In Vivo Analysis of the Mechanisms for Oxidation of Cytosolic NADH by Saccharomyces cerevisiaeMitochondria

Oxygen consumption by isolated mitochondria.To confirm the absence of the gene products (and any isoenzymes) in the deletion mutants, substrate-dependent oxygen consumption by isolated mitochondria was measured in wild-type S. cerevisiae and in isogenic gut2Δ, nde1Δ nde2Δ, andnde1Δ nde2Δ gut2Δ strains (Table3). External NADH dehydrogenase activity was completely abolished in the nde1Δ nde2Δ, andnde1Δ nde2Δ gut2Δ mutants. An earlier report (29) that mitochondria of an nde1Δ nde2Δmutant exhibited residual oxidation of NADH can be explained by the use of commercial NADH (19). This is contaminated with ethanol (24), which is readily oxidized by S. cerevisiaemitochondria. Glycerol-3-phosphate dehydrogenase activity was completely abolished in gut2Δ mutants.

To investigate whether deletion of the NDE and/orGUT genes affected other mitochondrial dehydrogenase systems, oxidation of some other substrates was tested (Table 3). Metabolism of a mixture of malate and pyruvate or of ethanol generates intramitochondrial NADH, which can be oxidized by the internal NADH dehydrogenase. Succinate is oxidized by an internal flavin adenine dinucleotide-dependent succinate dehydrogenase. These substrates were oxidized by mitochondria from wild-type as well as mutant strains. Whether the observed small quantitative differences (Table 3) are due to biological differences or to variations in the quality of the mitochondrial preparations is not clear.

Physiology of wild-type S. cerevisiae in aerobic, glucose-limited chemostat cultures.The physiological consequences of the different deletions were studied by growing wild-type and mutantS. cerevisiae strains in aerobic, glucose-limited chemostats at dilution rates varying from 0.05 to 0.38 h⁻¹ (Fig.2). If a chemostat culture is at steady state, the dilution rate is by definition equal to the specific growth rate. At specific growth rates below 0.30 h⁻¹, the wild-type strain CEN.PK113-7D (Fig. 2A) had a high biomass yield (0.5 g of biomass g of glucose⁻¹) and produced neither ethanol nor glycerol. Completely respiratory growth was also evident from the respiratory quotient (the specific CO₂ production rate,q_CO2, divided by the specific O₂ consumption rate, q_O2), which was 1.0 at specific growth rates below 0.30 h⁻¹(Fig. 3A). Above a dilution rate of 0.30 h⁻¹, ethanol was produced and the respiratory quotient increased to values above 1.0 as the q_CO2increased and the q_O2 decreased (Fig. 2A and 3A). Concomitantly, the biomass yield decreased sharply. Glycerol production was observed only at the highest dilution rate applied. The highest dilution rate at which wild-type cultures reached a steady state was 0.38 h⁻¹. Above this dilution rate, the wild-type strain washed out of the chemostat.

• Open in new tab
• Download powerpoint

Fig. 2.

Effects of dilution rate in an aerobic, glucose-limited chemostat culture on biomass yield (Y_XS) and specific production rate of ethanol and glycerol (respectively q_ethanol and q_glycerol) for wild-type S. cerevisiae CEN.PK113-7D (data taken from reference31) (A), gut2Δ mutant CEN.PK225-2C (B),nde1Δ nde2Δ mutant CEN.PK167-2B (C), and nde1Δ nde2Δ gut2Δ mutant CEN.PK263-5D (D). Apparently filled-in symbols are a result of overlapping data points from independent chemostat cultures.

• Open in new tab
• Download powerpoint

Fig. 3.

Effects of dilution rate in an aerobic, glucose-limited chemostat culture on the specific production rate of CO₂(qCO₂) and the specific consumption rate of O₂(qO₂) for wild-type S. cerevisiae CEN.PK113-7D (data taken from reference 31) (A),gut2Δ mutant CEN.PK225-2C (B), nde1Δ nde2Δmutant CEN.PK167-2B (C), and nde1Δ nde2Δ gut2Δ mutant CEN.PK263-5D (D). Apparently filled-in symbols are a result of overlapping data points from independent chemostat cultures.

Physiology of a gut2Δ mutant.In S. cerevisiae the glycerol-3-phosphate shuttle can, in principle, couple the oxidation of cytosolic NADH to the respiratory chain. To test whether its contribution is essential, the GUT2 gene, encoding the mitochondrial membrane-bound glycerol-3-phosphate dehydrogenase, was deleted. When the gut2Δ mutant was cultivated over a range of dilution rates, the measured metabolite production rates were essentially the same as those of the wild type (Fig. 2B and 3B). Aerobic fermentation set in at the same dilution rate as in the wild type (0.30 h⁻¹), but washout of biomass occurred at a lower dilution rate (0.33 versus 0.38 h⁻¹). Apparently, elimination of the mitochondrial membrane-bound glycerol-3-phosphate dehydrogenase hardly affected the respiratory capacity during aerobic glucose-limited growth. Thus, even if the glycerol-3-phosphate shuttle were active in a wild-type strain, in the mutant its contribution to the oxidation of cytosolic NADH could be taken over completely by other NADH-oxidizing systems.

Physiology of an nde1Δ nde2Δ mutant.Cytosolic NADH can also be oxidized by the external NDE1- andNDE2-encoded NADH dehydrogenases. In contrast to thegut2Δ mutant, the physiology of an nde1Δ nde2Δ mutant in aerobic, glucose-limited chemostat cultures differed substantially from that of the wild-type strain (Fig. 2C and3C). The nde1Δ nde2Δ mutant produced glycerol at all specific growth rates tested, albeit in small quantities at low dilution rates, whereas the wild-type strain produced glycerol only at a dilution rate of 0.38 h⁻¹. No ethanol production was observed up to a dilution rate of 0.23 h⁻¹. Above this dilution rate, ethanol was produced and, concomitantly, the specific glycerol production rate decreased, suggesting a decreased need for a cytosolic redox sink under these conditions. At the highest applied dilution rate, the glycerol production increased again. The respiratory quotient was 1.0 at low dilution rates but increased as production of glycerol and ethanol became apparent (Fig. 3C). Washout of biomass occurred at dilution rates above 0.33 h⁻¹. At low dilution rates, the biomass yield was almost as high as that of the wild type, but it decreased when glycerol and ethanol were being produced. Apparently, other NADH-oxidizing systems, such as the glycerol-3-phosphate shuttle, could not completely replace the external NADH dehydrogenases but had a sufficiently high capacity to prevent alcoholic fermentation at low dilution rates.

Physiology of an nde1Δ nde2Δ gut2Δ mutant.To investigate whether, in addition to the glycerol-3-phosphate shuttle and the external NADH dehydrogenases, other enzymes contribute to the respiration of cytosolic NADH, an nde1Δ nde2Δ gut2Δmutant was grown in aerobic, glucose-limited chemostat cultures. At dilution rates between 0.05 and 0.15 h⁻¹ (Fig. 2D and 3D), this mutant produced no ethanol but its specific glycerol production rate was high. The biomass yield of the nde1Δ nde2Δ gut2Δ mutant (0.3 g g⁻¹) was significantly lower than that of the wild type (0.5 g g⁻¹).

At dilution rates above 0.15 h⁻¹, no reproducible steady-state data were obtained: it took a long time (sometimes more than 50 volume changes) before cultures of the nde1Δ nde2Δ gut2Δ mutant became steady. Moreover, in multiple experiments at the same dilution rate, different apparent steady states were obtained with different biomass yields and different specific production rates of ethanol and glycerol. However, the calculated carbon recoveries were always within 95 to 102%. Qualitatively, ethanol was produced at dilution rates of 0.20 h⁻¹ and higher and the specific rate of glycerol production decreased. If the specific ethanol production rate was higher, the specific glycerol production rate was lower. This inverse relationship was also observed with thende1Δ nde2Δ mutant (Fig. 2C).

As was evident from the glycerol production, the simultaneous deletion of the NDE and GUT genes resulted in a situation where reoxidation of cytosolic NADH could no longer occur exclusively via respiration. At low dilution rates, however, the nde1Δ nde2Δ gut2Δ mutant could still grow without alcoholic fermentation. The fact that the observed glycerol production is not large enough to oxidize all cytosolic NADH (see below) indicates that in addition to the external NADH dehydrogenases and the glycerol-3-phosphate shuttle, alternative mechanisms must exist that can couple the reoxidation of cytosolic NADH to the mitochondrial respiratory chain.

Physiological significance of Nde1/2p and Gut2p.The production of glycerol by aerobic, glucose-limited chemostat cultures of the nde1Δ nde2Δ mutant (Fig. 2C) is indicative of redox stress. This observation shows that external NADH dehydrogenase is involved in respiratory glucose dissimilation by wild-type S. cerevisiae. A gut2Δ mutant did not exhibit such a phenotype (Fig. 2B). Apparently, either this system is less important in wild-type cells or its role can be taken over by the external NADH dehydrogenases and/or alternative systems. The observation that respiratory growth on glucose was more severely impaired in annde1Δ nde2Δ gut2Δ mutant than in an nde1Δ nde2Δ mutant confirms the earlier conclusion that the glycerol-3-phosphate shuttle can operate in S. cerevisiae(16).

Surprisingly, even at low specific growth rates, aerobic glucose-limited chemostat cultures of the nde1Δ nde2Δ gut2Δ mutant produced substantial amounts of glycerol. At a dilution rate of 0.10 h⁻¹, the glycerol yield on glucose was 0.63 mol mol⁻¹. This is higher than the glycerol yields reached with the classical sulfite process (ca. 0.5 mol mol⁻¹ [10]), in which NADH reoxidation via alcoholic fermentation is prevented by trapping acetaldehyde with sulfite. At low dilution rates, glycerol production by thende1Δ nde2Δ gut2Δ strain was not accompanied by alcoholic fermentation. This was unexpected, since earlier studies on wild-type S. cerevisiae (32, 38) indicated that as the oxygen feed to glucose-limited cultures is reduced, alcoholic fermentation replaces respiration as the preferred mode of glycolytic NADH reoxidation (38). In such cultures, glycerol formation occurred only when the oxygen feed became too low to reoxidize the NADH that was formed in biosynthetic reactions. Indeed, the specific rate of glycerol production in anaerobic glucose-limited cultures of wild-typeS. cerevisiae matches the theoretical rate required for reoxidation of “biosynthetic” NADH, while oxidation of glycolytic NADH can be accounted for by ethanol production (1, 30, 32).

Alternative mechanisms for oxidation of cytosolic NADH.One explanation for the lack of ethanol production by the nde1Δ nde2Δ gut2Δ mutant at low dilution rates is cytosolic conversion of glucose to equimolar amounts of pyruvate and glycerol (Fig. 4A). This results in a closed cytosolic redox balance, because each molecule of NADH formed in the production of pyruvate is reoxidized in the production of glycerol from glucose. Pyruvate can be oxidized further in the mitochondria, leading to the production of 1 mol of glycerol and 3 mol of CO₂ for each dissimilated mole of glucose. At a dilution rate of 0.1 h⁻¹, the specific production rate of glycerol in thende1Δ nde2Δ gut2Δ mutant was 1.2 mmol of glycerol g of biomass⁻¹ h⁻¹. The amount of glycerol formed according to the scheme depicted in Fig. 4A should correspond to the sum of the NADH formed during biomass synthesis and the pyruvate produced in glycolysis, and it can be calculated as follows. Assuming a constant biomass composition, the rate of glycerol production as a result of biomass formation in the mutant equals the value observed in anaerobic wild-type cultures: ca. 0.9 mmol of glycerol g of biomass⁻¹ h⁻¹ (32, 37). The amount of pyruvate formed during glucose catabolism in the mutant can be approximated from the rate of CO₂ production, since mitochondrial oxidation of 1 mol of pyruvate yields 3 mol of CO₂. The rate of CO₂ production by thende1Δ nde2Δ gut2Δ mutant at a dilution rate of 0.1 h⁻¹ equaled 3.9 mmol g⁻¹ h⁻¹(Fig. 3D), corresponding to approximately 1.3 mmol of pyruvate g⁻¹ h⁻¹. Thus, the total rate of glycerol production expected according to Fig. 4A would amount to 0.9 + 1.3 = 2.2 mmol g⁻¹ h⁻¹. Since the observed value was only 1.2 mmol g⁻¹ h⁻¹, glycerol production cannot be solely responsible for the oxidation of cytosolic NADH in the nde1Δ nde2Δ gut2Δ mutant. This signifies that in addition to the external NADH dehydrogenases and the glycerol-3-phosphate shuttle, at least one other mechanism must exist which can oxidize cytosolic NADH.

• Open in new tab
• Download powerpoint

Fig. 4.

Schematic representation of proposed metabolic pathways for the oxidation of cytosolic NADH in an S. cerevisiae nde1Δ nde2Δ gut2Δ mutant, grown in an aerobic, glucose-limited chemostat culture at low dilution rates. (A) Conversion of dissimilatory glucose into equimolar amounts of pyruvate and glycerol; (B) consumption of ethanol produced in cytosol by mitochondria; (C) ethanol-acetaldehyde shuttle.

An alternative explanation for the phenotype of the nde1Δ nde2Δ gut2Δ mutant at low specific growth rates is alcoholic fermentation in the cytosol followed by respiratory ethanol dissimilation via intramitochondrial dehydrogenases (Fig. 4B). For complete oxidation, the intermediate acetate first has to be exported to the cytosol, where it can be converted to acetyl coenzyme A (acetyl-CoA) by cytosolic acetyl-CoA synthase (6). Subsequently, acetyl-CoA should be transported back into the mitochondrion for its further oxidation via the tricarboxylic acid cycle (Fig. 4B). This mechanism, by which reducing equivalents from glycolytic NADH are transported into the mitochondrion via ethanol, cannot account for oxidation of the cytosolic NADH from biomass production, because cytosolic conversion of glucose into ethanol is redox neutral. Thus, the need to reoxidize the NADH arising from biomass synthesis might explain the glycerol production by the mutant. Consistent with this model, the glycerol production rate by the mutant at a dilution rate of 0.10 h⁻¹ (1.2 mmol of glycerol g⁻¹ h⁻¹) corresponded well to that of anaerobic glucose-limited chemostat cultures of wild-type S. cerevisiae (0.9 mmol g⁻¹ h⁻¹) (32, 37). Although Fig. 4B adequately describes the phenotype of the mutant at low specific growth rates, it cannot explain the reduced glycerol yields found at higher specific growth rates.

A third explanation for the lack of ethanol production by thende1Δ nde2Δ gut2Δ mutant at low dilution rates is the presence of a redox shuttle. An interesting option is the involvement of an ethanol-acetaldehyde shuttle (Fig. 4C) as originally proposed by von Jagow and Klingenberg (35). The enzymes required for this shuttle mechanism are the same as in the mechanism proposed above (Fig. 4B), but in a shuttle mechanism, catalytic amounts of ethanol and acetaldehyde suffice to transfer all cytosolic redox equivalents into the mitochondria. Consequently, operation of this shuttle also allows oxidation of the NADH from biomass production. When involvement of a redox shuttle is assumed, the observed glycerol production might be explained from an insufficient capacity of the shuttle to oxidize all cytosolic NADH.

Significance of mitochondrial redox shuttles.In vivo, mitochondrial ethanol consumption (Fig. 4B) and the ethanol-acetaldehyde shuttle (Fig. 4C) may occur simultaneously: in the latter mechanism, acetaldehyde returns to the cytosol, while in the former, it is oxidized to acetate inside the mitochondria. The contribution of the two mechanisms may depend on activities of key enzymes (in particular mitochondrial acetaldehyde dehydrogenase, which withdraws acetaldehyde from the shuttle cycle), as well as on the intracellular concentrations of acetaldehyde and ethanol. It is conceivable that at low dilution rates (≤0.15 h⁻¹), reoxidation of glycolytic NADH in the nde1Δ nde2Δ gut2Δ mutant involves ethanol consumption by the mitochondria, whereas at dilution rates of ≥0.20 h⁻¹, where ethanol appears outside the cells, the yeast gradually shifts to the ethanol-acetaldehyde shuttle. The reduced glycerol production by thende1Δ nde2Δ gut2Δ mutant at dilution rates above 0.15 h⁻¹, which coincided with an increased rate of ethanol production (data not shown), is consistent with a need for the presence of ethanol to obtain a functional shuttle. The nde1Δ nde2Δ mutant displayed similar behavior at dilution rates above 0.25 h⁻¹ (Fig. 2C).

A role of ethanol and acetaldehyde as redox carriers reconciles the observed phenotype of the nde1Δ nde2Δ gut2Δ strain with the previous observation that alcoholic fermentation is the preferred mode of glycolytic NADH reoxidation by S. cerevisiae under conditions where respiration is compromised (38). However, involvement of alternative redox shuttle mechanisms (Fig. 1) cannot be entirely excluded. S. cerevisiae has long been assumed to lack a malate-aspartate shuttle, because mitochondrial aspartate aminotransferase was thought to be absent (7, 13). Recently, a gene homologous to other aspartate aminotransferase genes and with a mitochondrial targeting sequence was found in S. cerevisiae (21), indicating that a malate-aspartate shuttle may exist in this yeast. A recent study on mutants lacking the cytosolic (Mdh2p) or the mitochondrial (Mdh1p) malate dehydrogenases (29) does not provide conclusive evidence for in vivo activity of the malate-aspartate shuttle, since both malate dehydrogenases have other functions in the cell. Moreover, in the ethanol-grown cultures of thende1Δ mdh1Δ and the nde1Δ mdh2Δ mutant studied in these experiments (29), the glycerol-3-phosphate shuttle is likely to contribute to the oxidation of cytosolic NADH (16). Another possible shuttle is the malate-oxaloacetate shuttle (3, 39). Recently, a mitochondrial oxaloacetate transporter has been found in S. cerevisiae (23). Whether this transporter works in vivo in the direction necessary for the shuttle remains to be elucidated.

The proposed involvement of mitochondrial alcohol dehydrogenase, either via an ethanol-acetaldehyde shuttle (Fig. 4C) or via ethanol-consuming mitochondria (Fig. 4B), can be tested by deleting the structural gene(s) encoding mitochondrial alcohol dehydrogenase in thende1Δ nde2Δ gut2Δ mutant. In such a strain, neither of the two mechanisms should be possible. While exploring this line of research, we obtained evidence that ADH3 is not the soleS. cerevisiae gene encoding mitochondrial alcohol dehydrogenase (B. M. Bakker et al., unpublished results). Experimental verification of an ethanol-acetaldehyde shuttle will therefore have to await the identification of the additional structural gene(s) encoding mitochondrial alcohol dehydrogenases.

Our study illustrates that quantitative physiological analysis of defined mutants in chemostat cultures is a useful approach for studies on the compartmentation of redox metabolism in S. cerevisiae. S. cerevisiae is very easily amenable to genetic modification and isolation of functional mitochondria is relatively straightforward. Therefore, the nde1Δ nde2Δ gut2Δ mutant described in this study offers an attractive system for in vivo and in vitro functional analysis of structural genes encoding corresponding redox enzymes from other organisms.