INTRODUCTION

Maintenance of water homeostasis is fundamental to most cells. Exposure to hyperosmotic conditions initiates responses serving to maintain an osmotic gradient across the cell membrane, appropriate for turgor and cell volume control. A common mechanism behind this adaptation involves intracellular accumulation of one or more nonpolar osmolytes. Among eukaryotic cells this response is best understood for the yeast Saccharomyces cerevisiae, which on exposure to dehydration stress initiates increased production and intracellular accumulation of glycerol (1, 2). Glycerol is produced through reduction of dihydroxyacetone phosphate to glycerol-3-phosphate and subsequent dephosphorylation to glycerol. The responsible enzymes are a NAD⁺-dependent glycerol-3-phosphate dehydrogenase (GPD)¹ and a poorly characterized glycerol-3-phosphatase. Transfer of cells to hyperosmolar conditions leads to increased production of GPD (3) due to transcriptional activation of one of the isogenes (GPD1) for the enzyme (4, 5, 6). In addition, membrane permeability to glycerol is diminished (7), presumably mainly due to inhibition of the facilitator that transports glycerol across the membrane (8). Under strongly hyperosmotic conditions, S. cerevisiae can accumulate glycerol up to molar concentrations (9). The osmotic induction of GPD1 appears to be controlled via an osmosensing pathway (6), involving cell surface osmosensors encoded by SHO1 (10) and a two-component system encoded by SLN1/SSK1 (11, 12). These detection systems are linked to a cascade of protein kinases encoded by SSK2, SSK22, PBS2, and HOG1, forming a mitogen-activated protein (MAP) kinase signaling pathway (10, 12, 13).

Osmotic control of glycerol production might also be exerted at the level of dephosphorylation of glycerol 3-phosphate. The early work by Tsuboi et al. (14, 15) indicated the existence of a phosphatase in S. cerevisiae having a specific role in glycerol production. These researchers described a partially purified phosphomonoesterase from bakers' yeast that displayed a high specificity toward glycerol 3-phosphate at pH 6.5. Using conditions that repress the ortho-phosphate repressible phosphatases and strains lacking various nonspecific phosphatases, we here confirm the existence of a phosphatase activity, specific for DL-glycerol 3-phosphate in extracts of S. cerevisiae. To functionally characterize this enzyme and gain further insight into its role in cellular osmoregulation we took on purification of the glycerol-3-phosphatase and unexpectedly detected two molecular forms of the enzyme, one of which shows constitutive expression, the other being induced by increased extracellular osmolarity. By partial amino acid sequencing of tryptic digests of the two molecular forms, the corresponding genes, previously reported as open reading frames without function, were identified. We propose different roles for the two glycerol-3-phosphatase isoenzymes that belong to a previously unrecognized family of low molecular weight phosphomonoesterases.

EXPERIMENTAL PROCEDURES

The genotypes of the S. cerevisiae strains used are shown in Table I. Strains were routinely grown in a medium composed of 1% yeast extract, 2% Bacto peptone, and 2% glucose (YPD) or in a synthetic yeast nitrogen base (YNB) medium (Difco), supplemented with 2% glucose and other necessary growth requirements (16) at 120 mg/liter. When appropriate, 10 mM inorganic phosphate (pH 5.5) was added to the YNB or YPD medium used (then named YNBP or YPDP medium) to ascertain repression of repressible phosphatases.

Table I. Yeast strains used Name Genotype Source or reference YS18 MAT his3 leu2 ura3-5 canR A. Hinnen YMR4 MAT his3 leu2 ura3-5 canR pho5,3::ura31 A. Hinnen SKQ2n MATa/ade1/+ ade2/+ his1/+ (38) YPH102a MATa ura3 leu2 his3 ade2 lys2 (36) JBY10 MATa ura3 leu2 his3 ade2 lys2 trp1 hog1-1::TRP1 (13) JBY43 MATa ura3 leu2 his3 ade2 lys2 trp1 pbs2-1::URA3 (13) SH611 SUC2 mal mel gal2 CUP1 S. Harashima SH3544 MATa pho8::URA3 ura3 his3-300 trp1-901 S. Harashima SH3598 MATa pho3-1 pho13::URA3 ura3-1,2 leu2-3,112 lys1 S. Harashima

Solid media contained 2% agar. Responses to osmotic stress were assessed in media adjusted with NaCl or sorbitol. Liquid cultures were incubated with rotary agitation at 30 °C. Growth was monitored by measuring optical density at 610 nm (A₆₁₀). For preparation of cell extracts cultures were harvested by centrifugation at A₆₁₀ = 0.5-1.

Cells were washed twice in ice-cold TRED buffer (10 mM triethanolamine, 1 mM EDTA, 1 mM dithiotreitol, pH 7.5) and resuspended in the same buffer plus 0.5 µl/ml of a protease inhibitor mix containing 70 mg/ml phenylmethylsulfonyl fluoride, 1.4 mg/ml leupeptin, and 1 mg/ml pepstatin. Extracts were prepared by disruption of cells using acid-washed glass beads, and the obtained extract was cleared by centrifugation (14,000 × g for 15 min). Extracts used to measure phosphatase activity were desalted by gel filtration according to procedures described previously (17). All operations were performed at 0-4 °C.

Protein concentration was routinely determined using the method of Bradford (Bio-Rad). Protein concentrations of purified protein fractions were estimated by determining the absorbance at 280 nm of concentrated samples in a 100-µl microcuvette. Values were compared with those of a bovine serum albumin standard and translated to protein concentration after correction for the difference in tryptophan content of the two proteins.

Glycerol-3-phosphatase was assayed essentially as described by Sussman and Avron (18). Cell-free extracts were routinely incubated in 20 mM Tricine-HCl (pH 7.0), 5 mM MgCl₂, and 10 mM DL-glycerol 3-phosphate in a total volume of 1.0 ml. For determination of kinetic parameters the DL-glycerol 3-phosphate concentration was varied within the concentration range of 0.5-32 mM. After starting the reaction, samples of 90 µl were withdrawn at time intervals and immediately added to 10 µl 50% HClO₄ to stop the enzyme reaction. Substrates other than glycerol-3-phosphate were added to a final concentration of 10 mM, and for these slower reactions samples were withdrawn at 60-s intervals for 6 min. Released inorganic phosphate was analyzed according to Ames (19), and the reaction rate was calculated from the slope of a linear plot of released phosphate versus time. To detect enzyme activity during the enzyme purification procedure, samples of 5 µl were incubated in 95 µl of assay buffer for 15 min, followed by phosphate analysis as described above.

Native molecular weight estimations were performed using PhastGel 8-25 gradient gels (Pharmacia LKB) for native PAGE. Strain YMR4 was cultured in YPDP and homogenized as described above, and 4.5 µl of extract was applied to each lane. The molecular weight standards used were those of the Pharmacia electrophoresis calibration kit. Two parallel gels were run in a PhastSystem (Pharmacia) as recommended by the manufacturer. One gel was stained for phosphatase using 10 mM DL-glycerol 3-phosphate in assay buffer containing 15 mM CaCl₂. Released phosphate was in situ precipitated as white calcium phosphate. The other gel was stained for proteins using the recommended Coomassie staining program.

Protein purification was carried out at 4 °C, and whenever possible samples were kept on ice. All chromatographic steps were performed on an fast protein liquid chromatography system (Pharmacia). Cell extracts (5-15 mg/ml protein) were loaded onto a 120-ml Superdex 200 column that was equilibrated with 10 mM TRED buffer. The column was eluted with the same buffer at a flow rate of 1 ml/min, and 1-ml fractions were collected. In this and subsequent chromatographic steps protein concentration was monitored by continuous UV absorbance at 280 nm, and the peak enzyme activity was localized by phosphatase assays. The most active fractions were pooled and applied to a 1-ml Mono-Q column equilibrated with 10 mM triethanolamine (pH 7.5) containing 1 mM dithiotreitol (TRD buffer). The column was washed with 45 ml of TRD buffer and eluted with a 30-ml linear gradient of 0-0.15 M NaCl in the same buffer followed by 5 ml of buffer containing 1 M NaCl. The fractions active for glycerol-3-phosphatase were reapplied to the same Mono-Q column and subjected to a consecutive round of ion exchange chromatography. This cycle was initiated by washing with 15 ml of TRD buffer adjusted to pH 7.0, followed by elution with 20 ml of a combined ion strength-pH gradient generated by linearly increasing the proportion of 10 mM Bis-Tris buffer (pH 5.5) containing 0.25 M KCl from 0 to 30% in the basal TRD buffer (pH 7.0), and finally terminated by elution with 5 ml of 100% Bis-Tris buffer. Purified enzyme fractions of the two isoforms of the phosphatase were taken from the peak fractions of first cycle of Mono Q chromatography, where the enzymes eluted at 63 mM NaCl (Gpp1p, see Fig. 4C) and at 35 and 57 mM NaCl (Gpp2 and Gpp1, respectively, Fig. 4D).

To examine the homogeneity of the purified enzyme fractions, samples were analyzed by SDS, 15% polyacrylamide gel electrophoresis, and proteins were visualized by silver staining (20). For microsequencing of the purified phosphatases, the stained band was excised with a scalpel and prepared for trypsin digestion as described previously (20).

Two-dimensional PAGE was run on a Millipore Investigator system as described previously (20). Dried gels were exposed to image plates and subsequently scanned in a PhosphorImager (Molecular Dynamics) at a resolution of 176 × 176 µm. Images were automatically processed and quantified by the two-dimensional software PDQuest (PDI) as described by Blomberg et al. (21). Individual quantification of resolved proteins were normalized to the total amount of radioactivity in all quantified spots of the gel and expressed as parts per million.

In gel trypsin digestion of purified phosphatase and separation of the generated peptides on reversed phase-HPLC (Smart system, Pharmacia), were performed as described elsewhere (20). N-terminal sequence analysis was acquired using a pulsed liquid sequenator 473A (Applied Biosystems) according to procedures recommended by the manufacturer. Fractions were applied to BioBrene conditioned filters in 20-µl portions that were dried under argon between applications.

RESULTS

To facilitate the search for a specific glycerol-3-phosphatase in crude extracts of S. cerevisiae, efforts were made to minimize the presence of various nonspecific alkaline and acid phosphatases. The acid phosphatases of S. cerevisiae encoded by the PHO5, PHO10, and PHO11 genes as well as the alkaline phosphatase encoded by the PHO8 gene are all repressed at the transcriptional level by inorganic phosphate (22, 23). To curb formation of these enzymes, cell cultures were grown with 10 mM inorganic phosphate. Enzyme activity measurements in extracts from such cultures indicated negligible contribution to the observed glycerol-3-phosphatase activity from residual activity of the repressible acid (Pho5p) and alkaline (Pho8p) phosphatases, since mutants lacking the corresponding genes (strain YMR4 and SH3544, respectively) demonstrated activities similar to those of the wild type strains (Table II). Neither was the analyzed glycerol-3-phosphatase activity notably influenced by the absence of the nonrepressible acid phosphatase encoded by PHO3 or the constitutive alkaline phosphatase coded for by PHO13 (Table II, strain SH3598).

Table II. Glycerol-3-phosphatase activity in crude extracts of various phosphatase-defective mutants of S. cerevisiae grown in YNB medium Values are means of three separate experiments ± S.D. Strain Relevant genotype Glycerol-3-phosphatase activity nmol min1 mg protein1 YS 18 Parent strain 68.0  ± 3.8 YMR4 pho5,3::ura31 70.2  ± 3.9 SH 611 Parent strain 73.1  ± 0.5 SH3544 pho8::URA3 67.6  ± 5.0 SH3598 pho3-1 pho13::URA3 62.5  ± 0.3

Examination of the substrate specificity of the detected enzyme activity in crude extract of wild type cells (Table III), showed preferential activity on glycerol 3-phosphate; less than 9% of that activity was exerted on glycerol 2-phosphate, glycerate 3-phosphate, glycerate 2,3-diphosphate, glucose 6-phosphate, and fructose-6 phosphate. This substrate profile was very similar with extracts of mutants lacking one (pho3, YMR4) or both (pho3 pho13, SH3598) of the constitutive acid and alkaline phosphatases (data not shown). From these results we conclude that it is possible to assay a specific glycerol-3-phosphatase activity in crude extracts of S. cerevisiae under the pH-neutral conditions used and that this activity is clearly distinct from that of previously described acid and alkaline phosphatases.

Table III. Phosphatase activity of crude extract and purified Gpp1p and Gpp2p on various organic phosphomonoesters Gpp1p and Gpp2p were purified by gelfiltration and Mono-Q chromatography. For each preparation the enzyme activity with DL-glycerol 3-phosphate was set to 100. The corresponding activities were as follows (in nmol min1 mg protein1): crude extract (68.0), Gpp1p (16170), and Gpp2p (2050). Substrate Relative enzyme activity Crude extract Gpp1p Gpp2p DL-glycerol 3-phosphate 100 100 100 Glycerol 2-phosphate 8.6 2.7 0.8 Glycerate 3-phosphate 3.0 0.01 0.2 Glycerate 2,3-diphosphate 2.8 0.4 1.0 Glucose 6-phosphate 3.7 0.5 0.5 Fructose 6-phosphate 3.7 0.1 1.0 2-Deoxyglucose 6-phosphate NDa 0.2 0.0 a ND, not determined.

To further characterize the glycerol-3-phosphatase, the molecular weight was estimated from the mobility of the enzyme on gel electrophoresis in a native gradient 5-25% polyacrylamide gel. Staining of gels for released phosphate from glycerol 3-phosphate revealed one band with a migrating position corresponding to a molecular mass of 30 kDa (Fig. 1).

Culturing cells with various concentrations of NaCl demonstrated that the total cellular activity of glycerol-3-phosphatase increased with increased extracellular salinity (Fig. 2). Substituting NaCl by osmotically equivalent concentrations of sorbitol gave a comparable increase, consistent with the supposition that the cells adjust their specific activity or enzyme concentration in response to changes of the surrounding osmolarity. The observed increase in activity was specific for glycerol 3-phosphate; no increased hydrolysis of the other organic phosphate monoesters used for the substrate specificity test was observed for cells grown at 0.7 M NaCl (data not shown), indicating that the demonstrated regulation is specifically exerted on the enzyme activity showing specificity for glycerol 3-phosphate.

strain YMR4 grown cultured in YPDP medium (0.4 MPa) adjusted to 1.9 MPa by the addition of 0.35 

sorbitol or to 3.3 MPa by the addition of 0.7 

To examine whether protein synthesis is required to achieve this increase in activity, cells were shifted from basal medium to medium containing 0.7 M NaCl in the presence or absence of cycloheximide. In the absence of inhibitor, enzyme activity increased to levels 1.5 and 1.7 times the original value after incubation for 60 and 110 min, respectively. This increase was completely prevented by the presence 100 µg/ml of cycloheximide (data not shown), indicating that the increase in glycerol-3-phosphatase activity requires protein synthesis.

Osmotic stress-induced glycerol production in S. cerevisiae requires the HOG1 and PBS2 genes, encoding MAP kinase homologues of the osmosensing and signaling HOG pathway (13). Examination of the specific activity of glycerol-3-phosphatase in extracts of pbs2 and hog1 mutants cultured at elevated salinities demonstrated that either defect in the signaling pathway caused a complete inhibition of the osmotic induction of the glycerol 3-phosphatase activity (Fig. 3).

To purify the glycerol-3-phosphatase we employed gel filtration and anion exchange chromatography. As a first step, extracts of cells grown in YNB medium and the same medium containing 1.4 M NaCl were subjected to Superdex 200 gel filtration (Fig. 4, A and B). The presence of a shoulder in the activity profile of eluate from extracts of cells grown at high salinity suggested two forms of the enzyme in salt grown cells. This assumption was reinforced by the subsequent Mono-Q ion chromatography step (Fig. 4, C and D) in which the proteins were eluted by a linear 0-0.15 M NaCl gradient. The phosphatase activity emerged as a single peak (form I) with extracts of nonstressed cells, whereas with extracts of cells grown at high salinity, an additional form (form II) was eluted at slightly lower salt concentration. The active Mono-Q fractions from cells cultured in 1.4 M NaCl media were further resolved by rechromatography on the same column, this time eluted with a combined pH and KCl gradient (see ``Experimental Procedures'' and Fig. 5A). By this step form I, hereafter called Gpp1p, was purified to electrophoretic homogeneity as revealed by silver-stained SDS gel electrophoresis (Fig. 5B), while the recovery of activity of form II by this procedure was insignificant. Enzyme activity data demonstrated that Gpp1p was not appreciably affected by changed salinity, while the second molecular form, hereafter called Gpp2p, increased strongly with salinity from an activity below detection level in nonstressed cells (Fig. 4, C and D). A quantitative summary of the purification of the two isoforms as performed with extract of cells grown at 1.4 M NaCl, is shown in Table IV.

Table IV. Purification of Gpp1p and Gpp2p from extracts of S. cerevisiae SKQ grown at 1.4 M NaCl Fraction Volume Protein Specific activity Total Activity Purification Yield ml µg nmol min1 mg protein1 nmol min1 -fold % Cell-free extract 2 27520 108 2973 1 100 Superdex 200 11 181.50 7063 1282 65 43 Mono-Q I GPP1 1 25.72 24029 618 222 21 GPP2 1 8.55 9400 80 87 2.7 Mono-Q II GPP1 1 4.97 12062 60 112 2 GPP2a a Activity not detectable.

Gpp1p was excised from SDS gels and cleaved by in-gel trypsinization. The generated peptides were separated on reverse phase HPLC, and selected fractions were subjected to N-terminal sequencing. The eight peptides examined revealed the sequences shown in Table V. When these peptides were compared with sequence data of the Swiss Protein Data Bank, full identity was established with an open reading frame recognized by the systematic sequencing of chromosome IX. The corresponding gene (YIL053W) which was designated GPP1, encodes a protein of 271 amino acids having a molecular mass of 30.4 kDa.

Table V. Amino acid sequences of peptides generated from the two isoforms of glycerol-3-phosphatase, resolved by protein chromatography or two-dimensional PAGE Numbers refer to the amino acid positions in the predicted protein sequences of the YIL053W (GPP1) and YER062C (GPP2) genes. Underlined residues mark positions of unique amino acids. Constitutive form (GPP1) Salt-induced form (GPP2) Purified by protein  NAALFDV(V)GXI 33-44  XXX 40-45 chromatography FAPDFAEEYVN 86-97 TDAIAK 58-64 WAVATXG 132-138 FAPDXX 65-71 WFL 146-151 (W)AVA(T)XG 111-117 XXRPYFITA 152-161 YFITAND 136-142a QGKP(H)PEPYLK 166-176 QG(K)P(H)PEPYLK 145-155 IGIATTFDLD 215-225 VGY 223-226 VGEYNAEXD 244-252 Extracted from preparative FAPDFAEEYV 86-96  NAALFDVDGTI 12-23 two-dimensional PAGE YGEHXIEVP 107-115 FAPDFA 65-71 VVVFEDAPAGIAAG 195-208 WF 125-128 XFITAND 136-142a VVVFEDAPA 174-182 VGYNAETDEV 223-233 a Sequence is preceded by E in YIL053W and would therefore not provide a site for trypsin.

SDS-polyacrylamide gel electrophoresis and trypsinization of Gpp2p, using the most active fractions from the Mono-Q chromatography (Fig. 4D), yielded seven sequences that showed striking similarity to Gpp1p, although some of the sequences displayed a slight but distinct sequence deviation from the corresponding regions of Gpp1p (Table V, unique amino acids underlined). Data base search revealed full identity with the primary structure of a gene (YER062C) identified by sequencing of chromosome V. This gene encodes a protein of 250 amino acids having a molecular mass of 27.8 kDa, which is slightly lower than for Gpp1p. This gene was designated GPP2.

Sequence comparisons between the GPP1 and GPP2 using the Gap and BestFit program from the Genetics Computer Group demonstrated strong homology, the inferred amino acid sequences showing 95% identity (Fig. 6). The only identified proteins to which Gpp1p and Gpp2p show significant identity (30-35%) are two phosphomonoesterases encoded by DOG1 and DOG2 (24, 25).

From the expression pattern and the theoretical M_r, the positions of Gpp1p and Gpp2p were tentatively localized on analytical two-dimensional PAGE gels. To confirm the identity, the located protein spots were excised from preparative gels, and internal peptides were generated and sequenced. The obtained amino acid sequences (Table V) unequivocally identified the candidate spots as the GPP1 and GPP2 encoded proteins. The position of the two isoforms are indicated by arrows in Fig. 7, A-C, which shows relevant portions of gels run with L-[S³⁵]methionine-labeled extracts of cells cultured in medium containing 0, 0.7, and 1.4 M NaCl. Computerized quantitation (Fig. 7, D and E) of the two forms confirmed previous observations; Gpp1p remains abundant independent of external salinity. This form comprises 0.2-0.3% of totally detected proteins as determined from the methionine incorporation. Normalizing obtained parts per million values for differences in methionine content, thus enabling stoichiometric comparisons, we found that Gpp1p was approximately equimolar to actin (Act1p) in cells grown in basal medium. Gpp2p increases, on the other hand, from a low level to become about as abundant as Gpp1p at the highest salinity (1.4 M NaCl).

  0.5 in YNB medium containing 0 

The purified isoforms showed insignificant activity on phosphomonoesters other than glycerol 3-phosphate (Table III). The stereospecificity of the enzyme for the D- and L-forms of glycerol 3-phosphate was examined by comparison of the enzyme activity obtained for the racemic substrate (DL-form) with that observed for the pure L-form (sn-glycerol 3-phosphate), which was anticipated to be the natural substrate for the enzyme. Surprisingly, both forms of the enzyme show essentially similar activities with DL- and L-glycerol 3-phosphate (data not shown), indicating no obvious stereoselectivity for either enantiomer. Hence, the enzymes should be named DL-glycerol-3-phosphatases. Using various concentrations of the DL-form, as the substrate, Eadie-Hofstee plots yielded apparent K_m of 3.1 mM for Gpp1p and of 3.9 mM for Gpp2p. The corresponding V_max values were 49 and 46 µmol/min/mg protein. The pH dependence profile showed an optimum at pH 6.5 for both enzymes (Fig. 8).

To further examine the effect of the HOG pathway on osmotic induction of the DL-glycerol-3-phosphatase activity, enzymes were purified from extracts of a hog1 and a parent strain that had been both incubated for 1 h at 0.7 M NaCl. The parent strain displayed clear Gpp2p and Gpp1p peaks after Mono-Q chromatography, while the hog1 mutant showed roughly the same amount of Gpp1p as the parent but insignificant Gpp2p enzyme (Fig. 9).

strain YPH 102a () and a congenic strain having a deleted

gene (JBY10) (), grown in YNB medium to

  1 and thereafter incubated for 1 h in the same medium containing 0.7 

DISCUSSION

This work reports the presence of two isoforms of DL-glycerol 3-phosphatase (EC) in S. cerevisiae. Peptide sequences obtained from purified isoforms demonstrated that the enzymes responsible for this activity are encoded by two distinct but highly identical genes, here named GPP1 and GPP2. These genes were previously sequenced as part of the systematic sequencing of the yeast genome but were reported as open reading frames without known function. The predicted amino acid sequences of GPP1 and GPP2 are 95% identical and show about 35% identity with that of DOG1, which was isolated by its ability to confer resistance to 2-deoxyglucose (24). This nonmetabolizable glucose analogue elicits glucose repression, and this effect is counteracted by overexpression of the DOG1 gene product which has a 2-deoxyglucose-6-phosphatase activity. Interestingly, Dog1p has a ``twin'' homologue encoded by the recently identified DOG2 gene (25), showing 92% amino acid identity to that of DOG1. The high similarity within each of the two gene pairs points to an origin by gene duplication and dispersal. The true physiological function of Dog1p and Dog2p is not yet established (24, 25). However, they clearly appear to have a role different from Gpp1p and Gpp2p, since these enzymes do not dephosphorylate 2-deoxyglucose 6-phosphate (Table IV) and Dog1p and Dog2p show insignificant activity on glycerol 3-phosphate (25, 26). Together these four enzymes constitute a previously unrecognized family of low molecular weight phosphomonoesterases, having molecular masses of approximately 30 kDa. Gel filtration and gradient gel electrophoresis under nondenaturing conditions indicate that Gpp1 and Gpp2 are catalytically active as monomers. Hence, the glycerol-3-phosphatase isozymes are clearly different from the nonspecific alkaline and acid phosphatases, which have subunit sizes of 57-66 kDa as determined by DNA sequencing and SDS gel electrophoresis (27, 28, 29, 30, 31), and which are generally active as oligomers (32).

The glycerol 3-phosphatase activity that was originally described in partially purified extracts of bakers' yeast (14) is by all likelihood identical to that of Gpp1p, since this enzyme is the predominant form in cells cultured at low osmolarity (Figs. 4 and 7). As indicated from subcellular fractionation of lysed spheroplasts, this activity has its main location in the cytosol (33, 34). The narrow substrate specificity of the DL-glycerol-3-phosphatase conforms with its location in the cytosol, where a strict specificity would be essential to avoid interference with the plethora of phosphorylated metabolites in the metabolic network. In contrast, the nonspecific phosphatases of acid and alkaline preferences are primarily confined to the vacuole or the periplasm of the cell (32).

Glycerol has dual functions in S. cerevisiae; it serves as a nontoxic redox sink during fermentation (35)² and as an osmoregulator during hyperosmotic stress (2). Osmoregulation in S. cerevisiae is achieved by varying the internal glycerol concentration to adjust the intracellular osmolarity in response to changes of the extracellular water potential. Exposed to hyperosmotic stress, the cells increase glycerol production and glycerol accumulation. A contribution from Gpp2p to this response is indicated by the finding that this isoform, which is the clearly minor form under nonstressed conditions, increases as cells are cultured at elevated osmolarity (Figs. 4 and 7). A strict requirement for a separate osmoregulated form of DL-glycerol-3-phosphatase in the osmoadaptation of S. cerevisiae remains, however, to be established, since the other GPP1 encoded isoform is abundantly present, constituting about 0.25% of total cellular protein, independent of external osmolarity (Fig. 7D).

Recent evidence indicates that the signaling mechanisms behind the glycerol response are dependent on a MAP kinase module involving the PBS2 and HOG1 gene products (13). These kinases appear to be activated via membrane-localized osmosensors (10, 12). According to this model, changes in external osmolarity generate an internal signal that is transmitted to the MAP kinase cascade. One of the targets of this signaling is GPD1, encoding glycerol-3-phosphate dehydrogenase, the enzyme constituting the first step in the glycerol biosynthetic pathway (5, 6). The rapid induction of GPD1 expression after a step increase in salinity is blocked in pbs2³ or hog1 (6) signaling-defective mutants. It would seem logical that the osmoregulated Gpp2p enzyme would be controlled coordinately with Gpd1p via the HOG-MAP kinase pathway. Indeed, such a coordinated control appears to be exerted, as evidenced by the finding that the Gpp2p activity does increase insignificantly or only slightly in an osmosignaling-defective hog1 mutant that is cultured at elevated salinities (Fig. 9). In analogy with the osmostimulated expression of the GPD1 one would expect that the osmotic control is due to transcriptional induction of the GPP2 gene. Hirayama et al. (37) very recently reported the cloning of seven hyperosmolarity-responsive (HOR) genes from S. cerevisiae by a differential screening method. Two of these genes, HOR2 and HOR7, were of unknown function. When inspecting the sequences of these genes, we observed that the HOR2 gene is identical to GPP2. Interestingly, Northern blot analysis showed that the expression of HOR2/GPP2 is induced by increased osmolarity and that this induction is inhibited in a hog1 strain. Hirayama et al. (37) also noted that HOR2 has a close homologue called RHR2, which is not subject to osmotic regulation. Since RHR2 is identical to GPP1, the results of these authors confirm on the transcriptional level, the observed regulatory pattern of GPP1 and GPP2.