For survival microorganisms should adapt continuously to changes in their environment. Most often this involves regulation of metabolic fluxes, i.e. to supply the extra ATP required to cope with stress and/or to produce compounds that are required under specific conditions, such as the compatible solute glycerol under hyperosmotic conditions. Flux regulation can be achieved by changing the concentrations of metabolites or by regulation of the concentrations of the enzymes involved. Many studies focus solely on metabolic regulation (Erasmus et al., 2003; Kresnowati et al., 2006) or rather on gene-expression regulation (Greatrix and van Vuuren, 2006; Marks et al., 2008). However, in reality gene expression and metabolism may contribute simultaneously (Rossell et al., 2005; ter Kuile and Westerhoff, 2001). Moreover, gene expression itself can be affected at several levels at the same time, for instance via transcription, mRNA degradation, protein synthesis and protein degradation (Daran-Lapujade et al., 2007; Garcia-Martinez et al., 2004; Kuhn et al., 2001; Molin et al., 2009; Vabulas and Hartl, 2005).
Induction of gene expression during the adaptation of the yeast Saccharomyces cerevisiae to an osmotic shock has been subject to extensive studies. S. cerevisiae accumulates the osmolyte glycerol intracellularly if the cell is exposed to high levels of salt or sorbitol (Posas et al., 2000). The production of glycerol from dihydroxyacetone phosphate is catalysed by the enzymes glycerol-3-phosphate dehydrogenase (EC number 18.104.22.168) and glycerol-3-phosphatase (EC number 22.214.171.124; Albertyn et al., 1994; Pahlman et al., 2001). The transcription of the genes encoding these enzymes is induced upon an osmotic shock via phosphorylation of the MAP-kinase Hog1p (Brewster and Gustin, 1994; Capaldi et al., 2008). Hog1p can be phosphorylated via two distinct pathways, the Sho1 and the Sln1 pathways (Chen and Thorner, 2007; Maeda et al., 1995; Posas et al., 1998; Tatebayashi et al., 2007). Upon phosphorylation Hog1P relocalizes to the nucleus and induces the expression of a great number of target genes, among which are GPD1, GPP1 and GPP2 encoding glycerol-3-phosphate dehydrogenase and glycerol-3-phosphatase, respectively. In addition, the glycerol concentration in the cell is increased by reduction of the leakage of glycerol from the cell. This is achieved by closure of the glycerol channel Fps1p (Hohmann et al., 2007; Luyten et al., 1995; Reed et al., 1987; Siderius et al., 2000).
The findings (i) that HOG1 is essential for survival under hyperosmotic conditions and (ii) that Hog1P induces expression of genes encoding glycerol-producting enzymes may suggest that this gene expression response itself is essential under hyperosmotic conditions (Posas et al., 2000). This idea was recently proven incorrect when two independent studies showed that translocation of Hog1P to the nucleus, induction of Hog1P dependent transcription (Westfall et al., 2008) and de novo protein synthesis (Mettetal et al., 2008) are not essential for survival of a hyperosmotic shock. Instead, Westfall et al. suggested that Hog1P plays a more direct and essential role in the induction of the glycerol flux. Mettetal et al. hypothesized, based on a mathematical model, that the observed gene-expression response is important during long-term adaptation to a strong osmotic shock, while metabolic adaptation is sufficient for the short-term response to a weaker osmotic shock. Clearly, the emphasis in previous studies on gene expression alone has distracted from other possible mechanisms of adaptation. The studies of Mettetal and Westfall show that there is an alternative adaptation mechanism, most probably via metabolic regulation. The mere existence of another mechanism, however, does not prove that this mechanism plays an important role if transcriptional regulation is properly in place. Especially for the long-term response, it might be just a secondary rescue mechanism, which is activated only when the gene-expression response is blocked. The studies of Mettetal and Westfall have been carried out in mutants or in the presence of inhibitors of protein synthesis. Here we raise the question as to how important the contributions by gene expression and metabolism actually are for the adaptation of wild-type yeast to a hyperosmotic shock under physiological conditions, and whether these contributions may be different for short- and long-term adaptation.
In this study we quantify the relative contributions of metabolic changes and gene-expression related changes in the osmo-response as a function of time, using time-dependent regulation analysis (Bruggeman et al., 2006; Rossell et al., 2005; ter Kuile and Westerhoff, 2001). In regulation analysis total flux regulation is dissected in a hierarchical regulation coefficient ρh and a metabolic regulation coefficient ρm. The hierarchical regulation coefficient ρh quantifies the extent to which changes in gene expression, resulting in alterations of Vmax, contribute to the change of flux. The metabolic regulation coefficient ρm quantifies the relative importance of metabolic interactions of the enzyme with its environment, e.g. through changes in concentrations of substrates, products and/or allosteric coefficients. In its original form the theory had only been derived for steady-state conditions. Subsequently, it has been extended to metabolic pathways outside steady state (Bruggeman et al., 2006). A recent application has proven the relevance of this extension, since the regulation coefficients indeed varied over time during starvation stress (van Eunen et al., 2009). The time-dependent form of the hierarchical regulation coefficient is defined as:
in which t is time and v(t) is the time-dependent flux through the enzyme. It has been derived (Bruggeman et al., 2006) that the two time-dependent regulation coefficients sum up to 1:
The derivation is based on the assumption that the two types of regulation are independent (Rossell et al., 2005). The hierarchical regulation coefficient ρh(t) can be readily measured by monitoring the flux v (in this study the rate of glyerol production) and the Vmax in time [equation (1)]. The metabolic regulation coefficient ρm(t) is then derived from the summation theorem [equation (2)].
In the study presented here, S. cerevisiae was cultured in well-defined, fully aerobic, glucose-limited chemostats. This allowed an accurate and reproducible analysis of fluxes and gene-expression levels (Piper et al., 2002). To test Mettetal's proposal that gene expression may become important for the long-term adaptation to osmotic shock, we followed regulation in time upon a hyperosmotic shock, according to the principles of time-dependent regulation analysis. We will show that metabolic regulation indeed precedes gene-expression regulation. Yet, even when gene-expression regulation sets in, metabolic regulation remains dominant.
Materials and methods
Strain and culturing conditions
Saccharomyces cerevisiae strain CEN-PK 113–7D (MATa MAL2–8cSUC2) was grown in an aerobic, glucose-limited chemostat at a dilution rate of 0.1/h and a culture volume of 1 l, in defined mineral medium containing glucose as the sole source of carbon and free energy (Verduyn et al., 1992). The pH was kept at 5.0 by titration with 2 M KOH (Van Hoek et al., 1998) and the temperature was kept at 30 °C. The feed medium contained 42 mM of glucose (7.5 g/l). Chemostats were stirred at a rate of 800 rpm and aerated at 0.5 l/min. Most equipment was acquired from Applikon (Schiedam, The Netherlands). The osmotic shock was applied by first removing 225 ml of culture from the chemostat and replacing it with 225 ml of medium with 4.4 M of sorbitol and without glucose. At the same time the feed medium was replaced by an otherwise identical feed medium containing 1 M of sorbitol. Thus the sorbitol concentration remained 1 M throughout the rest of the experiment.
The glycerol production was monitored in time before and after the osmotic shock. Samples for total glycerol (medium plus cells) were taken from the fermentor and glycerol was extracted by addition of perchloric acid (Rossell et al., 2005). The glycerol concentration was determined through an NADH-linked enzymatic assay as described by Wieland (1984). In brief, the assay was carried out in glycine/hydrazine buffer (240 mM/1.2 M; pH 9.8) in the presence of saturating amounts of NAD (2 mM), ATP (2 mM), MgSO4 (1 mM), glycerol-3-phosphate dehydrogenase (8 U/ml) and glycerol kinase (4 U/ml). The fluorescence was measured in a Novostar plate reader (470 nm). The specific flux was calculated from the measured glycerol concentrations by taking into account the dilution rate and the biomass density in the culture at each time point.
Cells were washed three times with a phosphate-free buffer (20 mM Tricine and 5 mM MgCl2 at pH 7.0). Cell-free extracts were prepared by bead-beating the samples (eight times for 10 s) with glass beads (425–600 µm) in a Fastprep machine (MP Biomedicals). In between the bursts the samples were cooled on ice water for 60 s. Salts and small metabolites were removed from the cell-free extract by size exclusion with Zeba columns (Pierce). The total protein content of the cell-free extract was measured by the BCA method (Pierce). The absorbance (570 nm) was measured in a Novostar plate reader (BMG Labtech, Germany). All enzyme-activity assay were performed twice.
Glycerol-3-phosphate dehydrogenase activity was measured in freshly prepared extracts through an NADH-linked assay, according to Blomberg and Adler with some slight adjustments (Blomberg and Adler, 1989). The assay was performed in 20 mM imidazole (HCl), 1 mM dithiothreitol, 0.6 mM MgCl2 and 0.1 mM NADH and at a pH of 6.5. Dihydroxyacetone phosphate was added as a start reagent at a concentration of 0.8 mM. The absorbance was measured in a Novostar plate reader (340 nm).
Glycerol-3-phosphatase activity was measured in 5- and 10-times diluted, freshly prepared extracts through a phosphate-release assay. In all steps of the analysis only phosphate-free plastics were used. The assay was performed in 20 mM Tricine–MgCl2 (pH 7.0) and the reaction was started with glycerol-3-phosphate (10 mM) (Sussman and Avron, 1981). Every 10 min a sample was taken and the reaction was immediately stopped with 50% HClO4 (w/v; final concentration 5% w/v). The inorganic phosphate released in each sample was measured with a mixture of molybdate and ascorbic acid (Ames, 1966). The samples were incubated for 60 min at 37 °C and absorbance was measured at 800 nm.
Electrophoresis and immunoblotting
Cells were washed twice in ice-cold water. Cells were broken by bead-beating the samples (three times for 20 s) with glass beads (425–600 µm) in a Fastprep (MP Biomedicals) in the presence of protease inhibitor (Sigma). In between the bursts the samples were cooled on ice water for 60 s. Tris–HCl (pH 7.0) was added (final concentration 5 mM) and the samples were centrifuged. The supernatant was collected and the total protein content was measured by the BCA method (Pierce). The absorbance (570 nm) was measured in a Novostar plate reader. SDS–PAGE loading buffer [66 mM Tris–HCl (pH 6.8), 3% (w/v) SDS, 5% (v/v) glycerol, 2% (v/v) β-mercaptoethanol, and 0.001% (w/v) bromophenol blue] was added. Samples containing 10 µg protein were run on 10% SDS–PAGE gels in Tris–glycine electrophoresis buffer. To prevent large differences in signal on the western blots (and therefore problems with linearity of the signal), half the amount of total protein was loaded for the 60 min sample. Proteins were transferred to PVDF membranes at 100 V for 1 h and detected with secondary antibodies linked to horseradish peroxidase and enhanced chemiluminescense. Quantification was performed on a Gel-Doc MultiImager (BIO-RAD, Tokyo, Japan).
Polyclonal antibody against Gpd1p was kindly provided by Professor Adler. This antibody did not show any signal on western blot in the gpd1Δ mutant (Valadi et al., 2004). As control, an antibody against Hsp90 was used (Cell signaling: catalogue no. 4874).
Sampling for RNA quantification was performed as described previously (Piper et al., 2002). RNA was extracted from the samples by the hot phenol method (Schmitt et al., 1990). Genomic DNA was removed by a DNAse I treatment (1 U, Ambion). The amount of RNA was measured with a Novostar plate reader (260 nm; BMG Labtech) and 1 µg of RNA was used in the cDNA reactions.
Oligonucleotide primers were designed to amplify an 80–120 bp amplicon. PDI1 (protein disulfide isomerase) was chosen as an internal standard. HTA1 (histone H2A) and HOG1 were used as additional negative controls, because we expected no increase in histones and no regulation of RNA of HOG1. STL1 (sugar transporter-like protein) was used as positive control, as it has been shown to increase upon an osmotic shock. Primers were designed with Primer Express software 1.0 (PE Applied Biosystems, Foster City, CA, USA). PCR reactions (10 µl) were set up and run as described by the manufacturer. In brief, the reactions contained 5 µl SYBR Green PCR Core Kit (Bioke), 3 pmol of each primer (Isogen or Biolegio) and 0.1 µl of cDNA template (equivalent to 1 ng RNA). Amplification, data acquisition, and data analysis were carried out in the 7900HT Fast Real-time PCR System (once at 2 min, 50 °C; 10 min, 95 °C; and 40 cycles at 95 °C, 15 s; 60 °C, 1 min; Applied Biosystems). The calculated cycle threshold values (Ct) were exported to Microsoft Excel for analysis via the ΔΔCt method. First the mRNA levels were normalized to that of protein disulfide isomerase (PDI1) in the same sample and then to the PDI1-normalized value in the steady-state sample according to:
in which ss and t refer to the steady state and the time after the addition of sorbitol, respectively.
Dissociation curve analysis (Dissociation Curves 1.0 f. software, PE Applied Biosystems, Foster City, CA, USA) of PCR products was performed to verify amplification of the correct product.
Time-dependent hierarchical regulation coefficients ρh(t) were calculated according to equation (1), in which v was taken to be the in vivo glycerol flux, time point zero was taken to be the steady state of the chemostat, prior to the addition of sorbitol, and t refers to the time after the addition of sorbitol. ρh(t) was calculated separately from three independent chemostat cultures and then averages and SEM were determined; ρm(t) was calculated from the summation theorem [equation (2)].
Glycerol production increases in chemostat cultures upon hyperosmotic shock
We first analysed the change of the glycerol flux in a glucose-limited chemostat after the addition of 1 M of sorbitol. To this end we measured the total glycerol concentration (intracellular and extracellular) in the cultures as a function of time. At steady state the glycerol concentration was very low (53 ± 6 µmol per liter culture volume; Figure 1A). After addition of sorbitol, the glycerol concentration increased almost exponentially up to 1.4 ± 0.2 mmol/l culture volume after 1 h. From these total glycerol concentrations and from the biomass density in the cultures the specific glycerol production flux was calculated (Figure 1B). The flux increased almost linearly from 0.87 µmol/min/g DW at time point 0 min to 32 µmol/min/g DW after 1 h.
Regulation of mRNA concentrations
The increase in the glycerol flux can be caused by an increased expression of the genes encoding the glycerol-producing enzymes or by the interaction of these enzymes with their metabolic environment. Gene expression is known to be involved in the adaptation of batch cultures to osmotic shock. We analysed the changes in the mRNAs encoding glycerol-3-phosphate dehydrogenase and glycerol-3-phosphatase in our glucose-limited chemostat cultures (Figure 2). Also in chemostats GPD1, GPP1 and GPP2 transcript levels were induced by the osmotic shock. GPD2 mRNA did not respond to the addition of sorbitol. The induced transcript levels reached a maximum at 20–30 min after the shock. STL1, which is known to be specifically regulated by Hog1P, was also induced (Posas et al., 2000). Together with the analysis of the glycerol flux, these results indicate that the chemostat cultures respond similarly to osmotic shock as batch cultures.
The induction of the osmosensitive transcripts was in the same range as the relative increase of the glycerol flux (approximately 30-fold). However, changes in mRNA levels are not always reflected by protein concentrations and/or enzyme activities and therefore we measured the latter as well (below).
Enzyme activities and concentrations
Subsequently, we analysed the changes in the maximal activities (Vmax) of the enzymes upon hyperosmotic shock. The activity of glycerol-3-phosphate dehydrogenase increased, after an initial decrease from 0.12 µmol/min/mg protein in steady state to 0.31 µmol/min/mg protein after an hour (Figure 3A). The initial decrease, if significant, may be caused by a denaturation of the enzyme. The activity of glycerol-3-phosphatase increased from 0.18 to 0.87 µmol/min/mg protein after 1 h. The absolute activity of glycerol-3-phosphate dehydrogenase and its increase was similar to that seen before in batch cultures (Blomberg and Adler, 1989).
As a control we also studied the protein concentrations of glycerol-3-phosphate dehydrogenase. A western blot of one of the two isoenzymes (Gpd1p) showed a somewhat higher increase in the protein concentration (6-fold at 60 min, Figure 3B) than was seen in the enzyme activity assay. Since activity (Vmax) measurements are more accurate than western blots and moreover since only GPD1 was quantified by western blots (whereas both isoforms will be responsible for the initial production of glycerol), further calculations were based on the enzyme activities [cf. equation (1)]. By doing this we treat possible posttranslational modifications as a part of the hierarchical regulation and confine the term metabolic to the effects of substrates, products and allosteric effectors.
Metabolic regulation precedes and dominates hierarchical regulation
To quantify to what extent the increase of the glycerol flux was due to the measured increase in the activities (Vmax) of the glycerol-producing enzymes and to what extent to metabolic interactions, time-dependent regulation analysis was applied to the flux and Vmax data. Hierarchical and metabolic regulation coefficients are depicted in Figure 4. For both enzymes ρm (the metabolic regulation coefficient) started close to 1 and ρh (the hierarchical regulation coefficient) close to zero. This indicates that the initial regulation of the glycerol flux was purely metabolic. After a few minutes the hierarchical regulation coefficient of both enzymes started to increase until it reached a value of approximately 0.2 between 20 and 60 min after the perturbation, while the metabolic regulation coefficient accordingly decreased to 0.8. This implies that even after 1 h, 80% of the increase in the glycerol flux is explained by metabolic changes in the cell, and 20% by induction of gene expression.
We note that the time between the last two time points is quite long (from 30 to 60 min). Since the glycerol flux as well as the Vmax values are still changing over this time period, we cannot exclude that a transient effect of gene-expression regulation is still hidden in this time course. However, if so, the effect is not sustained, and the overall regulation between 0 and 60 min [which we plot, cf. equation (1)] is predominantly metabolic.
In this study we have shown that the upregulation of the glycerol flux upon an osmotic shock of the yeast S. cerevisiae is dominated by metabolic regulation. The initial flux increase is regulated purely at the metabolic level, while the late response is jointly regulated by metabolism (80%) and gene expression (20%). This small contribution by gene expression is striking in the light of the literature, which is dominated by research on the Hog1P-dependent gene-expression response. The prevailing view is that protein synthesis and nuclear entry of Hog1P dominate the response of S. cerevisiae to osmotic stress (Klipp et al., 2005). In a recent study in which it was shown that gene-expression regulation is not essential for the adaptation to hyperosmotic stress per se (Mettetal et al., 2008), it was suggested that gene expression might play a role in the long-term adaptation (∼1 h) to high osmotic stress (0.5 M NaCl). In general it is thought that gene expression is more important for long-term responses, while metabolic regulation is a short-term regulation mechanism (Fell, 1996). In our study, the contribution of gene-expression indeed increased in time, but it remained small. Here we used 1 M of sorbitol, i.e. the osmotic equivalent of 0.5 M NaCl, and followed the cultures for 1 h. Yet, only a minor contribution by gene-expression (20%) was found in the late time points. Between 30 and 60 min the expression of Hog1-responsive genes has already stabilized or tends to decrease, suggesting that longer exposure will not reveal a more important role for Hog1-induced gene expression. Also the hierarchical regulation coefficient stabilizes within an hour. The Mettetal study differs from ours in that they subjected the cultures to an oscillating osmotic pressure, while we maintained a constant high pressure. Possibly the adapted gene-expression profile renders the cells more robust against fluctuating osmotic pressure or the used strain (Hog1p-YFP mutant of BY4741) may have a different strategy of coping with osmotic stress.
Although we measured the protein concentration of Gpd1p as well as the total activity (Vmax) of glycerol-3-phosphate dehydrogenase, we calculated the regulation coefficients solely on the basis of the Vmax. Had we calculated the hierarchical regulation coefficient for glycerol-3-phosphate dehydrogenase based on the westerns for Gpd1p, the hierarchical regulation coefficient would have increased to approximately 0.5 at 60 min. The difference between the Vmax and the western data would then be attributed to posttranslational modification. For this explanation to be valid we would need to assume that Gpd1p carries all the flux and Gpd2p plays no role even before the osmotic shock. However, at the mRNA level we observe a basal expression of both GPD1 and GPD2, suggesting that they share the initial flux, prior to the shock. Since the Vmax includes contributions by both isoenzymes, the fact that only one of the two responds to the shock explains at least partially why the induction of the Vmax is less than that of Gpd1p. Based on the available data, we cannot exclude a small contribution by regulation of posttranslational modification. Because of the relatively low resolution of current proteomics techniques, this would be difficult to quantify now (Daran-Lapujade et al., 2007), but rapid developments in proteomics promise sufficient resolution in the near future.
Most studies of the osmotic stress response were done in batch cultures at glucose excess (Blomberg and Adler, 1989). We wondered whether the distribution of regulation of the glycerol flux would be different in chemostat and batch cultures. Only few studies actually measure glycerol production flux. Even the comprehensive, quantitative study by Klipp et al. (2005) did not report the glycerol flux, but only the intracellular and extracellular glycerol concentrations relative to those at a certain time point. Obviously, the concentrations are functionally important for the protection against an added osmolyt. Yet, to understand the relative importance of gene expression and metabolism, information about the fluxes is required [cf. equation (1)]. Blomberg and Adler (1989) showed that at 28 mM glucose the basal glycerol-production rate is already high and the glycerol flux is only upregulated by a factor of 3 (Blomberg and Adler, 1989), while we showed a factor of over 30 in glucose-limited chemostats in which the residual glucose concentration is typically below 1 mM (Postma et al., 1989). The high basal glycerol flux in batch cultures may be attributed to glucose repression of the respiratory chain, which will cause a partial routing of biosynthetic NADH via the glycerol pathway (Bakker et al., 2001). Blomberg and Adler also plotted the logarithm of the flux against the logarithm of the Vmax of glycerol-3-phosphate dehydrogenase in glucose-excess batch cultures. From this plot a hierarchical regulation coefficient ρh of 1.7 can be calculated (the inverse of the slope), implying a metabolic regulation coefficient ρm of − 0.7. Apparently, the gene-expression regulation is sufficient to sustain the small upregulation of the glycerol flux under glucose-excess conditions. Under these conditions even a counteracting metabolic response is observed as reflected by the negative metabolic regulation coefficient. As yeasts in nature and in industrial applications often encounter nutrient-poor conditions or alternating nutrient concentrations, it becomes very relevant to study the osmotic response under different nutrient conditions.
The question now arises how the strong metabolic regulation of the glycerol flux is brought about by the cell. It is quite likely that Hog1p plays a role in the metabolic regulation of the glycerol flux, as in the hog1Δ strain the upregulation of glycerol production is strongly reduced as compared with in the wild-type, although not completely abolished (Albertyn et al., 1994). The essential role of Hog1p may reside in its kinase activity (Reiser et al., 1999). Which (cytosolic) proteins are Hog1p targets is largely unknown (but see Reiser et al., 1999; Westfall et al., 2008). A direct phosphorylation of the Gpd or Gpp proteins by Hog1-P cannot explain our results as we have measured the catalytic activity (Vmax) of these enzymes in the presence of phosphatase inhibitors and and even then the upregulation could not explain the increase in the glycerol flux. It is therefore more likely that phosphorylation of other target proteins leads to alterations in metabolite concentrations which directly affect the activity of Gpd and Gpp. Such a mechanism would explain why the upregulation of the flux is slower than one should expect from direct metabolic regulation (cf. Figure 1). In addition, the enzyme Gpd has recently been shown to be redistributed upon osmotic stress (Jung et al., 2010). This redistribution of the enzyme Gdp may be another regulatory mechanism that influences the glycerol flux. In our analysis, regulation due to redistribution would be observed as metabolic regulation, since it does not affect the extractable enzyme activity.
The initial response which is completely determined by metabolic regulation must be the result of very rapid changes in metabolite concentrations. Ongoing work in our group aims to unravel the metabolic regulation of the glycerol flux. Considering that the final 30-fold induction of the flux is 80% caused by alterations in metabolite concentrations, either very large concentration changes or very sensitive enzymes are required. A further complication arises from the fact that the very first changes may simply result from the rapid shrinking of the cells. The change of the cell volume will cause shifts in binding and reaction equilibria. For instance, a large part of the NADH in the cell is bound to proteins. Cell shrinking will shift the equilibrium further towards binding, initially leading to a lower amount of free NADH per cell, and yet a higher concentration. NADH is one of the substrates of Gpd. An increase in NADH level therefore will stimulate the production of glycerol. A direct measurement of the free NADH/NAD ratio in the yeast cytosol is now feasible (Canelas et al., 2008). It has also been reported that mitochondrial respiration is compromised by osmotic shock, probably via a change of the mitochondrial membrane structure (Mathai et al., 1993). A decrease in the respiratory capacity should lead to an increased NADH concentration, which in turn should stimulate the glycerol production. This is qualitatively consistent with our observation that the oxygen consumption flux decreases upon addition of sorbitol (not shown). We cannot distinguish, however, whether the increase of the glycerol flux is a result of the decreased oxygen flux or vice versa, since the redox balance must be maintained under all conditions. Alternatively, rerouting of the carbon-flux by altering GAPDH and TPI levels has been shown to occur as a result of oxidative stress (Ralser et al., 2007). A reduction in the glyceraldehyde 3-phosphate-dehydrogenase and triosephosphate-isomerase activity may result in higher dihydroxyacetone phosphate levels and in this way lead to activation of the glycerol flux during osmotic stress. The kinetic model of glycerol synthesis by Cronwright et al. (2002) predicts that the glycerol flux is most sensitive to changes in the concentrations of dihydroxyacetone phosphate, ATP and ADP. In the light of the large metabolic regulation to be explained, it should, however, not be excluded that unknown metabolic regulators exist.
Finally, we measured a much larger increase in the mRNA levels than in the enzyme activities. This discrepancy is often seen. One of the explanations may be that the translation rate does not depend linearly on the level of a certain transcript. However, a specific regulation of protein synthesis or protein degradation may also be involved. Protein degradation and synthesis have been shown to be regulated by osmotic stress (Mao et al., 2008).
In conclusion, our study urges a redirection of systems biology research on osmotic stress response towards (i) the nature of the metabolic regulation and (ii) the interplay of gene expression and metabolism as a function of the degree of osmostress, the nutrient supply and/or the combination with other types of stress.
We would like to give our thanks to the sampling team, and to MSc students, for pilot experiments. The work was supported by IOP Genomics.