Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
The accumulation of compatible solutes, such as glycerol, in the yeast Saccharomyces cerevisiae, is a ubiquitous mechanism in cellular osmoregulation. Here, we demonstrate that yeast cells control glycerol accumulation in part via a regulated, Fps1p-mediated export of glycerol. Fps1p is a member of the MIP family of channel proteins most closely related to the bacterial glycerol facilitators. The protein is localized in the plasma membrane. The physiological role of Fps1p appears to be glycerol export rather than uptake. Fps1Δ mutants are sensitive to hypo-osmotic shock, demonstrating that osmolyte export is required for recovery from a sudden drop in external osmolarity. In wild-type cells, the glycerol transport rate is decreased by hyperosmotic shock and increased by hypo-osmotic shock on a subminute time scale. This regulation seems to be independent of the known yeast osmosensing HOG and PKC signalling pathways. Mutants lacking the unique hydrophilic N-terminal domain of Fps1p, or certain parts thereof, fail to reduce the glycerol transport rate after a hyperosmotic shock. Yeast cells carrying these constructs constitutively release glycerol and show a dominant hyperosmosensitivity, but compensate for glycerol loss after prolonged incubation by glycerol overproduction. Fps1p may be an example of a more widespread class of regulators of osmoadaptation, which control the cellular content and release of compatible solutes.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
It is likely that all types of cells, even those of multicellular organisms, have developed mechanisms to adapt to changes of the external osmolarity and hence are able to respond to altered water availability. Osmotic adaptation not only plays a role in the response to osmotic stress but is also central to many developmental and morphogenic processes (Dale, 1986; Harold, 1990; Howard and Valent, 1996). A ubiquitous strategy in osmoregulation probably used by all cells is the accumulation of one or several compatible solutes to control the osmolarity of the cytosol (Yancey et al., 1982). The solutes used are species specific and range from ions to amino acids and their derivatives, as well as sugars and sugar alcohols. The unicellular eukaryotic model organism Saccharomyces cerevisiae (baker's yeast) uses glycerol (Brown, 1976).
Although very little is known about mechanisms controlling gene expression under anaerobic conditions, several signalling pathways have been reported to be stimulated by altered medium osmolarity. Osmotic induction of GPD1 and GPP2 expression is controlled partly by the high osmolarity glycerol (HOG) response pathway (Brewster et al., 1993; Albertyn et al., 1994; Hirayama et al., 1995; Norbeck et al., 1996; Rep et al., 1999). Another MAP kinase pathway (Treisman, 1996), the protein kinase C (PKC) pathway, can be stimulated by hypo-osmotic shock (Davenport et al., 1995), but does not appear to be involved in the control of GPD1 expression (Rep et al., 1999). Recently, it has been shown that a hypo-osmotic shock leads to a rapid, transient increase in the intracellular calcium level, but the pathway(s) controlling this effect is unknown (Batiza et al., 1996).
Although the regulation of glycerol production can account, at least partly, for the accumulation of the compatible solute after a hyperosmotic shock, additional levels of control are needed when the external osmolarity decreases and the cell has to dispose of osmolytes to avoid excessive turgor pressure. For example, we have shown previously that yeast cells rapidly release accumulated glycerol after a hypo-osmotic shock (Luyten et al., 1995). Rapid export mechanisms of compatible solutes have also been described for bacteria and for mammalian cells (Koo et al., 1991; Strange et al., 1994; Kwon and Handler, 1995; Ruffert et al., 1997; Kirk and Strange, 1998), and appear to be involved in the process of regulated volume decrease after the initial cell swelling caused by a hypo-osmotic shock. Although proteins such as MscL in Escherichia coli and ClC-2 in mammalian cells are candidate solute exporters, the molecular nature of the efflux pathways is unknown (Kirk and Strange, 1998). The rapid glycerol release by yeast after a hypo-osmotic shock has been shown to be dependent on Fps1p (Luyten et al., 1995).
The FPS1 gene was isolated as a suppressor of the growth defect of a mutation affecting the control of yeast glycolysis (Van Aelst et al., 1991), and it is most closely related to bacterial glycerol facilitators such as GlpF from E. coli (Heller et al., 1980; Muramatsu and Mizuno, 1989). We have shown that Fps1p mediates both uptake and export of glycerol from yeast cells (Luyten et al., 1995; Sutherland et al., 1997). In this work, we demonstrate that the physiological role of Fps1p is to regulate export rather than uptake of glycerol by describing growth phenotypes associated to the lack of proper glycerol export in fps1Δ mutants. Mutants lacking Fps1p are sensitive to hypo-osmotic shock, a phenotype which demonstrates that rapid solute release is an essential step in the recovery from a sudden drop in medium osmolarity. The same mutants also grow poorly under anaerobic conditions when glycerol production is stimulated for redox regulation.
Relatively little is known about the molecular mechanisms of MIP channel regulation. Kidney AQP2, which is required for the concentration of primary urine, appears to be controlled by vasopressin and protein kinase A at the level of membrane localization (Knepper and Inoue, 1997; Mulders et al., 1997a). Plant aquaporins are regulated at the level of gene expression by drought stress and developmental cues (Yamada et al., 1995; Kaldenhoff et al., 1996). In addition, evidence has been presented suggesting that the transport activity of at least some aquaporins might be directly regulated via phosphorylation, as shown in studies involving heterologous expression in Xenopus oocytes (Maurel et al., 1995) and recently also in native tissue (Johansson et al., 1998).
In this work, we demonstrate that Fps1p-mediated glycerol export in S. cerevisiae is rapidly regulated by external osmolarity. We also describe a domain N-terminal to the first transmembrane domain, which is required for closing of the channel. Deletion of this domain renders transport function constitutive, resulting in loss of glycerol from the cell and sensitivity to high osmolarity.
Fps1p is a plasma membrane protein
The function of Fps1p in glycerol uptake and efflux and the presence of six putative TMDs suggested that Fps1p is localized in the yeast plasma membrane. To detect Fps1p, we have tagged it at the C-terminus with a c-myc epitope and transformed the construct on a multicopy plasmid into yeast cells. The tagged Fps1p is functional because it restores glycerol transport characteristics (data not shown). Fractionation of a yeast membrane preparation and subsequent Western blot analysis of specific sucrose gradient fractions showed that Fps1p cofractionated with a marker protein, the plasma membrane H+-ATPase Pma1p (Serrano et al., 1986; Serrano, 1989; Fig. 1A). In addition, immunofluorescence studies on yeast cells transformed with a plasmid carrying the FPS1 gene with the C-terminal tag resulted in a signal predominantly at the cell surface (Fig. 1B). Interestingly, and in contrast to Pma1p, the Fps1p signal was not evenly distributed over the cell surface but rather appeared in patches. We also note that a portion of Fps1p appeared to be associated with intracellular structures, possibly the endoplasmic reticulum.
fps1Δmutants show phenotypes related to glycerol export
To study the physiological importance of Fps1p, we have tested the effects of deletion of FPS1 under growth conditions in which glycerol uptake or efflux could be expected to be important for growth or survival. Yeast can use glycerol as a source for carbon and energy (Gancedo et al., 1968). Mutants lacking Fps1p did not show a growth defect on medium containing glycerol as a sole carbon source, suggesting that other transport proteins are involved in the uptake of glycerol for its utilization (Lages and Lucas, 1997; Sutherland et al., 1997; Fig. 2A).
We then tested for growth phenotypes under conditions in which glycerol export is required. During anaerobic growth, glycerol production is essential for redox-regulation by reoxidizing excess NADH produced during biomass production (Ansell et al., 1997). Under these conditions, fps1Δ mutants accumulated large amounts of glycerol inside the cell (Fig. 2B), grew much slower than the wild type (Fig. 2C) and took much longer to consume the sugar substrate (Fig. 2D). This growth defect is probably because of osmotic problems caused by the high intracellular glycerol levels.
During adaptation to hypo-osmotic shock, yeast cells reduce their intracellular glycerol content by 50–75% within 3 min of the shock, whereas mutants lacking Fps1p require more than 1 h to achieve the same glycerol loss (Luyten et al., 1995). fps1Δ mutants survived a hypo-osmotic shock at a 50- to 100-fold lower proportion than wild type (Fig. 2E). In addition, as apparent from the smaller colonies formed by surviving mutant cells, fps1Δ mutants also resumed growth much more slowly (Fig. 2E). These defects were dramatically aggravated by additional deletion of the SLT2/MPK1 gene (Fig. 2E). Slt2/Mpk1p is the terminal protein kinase of the PKC-signalling cascade and loss of Slt2/Mpk1p causes defects in cell wall assembly, making cells more vulnerable to hypo-osmotic conditions (Martin et al., 1993). We conclude that fps1Δ is a hypo-osmosensitive mutant. However, fps1Δ mutants did not require an osmotic stabilizer for growth on low osmolarity medium, suggesting that Fps1p is specifically required for the adaptation after a sudden drop in osmolarity (Fig. 2F).
Fps1p-mediated glycerol transport has different kinetics for efflux and import
The physiological importance of Fps1p for glycerol export rather than uptake could indicate that the transport rate depends on the transport direction. To test this, we monitored the uptake of 100 mM radioactive glycerol as well as the efflux of glycerol from cells loaded with 100 mM radioactive glycerol and then diluted 100-fold with buffer lacking glycerol. The rate of glycerol efflux, apparent as the slope of the curve in Fig. 3A, was clearly higher than that for uptake, especially when taking into consideration the first 60–80 s when both uptake and efflux seem to be linear (Fig. 3A and our unpublished results). We estimated the intracellular glycerol content after loading to be about 75 mM (data not shown), and hence the higher efflux rate cannot be due to a higher initial intracellular than extracellular glycerol concentration. These data suggest that the transport rate for glycerol indeed depends on the transport direction.
We then used loading of cells with radioactive glycerol and subsequent dilution in buffer without glycerol to monitor the difference between the wild-type and the fps1Δ mutant. As shown previously (Luyten et al., 1995; see also Fig. 4) for osmotically induced glycerol export, glycerol efflux driven by the concentration gradient alone also depended on the presence of FPS1 (Fig. 3B). The difference between wild type and mutant is most pronounced for the efflux of some 60% of the glycerol within the first 2 min after dilution, whereas the subsequent slow glycerol efflux is similar in wild type and mutant. Because the initial glycerol efflux is fully dependent on Fps1p and also the glycerol uptake within short time scales is Fps1p-dependent (Fig. 4; Sutherland et al., 1997), we conclude that the transport monitored in Fig. 3A is largely or completely due to Fps1p. Thus, it appears that Fps1p-mediated glycerol transport has different kinetics for uptake and efflux.
Control of the glycerol transport rate
The rate of glycerol uptake is strongly reduced during growth at high osmolarity, indicating that yeast cells possess mechanisms that control the transport rate (Luyten et al., 1995; Sutherland et al., 1997). To elucidate the kinetics of this regulation, we shifted cells to 5% NaCl, took samples at the intervals indicated and determined the glycerol uptake rate (Fig. 4). fps1Δ mutants showed a low glycerol uptake rate in the presence and in the absence of salt. The data obtained for the wild-type cells suggest a very rapid control mechanism for glycerol transport. Within 1 min of salt addition, the glycerol transport rate was reduced to levels similar to those observed for a fps1Δ mutant (Fig. 4A). A shift to higher sucrose rather than NaCl concentrations at the same water activity caused the same reduction in glycerol transport rate, demonstrating that the effect is osmotic and not salt dependent (data not shown). In contrast, when cells were transferred from high to low osmolarity, glycerol transport was rapidly restored, as documented by the release of 75% of the accumulated glycerol within less than 3 min (Fig. 4B). These rapid changes in the glycerol transport rate suggest the existence of a transport protein whose opening and closing is rapidly regulated by osmolarity. Because fps1Δ mutants do not show these rapid changes in transport, Fps1p is a likely candidate for such a protein.
The rapid control of glycerol transport is not mediated by known osmosensing signalling pathways
Three signalling pathways in S. cerevisiae are known to respond rapidly to altered osmolarity. The HOG MAP kinase pathway is induced by an increase in external osmolarity (Brewster et al., 1993) and the PKC MAP kinase pathway is stimulated by a drop in osmolarity (Davenport et al., 1995). In addition, hypo-osmotic shock stimulates a transient increase in intracellular calcium by an unknown pathway (Batiza et al., 1996). Using mutants lacking the terminal protein kinases of the HOG- and the PKC-pathway, Hog1p and Mpk1p respectively, we have investigated the possible involvement of these two pathways in the regulation of glycerol transport (Fig. 5). The hog1Δ mutant was able to reduce the glycerol transport capacity with a similar time course and to a similar extent, five- to 10-fold, as the wild type (Fig. 5A). These data indicate that the HOG pathway is not involved in the rapid regulation of glycerol transport. We note, however, that in a hog1Δ mutant the glycerol uptake rate was about threefold higher than in the wild type under all growth conditions. Because gpd1Δ mutants, which like the hog1Δ mutant also produce much less glycerol, showed the same uptake characteristics as the wild type (data not shown), the reason for the behaviour of the hog1Δ mutant is probably not related to the cellular glycerol content, but might be a consequence of altered osmoregulation in a hog1Δ mutant even under normal growth conditions.
The rapid release of accumulated glycerol by a mpk1Δ mutant was indistinguishable from that of a wild-type strain, indicating that the PKC pathway is not involved in the control of the rate of glycerol transport. Because the growth medium contained 0.5 M sorbitol to prevent lysis of the mpk1Δ mutant, less glycerol was lost after a downshock from 5% NaCl to a medium with a low NaCl content (Fig. 5B) compared with the experiment described in Fig. 4B. Consistent results were obtained with bck1Δ and pkc1Δ mutants, which are defective in protein kinases upstream of Mpk1p in the PKC pathway (Levin et al., 1990; Lee and Levin, 1992; data not shown).
Gadolinium ions inhibit stretch-sensitive channels (Yang and Sachs, 1989). They have been shown to block both the transient stimulation of the intracellular calcium content in yeast after a hypo-osmotic shock (Batiza et al., 1996) and the release of solutes in bacteria in some (Berrier et al., 1992; Schleyer et al., 1993), but not all, systems (Ruffert et al., 1997). Gadolinium had no effect on glycerol accumulation or release even at concentrations five times higher than those needed to inhibit the calcium peak (Fig. 5C). Furthermore, inhibition of calmodulin by calmidazolium (200 μg ml−1) or addition of the calcium ionophore A23187 (0.1 mM) in the presence of 1 mM external calcium did not have any effects on the maintenance of a high concentration of glycerol before a hypo-osmotic shock, or on the release of glycerol upon hypo-osmotic stress (data not shown).
Phospholipase C is believed to operate upstream of both PKC and calcium signalling. This enzyme produces inositol trisphosphate, a well-known second messenger (Strange, 1994). Upon hypo-osmotic shock, the alga Dunaliella salina increases the rate of polyphosphoinositide hydrolysis (Einspahr et al., 1988). However, in our system, the presence of products or inhibitors of phospholipase C such as 0.75 mM diacylglycerol or 1 mg ml−1 compound 48/80 (Brandão et al., 1994) did not affect the ability to accumulate or release glycerol (data not shown). Similarly, activation of protein kinase A by addition of 2 mM dinitrophenol (Thevelein et al., 1987) had no effect on glycerol release or maintenance (data not shown). We also found that glycerol release took place normally, even after depleting the cell of ATP by inhibiting glycolysis with iodoacetamide (Griffioen et al., 1994) or by the addition of 100 mM glucose to a tps1Δ strain (Hohmann et al., 1993). Thus, it appears that control of glycerol transport does not require ATP as an energy source. Furthermore, substances causing membrane stretching such as chlorpromazine (250 μM) and trinitrophenol (5 mM; Martinac et al., 1990) did not provoke glycerol release, although chlorpromazine seemed to enhance the glycerol efflux rate upon hypo-osmotic shock (data not shown).
In mouse zygotes, it has been shown that hypo-osmostress-induced regulatory volume decrease is sensitive to both K+ and Cl− channel blockers (Seguin and Baltz, 1997), and the volume-sensitive organic osmolyte channel may also mediate the release of Cl− (Perlman et al., 1986). Addition of the chloride channel inhibitor DIDS (2.5 mM), or the K+ channel blocker barium (100 mM) to yeast cells had no effect on the release of glycerol. In addition, the potassium ionophore valinomycin (50 μM) was unable to provoke or inhibit glycerol release. Glycerol release after a hypo-osmotic shock was also not affected in the presence of a high concentration of extracellular potassium (210 mM or 425 mM), suggesting that potassium fluxes are probably not directly involved in the control of Fps1p-mediated glycerol release (data not shown).
Another signal transduction pathway has been proposed recently for animal cells. This would comprise a PI3-kinase, a tyrosine kinase and a G-protein. This cascade would regulate chloride conductance and also provoke transient reorganizations of the cytoskeleton upon hypo-osmotic shock (Lo et al., 1995; Sadoshima et al., 1996; Tilly et al., 1996). However, in our system, inhibition of PI3-kinases (by 10 μg ml−1 wortmannin) or tyrosine kinases (by 200 μM genistein) did not affect glycerol release (data not shown). Likewise, drugs affecting the cytoskeleton (200 μM latrunculin A; Ayscough et al., 1997) or G-proteins (6 μg ml−1 cholera toxin and 100 μM AlF4−; Kahn, 1991; Schliess et al., 1995) did not have any effects in our test system. Unexpectedly, MIP-channel inhibitors such as 1 mM mercury, 1 mM phloretin or 1 mM Cu2+ did not have any effect on glycerol release either. It should be noted, however, that Fps1p does not possess the characteristic cysteine residues known to be involved in mercury-mediated inhibition of many MIP channels (Mulders et al., 1997b).
Expression of FPS1 is not controlled by osmolarity
The rapid kinetics of the regulation of the glycerol transport rate suggested that regulation of gene expression does not account for this effect. It may, however, be involved in a long-term adaptation of glycerol transport to altered osmolarity. We have analysed expression of the FPS1 gene after an increase in medium osmolarity to 5% NaCl. In contrast to expression of GPD1, the gene encoding the osmoregulated glycerol-3-phosphate dehydrogenase (Albertyn et al., 1994), expression of FPS1 was not altered by an increase in external osmolarity (Fig. 6). A slight decrease in the FPS1 mRNA level was, however, observed after prolonged exposure to high osmolarity. Expression of FPS1 was not affected by the carbon source (glucose versus glycerol, data not shown).
The glycerol facilitator from E. coli mediates unregulated glycerol transport in yeast
Apparently, yeast glycerol transport is controlled by an as yet uncharacterized osmoregulated mechanism. In order to investigate the control of Fps1p further, we first decided to examine whether a related heterologous protein could be controlled by osmolarity in yeast. Fps1p is most homologous to the glycerol facilitator GlpF from E. coli (30.5% sequence identity), but this homology is restricted to the core of the protein with the six putative TMDs. Fps1p (669 amino acids) is also much longer than GlpF (281 amino acids) and this size difference is mainly due to long N- and C-terminal hydrophilic extensions of Fps1p (Van Aelst et al., 1991). When yeast cells expressed glpF instead of FPS1, glycerol transport was restored and in fact was about 2.5-fold higher than in the wild type (Fig. 7). This difference might be due to higher expression of glpF or to different transport characteristics of the gene products of glpF and FPS1.
When cells producing GlpF instead of Fps1p were exposed to an increased medium osmolarity, the glycerol transport rate was transiently somewhat reduced, but after 10 min returned to approximately the same level as before the osmotic shock (Fig. 7). This is in clear contrast to the strong and sustained downregulation of glycerol transport observed in yeast cells possessing Fps1p. Thus, it appears that glpF expressed in yeast confers osmotically insensitive glycerol transport.
The unique N-terminal extension of Fps1p is required for regulated glycerol transport
Because the most significant difference between Fps1p and GlpF are the long N- and C-terminal extensions, we decided to study the effects of deletion of these extensions on the control of glycerol transport.
Using a PCR approach, the N- or the C-terminal extension of Fps1p was deleted and each construct was transformed into a fps1Δ strain (Fig. 8A). Deletion of the C-terminus apparently resulted in an entirely inactive channel because that construct did not alter glycerol accumulation after increased osmolarity and could not mediate glycerol release after a hypo-osmotic shock (data not shown). A possible explanation could be that the truncated protein is not properly localized to the plasma membrane. Deletion of the N-terminal extension (construct fps1-Δ1 ), however, caused an inability to accumulate significant glycerol levels within 3 h after the upshift (Fig. 8B). Only after 1 day of incubation in the presence of salt was a significant glycerol accumulation observed (see below).
The truncated construct was expressed (Fig. 8C) and at least partially localized at or in the plasma membrane (data not shown). Investigation of the glycerol transport rate showed that the defect in glycerol accumulation could partially be attributed to an inability to rapidly reduce transport at high osmolarity (Fig. 8D). The construct lacking the N-terminus of Fps1p behaved very much like E. coli GlpF and mediated a very high level unregulated glycerol transport. This suggests that the transport function of Fps1p became constitutive in this mutant.
To analyse which parts of the 250 amino acid N-terminal extension are required for Fps1p regulation, we have deleted different segments (Fig. 8A) of this region and performed the same type of experiments as described above (Fig. 8). Remarkably, large parts of the N-terminal extension did not seem to be required for regulation. Mutant fps1-Δ4, which lacks amino acids 13–144, appeared to be regulated normally (Fig. 8D). This deletion also encompasses a stretch of 18 asparagines within 20 amino acids (residues 50–69; Van Aelst et al., 1991), which therefore does not seem to play a role in regulation of glycerol retention. However, a segment close to the first transmembrane domain appears to contain sequences required for regulation of Fps1p, as shown by the failure of mutant fps1-Δ5 (lacking residues 151–230) to mediate regulation of glycerol transport (see below).
Different truncations of the N-terminus caused defects in glycerol accumulation to different extents (Fig. 8B). The intracellular level of glycerol attained after 3 h by the constitutively active mutants (fps1-Δ1, fps1-Δ3, fps1-Δ5 ) appeared to correlate inversely with the expression level of the corresponding construct (Fig. 8C), consistent with higher levels of a constitutively active Fps1p and resulting in more efficient glycerol export.
Glycerol uptake experiments after an osmotic shock with 5% NaCl confirm the inability of mutations 1, 3 and 5 (Fig. 8D and data not shown) to mediate a downregulation of the transport rate. The truncation mutants that retained the ability to accumulate glycerol (2 and 4) within the first 3 h after the addition of salt also mediated the downregulation of the glycerol uptake capacity with the same kinetics as wild-type Fps1p (Fig. 8D).
Remarkably, the strains carrying the truncated Fps1p that mediated constitutive glycerol export were able to accumulate almost as much glycerol as the wild type after extended incubation (Fig. 8B). When these cells were then transferred to medium with lower osmolarity, they were able to dispose of the accumulated glycerol exactly like the wild type (Fig. 8E). This observation demonstrates that the ability to accumulate glycerol was not due to loss of the plasmid or degradation of the export channel. Indeed, these cells leaked glycerol at a rate three- to fourfold higher than wild-type cells even after 24 h incubation in high salt, indicating that at this time point the transport rate was still high. Furthermore, the cells expressing fps1-Δ1 exhibited a glycerol production rate ≈five- to sixfold higher than that of the wild-type cells at this time point. This suggests that the cells compensate for glycerol loss by enhanced glycerol production.
Defect in the control of Fps1p-mediated glycerol export causes sensitivity to high osmolarity
Because cells expressing the FPS1 mutations 1, 3 and 5 did not accumulate glycerol as rapidly as wild-type or fps1Δ cells, we suspected those cells to be sensitive to high osmolarity. This was indeed the case (Fig. 9A). The degree of sensitivity also correlated well with the amount of glycerol accumulated after 3 h (see Fig. 8B). As expected for a gain-of-function mutation, the osmosensitive phenotype conferred by the truncated Fps1p was dominant (Fig. 9B).
Expression of E. coli glpF instead of FPS1 did restore glycerol permeability but not the downregulation after an osmotic shock (Fig. 7). Consistent with this observation, strains expressing glpF instead of Fps1p are as sensitive to high osmolarity as strains expressing the truncation construct fps1-Δ1 (Fig. 9C).
The central importance of Fps1p for yeast osmoregulation is demonstrated by the fact that different mutations in FPS1 can cause sensitivity to either increased or decreased external osmolarity. Deletion of the entire gene resulted in an inability to properly dispose of accumulated glycerol, and hence in sensitivity to a hypo-osmotic shock, which was manifested by a significantly lower proportion of surviving cells and also a slower growth resumption. Deletion of the N-terminal extension of Fps1p resulted in an inability to quickly reduce the glycerol transport rate, in a diminished cellular glycerol content during the early stages of adaptation, and consequently in slower growth at high osmolarity.
Adaptation to a hypo-osmotic shock
A hypo-osmotic shock results in water uptake by the cell and hence in cell swelling. Thus, in order to adapt to the new condition and to regain normal turgor, the cell has to diminish its solute content.
Regulated export of solutes in adaptation to hypo-osmotic shock has been described in bacteria and in mammalian cells. Upon moderate hypo-osmotic shock, bacteria specifically release K+ ions, trehalose, glutamate, proline and glycine betaine (Koo et al., 1991; Lamark et al., 1992; Schleyer et al., 1993; Glaasker et al., 1996; Ruffert et al., 1997). The gene for one such mechanosensitive channel protein, E. coli MscL, has recently been isolated, but its actual role in osmoregulation and in solute export is unclear because mutants lacking that protein do not show any phenotype and are still able to release osmolytes upon hypo-osmotic shock (Sukharev et al., 1997). Thus, the molecular nature of solute exporters in bacteria is not known. The situation is similar in the mammalian system. After the initial cell swelling upon hypo-osmotic shock, mammalian cells decrease their volume by controlled efflux of both organic and inorganic solutes (Kwon and Handler, 1995; Kirk and Strange, 1998; Lang et al., 1998). So far, the molecular nature of the proteins mediating solute export remains elusive (Kwon and Handler, 1995; Kirk and Strange, 1998), although the swelling-sensitive chloride channel ClC-2 seems to be a good candidate (Jordt and Jentsch, 1997).
Apparently, yeast cells also have mechanisms to dispose of osmolytes, specifically glycerol, and they can reduce their cellular glycerol content very rapidly (Luyten et al., 1995; this work). The process is complete within 2–3 min and results in a reduction of the cellular glycerol content by about 75%. We have found a protein, Fps1p, that is clearly required for the rapid glycerol release. We have also been able for the first time to demonstrate by genetic evidence that the release of osmolytes is essential for the adaptation to hypo-osmotic shock because yeast fps1Δ mutants are sensitive to a hypo-osmotic shock.
In bacteria, some evidence has been provided that active transport processes are involved at least in later stages of osmolyte release (Glaasker et al., 1996). Because depletion of the cellular ATP and also membrane depolarization with dinitrophenol did not affect glycerol accumulation or extrusion, export in yeast is unlikely to be an active transport process. S. cerevisiae has an active transport system for glycerol, but this is probably not involved in osmoregulation (Lages and Lucas, 1997; Sutherland et al., 1997). Also, none of the known MIP channel proteins has so far been reported to mediate active transport.
Our finding that a MIP channel is required for solute export in yeast opens up the possibility that at least some of the long-sought solute exporters in other organisms might also belong to this ancient protein family. A number of the mammalian and plant aquaporins are known to transport not only water but also osmolytes such as glycerol and urea (Froger et al., 1998). Although glycerol is not known to play any role in osmoregulation of mammals and plants, these findings could be indicative of a broader substrate specificity that is also relevant to osmolytes. Interestingly, E. coli mutants lacking the MIP channel AqpZ have recently been reported to be sensitive to hypo-osmotic conditions, an observation that is difficult to explain on the basis of water transport. Solute export has not been tested in that mutant, probably because AqpZ seems to be quite specific to water in the Xenopus oocyte heterologous test system (Calamita et al., 1995; 1998).
Adaptation to higher osmolarity
Through analysis of mutants defective in glycerol synthesis, we have previously demonstrated that the production of this compatible solute is an essential step in adaptation of yeast to higher osmolarity (Albertyn et al., 1994; Ansell et al., 1997). The production of glycerol appears to be regulated in part at the level of gene expression (Albertyn et al., 1994; Hirayama et al., 1995; Norbeck et al., 1996). Here, we demonstrate that the retention of glycerol is another level of regulation. That simple leakage through the bilayer is rather inefficient is demonstrated by the fact that under anaerobic conditions, when glycerol is produced for redox regulation, the lack of Fps1p results in dramatic glycerol accumulation and in a growth defect. Establishing a channel for glycerol export of course necessitates control of this export path. Uncontrolled glycerol extrusion through the channel leads to glycerol loss and hence to osmosensitivity, as demonstrated by the phenotype of mutants lacking the N-terminal regulatory domain of Fps1p or of cells expressing E. coli GlpF. Thus, we can describe adaptation of yeast cells to high osmolarity via glycerol accumulation as at least a two-step process: immediate closure of plasma membrane channels to ensure retention of the glycerol whose production is then increased in the second phase.
Remarkably, after a 24 h adaptation period, cells carrying the N-terminally truncated Fps1p accumulate as much glycerol as wild-type cells. Glycerol accumulation in those cells is not due to loss or rearrangement of the plasmid carrying the FPS1 construct because these cells retain their ability to rapidly dispose of accumulated glycerol after a hypo-osmotic shock. Certainly, the truncated version of Fps1p is not properly closed at this stage as even fully adapted cells carrying this construct lose glycerol at a rate of about threefold higher than wild-type cells. Apparently, constitutive loss of glycerol is compensated by high level production, which even after 24 h of incubation in high salt medium occurred at a rate of about fivefold higher in the mutant than wild type. In addition to a very high rate of glycerol production, a reduction of passive diffusion through the membrane bilayer, for example via altering the fatty acid and sterol composition of the membrane (Sutherland et al., 1997), might contribute to glycerol accumulation in cells carrying the N-terminal-deficient Fps1p.
Glycerol metabolism is essential for anaerobic growth of yeast when a surplus of NADH cannot be reoxidized by respiration (Albers et al., 1996; Ansell et al., 1997). The glycerol produced under such conditions would lead to an increase in turgor if not effectively exported. Fps1p is required for glycerol export under these conditions too, and mutants lacking Fps1p grew poorly under anaerobiosis. Interestingly, the highest internal glycerol levels are seen quickly after the shift to anaerobic conditions. Thus, Fps1p is needed, as after a hypo-osmotic shock, early in the adaptation to the new condition whereas slower diffusion through the membrane or other less efficient export pathways seem to be involved in later stages of the adaptation.
Direction of glycerol transport
Our data provide evidence that the Fps1p-mediated glycerol export is more efficient than uptake. This would make sense because we could only find growth phenotypes for mutants lacking Fps1p related to glycerol export but not to uptake. MIP channels have a symmetrical topology, and in fact the first and the second half of the proteins share sequence homology (Park and Saier, 1996; Froger et al., 1998). Hence, it is assumed that the transport function of MIP channels is bi-directional, probably with the same efficiency. Interestingly, Fps1p differs from the other MIP channels in the well-conserved NPA motifs in loops B and E, which are NPS and NLA, respectively, in Fps1p. The NPA boxes are thought to be directly involved in the formation of the channel (Walz et al., 1997). Whether these differences have any influence on the transport function or direction is under study.
Mechanisms controlling Fps1p function
We show here that rapid regulation of Fps1p-mediated glycerol transport requires a unique N-terminal hydrophilic extension. This extension, as well as the C-terminal one, are absent from other glycerol facilitators such as GlpF from E. coli. The latter protein can, when expressed in yeast, mediate glycerol transport into and out of the cell (Luyten et al., 1995; this work), but apparently this transport function cannot be significantly regulated by external osmolarity. Whether the N-terminal extension of Fps1p is not only necessary but also sufficient for mediating regulation or whether the architecture of Fps1p and GlpF are fundamentally different is under study.
The transport function of mechano- or swell-sensitive channels, such as MscL from E. coli and mammalian ClC-2, have been reported to be controlled via extensions (Jordt and Jentsch, 1997; Sukharev et al., 1997). In the case of ClC-2, a model has been presented in which parts of the extension interact with the channel-forming domains of the protein blocking transport by plugging a ‘lid’ or ‘ball’ onto the channel. The extensions of Fps1p do not show any homology with those of MscL or ClC-2, nor does it show any similarity to other proteins in the current databases. The surprising observation that most of the N-terminal extension of Fps1p can be deleted without affecting control of glycerol transport also indicates that the control mechanism of Fps1p might be different from that of ClC-2, in which sequences located well upstream of the first transmembrane domain are required for regulation.
The N-terminal extension could be the target for signalling pathways. However, we have not found evidence for the involvement of any of the known yeast osmosensing signalling pathways, i.e. the HOG- and the PKC-pathway and the pathway controlling the calcium pulse triggered by hypo-osmotic shock (Brewster et al., 1993; Davenport et al., 1995; Batiza et al., 1996), in regulating glycerol transport or retention. Using a variety of drugs, we also could not find evidence for the involvement of a number of other signalling pathways. Finally, the downregulation of Fps1p-mediated glycerol transport occurs very rapidly and is almost complete within 15 s, whereas stimulation of the HOG pathway by high osmolarity occurs in a minute time scale (Maeda et al., 1995; Siderius et al., 1997). This does not exclude the involvement of signalling pathways in the control of Fps1p, but seems to make such an involvement unlikely.
If Fps1p itself senses changes in osmolarity controlling glycerol transport, the N-terminal and possibly the C-terminal extensions might be involved in the sensing process. Because the extensions of Fps1p are predicted to face the cytoplasmic site of the plasma membrane, this would then suggest that changes in osmolarity could be sensed within the cell. Stretches imposed on the cytoskeleton have been discussed as signals generated by osmotic shock (Brewster and Gustin, 1994). However, latrunculin A, a drug affecting the cytoskeleton, did not cause any apparent changes to Fps1p control.
Remarkably, it appears as if Fps1p occurs in patches on the cell surface rather than being evenly distributed over the plasma membrane. Whether this effect plays any role in glycerol export or its regulation is presently not known. We note, however, that mammalian AQP4 has been found to form orthogonal arrays in kidney, brain and muscle (Verbavatz et al., 1997). Thus, such a clustered arrangement might have physiological significance and lead to the formation of specific sites for solute transport in the plasma membrane.
Regulatory effects mediated by Fps1p
If Fps1p were an osmosensor, it could regulate other proteins and/or signalling pathways. Indeed loss of Fps1p has been shown to affect different cellular processes. Fps1p is required for cell fusion during mating (Philips and Herskowitz, 1997), and clear evidence has been provided that this is related to osmoregulation. Before cell fusion, the cell wall has to be digested locally to allow membrane fusion to occur. Apparently, the cell must prevent bursting under these conditions and this requires Fps1p. However, mating occurs under normal growth conditions when yeast cells accumulate very little glycerol, and hence it is unclear whether Fps1p truly exports some solutes under these conditions. Thus, the precise role of Fps1p during cell fusion is not clear.
Deletion of FPS1 leads to a prolonged phosphorylation of the Mpk1p kinase in the PKC pathway after a hypo-osmotic shock (Davenport et al., 1995; M. C. Gustin, personal communication). Deletion of FPS1 also slightly diminishes osmotic induction of GPD1 expression. Conversely, transformation with the truncated version of FPS1 causing constitutive glycerol export results in a stronger and more sustained stimulation of GPD1 expression and of the HOG pathway after a hyper-osmotic shock (unpublished data). All these observations, however, may be explained by osmotic effects due to altered cellular glycerol content conferred by the FPS1 mutation, and hence do not necessarily point to a direct regulatory role of Fps1p.
Regulation of other MIP channels
The topology of Fps1p with its two long hydrophilic extensions in addition to the core of six transmembrane domains is very uncommon among the channels of the MIP family that have been identified so far. Only two other known proteins appear to have such extensions which, however, are not homologous to those of Fps1p: Yfl054p, a S. cerevisiae protein of unknown function (André, 1995), and BIB, the big brain protein of Drosophila (Rao et al., 1990; Burris et al., 1998). The transport function of these two proteins and hence if it is regulated is unknown.
Known regulated MIP channels include mammalian AQP2 and spinach PM28 A. Kidney AQP2 is regulated at the level of localization by hormonal stimuli and ultimately by protein kinase A (Knepper and Inoue, 1997). It is not known whether the transport function as such is also regulated. Spinach PM28 A is phosphorylated at two sites in vivo and mutations of those sites render the protein hyperactive in a Xenopus oocyte system (Johansson et al., 1998). One phosphorylation site is located within a cytoplasmic loop and the second one in the (small) C-terminal extension. This might indicate that the regulatory mechanisms of PM28 A are different from those of Fps1p. In conclusion, elucidation of the molecular details of the regulatory mechanisms of Fps1p-mediated glycerol transport may provide novel insight into the control of channel function. Furthermore, this knowledge may lead to rational design of effectors of MIP channel function, which in turn should be of interest for the clinical and pharmaceutical intervention into human osmoregulation in which aquaporin water channels seem to play a central role (Knepper et al., 1997).
Yeast cells were routinely grown in medium containing 2% peptone and 1% yeast extract supplemented with 2% glucose as carbon source (YEPD). Selection and growth of transformants carrying a replicating plasmid was performed in yeast nitrogen base medium (YNB; Sherman et al., 1983).
Plate growth assays were performed by pregrowing the cells either in medium without salt (for hyperosmotic shock) or in medium supplemented with 1 M sorbitol (for hypo-osmotic shock). Cells were resuspended in the same medium to an OD600 of 1.0. A 10-fold serial dilution of this culture was made and 5 μl of each dilution were spotted onto agar plates supplemented with 2% glucose and 5% NaCl (hyperosmotic shock) or without osmoticum (hypo-osmotic shock). Growth was monitored after 2–3 days at 30°C.
To apply a hyperosmotic shock in liquid medium, cells were pregrown in medium supplemented with 2% glucose to an OD600 of 0.5–1.0. The cells were then sedimented and resuspended in medium with 5% NaCl (final concentration). Samples were taken at the time points indicated in the figures. For hypo-osmotic shock in liquid cultures, cells were pregrown in medium supplemented with 2% glucose to an OD600 of ≈1.0, sedimented and resuspended in medium containing 5% NaCl (final concentration), and thereafter incubated for another 2 h with vigorous shaking. The culture was then diluted 10 times with medium without salt. In case of the mpkΔ, bck1Δ and pkc1Δ mutants, all media contained 0.5 M sorbitol to stabilize the cells. Samples were taken at the time points indicated in the figures.
To test the effects of different compounds on glycerol accumulation and release, cells were prepared as above (hypo-osmotic shock) and the compound was added 10 min before the downshock was applied. Samples for glycerol determination were taken as above. The concentrations of the compounds used were: 50 mM gadolinium, 200 μg ml−1 calmidazolium, 0.1 mM A23187 in the presence of 1 mM calcium, 0.75 mM diacylglycerol, 1 mg ml−1 compound 48/80, 2 mM dinitrophenol, 0.1 mM iodoacetamide, 250 μM chlorpromazine, 5 mM trinitrophenol, 200 μM latrunculin A, 6 μg ml−1 cholera toxin, 100 μM AlF4−,1 mM Hg2+, 1 mM phloretin, 1 mM Cu2+, 2.5 mM DIDS (4,4′-diisothiocyanate stilbene-2,2′-disulphonic acid), 100 mM Ba2+, 50 μM valinomycin, 10 μg ml−1 wortmannin and 200 μM genistein.
Anaerobic growth was performed in CBS medium (mineral medium; Franzén et al., 1996), but with a glucose concentration of 20 g l−1 instead of 40 g l−1. Tryptophan, histidine, leucine, adenine and uracil (Sigma) were added to a final concentration of 120 mg l−1. Wild-type and fps1Δ strains were grown in a fermenter (Belach Bioteknik) with a working volume of 2.5 l at 30°C, pH 5.0, 400 r.p.m. and flushing with pure nitrogen (0.25 v/v min−1). Glucose, ergosterol, Tween 80 and the appropriate nutrients were added to the fermenter after sterilization. Growth was followed by measuring the absorbance at 610 nm. Samples of 1.5 ml were taken to determine glucose and glycerol concentration.
Partial truncations of the FPS1 gene were constructed by using the polymerase chain reaction (PCR) with YEpFPS1 as template. The plasmid was completely amplified except for the region to be deleted. All primers contained a SacII site and three additional nucleotides at their 5′ end. The SacII site was used for religation of the plasmids, resulting in the insertion of a proline and an arginine residue at the site of the deletion. The truncations resulted in the removal of amino acids 13–230 (YEpfps1-Δ1 ), 13–69 (YEpfps1-Δ2 ), 76–230 (YEpfps1-Δ3 ), 13–144 (YEpfps1-Δ4 ), 151–230 (YEpfps1-Δ5 ) and 535–649 (YEpfps1-ΔC ).
YEpmyc 181 is a 2μLEU2 plasmid allowing the attachment of the c-myc epitope to the C-terminus of a protein (Reisdorf et al., 1993). FPS1 was cloned into YEpmyc 181 by PCR with YEpFPS1 as template. The reverse primer was designed to replace the stop codon with a XbaI site. The PCR product was digested with Pst I/XbaI and the resulting 1.1 kb fragment was subcloned into pUC18. To match the restriction sites with the multiple cloning site of YEpmyc 181, a 2.6 kb Sal I/Pst I fragment was subcloned into pRS405 (Sikorski and Hieter, 1989). The Pst I/XbaI fragment from pUC18 was subcloned into the pRS405 plasmid that contained the Sal I/Pst I fragment to create the complete FPS1 gene without stop codon. The 3.7 kb Sal I/XbaI fragment was then subcloned into YEpmyc 181, generating the epitope-tagged YEpmyc-FPS1 plasmid.
Epitope tagging of the N-terminal-truncated versions of FPS1 was achieved by replacing a XhoI/KpnI fragment from YEpmyc-FPS1 with the corresponding XhoI/KpnI fragment from YEpfps1-Δ1, YEpfps1-Δ2, YEpfps1-Δ3, YEpfps1-Δ4 and YEpfps1-Δ5, thus generating the plasmids YEpmyc-fps1-Δ1, YEpmyc-fps1-Δ2, YEpmyc-fps1-Δ3, YEpmyc-fps1-Δ4 and YEpmyc-fps1-Δ5.
pRS406 is an integrative vector carrying URA3 as a selectable marker (Sikorski and Hieter, 1989). Full-length FPS1 as well as the truncated versions of the gene were cloned into the pRS406 vector by inserting a BamHI/HindIII fragment from YEpFPS1, YEpfps1-Δ1, YEpfps1-Δ2, YEpfps1-Δ3, YEpfps1-Δ4 and YEpfps1-Δ5 into the complementary sites of pRS406, thus generating plasmids YIpFPS1, YIpfps1-Δ1, YIpfps1-Δ2, YIpfps1-Δ3, YIpfps1-Δ4 and YIpfps1-Δ5.
The pRS406 vectors containing either full-length or truncated versions of FPS1 were linearized with Bgl II and integrated into the yeast genome of YSH294 upstream from fps1Δ::LEU2, generating strains YMT100–YMT105 (see Table 1). Single copy integration was confirmed by Southern analysis.
The construction of YEpGlpF (YEplac 195), containing the gene encoding the E. coli glycerol facilitator GlpF expressed under the control of the yeast PGK1 promoter, has been described previously (Luyten et al., 1995).
Yeast transformations were performed by the lithium acetate method (Ito et al., 1983) and plasmids were selected and propagated in E. coli TOP10F′.
For all gene cloning experiments, standard techniques were applied as described in (Sambrook et al., 1989). DNA fragments and PCR products were purified using the Qiagen kits.
Glycerol transport experiments
To determine glycerol uptake, cells were pregrown in YNB supplemented with 2% glucose to an OD600 of ≈2.0. The cells were harvested, washed and resuspended in ice-cold MES buffer (10 mM MES, pH 6.0) to a density of 40–60 mg cells ml−1. Glycerol uptake was measured by adding glycerol to a final concentration of 100 mM ‘cold’ glycerol plus 40 μM [14C]glycerol (153 mCi mmol−1; Amersham) in a total volume of 50 μl. After exposing these cells for 10 s, the reaction was stopped by diluting with 5 ml ice-cold water. The cells were filtered (Whatman GF/C filter, 25 mm), washed twice with ice-cold water and the radioactivity was counted in a liquid scintillation counter. The radioactivity retained on the filter was taken as a measure of the glycerol taken up by the cells.
To determine the glycerol uptake during hyperosmotic stress, MES supplemented with NaCl was added to the cell suspension to give a final salt concentration of 5%. Time samples were taken as indicated in the figures.
For efflux measurements, cells were prepared as for uptake experiments. Wild-type and fps1Δ mutant cell suspensions were loaded by exposure to 100 mM (specific activity 14–250 mCi mmol−1) and 300 mM (specific activity 14 mCi mmol−1) [14C]glycerol, respectively, for 30 min at 30°C in MES buffer, as described for uptake experiments. Different initial concentrations of glycerol were used to achieve similar levels of intracellular glycerol in both strains because the fps1Δ mutant accumulated glycerol less efficiently than the wild type.
All transport experiments were repeated at least three times to confirm the trends observed. Representative results are shown.
Glycerol production and leakage
To measure the glycerol production rate, cells were prepared as for hyperosmotic shock. Samples were taken for glycerol determination (1.5 ml) as well as to determine the OD at 600 nm. The glycerol production rate was calculated as the change in total (intracellular plus extracellular) glycerol per OD cells per minute (μM glycerol per OD cells min−1).
To determine the glycerol leakage rate, cells were subjected to a hyperosmotic shock and continued to grow in medium containing 5% NaCl for either 5 or 24 h, when the different strains had either the same glycerol production rate (5 h) or the same intracellular glycerol content (24 h). Thereafter, the cells were sedimented and resuspended in fresh, salt-containing medium. Samples of 1.5 ml were taken each 15 min, sedimented and the glycerol content of the supernatant was determined. The glycerol leakage rate was calculated as the change in extracellular glycerol per OD cells per minute (nM glycerol per OD cells min−1).
Membrane preparation and Western analysis
Cells were grown in YNB supplemented with 2% glucose to OD600 = 1.5–2.0, harvested and resuspended in homogenization buffer (50 mM tris-HCl, pH 7.5, 0.3 M sucrose, 5 mM EDTA, 1 mM EGTA, 5 mg ml−1 bovine serum albumin, 2 mM DTT, 1 mM phenylmethylsulphonylfluoride and 2 μM pepstatin A). The cells were disrupted by vortexing with glass beads and the lysate was centrifuged at 10 000 × g for 10 min at 4°C. The resulting supernatant was centrifuged at 100 000 × g for 60 min and the microsomal membrane pellet was resuspended in membrane wash buffer [10 mM tris-HCl, pH 7.0, 1 mM EGTA, 1 mM DTT and 20% (v/v) glycerol].
To prepare plasma membranes, the microsomal membranes were placed on top of a 43.5–53.3% discontinuous sucrose gradient and centrifuged at 100 000 × g in a swing bucket SW55Ti rotor for 3 h at 4°C. The light-density fraction and the interphase between the two sucrose layers were collected and centrifuged at 100 000 × g for 60 min and the pellets were resuspended in membrane wash buffer.
Five micrograms of total protein from the microsomal membranes, the light-density fraction and the interphase were separated by SDS–PAGE and were blotted onto nitrocellulose filters (HybondC extra, Amersham). The filters were blocked with 2% skimmed milk (Difco) in PBST [8 g l−1 NaCl, 0.2 g l−1 KCl, 1.44 g l−1 Na2HPO4, 0.24 g l−1 KH2PO4, pH 7.4, and 0.05% (v/v) Tween 20]. The membranes were probed overnight with a mouse monoclonal IgG antibody against the c-myc tag attached to Fps1p (9E10, Santa Cruz Biotechnology), and with a rabbit polyclonal IgG antibody against the plasma membrane H+-ATPase Pma1p (kindly provided by D. Seto-Young and D. Perlin). The antisera were applied at 1: 1000 (c-myc tag) and 1: 2000 (Pma1p) dilutions. After washing the filters in PBST, the membranes were incubated for 1 h with secondary antibody (alkaline phosphatase-conjugated goat anti-rabbit IgG 1: 10 000 or alkaline phosphatase-conjugated goat anti-mouse IgG 1: 1000 respectively) in PBS–Tween 20.
For detection, the membrane filters were incubated with 50 mg of 5-bromo-4-chloro-3-indolyl phosphate and 75 mg of nitroblue tetrazolium salts per ml.
Protein was quantified using the method of Bradford (Bradford, 1976) with bovine serum albumin as standard.
Indirect immunofluorescence was performed on spheroplasts following the procedure of Pringle et al. (1991). Antibody incubations were carried out in PBS + 0.4% dried milk, primary antibodies were incubated overnight at 4°C and secondary antibodies for 2 h at room temperature. Mouse monoclonal anti-c-myc (9E10, Santa Cruz Biotechnology) and rabbit anti-Pma1p (kindly provided by D. Seto-Young and D. Perlin) were used as primary antibodies and fluorescein-conjugated anti-mouse IgG and anti-goat IgG, respectively, as secondary antibodies.
Cells were pregrown to an approximate OD600 of 1.0, harvested and resuspended in YNB + 2% glucose + 5% NaCl. Total RNA was isolated at the time points indicated in Fig. 6 and separated by electrophoresis as described previously (de Winde et al., 1996). Blots were hybridized with 32P-labelled fragments of FPS1, GPD1 (encoding one of the NAD-dependent glycerol 3-phosphate dehydrogenase isoenzymes) and IPP1 (encoding inorganic pyrophosphatase) in buffer containing 7% SDS, 0.5 M sodium phosphate buffer, pH 7.0 and 1 mM EDTA. The signal was quantified using a phosphorimager (Fuji, BAS-1000).
Glucose and glycerol concentrations were determined enzymatically using commercial glucose and glycerol determination kits (Boehringer Mannheim) as described previously (Albertyn et al., 1994).
Dry weight was determined by harvesting cells by filtration. The filters with the cells were dried at 80°C overnight.
The data presented for biochemical determinations show the results of one typical experiment out of at least three independent experiments giving consistent results.
*Present address: Institute for Wine Biotechnology, University of Stellenbosch, Private Bag XI, 7602 Matieland, South Africa
**Present address: Department of Microbiology, University of Stellenbosch, Private Bag XI, 7602 Matieland, South Africa
We thank W. Verheyden for help with the glycerol measurements, D. Perlin and D. Seto-Young (Public Health Research Institute, New York) for the Pma1p antibody, M. Sanders and P. Crews (University of California at Santa Cruz) for latrunculin A, members of the laboratory of M. Boutry (Université Catholique de Louvain-la-Neuve) for help with the immunofluorescence technique, the Laboratorium voor Tropische Plantenteelt (KUL) for the use of the fluorescence microscope, Martijn Rep (Leuven) for helpful suggestions throughout this work and Roslyn Bill (Göteborg) for critical comments on the manuscript. We acknowledge support from the Commission of the European Union via contracts ERB-CHRX-CT93-0265 to J.M.T. and J.R., BIO4-CT95-0161 to J.M.T. and S.H., FMRX-CT96-0007 to J.M.T. and ERB-4061-CT97-0406 to S.H., from the Ministry of the Flemish Community (International Scientific Cooperation between Flanders and South Africa, contract BIL96/27 to J.M.T. and B.A.P.), from the Foundation for Research Development (FRD) to B.A.P. and S.G.K. and from MEC (Spain) to J.R. (contract PB95-0976).