⁻¹ after 48 h, and at the end of the 96-h anaerobic experiment the specific lactate production rate had even decreased further to 0.22 ± 0.02 mmol g of biomass⁻¹ h⁻¹. The overall yield of lactate on glucose during this 96-h anaerobic experiment was 1.66 ± 0.04 mol of lactate mol of glucose⁻¹. Glycerol (0.09 mol of glycerol mol of glucose⁻¹) and carbon dioxide (0.07 mol of CO₂ mol of glucose⁻¹) were the main by-products of the anaerobic fermentation. During the experiment the biomass concentration decreased from 4.99 ± 0.02 to 3.23 ± 0.08 g of biomass liter⁻¹. The relatively low concentration of storage carbohydrates (around 5% [wt/wt]) in aerobic chemostat cultures at a dilution rate of 0.1 h⁻¹ (24) makes it unlikely that only their consumption was responsible for the large decrease in the biomass concentration. Apparently, other cell constituents also were subject to turnover.

FIG. 3.

Metabolic response of S. cerevisiae RWB850-2, pregrown in aerobic, glucose-limited chemostat cultures, to a 200 mM glucose pulse under strictly anaerobic conditions. The graph shows the averages and mean deviations of two independent replicate experiments. The open circles indicate glucose concentration. The closed symbols indicate the lactate concentration (circles) and the culture dry weight (squares).

To minimize the complexity of the experimental system, C₂ compounds (to meet the C₂ requirement of pyruvate-decarboxylase-negative S. cerevisiae) or oleate and ergosterol (to meet the anaerobic growth requirements of S. cerevisiae) were not included in these long-term anaerobic incubations. It was anticipated that endogenous stores of C₂ compounds and anaerobic growth factors should allow for at least one biomass doubling after the switch to anaerobic conditions. Indeed, combined addition of ethanol, Tween 80 (as a source of oleate), and ergosterol in anaerobic control experiments did not result in growth or enhanced lactate production after up to 7 days of incubation.

Physiology during a gradual change from aerobic growth to a severely limited oxygen supply.

The experiments described above demonstrate that lactate production is stimulated by oxygen, whereas growth of the engineered strain is totally oxygen dependent. To quantify the relationship between oxygen availability and growth, S. cerevisiae RWB850-2 was subjected to a range of oxygen supplies. Because analysis of steady-state cultures under different degrees of oxygen limitation is laborious, an alternative approach was used. A defined, continuously decreasing oxygen feed regimen was applied to carbon-limited continuous cultures by changing the mixture of nitrogen and air in the inlet gas (see Materials and Methods). During the 200-h experiment the air supply was linearly decreased from 200 ml of air min⁻¹ per 500 ml of total gas min⁻¹ at the start to 0 ml of air min⁻¹ per 500 ml of total gas min⁻¹ at 200 h.

During the first 40 h of the experiment (Fig. 4), the air supply was still sufficient for fully respiratory metabolism, as indicated by the high biomass dry weight and the absence of lactate production. After 40 h, the dissolved oxygen concentration decreased to below the detection limit. In agreement with the onset of oxygen limitation, lactate production immediately set in and the culture dry weight started to decrease (Fig. 4A). During the remaining 160 h of the oxygen gradient, the air supply was further decreased to zero. After some initial fluctuations caused by small amounts of foam affecting the culture volume and therefore the oxygen transfer, the culture dry weight decreased linearly (Fig. 4A). In line with the decreasing culture dry weight, the consumption of acetic acid decreased with decreasing oxygen supply, resulting in accumulation of acetate in the supernatant. Pyruvate was produced throughout the experiment, probably reflecting a competition for glycolytic NADH between mitochondrial respiration and the heterologous LDH. No formation of glycerol was observed throughout the experiment.

Starting with the first appearance of lactate, the biomass yield on glucose decreased linearly with decreasing oxygen feed (Fig. 4B). The increasing specific glucose consumption rate also reflected the decreasing biomass yield on glucose. As the culture became progressively more oxygen limited, the specific lactate production rate increased to above 6 mmol g of biomass⁻¹ h⁻¹ (Fig. 4B). Surprisingly, the specific oxygen consumption rate of RWB850-2 remained high, ranging between 3.1 and 2.6 mmol g of biomass⁻¹ h⁻¹, as long as no residual glucose was detected in the supernatant (Fig. 4B). Consequently, the biomass yield on oxygen was constant throughout the oxygen-limited phase of the four independent replicate experiments (30.1 ± 2.8 g of biomass per mol of oxygen) and therefore was independent of the air supply (Fig. 5). Because homofermentative lactate production does not lead to carbon dioxide formation, the respiratory quotient was 1.04 mol of CO₂ produced per mol of O₂ consumed.

FIG. 5.

Biomass yields on oxygen during defined oxygen limitation gradients of S. cerevisiae RWB850-2. The relative contributions of air and nitrogen to the inlet gas mixture are depicted below the graphs. Measurements were obtained from four independent replicate experiments. The average of these values is indicated by the line. The larger deviation of the mean at the end of the experiment (more severe oxygen limitation) is due to measurement errors as a result of low volumetric oxygen consumption rates and biomass concentrations.

Between 140 and 150 h after the start of the gradient, residual glucose was detected in the supernatant of the culture (Fig. 4A). The resulting nonlinearity hampered the calculation of the rates for the remainder of the experiment. Shortly after this appearance of glucose in the supernatant the air supply finally dropped to zero, and RWB850-2 was washed out completely from the culture. The isogenic wild type still grew in the absence of aeration, with a biomass yield on glucose of 0.09 g of biomass g of glucose⁻¹. Apparently, under these experimental conditions entry of oxygen into the fermentors was sufficient for synthesis of the anaerobic growth factors oleate and ergosterol.

Steady-state analysis of cultures with a limited oxygen supply.

Oxygen-limited steady-state chemostat cultivation of RWB850-2 was used to further investigate the apparent independence of the biomass yield on oxygen from the aeration rate and thereby from the specific lactate production rate. As seen in the oxygen gradient cultures, an inverse relationship between oxygen supply and specific rate of lactate fermentation was observed in the steady-state cultures. In agreement with this, the biomass yield on glucose decreased with increasing lactate production rate (Fig. 6). However, as expected from the oxygen gradient experiments, the biomass yield on oxygen (34.1 ± 2.8 g of biomass per of mol oxygen) was independent of the specific lactate production rate and therefore of the oxygen supply (Fig. 6). Steady-state cultures could be obtained at an air supply as low as 80 ml of air per 500 ml of total gas mixture. This resulted in a culture dry weight of 1.7 g liter⁻¹. Attempts to reduce the air supply to below 80 ml of air per 500 ml of total gas mixture irrevocably resulted in complete wash out of the culture. Apparently, the engineered strain was not capable of oxygen-limited growth at elevated residual glucose concentrations.

FIG. 6.

Biomass yields on oxygen and glucose of steady-state chemostat cultures of S. cerevisiae RWB850-2 with different regimens of limited oxygen supply (in all cases the dissolved oxygen concentration was below 0.1% of air saturation). The cultures were run at a dilution rate of 0.10 h⁻¹ on a mixture of glucose and acetate (10% of total substrate carbon). With decreasing air supply to the fermentor, the specific lactate production rate (x axis) increased. The biomass yield on glucose (□) is linearly correlated to the specific lactate production rate. In contrast, the biomass yield on oxygen (▪) is independent of the specific lactate production rate.

DISCUSSION

Fermentation rates in wild-type and homofermentative lactate-producing S. cerevisiae.

Among yeasts, S. cerevisiae is known both for its unique ability to grow fast under aerobic as well as anaerobic conditions (41) and for its rapid conversion of glucose to ethanol. Specific ethanol production rates of up to 29 mmol g of biomass⁻¹ h⁻¹ have been reported for wild-type S. cerevisiae CEN.PK 113-7D (isogenic to the engineered strain used in this study) in anaerobic batch fermentations, with a maximum specific growth rate of 0.33 h⁻¹ (5). In contrast, our results demonstrate that pyruvate-decarboxylase-negative S. cerevisiae expressing a heterologous lactate dehydrogenase cannot grow anaerobically. Furthermore, the highest specific lactate production rate observed under anaerobic conditions in the presence of excess glucose (0.8 mmol g of biomass⁻¹ h⁻¹) was almost 36-fold lower than the specific ethanol production rate described above. If homolactic fermentation were energetically equivalent to alcoholic fermentation (i.e., yielding 1 mol of ATP per mol of fermentation product), this rate of lactate fermentation would be barely sufficient to meet the estimated maintenance energy requirement of 0.5 mmol of ATP g of biomass⁻¹ h⁻¹ (32, 42). Enzyme assays in cell extracts consistently revealed LDH activities above 25 μmol · min⁻¹ · mg of protein⁻¹, which renders it unlikely that the low rates of anaerobic lactate fermentation were due to an insufficient expression level of the heterologous protein. A shortage of fructose-1,6-diphosphate in the cells would prevent full activation of the L. casei lactate dehydrogenase used in this study (20). To test this possibility, a steady-state chemostat cultivation with limited air supply was repeated with a pyruvate decarboxylase-negative strain expressing a Lactobacillus plantarum lactate dehydrogenase that does not require fructose-1,6-diphosphate (20). Because the results were the same as those with RWB850-2 (data not shown), a shortage of fructose-1,6-diphosphate is probably not the cause of the low lactate formation rate.

The immediate increase of the lactate fermentation rate to 2.0 mmol of lactate g of biomass⁻¹ h⁻¹ upon aeration of the anaerobic cell suspensions indicates that the complete absence of anaerobic growth was not due to an intrinsically limited capacity of the lactate fermentation pathway. Indeed, the highest specific lactate production rate observed with RWB850-2, isogenic to CEN.PK 113-7D, was 6.3 mmol g of biomass⁻¹ h⁻¹ during the defined aeration gradient experiments (Fig. 4B). This was, however, not a sustainable rate, because the highest specific lactate production rate observed in oxygen-limited steady-state chemostat cultures was only 3.9 mmol g of biomass⁻¹ h⁻¹ (Fig. 6).

Stimulation of fermentation rates by oxygen.

Stimulation of glucose fermentation by oxygen is a common phenomenon among yeasts but is absent in wild-type S. cerevisiae (33). In S. cerevisiae, stimulation of alcoholic fermentation by oxygen is observed when it is engineered for xylose fermentation via introduction of heterologous xylose reductase and xylitol dehydrogenase (4). In the cases examined thus far, stimulation of sugar fermentation by oxygen in wild-type or engineered yeasts was shown to be due to a redox imbalance (10, 33). In such cases oxygen serves as a redox acceptor, a function that can also be fulfilled by other compounds, such as acetoin (33, 37). However, during anaerobic alcoholic fermentation on glucose, wild-type S. cerevisiae does not display a redox imbalance. When NADH is in excess it is channeled to glycerol production (37). Because lactic acid fermentation via NAD⁺-dependent LDH is, in a redox-wise manner, identical to alcoholic fermentation (Fig. 1) and because the glycerol pathway was not affected in RWB850-2, it is highly unlikely that a redox imbalance is the cause of the low lactate production rate of the engineered strain. Instead, the stimulatory effect of oxygen on lactate fermentation probably reflects an energetic constraint.

Energetics of homofermentative lactate production by engineered S. cerevisiae: absence of net ATP formation.

The biomass yield on oxygen of wild-type S. cerevisiae CEN.PK 113-7D, which is isogenic to RWB850-2, is approximately 36 g of biomass mol of oxygen⁻¹ under aerobic glucose-limited conditions (8). Under conditions with a limited oxygen supply in otherwise glucose-limited chemostat cultures, wild-type S. cerevisiae displays respirofermentative metabolism (43). At all but extremely low air supplies, oxygen has a mainly catabolic role in S. cerevisiae and, as the oxygen supply decreases, a larger fraction of the glucose is fermented to ethanol. Alcoholic fermentation yields ATP via substrate-level phosphorylation. Therefore, wild-type S. cerevisiae requires less oxygen to form a certain amount of ATP (and thus biomass) as more glucose is fermented to ethanol. Consequently, a decreasing oxygen supply leads to a decrease of the specific rate of oxygen consumption and an increase of the biomass yield on oxygen. A graphical representation of this phenomenon has been presented by Fiechter et al. (18). Experimental data from Costenoble et al. (14) on microaerobic glycerol formation by S. cerevisiae CBS 8066 allow for the calculation of the biomass yields on oxygen. These yields range from 92 g of biomass mol of oxygen⁻¹ under moderate oxygen limitation to values as high as 540 g of biomass mol of oxygen⁻¹ under stringently oxygen-limited conditions. Similar extreme values can be calculated from a study by Weusthuis et al. (43).

The relationship between oxygen supply and biomass yield found with engineered S. cerevisiae under conditions of limited oxygen supply differed drastically from that of wild-type S. cerevisiae. Instead of steadily increasing with decreasing oxygen supply, the biomass yield on oxygen of RWB850-2 (30.1 ± 2.8 g of biomass per mol of oxygen in the oxygen gradient cultures, 34.1 ± 2.8 g of biomass per mol of oxygen in steady-state cultures) was independent of the oxygen supply (Fig. 5 and 6).

How can this constant biomass yield on oxygen be explained? A varying of biomass yield on ATP which, at every oxygen feed tested, exactly balances the ATP generated by lactate fermentation to result in a constant biomass yield on oxygen is highly unlikely. Furthermore, the biomass yield on oxygen was constant over a wide range of specific lactate production rates and lactate concentrations. The lack of a correlation with the extracellular lactate concentration argues especially strongly against the involvement of lactate toxicity (as a result of either weak-acid uncoupling or intracellular accumulation). The remaining and plausible explanation is that, in the engineered strain, the overall conversion of glucose into extracellular lactate does not yield ATP. Because the biochemical pathways for the uptake of glucose and its subsequent conversion into either intracellular lactic acid or ethanol are identical in both redox as well as ATP balances (Fig. 1), the difference has to be sought in product export. Ethanol diffuses freely through the yeast plasma membrane. However, at the near-neutral intracellular pH in yeast, lactate will be predominantly in the anionic form (the most commonly cited pK_a value for lactic acid in chemical handbooks is 3.87). It is unlikely that this polar molecule will diffuse through the cell membrane of S. cerevisiae (13). The importer of lactic acid is known to be the proton symporter JEN1 (12). However, the transporter and the mechanism for export of this acid are still unknown. Deletion of JEN1 and the putative monocarboxylate transporter genes MCH1 to -5 did not affect lactate excretion in wild-type S. cerevisiae (27). Particularly because lactic acid production is possible at pH values well below the pK of the acid (30), it seems likely that in S. cerevisiae the export of lactic acid from the cell requires ATP. This ATP requirement may be direct, involving an ATP-driven primary transport mechanism. Such a mechanism, involving the ABC-type transporter encoded by the PDR12 gene (16), is involved in export of the organic acids benzoate and sorbate. Alternatively, an ATP requirement for lactate export may involve a secondary transport mechanism in combination with the plasma-membrane ATPase, in which case ATP indirectly supplies the energy for the translocation process via the generation of a proton-motive force across the plasma membrane. A net requirement of one ATP per molecule of lactate would exactly compensate for the single ATP produced per lactate in glycolysis.

A zero net ATP yield from lactate fermentation explains the physiology of the engineered strain under anaerobic conditions. When fermentation does not result in a positive ATP balance, cells cannot meet the ATP required for maintenance, thus explaining the massive turnover of biomass in the long-term experiments (Fig. 3). A zero net ATP yield offers a plausible explanation for the low lactate production rates under anaerobic conditions: the resulting low cytosolic ATP concentration will limit the activity of the kinases in the upper part of glycolysis. Furthermore, the stimulatory effect of oxygen can be adequately explained from the alleviation of this restriction by the provision of ATP through oxidative phosphorylation.

Implications for industrial lactic acid production.

This study demonstrates that the present homofermentative lactate-producing S. cerevisiae strains require oxygen for the generation of ATP. This ATP is needed either for growth and, even under nongrowing conditions, for meeting maintenance energy requirements. In any industrial application of homofermentative lactate-producing S. cerevisiae strains, their requirement for aeration will have serious consequences for both reactor design and process economics. This creates an incentive for the construction of homofermentative lactate-producing S. cerevisiae strains that do not require respiration.