The two isoenzymes for yeast NAD+‐dependent glycerol 3‐phosphate dehydrogenase encoded by GPD1 and GPD2 have distinct roles in osmoadaptation and redox regulation

Ricky Ansell, Katarina Granath, Stefan Hohmann, Johan M. Thevelein, Lennart Adler

Author Affiliations

  1. Ricky Ansell2,
  2. Katarina Granath1,
  3. Stefan Hohmann3,
  4. Johan M. Thevelein3 and
  5. Lennart Adler1
  1. 1 Department of General and Marine Microbiology, Lundberg Laboratory, Gothenburg University, Medicinaregatan 9C, S‐41390, Gothenburg, Sweden
  2. 2 Laboratorium voor Moleculaire Celbiologie, Katholieke Universiteit Leuven, Kardinaal Mercierlaan 92, B‐3001, Leuven‐Heverlee, Flanders, Belgium
  3. 3 National Laboratory of Forensic Science, S‐58194, Linküping, Sweden


The two homologous genes GPD1 and GPD2 encode the isoenzymes of NAD‐dependent glycerol 3‐phosphate dehydrogenase in the yeast Saccharomyces cerevisiae. Previous studies showed that GPD1 plays a role in osmoadaptation since its expression is induced by osmotic stress and gpd1Δ mutants are osmosensitive. Here we report that GPD2 has an entirely different physiological role. Expression of GPD2 is not affected by changes in external osmolarity, but is stimulated by anoxic conditions. Mutants lacking GPD2 show poor growth under anaerobic conditions. Mutants deleted for both GPD1 and GPD2 do not produce detectable glycerol, are highly osmosensitive and fail to grow under anoxic conditions. This growth inhibition, which is accompanied by a strong intracellular accumulation of NADH, is relieved by external addition of acetaldehyde, an effective oxidizer of NADH. Thus, glycerol formation is strictly required as a redox sink for excess cytosolic NADH during anaerobic metabolism. The anaerobic induction of GPD2 is independent of the HOG pathway which controls the osmotic induction of GPD1. Expression of GPD2 is also unaffected by ROX1 and ROX3, encoding putative regulators of hypoxic and stress‐controlled gene expression. In addition, GPD2 is induced under aerobic conditions by the addition of bisulfite which causes NADH accumulation by inhibiting the final, reductive step in ethanol fermentation and this induction is reversed by addition of acetaldehyde. We conclude that expression of GPD2 is controlled by a novel, oxygen‐independent, signalling pathway which is required to regulate metabolism under anoxic conditions.


The transition from aerobic to anaerobic growth conditions is accompanied by dramatic changes in cellular metabolism (Spiro and Guest, 1991; Zitomer and Lowry, 1992). Aerobic growth involves, to different extents depending on the organism, respiration where substrate oxidation is linked to the reduction of oxygen or an alternative electron acceptor via proton‐translocating electron transport. The generated proton‐motive force is used to drive the energy requiring processes. In fermentation, by contrast, energy is derived exclusively from substrate‐level phosphorylation and catabolism has to proceed in a redox‐balanced mode such that part of the substrate is used as a redox sink and sacrificed by excretion. In order for the cell to be able to shift swiftly between these two physiological states, the metabolism has to be sensitively attuned to the ambient redox state by a genetic regulatory programme that selects the optimal metabolic mode for any given environmental condition.

The yeast Saccharomyces cerevisiae responds to oxygen limitation by increased expression of a group of hypoxic genes which are all directly repressed under aerobic conditions by Rox1 (Deckert et al., 1995a,b). Expression of ROX1 is regulated by oxygen/haem availability mediated by the haem‐dependent transcriptional activator Hap1. Another nuclear protein, Rox3, has also been implicated in regulation of the haem‐repressed genes (Rosenblum‐Vos et al., 1991) and has recently been proposed to be involved in a general stress response (Evangelista et al., 1996). Little is known, however, about the control of gene expression in S.cerevisiae under strictly anaerobic conditions. Using differential hybridization, Zitomer and Lowry (1992) reported cloning of two sets of genes that were expressed ‘early’ and ‘late’, respectively, after shift to anoxic conditions. The expression of these genes was found to be haem‐ and Rox1‐independent, but their regulation has not yet been further described.

In cellular redox reactions the NAD+/NADH couple plays a vital role as a reservoir and carrier of reducing equivalents. For catabolic reactions to proceed, the ratio between the two pyridine nucleotides has to be set high. Under aerobic conditions, oxidation of the NADH generated in the cytosol is assumed to occur via a shuttle system (Dawson, 1979), which transfers the reducing power to the mitochondrial electron transport chain. In S.cerevisiae, mitochondrial NADH oxidases that face the cytosol may also contribute to maintaining a NAD+/NADH ratio that assures the continuous operation of central metabolism (de Vries and Marres, 1989).

Anaerobic conditions require production of endogenous electron acceptors. Theoretical considerations (Nordstrüm, 1966; Lagunas and Gancedo, 1973; Van Dijken and Scheffers, 1986) and experimentally determined mass balances (Nordstrüm, 1968; Van Dijken and Scheffers, 1986) indicate that glycerol production serves as a redox valve to dispose excess reducing power in S.cerevisiae. Production of glycerol involves reduction of dihydroxyacetone phosphate to glycerol 3‐phosphate via glycerol 3‐phosphate dehydrogenase and dephosphorylation of glycerol 3‐phosphate by a specific glycerol 3‐phosphatase (Norbeck et al., 1996). Thus, glycerol production is an energetically wasteful process. The NAD+‐dependent glycerol 3‐phosphate dehydrogenase catalysing the first step in glycerol production is encoded by two isogenes GPD1 and GPD2 (Larsson et al., 1993; Albertyn et al., 1994; Eriksson et al., 1995). Expression of GPD1 increases in cells exposed to raised external osmolarity which causes an increased production and intracellular accumulation of glycerol (André et al., 1991; Albertyn et al., 1994; Eriksson et al., 1995). This osmoregulatory response serves to counter the dehydration and loss of cell volume that is experienced by cells subject to increased osmotic stress. The osmotic control of GPD1 expression is possibly executed via a recently described osmosensing signal transduction cascade, called the HOG pathway, involving homologues of mitogen‐activated protein (MAP) kinases and of a bacterial phosphorelay system (Boguslawski, 1992; Brewster et al., 1993; Albertyn et al., 1994; Maeda et al., 1994, 1995; Posas et al., 1996; Hohmann, 1997). The GPD2 gene which encodes a protein highly homologous to the GPD1 product is, however, not under osmotic control and deletion of GPD2 yielded no obvious phenotype (Eriksson et al., 1995).

In this paper we show that GPD2 is subject to redox control and induced by anoxic conditions, and that regulation of GPD2 expression appears to be independent of ROX1 and ROX3. Deletion of GPD2 results in defective growth under anaerobic conditions, while deletion of both GPD1 and GPD2 causes a complete block in glycerol production, failure to grow at all under anoxic conditions and strong osmosensitivity.


To examine the physiological role of the two isogenes encoding glycerol 3‐phosphate dehydrogenase, the properties of a tetrad containing isogenic spores of GPD1 GPD2 (wild‐type), gpd1Δ GPD2, GPD1 gpd2Δ and gpd1Δ gpd2Δ were investigated.

Growth under osmotic stress

Cultured at high salinity (0.7 M NaCl) the gpd1Δ strain showed a markedly reduced salt tolerance, while the gpd2Δ strain behaved like the wild‐type strain (Figure 1A), consistent with previous observations (Larsson et al., 1993; Albertyn et al., 1994; Eriksson et al., 1995). Deletion of both GPD1 and GPD2 resulted in a very severe growth defect at high salinity (Figure 1A).

Figure 1.

Growth of the wild‐type strain (▴),and the gpd1Δ (▵), gpd2Δ (▪) and gpd1Δ gpd2Δ (□) mutants in (A) YNB medium supplemented with 0.7M NaCl or (B) YNB medium under anaerobic conditions. (C) Anaerobic growth of the gpd1Δ gpd2Δ mutant in YNB medium, without (□) or with (□) the addition of 10 mM acetaldehyde (time of addition is indicated by an arrow).

Growth under anaerobic conditions

Incubation under anaerobic conditions revealed a growth pattern for the single mutants that was opposite to that seen at high osmolarity (Figure 1B). Under anoxic conditions, the gpd1Δ mutant exhibited wild‐type growth behaviour, while the gpd2Δ strain showed a reduced growth rate. This is consistent with results from experiments where growth conditions were switched from aerobic to anaerobic (Bjürkqvist et al., 1997). Growth of the gpd1Δ gpd2Δ strain was completely inhibited under anoxic conditions. This mutant also did not grow aerobically in the presence of 10 mM antimycin A, whereas growth of the wild‐type strain and the single mutants was unaffected by this respiration inhibitor (data not shown). Since cells growing in a rich medium are expected to produce less reducing equivalents than in a minimal medium (Nordstrüm, 1966; Lagunas and Gancedo, 1973; Van Dijken and Scheffers, 1986), we have examined the growth behaviour of the mutants also in rich medium. Indeed, the anaerobic growth defect of the gpd2Δ strain was not observed in rich YPD medium (see Figure 5). In contrast, the growth inhibition of the double mutant under anaerobic conditions was not relieved in rich medium. However, the reduced growth rate of the double mutant observed under aerobic conditions was reverted to wild‐type behaviour in YPD medium. Spores of the double mutant resumed growth more slowly than wild‐type and single mutants on YPD medium, even under aerobic conditions.

The external addition of acetaldehyde (10 mM), an effective oxidizer of NADH (Holzer, 1961), rescued anaerobic growth of the gpd1Δ gpd2Δ double mutant (Figure 1C). These results indicate that the anaerobic growth inhibition of the double mutant is due to a defective redox regulation, resulting in accumulation of NADH and lack of NAD+ in the cytosol.

Glycerol production and NAD(P)H levels

Analysis of glycerol production by wild‐type and mutant strains during aerobic or anaerobic conditions in the presence or absence of 0.7 M NaCl, revealed different patterns for all strains (Figure 2). The induction of glycerol production observed in wild‐type cells after transfer to increased osmolarity is completely absent in the gpd1Δ mutant during aerobic as well as anaerobic conditions. The strong increase in glycerol synthesis occurring after a shift to anaerobic conditions is similar to that of the wild‐type strain. The gpd2Δ strain displays, on the other hand, osmotic induction of glycerol production similar to that of the wild‐type cells, although the absolute levels are lower in the mutant strain. Anaerobic induction of glycerol production also occurs in this strain to a relative extent similar to that in the wild‐type. However, the glycerol levels achieved by the gpd2Δ mutant are considerably lower than in the reference strains. The gpd1Δ gpd2Δ strain showed no detectable glycerol production under any of the conditions tested.

Figure 2.

Total glycerol production by (A) wild‐type, (B) gpd1Δ, (C) gpd2Δ and (D) gpd1Δ gpd2Δ strains cultured in YNB medium with (+) or without (−) 0.7 M NaCl under aerobic (O2) or anaerobic (N2) conditions. Glycerol was analysed using extracts of overnight YNB cultures that were diluted to A610 ∼1 by fresh medium and incubated for 4 h under the described conditions.

Measurements of the intracellular NAD(P)H levels after transfer of wild‐type and gpd1Δ gpd2Δ double mutant cells to anaerobic conditions confirmed the strongly diminished ability of this strain to reoxidize NADH to NAD+ (Figure 3). The NAD(P)H levels attained by the double mutant under anaerobic conditions were ∼10‐fold higher than in wild‐type cells. The external addition of acetaldehyde (10 mM), which rescued growth of the gpd1Δ gpd2Δ double mutant under anaerobic conditions (Figure 1C) also restored the NAD(P)H concentration to wild‐type levels (Figure 3). These results show that the excess of reduced pyridine nucleotide in the cell is NADH rather than NADPH, since the reduction of acetaldehyde to ethanol occurs via alcohol dehydrogenase which is strictly dependent on NADH (Racker, 1955). These observations also indicate that the anaerobic growth inhibition of the double mutant is due to an altered redox state of the pyridine nucleotides, rather than to the inability to produce glycerol per se. This notion is further supported by the observation that anaerobic growth as well as ethanol and CO2 production of the double mutant could be rescued also by the external addition of acetoin instead of acetaldehyde, and measurement of culture fluorescence indicated that the NAD(P)H level decreased upon addition of acetoin (Bjürkqvist et al., 1997).

Figure 3.

Intracellular NAD(P)H concentration in wild‐type (dark bars) and gpd1Δ gpd2Δ (grey bars) strains transferred from aerobic to anaerobic conditions and incubated for the indicated times. Acetaldehyde was added to 10 mM final concentration after 3 h of anaerobic incubation and samples taken after an additional incubation period of 2 h.

Km values of Gpd1 and Gpd2

To examine whether the enzyme encoded by the GPD2 gene has substrate affinities different from that of the GPD1‐encoded dehydrogenase, enzyme activities were determined for different substrate concentrations in extracts of cells in which the GPD1 gene was deleted. Using Lineweaver‐Burk plots, KmNADH was estimated as 0.018 mM and KmDHAP as 0.86 mM for the Gpd2 enzyme, which compares well with the kinetic parameters previously reported for purified Gpd1 dehydrogenase, being 0.018‐0.023 mM for NADH and 0.37‐0.54 mM for dihydroxyacetone phosphate (Chen et al., 1987; Albertyn et al., 1992). Obviously, the different phenotypes of the gpd1Δ and gpd2Δ mutants cannot be explained by differences in substrate affinities of the two isoforms.

Gene expression

To analyse the effect of anaerobiosis on the expression of GPD1 and GPD2, the changes in transcript levels were analysed by Northern blotting using RNA from wild‐type cells before and after a shift to anaerobic conditions (Figure 4A). The transcript levels of GPD1 were essentially indifferent to such transfer, whereas the GPD2 mRNA levels were markedly induced, reaching a 9‐fold induction 2 h after transfer to anoxic conditions. This high induction level was, however, only transient. After ∼3‐5 h, when a culture shifted to anaerobic conditions commenced growth (Figure 1B), the GPD2 mRNA level had fallen back to ∼2‐ to 3‐fold compared with aerobically cultured cells. A similar pattern of anaerobic induction of GPD2 was observed in a strain lacking the GPD1 gene (data not shown), explaining the similar anaerobic induction of glycerol production in wild‐type and gpd1Δ strains (Figure 2A and B). In cells deleted for the GPD2 gene, the expression of GPD1 becomes responsive to anoxia, showing a slightly increased expression on transfer to anaerobic conditions (data not shown). This is in contrast to the situation in wild‐type cells, but in full agreement with the slight increase in glycerol production that occurs in gpd2Δ cells shifted to anaerobic conditions (Figure 2C), indicating that GPD1 can partly substitute for a defective GPD2 gene. To further examine the mechanism behind GPD2 induction, we followed the transcript level in wild‐type cells grown aerobically in a bioreactor following exposure to bisulfite. Bisulfite traps acetaldehyde and blocks its reduction to ethanol, causing an enhanced NADH/NAD+ ratio and increased glycerol production (Neuberg, 1945; Holzer et al., 1963). The fixation of acetaldehyde by added bisulfite caused an ∼2‐fold induction of GPD2 (Figure 4B), while the subsequent external supply of acetaldehyde produced an ∼2‐ to 3‐fold down‐regulation of GPD2 expression (data not shown). These results indicate that transcriptional regulation of GPD2 does not respond directly to changes in the ambient oxygen availability, but senses changes in the redox state of an internal reporter system which is directly or indirectly linked to the NADH/NAD+ ratio.

Figure 4.

Northern blot analysis of the level of GPD1 and GPD2 mRNA from wild‐type cells (A) incubated in YNB media for 4 h and then shifted to anaerobic conditions; (B) cultured in a bioreactor with controlled aeration and then furnished with 2.5 mM bisulfite or (C) incubated for 4 h with various concentrations of NaCl. Cells were sampled for RNA extraction at the indicated times. Actin mRNA served as internal control. 15 μg of RNA was applied to each gel lane. Time scale refers to time of sampling after the shift in growth conditions.

Figure 5.

Growth of the gpd2Δ strain in YNB medium under aerobic (▴), anaerobic (▵), anaerobic + 0. 7 M NaCl (▪) or anaerobic conditions in YPD medium (□).

We also examined the effect of increased extracellular salinity on the expression of GPD1 and GPD2 (Figure 4C). Northern blotting of RNA from wild‐type cells incubated for 4 h in medium containing NaCl in a concentration range 0‐1.8 M, demonstrated that GPD1 expression increased in a linear fashion with increasing osmotic stress, as previously observed using a reporter gene system (Eriksson et al., 1995). Quantitation revealed that the transcript levels had increased 35‐fold over the basal level at the highest salinity. The transcript levels of the isogene GPD2 were not induced by osmotic stress but, on the contrary, decreased with increased salt stress. This inverse regulation of GPD2 relative to that of GPD1 is also consistent with sensitivity of GPD2 expression to the NADH/NAD+ ratio. The large amounts of glycerol that result from enhanced GPD1 expression under high osmolarity conditions can be expected to repress GPD2 expression by decreasing this ratio.

GPD1 and GPD2 can replace each other under anaerobic salt stress or by overexpression

The anaerobic growth defect of a gpd2Δ strain in defined minimal medium does not appear in cultures containing 0.7 M NaCl (Figure 5), indicating that the osmotic induction of GPD1 relieves the redox constraints of strains lacking GPD2. Conversely, anaerobic incubation of a gpd1Δ strain reverses its inability to grow at 1 M NaCl (data not shown). Hence, the enzymes encoded by GPD1 and GPD2 appear to be able to substitute for each other, depending on the growth conditions. This was further supported by expression of GPD1 or GPD2 from a multicopy plasmid in the gpd1Δ gpd2Δ double mutant. When overexpressed by such means, any of the two isogenes relieves the growth defect of the mutant on high‐salinity media or under anoxic conditions (data not shown). Overexpression of either gene causes a strong overproduction of glycerol (Figure 6) which is apparently sufficient to rescue the growth defects of the double mutant under either stress condition. The functional overlap between the two GPD gene products is also supported by the fact that the sensitivity of gpd1Δ to high osmolarity or gpd2Δ to anoxic conditions is aggravated in a gpd1Δ gpd2Δ double mutant. The fact that multicopy expression of GPD1 does not further enhance glycerol production under high osmolarity conditions (Figure 6, upper panel, columns A and B), indicates that factors different from mRNA or enzyme levels limit production under these circumstances.

Figure 6.

Total glycerol production by (column A) wild‐type and the gpd1Δ gpd2Δ strain transformed with (column B) GPD1 or (column C) GPD2 in the multicopy YEplac181 plasmid. Glycerol was analysed using cultures incubated for 4 h in YNB medium with (+) or without (−) 0.7 M NaCl under aerobic (O2) or anaerobic (N2) conditions.

Anaerobic induction of GPD2 is independent of the HOG pathway or known regulatory proteins involved in oxygen‐dependent gene expression

The osmosensing HOG signalling pathway controls the osmotic induction of the GPD1 gene (Albertyn et al., 1994). To examine whether this pathway is involved also in the control of the GPD2 expression, we followed the anaerobic induction of this gene in a hog1Δ strain (Figure 7A). The pattern of induction was clearly independent of an intact HOG pathway.

Figure 7.

Northern analysis of the level of GPD2 mRNA in a hog1Δ strain incubated in YNB media for 4 h and then shifted to anaerobic conditions using actin mRNA as internal control or during aerobic (O2) and anaerobic (N2) conditions in strains deleted for ROX1 or carrying an inactivating mutation in ROX3. RNA was isolated 4 h after transfer of cultures from aerobic to anaerobic conditions using strain RZ53‐6, its congenic rox1Δ strain, strain aGH1 and its congenic rox3 strain. Actin mRNA was used as internal control.

In S.cerevisiae, the ROX1 gene encodes a haem‐controlled repressor of hypoxic genes that is active only under aerobic, but not anaerobic, conditions (Zitomer and Lowry, 1992). The ROX3 gene encodes a nuclear protein which initially has been proposed to be involved in control of oxygen‐dependent gene expression (Rosenblum‐Vos et al., 1991). Recently, Rox3 has been shown to be involved in a general response system to diverse stress conditions (Evangelista et al., 1996). Anaerobic induction of GPD2 seems to be independent of ROX1, since neither the kinetics (data not shown) nor the final level of induction (Figure 7B) were affected by deletion of ROX1. The aerobic expression of GPD2 was also not significantly increased in a rox1Δ strain, as would be expected from the mode of action of Rox1p as a transcriptional repressor. Similarly, inactivation of Rox3p did not appear to affect the capacity for anaerobic induction of GPD2. Since the rox3 mutant was, for unknown reasons, unable to grow in defined minimal media, the effect on GPD2 expression of the Rox3 factor was examined in rich YPD media.


One enzyme, two functions, two genes

The two isogenes GPD1 and GPD2, encoding glycerol 3‐phosphate dehydrogenase in glycerol metabolism of S.cerevisiae, fulfil distinctly different functions in cellular physiology. Expression of GPD1 is induced under hyperosmotic stress, probably via an osmosensing signal transduction pathway (Brewster et al., 1993; Albertyn et al., 1994) and mutation of GPD1 results in osmosensitivity (Larsson et al., 1993; Albertyn et al., 1994). In this study, the expression of the GPD2 gene was found to be subject to anaerobic induction and the gpd2Δ mutant proved affected in growth when oxygen was excluded. Hence, the two GPD isogenes are regulated and used for clearly different physiological purposes. There is no evidence that the two gene products are functionally different: there is a high identity at the amino acid level (Eriksson et al., 1995) and the isoforms have similar affinity for their substrates. Apparently, the two isoenzymes are only separated by the regulatory context under which they are produced. This is further demonstrated by multicopy overexpression where either isogene can fulfil both functions. In addition, application of the appropriate inducing conditions for one isoenzyme could also suppress the growth defect conferred by deletion of the other isogene. Thus, when expressed at a high enough level, either isoform is capable of substituting for the other.

Recently, it has been reported that the fission yeast Schizosaccharomyces pombe also possesses a pair of GPD genes. One of the two, gpd1, is induced by osmotic stress but the function of the second gene could not be determined (Ohmiya et al., 1995). Given the similar fermentative physiology of the two divergent yeasts, it appears likely that the S.pombe gpd2 gene resembles S.cerevisiae GPD2 in its physiological role.

Apparently, yeasts have chosen to duplicate the GPD genes and to have two genes controlled by different physiological triggers rather than to control the expression of one and the same gene by different pathways. It is unclear whether this reflects a general evolutionary strategy, or may differ from gene to gene. However, isogenes are abundant in the S.cerevisiae genome (Goffeau et al., 1993) and in most cases the specific function of one or several of the isoforms is not known. Hence, clues to the physiological role of other isogenes might be revealed in some instances by examining the anaerobic phenotype of their null mutations. Additional examples where two isogenes have partially overlapping function but are controlled by entirely different mechanisms are glucan synthase (Mazur et al., 1995), iso‐cytochrome C (Laz et al., 1984; Evangelista et al., 1996) and plasma membrane ATPase (Serrano, 1991).

Glycerol metabolism

Production of glycerol is a wasteful process that is associated with ATP expenditure and loss of a reduced three‐carbon compound that can not be re‐utilized under anaerobic conditions. The physiological raison d'être for the importance of glycerol production in yeast metabolism resides in its involvement in osmoregulation, where it serves as primary compatible solute (Blomberg and Adler, 1992; Mager and Varela, 1993) and in its capacity to serve as an innocuous, pH‐neutral sink for reducing equivalents (Nordstrüm, 1968; Oura, 1977). The phenotype of the gpd1Δ gpd2 mutant clearly illustrates the importance of these functions. This mutant, which is incapable of producing detectable amounts of glycerol, is not only strongly sensitive to osmotic stress, but also incapable of growing under anaerobic conditions. Remarkably, however, the gpd1Δ gpd2Δ double mutant could grow under aerobic conditions, despite its blocked pathway to glycerol 3‐phosphate, the precursor of glycerolipids. Presumably, the acyl dihydroxyacetone pathway enzymes of S.cerevisiae (Racenis et al., 1992) provide sufficient activity for alternative biosynthesis of phosphatidic acid.

Redox regulation

Transfer of facultative anaerobes from aerobic to anaerobic conditions elicits large‐scale changes in the pattern of gene expression. When Escherichia coli is shifted from fully aerobic to oxygen‐limited environments, expression of at least 50 genes is induced (Smith and Neidhardt, 1983; Clark, 1984). Two global transcriptional regulators play a major role in redox regulation in this organism. One of these, FNR, is an iron‐dependent DNA‐binding protein which regulates transcription of many anaerobically important genes (Spiro and Guest, 1991; Iuchi and Lin, 1993; Bell and Busby, 1994). The other protein involved in anaerobic gene regulation is ArcA, which serves as repressor of several genes involved in aerobic metabolism. ArcA is one of the constituents in a two‐component system for which the other protein, ArcB, is a membrane‐associated histidine kinase that activates ArcA by phosphorylation (Iuchi and Lin, 1991, 1993). It is speculated that ArcB signals to ArcA in response to the redox state of a component in the electron carrier chain. The signal to which FNR responds has not yet been identified.

In mammalian cells, the activities of two global transcriptional factors, NF‐κB (Baeuerle, 1991) and the Fos‐Jun heterodimer AP‐1 (Morgan and Curran, 1991), appear to be subject to redox control. In vitro, the DNA‐binding activity of AP‐1, is inhibited by oxidation of a conserved cysteine in the DNA‐binding domain of the two proteins (Abate et al., 1990). DNA binding is regained by the presence of reducing agents or by a nuclear factor, denoted redox factor‐1, Ref‐1 (Xanthodaikis and Curran, 1992), a homologue of which has also been described in plants (Babiychuck et al., 1994). The physiological relevance of Ref‐1 and the redox‐modulated cysteines of AP‐1 is obscure, but the observations suggest that redox cycles may be important mechanisms for the regulation of these transcription factors.

Little is understood about the regulation of functions specifically required for anaerobic growth of S.cerevisiae. A set of hypoxic genes encoding selected enzymes in haem, sterol and fatty acid biosynthesis is repressed under aerobic conditions and induced when oxygen is limiting (Zitomer and Lowry, 1992). It is assumed that increased expression of these genes is required to compensate for restrictive oxygen availability under hypoxic conditions. Other genes induced at low oxygen tension encode replacement functions that might increase the respiratory capacity of the cells under oxygen limitation, such as an alternative subunit V of cytochrome oxidase and an alternative mitochondrial ATP/ADP translocator (Zitomer and Lowry, 1992). The availability of oxygen appears to be sensed through the haem level, the synthesis of which is oxygen dependent. Haem activates the transcription of ROX1, encoding a hypoxic repressor (Balasubramanian et al., 1993; Deckert et al., 1995a). The ROX1‐controlled genes encode products that appear to be required only when oxygen is limiting. The mechanisms regulating genes serving more general metabolic functions under anaerobiosis are entirely unknown. Using differential hybridization, Zitomer and Lowry (1992) reported isolation of 15 genes that were induced by anoxic conditions. Of these, the four studied in more detail were induced by anaerobiosis in a ROX1‐independent fashion, but no function was associated with any of the isolated genes. It is thus of particular interest that the GPD2 gene, which encodes a dehydrogenase that plays a distinct role for anoxic redox regulation, is induced by anaerobiosis. A specific regulatory mechanism, independent of ROX1 and ROX3, should exist to achieve this induction, since anoxic induction of GPD2 occurred in a ROX1‐ as well as a ROX3‐deficient background. ROX3 encodes a nuclear protein that appears to be involved in the control of anaerobic expression of haem‐regulated genes (Rosenblum‐Vos et al., 1991), and which participates also in mediating diverse stress responses in yeast (Evangelista et al., 1996). It is also clear that the osmosensing HOG pathway (Brewster et al., 1993), controlling the osmotic induction of GPD1 (Albertyn et al., 1994), does not compromise the anoxic induction of GPD2. This indicates that the enhanced production of glycerol 3‐phosphate dehydrogenase resulting from both osmotic and anaerobic stress is not only controlled by distinct isoenzymes, but is also subject to functionally discrete regulatory mechanisms.

Hence, the anaerobic induction of GPD2 seems to occur by a novel mechanism, independent of the known hypoxic gene regulators. Preliminary evidence suggests that oxygen itself is not directly involved as a regulatory signal. Instead, the redox state of the cytosol appears to be sensed, perhaps at the level of the NADH/NAD+ ratio. Support for this interpretation comes from the observation that, also under aerobic conditions, the GPD2 mRNA level responds to regimes that enhance (addition of bisulfite) or decrease (addition of acetaldehyde) the cytosolic NADH/NAD+ ratio. While this ratio plays a vital role in sustaining and coordinating the catabolic reactions of the cell, very little is still known about the way the levels are set. GPD2 expression appears to be an excellent starting point to study the molecular biology of redox sensing and signalling. Analysis of the GPD2 promoter should lead to the identification of redox‐responsive regulatory elements. Presently, we are studying mutants derived from a gpd1Δ deletion strain which show elevated expression of GPD2 with the idea to identify novel genes in the redox regulatory system.

Material and methods

Yeast strains and growth conditions

The S.cerevisiae strains used in this study are described in Table I. To construct a tetra‐type tetrad (wild‐type, gpd1Δ, gpd2Δ and gpd1Δ gpd2Δ) a linearized fragment containing the disrupted GPD2 gene (Eriksson et al., 1995) was transformed into the wild‐type strain W303‐1A. The generated GPD2 null mutant was confirmed by Southern analysis and mated with a previously constructed GPD1 null mutant of opposite mating type in W303‐1A background (Albertyn et al., 1994). Strain construction, mating, sporulation and tetrad analysis followed standard protocols (Sherman and Hicks, 1991). The S.cerevisiae RZ53‐6 and aGH1 strains were kindly provided by Dr R.S.Zitomer.

View this table:
Table 1. Yeast strains used in this study

Cells were cultured at 30°C in defined minimal medium [0.7% Bacto Yeast Nitrogen Base (YNB), without amino acids], supplemented with 2% glucose, and 120 μg/ml of appropriate nutrients, or in a complex YP medium with 2% glucose and 120 μg/ml adenine. Liquid cultures were routinely grown in 250 ml Erlenmeyer flasks on a rotary shaker at 225 r.p.m. Media used for anaerobic incubations were supplemented with ergosterol (10 mg/l) and Tween 80 (420 mg/l) (Andreasen and Stier, 1953, 1954). Anaerobic growth was performed in 100 ml serum flasks or 15 ml anaerobic culture tubes (Bellco Glass Inc., USA) equipped with rubber stoppers that were tightened with aluminium or screw‐cap sealing. Before anaerobic incubation, culture media were flushed for 3 min with nitrogen gas (AGA plus, Sweden) containing <5 p.p.m. of oxygen.

Plasmids were selected and propagated in E.coli DH5α or TG1. Growth was followed by optical density measurements at 610 nm (A610).

Plasmid constructs

The GPD1 and GPD2 deletion constructs used were as previously described (Albertyn et al., 1994; Eriksson et al., 1995). To overexpress GPD1 or GPD2 in the gpd1Δ gpd2Δ double mutant, the 2.9 kb ClaI‐SwaI fragment of GPD1 (Larsson et al., 1993) or the 2.5 kb DraI‐PstI fragment of GPD2 (Eriksson et al., 1995) were cloned into the multicopy vector YEplac181 (Gietz and Sugino, 1988) and transformed by the LiAc method into W303‐1A gpd1Δ gpd2Δ strain. All cloning procedures followed standard protocols (Sambrook et al., 1989).

Northern blotting

Total RNA was isolated by standard procedures from overnight cultures grown to A610 = 1‐1.5 and transferred to an equal volume of fresh minimal media with or without 0.7 M NaCl. Anaerobic cultures were contained in anaerobic flasks that were incubated with shaking as the aerobic cultures. For examining the effects of bisulfite addition, cells were grown in YNB medium in a Belach bioreactor (Belach Bioteknik AB, Sweden) with a working volume of 2.5 l. Aeration was adjusted to give oxygen levels >60% of air saturation. Bisulfite was added to a concentration of 2.5 mM and to revert the effect of bisulfite, acetaldehyde was added to the same concentration.

RNA samples were screened on ethidium bromide‐agarose gels and quantified by spectrophotometry, using the Beckman DU65 (Beckman Instruments) nucleic acid program. Samples containing 15 μg of total RNA were denatured and run on low‐formaldehyde (2.5% v/v) agarose gels at 10 V/cm for 75 min, tested for quality at 254 nm on TLC plates, and blotted to positively charged nylon membranes (Boehringer Mannheim) by capillary transfer, using 10× SSC as transfer buffer. Blotted filters were crosslinked by 1‐min exposure to low‐wavelength UV and baked at 80°C for 2 h.

Prehybridizations were performed for 3‐4 h at 55°C in 5× SSC, 10 mM sodium phosphate, pH 6.5, 10× Denhardt's solution, 2% SDS and 10 mg/ml herring sperm DNA.

Hybridizations were performed at 55°C for 18‐20 h using the same solution supplemented with 10% dextran sulfate and having labelled oligonucleotides added at 5 ng/ml. The filters were washed twice in 1× SSC/1% SDS, for 20 min at room temperature and once for 15 min at the hybridization temperature. The filters were analysed by using a Molecular Dynamics PhosphorImager. Membranes were stripped for re‐hybridization by shaking in sterile water at 80°C for 10 min.

The oligonucleotides used were 5′‐labelled with 25 mCi [γ‐32P]ATP (Amersham) and 5 U poly‐nucleotide kinase (Boehringer Mannheim) per 50 ng probe, left at 37°C for 30 min. Unincorporated ATP was displaced using a Sephadex G50 (Pharmacia) mini‐column.

Sequences of oligonucleotides used were: 5′‐TGTACTATTGGAGCGAAAACTTCT‐3′ to probe for GPD1 mRNA, and 5′‐GGTCCTCATGACAGTGTTTGTGCT‐3′ to assay for GPD2 mRNA. The control ACT1 probe (a gift from Ines Eberhardt, K.U.Leuven) was 5′‐ATCGATTCTCAAAATGGCGTGAGG‐3′. The specificities of the oligonucleotides used for probing GPD1 and GPD2 mRNA were controlled using the gpd1Δ and gpd2Δ strains (Eriksson et al., 1995), as well as the gpd1Δ gpd2Δ mutant where no GPD transcript was detected.

Enzyme activity measurements

To determine Km values for the Gpd2 enzyme, a crude extract of gpd1Δ cells was prepared and desalted by gel filtration (Blomberg and Adler, 1989). Enzyme activity was assayed in 20 mM MES buffer, pH 6.5, containing 10 mM Mg2+. Dihydroxyacetone phosphate and NADH were varied within the concentration range 0.2‐4 mM and 0.01‐0.1 mM, respectively.

Glycerol analysis

To analyse the capacity for glycerol production overnight cultures were grown to A610 = 1‐1.5 and transferred to an equal volume of fresh minimal media with or without an NaCl addition to give a final concentration of 0.7 M. Samples of 1.5 ml were taken 4 h after transfer of cultures to fresh medium. The sampled cell suspensions were analysed for total (intra‐ plus extracellular) glycerol by boiling for 10 min before clearing by centrifugation and determination of the glycerol content using a commercial kit (Boehringer‐Mannheim). The amount of glycerol was related to cell mass by the optical density measurements of the cultures at the time of harvest (A610 = 1 equals 0.3 mg cells dry weight/ml).

NADH analysis

Overnight cultures grown in Erlenmeyer flasks on a rotary shaker were diluted by fresh medium to A610 ∼1 and 5 ml were transferred to anaerobic culture tubes and samples taken at time intervals for NADH estimation. Before NADH extraction, the cultures were centrifuged at 4°C for 10 min at 3000 g and 4 ml of the supernatant was removed from the inverted culture tube using a needle and syringe. The remaining cell suspension (1 ml) was extracted by anaerobically adding 250 μl of degassed 2 M KOH in 90% (v/v) methanol and incubating for 1 min at 55°C. After cooling on ice, the tubes were opened and each sample was extracted with 2 ml of diethyl ether for 5 min. Samples were then brought to pH 8.4 using 1 M triethanolamine‐HCl, pH 5.5, and thereafter centrifuged at 4°C for 3 min at 10 000 g, before freezing in liquid nitrogen. Samples were analysed for NAD(P)H using a bioluminescence assay with luciferase/NAD(P)H:FMN oxidoreductase (Boehringer‐Mannheim Biochimica Cat. No. 567 728). Since the kit is no longer commercially available the required reagents and standard solutions were prepared according to the description of the Boehringer test protocol. Luminescence was measured in a Packard Pico Lite Luminometer.


We are grateful to Agneta Danielsson and Cecilia Eriksson for skilful help with the NAD(P)H measurements, and to Hadi Valadi for expert assistance with the bioreactor experiments. This work was supported by grants from the Swedish National Board for Natural Sciences (NFR), the Swedish National Board for Technical Development (TFR), and the EU programs BIO‐CL 950161 and ERB4061 PL95‐0014.