Rap1p-binding Sites in the Saccharomyces cerevisiae GPD1 Promoter Are Involved in Its Response to NaCl*

Mechanisms involved in transcriptional regulation of the osmotically controlled GPD1 gene inSaccharomyces cerevisiae were investigated by promoter analysis. The GPD1 gene encodes NAD+-dependent glycerol-3-phosphate dehydrogenase, a key enzyme in the production of the compatible solute glycerol. By analysis of promoter deletions, we identified a region at nucleotides −478 to −324, in relation to start of translation, to be of great importance for both basal activity and osmotic induction ofGPD1. Electrophoretic mobility shift and DNase I footprint analyses demonstrated protein binding to parts of this region that contain three consensus sequences for Rap1p (repressoractivator protein 1)-binding sites. Actual binding of Rap1p to this region was confirmed by demonstrating enhanced electrophoretic mobility of the protein-DNA complex with extracts containing an N-terminally truncated version of Rap1p. The detected Rap1p-DNA interactions were not affected by changes in the osmolarity of the growth medium. Specific inactivation of the Rap1p-binding sites by a C-to-A point mutation in the core of the consensus showed that this factor is a major determinant ofGPD1 expression since mutations in all three putative binding sites for Rap1p strongly hampered osmotic induction and drastically lowered basal activity. We also show that the Rap1p-binding sites appear functionally distinct; the most distal site (core of the consensus at position −386) exhibited the highest affinity for Rap1p and was strictly required for low salt induction (≤0.6 mNaCl), but not for the response at higher salinities (≥0.8m NaCl). This indicates that different molecular mechanisms might be operational for low and high salt responses of theGPD1 promoter.

Stress-activated signaling pathways in eukaryotic organisms are presently attracting much interest. The genetically tractable yeast Saccharomyces cerevisiae has proven particularly useful in identifying signal transduction components and in unraveling their function, which has generated a wealth of molecular information over the past few years (1)(2)(3). The HOG (high osmolarity glycerol response) mitogen-activated protein kinase pathway is a prominent signal transduction pathway responding to osmotic stress. The HOG1 and PBS2 genes, which code for a mitogen-activated protein kinase and its regulatory mitogen-activated protein kinase kinase, respectively, represent two central components of this pathway that were identified by analysis of yeast mutants sensitive to high salt concentrations (4). More recent work identified several upstream components of two distinct branches of the pathway affecting Hog1p phosphorylation (5)(6)(7)(8). Another signaling pathway of general importance in modulating various cellular activities, i.e. the cAMP-dependent protein kinase A pathway, influences the expression of some of the stress-regulated genes (1,9) by opposing the effects from signaling in the HOG pathway (10). There is, moreover, evidence for the existence of an osmostress-activated signaling pathway, involving calcineurin, that does not respond to general osmotic stress, but to high salt concentrations (11). In addition, phosphatidylinositol 3,5bisphosphate rapidly accumulates in yeast cells during hyperosmotic stress (12,13), suggesting the involvement of a so far uncharacterized phosphoinositide pathway in the yeast stress response. Clearly, yeast appears to coordinately activate various signal transduction pathways when confronted with osmotic challenges.
Central to the understanding of the overall cellular function of signaling pathways is the characterization of the regulatory elements of their target genes and the mechanism by which the transcription of these genes are controlled. However, only in a limited number of cases have the molecular details operational at the osmoregulated promoters been unraveled. It was recently shown that the ENA1 gene, encoding a P-type ATPase involved in the extrusion of Na ϩ from the yeast cytoplasm, is regulated by a derepression mechanism. The repressor that binds to the ENA1 promoter was identified as Sko1p (14), and the repression effect is dependent on the integrity of the Ssn6p-Tup1p corepressor complex, which appears to be relieved by a HOG1-dependent mechanism. This corepressor has also been implicated in the control of the HAL1 gene and a number of other stress-regulated promoters, indicating derepression as a more general molecular mechanism for stress induction (15). The pentanucleotide element CCCCT, being the core consensus of the stress-responsive element (STRE), 1 has been implicated in transcriptional activation of numerous genes during general stress conditions (1). This element responds to osmotic shock mediated via the HOG module. Two transcription factors, Msn2p and Msn4p, have been shown to bind to STRE and to be instrumental for STRE-activated transcription (16). The control of the Msn2p/4p-dependent gene activation by the HOG pathway appears to be influenced by the phosphorylation state of the transcription factors and their subsequent nuclear localization (17).
Elevated glycerol production is a prerequisite for the adaptation of S. cerevisiae to hyperosmotic stress, and several investigators have identified glycerol-3-phosphate dehydrogenase as the key enzyme in the glycerol synthesis pathway (18 -21). The principal mechanism for increasing glycerol production is increased expression of the GPD1 gene, and evidence has accumulated to indicate the GPD1 promoter as a target for the diverse signaling pathways responding to cellular dehydration (22). Salt-induced GPD1 expression is not dependent on a functional Sko1p, 2 Msn2p/4p (23), or functional STREs. 3 However, it was recently proposed that two putative transcription factors with slight sequence similarity to Gcr1p, an important regulatory factor of glycolytic gene expression, are involved in the hyperosmotic regulation of GPD1 (23). The most prominent of these two candidates, Hot1p (high osmolarity-induced transcription), was shown to be essential for full level response to salt stress; the salt-induced transcriptional response in a hot1⌬ mutant was ϳ40% of the wild-type level. Simultaneous deletion of MSN1, encoding the other Gcr1p sequence homolog, resulted in a further reduction in salt-induced GPD1 expression. However, irrespective of a combination of a number of gene knockouts involving also HOG1, MSN2, and MSN4, the GPD1 gene remained salt-responsive, although to a much decreased extent. Thus, although Hot1p and Msn1p are involved in the osmostress-mediated transcriptional activation of GPD1, additional mechanisms are apparently in operation.
In this work, we report experimental evidence for the repressor/activator protein Rap1p (25) as an important determinant of both the basal and salt-induced transcriptional activities of the GPD1 promoter. We also present evidence for the binding of Rap1p to neighboring binding sites that appear involved in mediating regulatory effects upon dehydration via different mechanisms, depending on the magnitude of stress. This is the first time that this multifunctional transcription factor is being implicated in the osmostress response. It is hypothesized, based on the importance of Rap1p in the Gcr1p-mediated induction of glycolytic genes, that Rap1p might interact directly or indirectly with Hot1p and/or Msn1p and thereby facilitate their binding and subsequent activation of the GPD1 promoter.
GPD1 Gene Promoter Constructs-The plasmids constructed and the synthetic oligonucleotides (PCR primers) utilized in this work are listed in Table I. Oligonucleotides were purchased from Scandinavian Gene Synthesis AB (Köping, Sweden) and used without further purification. All basic recombinant DNA techniques were performed according to standard procedure (31) if not otherwise stated. Restriction enzymedigested DNA was purified by agarose electrophoresis followed by ␤-agarase treatment (GELase, Epicentre Technologies Corp.; and agarase, Roche Molecular Biochemicals, Mannheim, Germany) of the excised agarose-embedded DNA and subsequent ethanol precipitation.
The following procedure was applied in the construction of the plasmid pO54 (32). A 1449-bp SspI/SspI GPD1 promoter fragment was isolated, ligated to a BamHI linker (CGGATCCG; Sigma catalog no. L6888), and inserted into the BamHI site of the integrative CAT reporter plasmid YIp5-32cat (33). This procedure yielded a GPD1 promoter-CAT reporter fusion (pGPD1-CAT) with the following junction sequence: Ϫ20(GPD1) CAAATCGGATCCGAGATTTTCAAGGAGCTAAGG-AAGCTAAAATGGAG ϩ6(CAT) (junction verified by sequencing; the GPD1 promoter is in boldface, the BamHI linker is underlined, and the CAT gene is in italic with the coding sequence indicated in boldface). Plasmid pPE110 (GPD1-(Ϫ687 to Ϫ16)) was constructed by inserting a 1.64-kb StuI/XbaI fragment from pO54, containing a 670-bp proximal GPD1 promoter fragment (nucleotides Ϫ687 to Ϫ16 upstream from the GPD1 translational start codon) and the CAT gene, into SmaI/XbaIcleaved pRS316 (26). The pPE110a plasmid is a pPE110 derivative with a XhoI/KpnI deletion in the multiple cloning site. pPE110b and pPE110c are pPE110a derivatives in which the GPD1 promoter fragment has been exchanged for the corresponding fragments from pPE103a and pPE110b, respectively (introducing unique KpnI sites at positions Ϫ581 and Ϫ483 in the promoter).
Plasmid pPE110R1 (GPD1-(Ϫ687 to Ϫ16) with a C-to-A exchange at nucleotide Ϫ386) was made by swapping the promoter fragment ClaI/ XmaI (nucleotides Ϫ687 to Ϫ378) from pPE110 with a PCR-generated fragment. Primer pair PCR9/PCR10 (utilizing pO54 as a substrate) was used to generate a 310-bp fragment that was digested with ClaI/XmaI and ligated into pPE110. Plasmid pPE110R3 (GPD1-(Ϫ687 to Ϫ16) with C-to-A exchanges at nucleotides Ϫ386, Ϫ371, and Ϫ358) was made using the same strategy. An XmaI/BamHI promoter fragment (nucleotides Ϫ378 to Ϫ16) in pPE110R1 was swapped with a PCR-generated fragment (primer pair PCR18/PCR3) digested with the same enzymes. This generates three-point mutations in the presumed Rap1p-binding sites. The full GPD1 promoter sequence of all constructs was checked by sequencing using the sequencing kit BigDye (Perkin-Elmer catalog no. 4303149), and the PCR-generated products were analyzed by the BM enheten (Enheten för biomolekylä r service, University of Lund, Lund, Sweden).
CAT Enzyme-linked Immunosorbent Immunoassay-The CAT protein amount was measured by an immunological assay (Roche Molecular Biochemicals catalog no. 1 363 727) according to the instructions given by the manufacturer. The assay has earlier been checked for quantitative reliability (34). The cell pellet (ϳ1 ϫ 10 8 cells) was chilled on ice, washed once with ice-cold 0.1 M Tris (pH 7.8), resuspended in 500 l of lysis buffer (0.4 M MOPS and 1% Triton X-100 (pH 6.5)) supplied with the CAT enzyme-linked immunosorbent immunoassay kit, and then disrupted by vortexing with 0.4 g of glass beads (0.5-mm diameter) for 4 ϫ 30 s with intermediate incubations on ice for at least 1 min. The cell extract was centrifuged at 13,200 ϫ g for 5 min at 4°C, and the supernatant was frozen at Ϫ20°C until analyzed. The CAT data obtained were divided by the total protein concentration in each sample as determined by a Lowry-based protein assay kit (Sigma catalog no. 5656) using bovine serum albumin as the standard and indicated as parts/million.
Electrophoretic Mobility Shift Assays (EMSAs)-Cells from a 0.5-liter culture (A 610 Ϸ 1) were harvested by centrifugation (2600 ϫ g, 10 min, 4°C) and washed once with 15 ml of H-buffer (200 mM Tris-HCl (pH 8.0), 10% (w/v) glycerol, 10 mM MgCl 2 , and 1 mM dithiothreitol). The cells were resuspended in 1 ml of H-buffer containing protease inhibitors (35 g/ml phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, and 1.4 g/ml leupeptin). The cell suspension was transferred to 1.5-ml microcentrifuge tubes; and after centrifugation (20,000 ϫ g, 3 min, 4°C), the cell pellets were stored in Ϫ20°C. Cells were thawed at 0°C and mixed with 1 ml of H-buffer (supplemented with protease inhibitors) and 1 g of acid-washed glass beads (0.5-mm diameter) in precooled homogenization tubes. Homogenization of the cells was performed by vortexing in a Vibrogen-Zellmü hle (Edmü nd Bü hler, Tü bingen-Weilheym, Germany) for 10 min at 4°C until Ն95% of the cells were broken (checked by 10% (w/v) nigrosin staining). The homogenates were transferred to 1.5-ml vials and centrifuged (20,000 ϫ g, 5 min, 4°C). The supernatant was divided into aliquots and frozen in liquid nitrogen. The protein concentration was determined using the Bio-Rad D c protein assay with bovine serum albumin as the standard.
The DNA fragments used as specific competitors and radioactive probes in EMSA were isolated from the cloning vectors by restriction enzyme cleavage and agarose gel electrophoresis. The agarose-embedded DNA was cut out from the gel and purified by ␤-agarase treatment as described above. DNA probes were prepared by a fill-in reaction using the Klenow fragment of DNA polymerase in the presence of [␣-32 P]dATP (Amersham Pharmacia Biotech catalog no. AA0004, Buckinghamshire, United Kingdom). Unincorporated nucleotides were removed by purification with the Wizard DNA clean-up system (Promega), and the eluted DNAs were ethanol-precipitated. The activity of the probe was ϳ10 5 cpm (Cerenkov)/ng in a typical case. The binding reaction contained, in a final volume of 10 l, the following components: 0.5-1 ϫ 10 5 cpm of DNA probe, 6 g of poly(dI-dC) (Roche Molecular Biochemicals), 10 -20 g of crude whole cell protein extracts, a 100 -400-fold higher molar concentration of unlabeled competitor DNA where indicated, and binding buffer (20 mM Hepes (pH 7.9), 125 mM KCl, 0.5 mM EDTA, 12% glycerol, and 1 mM dithiothreitol). The binding reaction was gently mixed and incubated for 15 min in 30°C before loading on a 4% polyacrylamide gel (30:0.8 acrylamide/bisacrylamide, prerun at 8 V/cm for 1 h at 4°C) in Tris/glycine buffer (50 mM Tris (pH 8.6), 100 mM glycine, and 2 mM EDTA) and run at 13 V/cm for ϳ2 h at 25°C. The gels were dried, and PhosphorImager plates were exposed and subsequently scanned in a Molecular Dynamics PhosphorImager. The images were processed with the aid of computer software from Adobe Systems Inc. (Photoshop Version 2.0.1) and Deneba Systems Inc. (Canvas Version 3.5).
DNase I Footprint Experiments-For the DNase I footprint experiment, plasmid pPE130 (pGPD1-(Ϫ478 to Ϫ16)) was cleaved with HindIII, labeled (Klenow fill-in reaction) with [␣-32 P]dATP, purified with the Wizard DNA clean-up system, cleaved with BamHI, and ethanol-precipitated, and the resulting fragments were electrophoretically separated on a 1.5% agarose gel. The fragment of interest was electrophoresed into a DEAE membrane (Schleicher & Schü ll) and eluted according to a standard procedure (31). 1 ϫ 10 4 cpm (Cerenkov) of the probe were used for each binding reaction and for a Maxam-Gilbert sequencing reaction (31). The probe was mixed with binding buffer (supplemented with 2 mM MgCl 2 and 10 mM CaCl 2 ), 4 g of poly(dI-dC), and 20 g of protein extract (except for the control sample) to a final volume of 20 l. The mixture was incubated for 15 min at 30°C before the addition of 2 l of DNase I (5120 units/ml; 2560 units/ml for the control sample without protein). After a 120-s incubation at 30°C, the reaction was stopped by adding 2 l of 0.5 M EDTA and 2 g of poly(dI-dC), and the samples were put on ice. A 1-min extraction with a phenol/chloroform/isoamyl alcohol mixture (25:24:1; v/v) followed; the DNA was ethanol-precipitated twice; and the pellet was washed once with 70% ethanol. The DNA was vigorously resuspended in 3 l of water and incubated for 5 min at 60°C before 3 l of formamide dye solution were added; and the samples were frozen for later use. A 6% denaturing polyacrylamide gel (19:1 acrylamide/bisacrylamide) was prepared and prerun for 1 h, and the samples were applied after a 5-min denaturation step at 90°C in a water bath. The gels were run at 90 watts for ϳ2 h at ϳ50°C, dried, and exposed to a PhosphorImager plate for 11 days. After scanning of the plate in the PhosphorImager, the image was analyzed by the following computer software: PDQuest (ball background subtraction, Protein Data Bases Inc.), Photoshop Version 2.0.1, and Canvas Version 3.5.

RESULTS
Promoter Regions of Importance in the Osmotic Induction of the GPD1 Gene-Transcription of the GPD1 gene is enhanced by increased osmolarity, instigated by either NaCl or sorbitol additions, as shown previously either by Northern analysis or by the use of a GPD1 promoter-CAT reporter fusion (22,32). To identify the element(s) of the GPD1 promoter responsible for the osmotically controlled transcriptional activation, a series of 5Ј-promoter deletions were constructed starting at nucleotide Ϫ687 in relation to the start of translation. The promoter constructs were fused to a CAT reporter on a YCp vector (pPE110 -114) ( Table I) and introduced into strains YPH499 and W303-1A. This reporter construct reflects well the salt response of the normal chromosomal GPD1 gene 3 and thus appears to contain all relevant regulatory sequences. Transformants were cultured to the mid-exponential growth phase at different salinities (0, 0.5, and 1.0 M NaCl), and the resulting amount of CAT protein was measured in crude protein extracts by an immunoassay (Fig. 1). Extensive 5Ј-deletions of the promoter only slightly affected the transcriptional activity during growth in basal medium; ϳ75% of the GPD1 promoter activity remained (7.1 to 5.2 ppm) even for the deletion to position Ϫ322 (construct pPE112). Cells growing exponentially in 1 M NaCl medium exhibited for the full-length promoter (pPE110, nucleotides Ϫ687 to Ϫ16) almost 3-fold higher levels of transcriptional activity than when cultured without salt addition, and ϳ50% of this salt-induced level remained for the most extensive deletion (19.0 to 9.0 ppm). The greatest reduction in saltinduced levels was observed for deletions beyond position Ϫ478; however, even the promoter fragment containing only the most proximal 322 nucleotides still displayed osmotic stimulation. Thus, these 5Ј-deletion studies indicate that the GPD1 promoter does not contain one exclusive element involved in the osmotic responsiveness.
To further substantiate which regions are implicated in the salt-induced regulation of the promoter, internal deletions were made (pPE115-120) (Table I), and the transcriptional activity was again analyzed during exponential growth (Fig. 1). It was apparent that constructs harboring internal deletions exhibited more severe promoter activity defects than the 5Јdeletions. For example 5Ј-deletion down to nucleotide Ϫ322 only slightly reduced the basal activity (75% remaining), whereas the constructs encompassing internal deletions of the region proximal to nucleotide Ϫ478 displayed at the most only ϳ20% of the activity of the full-length promoter. It was also evident that deletions of region Ϫ377 to Ϫ322 strongly affected the general level of activity of the promoter while still responding to an osmotic upshift, in particular to low salinity (0.5 M NaCl). This was also seen for the more distal region Ϫ482 to Ϫ378, although the response to salt for this construct was less significant. Most strikingly, when both these regions (pPE118, nucleotides Ϫ483 to Ϫ322) were deleted, the GPD1 promoter appeared almost completely inactive under any growth condition. It was also notable that more extensive deletions, nucleotides Ϫ580 to Ϫ323 or Ϫ687 to Ϫ323, regained some of the activity for the promoter. It thus appears as if the region upstream from nucleotide Ϫ478 might harbor repressor(s), whereas region Ϫ478 to Ϫ324 might be the site for binding of protein(s) acting as a repressor of the upstream repressor(s). However, we cannot exclude position effects by bringing the vector DNA closer to the transcriptional initiation site in these deletion constructs.
Protein-DNA Interactions at the GPD1 Promoter-The promoter region spanning nucleotides Ϫ687 to Ϫ16 was analyzed for protein-DNA interactions utilizing different parts of the GPD1 promoter in an EMSA. Protein extracts from the two strains YPH499 and W303-1A were tested for all indicated probes, yielding indistinguishable results. Three major interactions were detected along the examined promoter region (Fig.  2), and all these protein-DNA interactions were clearly specific since they could be competed out with an excess of nonradioactive competitor fragments covering the site of interaction. The fragment encompassing nucleotides Ϫ687 to Ϫ478 (probe I) revealed one interaction (designated I*) ( Fig. 2A). This binding could be out-competed with an excess of unlabeled fragment covering the same range of the promoter, but not by a fragment spanning nucleotides Ϫ577 to Ϫ377 (Fig. 2A, lane 3), thus localizing this interaction to the region between nucleotides Ϫ687 and Ϫ578. The strongest protein-DNA interaction observed along the GPD1 promoter was located between positions Ϫ478 and Ϫ378 (designated II*) (Fig. 2B). A less prominent interaction (designated III*) (Fig. 2C) was located at positions Ϫ377 to Ϫ324. This latter localization was accomplished using  probe Ϫ377 to Ϫ16 (probe III) in combination with competition with either fragment Ϫ377 to Ϫ16 (competitor III) or fragment Ϫ324 to Ϫ16 (competitor V) (Fig. 2C, lanes 4 and 6, respectively), where the latter competitor could not rival the III* interaction. However, a competitor DNA fragment containing sequence Ϫ478 to Ϫ377 (competitor II) totally abolished this III* interaction, indicating overlapping DNA element specificity for the protein(s) binding to probes II and III (Fig. 2C, compare  lanes 2 and 7). These two sites, apparently harboring binding potential for the same protein(s), exhibited clearly different binding affinities for this factor. Thus, the region of the GPD1 promoter exhibiting the greatest impact on both basal and salt stress-induced promoter activities, region Ϫ478 to Ϫ324 (Fig.  1), exhibited two strong protein-DNA interactions, potentially resulting from binding of the same protein(s).

5Ј-ggt ggtacc GTTTTTCCGATGTAGAAGTAG-3Ј
No Salt-dependent Binding to the GPD1 Promoter Can Be Detected-The detection of protein-DNA interactions prompted gel shift investigations for any salt-dependent complex. Cells were grown at different salinities (0, 0.35, 0.7, 1.0, and 1.4 M NaCl) before harvest and protein extract preparation. However, no salt stress-specific protein-DNA interactions were detected by the EMSA analysis, even when utilizing an extract from cells grown in 1.4 M NaCl medium (data not shown). This was established using probe Ϫ478 to Ϫ323 (probe VI) (Fig. 2), encompassing binding sites for two of the factors mentioned above (forming complexes II* and III*), or using probe Ϫ687 to Ϫ478 (probe I) (Fig. 2), revealing complex I* formation within region Ϫ687 to Ϫ577.
Identification of Rap1p as the Major Binding Activity-To locate more precisely the sites of the most proximal protein-DNA interactions detected by the EMSA analysis, region Ϫ478 to Ϫ16 of the GPD1 promoter was analyzed in a DNase I footprint assay. The only interaction that was experimentally observed was positioned between nucleotides Ϫ394 and Ϫ374 (Fig. 3). This fragment contains a sequence element well in agreement with the complement of the published Rap1p-binding consensus sequence (5Ј-(A/G)T(A/G)CACCCANNC(C/A)CC-3Ј) (35). To establish whether the binding activity of the protein detected in the footprint assay varied depending on the osmotic stress situation, extracts from cells grown in media with different salinities were analyzed. A slight difference in the 5Јportion of the resulting footprint could be seen (Fig. 3) when comparing extracts from cultures grown in 0, 0.7, and 1.4 M NaCl. However, this difference could not be verified in later experiments. To exclude that the detected footprint was caused by a Rap1p-related protein with the same binding specificity, we utilized protein extracts from a strain with the wild-type RAP1 gene exchanged for a truncated gene. This curtailing results in a functional Rap1p with a 230-amino acid deletion in the nonessential N terminus (⌬230-Rap1p) (28). When using extracts from the ⌬230-Rap1p strain, the major II* interaction between nucleotides Ϫ478 and Ϫ323 was shifted to a faster mobility (Fig. 4A). Similarly, the mobility of the complex formed from the interaction in region Ϫ377 to Ϫ16 was altered (Fig. 4B). These results clearly indicate that the protein-DNA interactions recorded at nucleotides Ϫ394 to Ϫ374 (interaction II*) and Ϫ377 to Ϫ324 (interaction III*) both involve binding of Rap1p. Interaction I* between nucleotides Ϫ687 to Ϫ577 was, however, not affected in the ⌬230-Rap1p strain (Fig. 4C), proving this protein-DNA interaction to be distinct from the other two more proximal interactions and not involving Rap1p. This distal factor will here be designated GUP for GPD1 promoter upstream binding protein(s).
The Rap1p-binding Sites are Functional in Vivo and Are Differentially Involved in Low and High Salt Response Mechanisms-To provide in vivo evidence for the functionality of the high affinity Rap1p interaction (interaction II*) between positions Ϫ394 and Ϫ374 (indicated as RAP1a in Fig. 5A), one essential nucleotide in the center of the Rap1p-binding consensus sequence, C Ϫ386 , was exchanged for an A (Fig. 5A). This nucleotide at the core of the consensus sequence is vital for Rap1p binding (35). In accordance, the in vitro Rap1p interaction with the mutated binding site, RAP1a (C386A), was abolished (Fig. 5B). In initial in vivo experiments, it was evident that the GPD1 promoter containing the mutated RAP1a (C386A) site was responding to high (but not low) concentrations of NaCl. Therefore, to more precisely locate the NaCl threshold value for this RAP1a site-dependent induction of GPD1, the effect of stepwise increase in salinity was monitored (Fig. 5C). It was evident that the wild-type promoter responded almost linearly to increases in NaCl over the whole range of concentrations tested, whereas the GPD1 promoter containing the mutated RAP1a (C386A) site exhibited no salt-enhanced activity until Ն0.8 M NaCl. Apparently, there is a clear mechanistic threshold between 0.6 and 0.8 M NaCl, and a functional Rap1p site positioned around nucleotide Ϫ386 is a prerequisite for the stress control of the promoter at low salt concentrations.
The promoter fragment containing the mutated RAP1a site never reached wild-type levels of induction at higher salinities; however, the level of induction (ϳ4-fold during growth in 1.4 M NaCl medium) was of roughly the same magnitude as for the wild-type promoter (Fig. 5C).
The exact position of the weak Rap1p site between nucleotides Ϫ377 and Ϫ324 (interaction III*) remains to be determined since there are two theoretical locations (indicated as RAP1b and RAP1c in Fig. 5A). The in vivo importance of these more proximal binding sites (RAP1b and RAP1c) was thus examined in combination by totally abolishing Rap1p binding by C-to-A mutations in all three putative core regions (RAP1a (C386A), RAP1b (C371A), and RAP1c (C358A)) (Fig. 5A). A functional role was confirmed by the fact that the triple binding site-mutated GPD1 promoter exhibited low basal activity and only slight salt induction, even during growth in 1.4 M NaCl medium (Fig. 5C). Apparently, binding of Rap1p to the RAP1b and/or RAP1c sites is a prerequisite for induction at high salinity, at least in the construct where binding to the strong RAP1a site is abolished.
The salt-induced response of the GPD1 promoter at salinities below 0.6 M NaCl is exclusively dependent on the most distal and strongest Rap1p-binding site. However, at higher salinities, one (or both) of the weaker and more proximal Rap1pbinding sites can partially compensate for the loss of the strong RAP1a site. Furthermore, the data support the existence of a promoter were used as probe ( 32 P-labeled). A, probe I, nucleotides Ϫ687 to Ϫ478; B, probe II, nucleotides Ϫ478 to Ϫ378; C, probe III, nucleotides Ϫ377 to Ϫ16 (nucleotide positions in relation to the GPD1 translational start site). The different competitor DNA fragments are indicated at the top: competitor I, nucleotides Ϫ687 to Ϫ483; competitor II, nucleotides Ϫ478 to Ϫ378; competitor III, nucleotides Ϫ377 to Ϫ16; competitor IV, nucleotides Ϫ576 to Ϫ378; competitor V, nucleotides Ϫ324 to Ϫ16; and competitor VI, nucleotides Ϫ478 to Ϫ323. The strongest protein-DNA interaction for each probe is indicated by I*, II*, and III*. Weaker interactions are indicated only by arrows, and FP denotes free probe. The addition of specific unlabeled competitor (comp.) DNA (probe number and amount, as x-fold excess compared with the labeled probe, are indicated in each panel) allowed better positional mapping of the interactions and, in addition, also indicated the sequence specificity of the complex. Protein extracts (10 -20 g) from exponentially growing cells were applied as indicated. Extracts from either strain YPH499 or W303-1A or from cells grown in different growth medium (yeast nitrogen base or yeast extract/peptone) were investigated, yielding identical results. threshold value in the range 0.6 -0.8 M NaCl that distinguishes two different induction mechanisms at this promoter during saline growth. DISCUSSION Complex Regulation of the GPD1 Promoter-Insights into the mechanisms involved in the control of the GPD1 promoter will provide a major key to our general understanding of hyperosmotic stress-induced gene regulation. Mechanisms in operation at this promoter have the potential of being dehydration-specific since other types of stresses such as heat, starvation, etc., that strikingly affect the expression of many other osmostress-controlled genes like CTT1, HSP12, and HSP104 (10, 36) do not strongly influence the expression of GPD1 (37,38). 3 Here we provide evidence that a number of different promoter elements are involved in the transcriptional regulation of GPD1 during saline conditions, our main contri-bution being the identification and functional characterization of Rap1p-binding sites. Rap1p binds to at least two sites in the region Ϫ400 to Ϫ350 nucleotides upstream from the translational start site, and functional Rap1p-binding sites appear to be a prerequisite for proper osmotic control of GPD1 transcription.
We also provide data regarding the importance of some other regions of the promoter under basal growth conditions and high salinity stress. The presence of repressor element(s) whose function is counteracted by binding of Rap1p to the promoter is suggested by the finding that internal deletion of nucleotides Ϫ478 to Ϫ322 almost completely abolished promoter activity under any growth condition, whereas a 5Ј-deletion to position Ϫ322 only marginally influenced the promoter activity. Further support that Rap1p binding per se might negatively influence the activity of the putative repressor elements comes from the observation that the single and triple Rap1p site mutations give a low basal activity of the promoter (Fig. 5C). Apparently, Rap1p will repress the activity of these elements when bound to the promoter. It has been suggested that a number of osmostress-induced genes, including GPD1, are controlled by an SSN6/TUP1-mediated repression mechanism that is lifted in the presence of osmotic stress (15). This mechanism was shown to be operational for the HAL1 gene through a defined upstream repressing sequence element (15), whereas for the ENA1 gene, it was demonstrated that Sko1p apparently tethers the Ssn6p-Tup1p corepressor to the promoter (14). The GPD1 promoter does not appear to be controlled by such a salt-induced derepression mechanism since none of the 5Ј-or internal promoter deletions exhibited any substantially increased basal activity. This lack of evidence for salt-induced derepression of the GPD1 promoter via Ssn6p-Tup1p is at variance with earlier results (15), but in accordance with a more recent report (23).
Proximal to the Rap1p-binding sites are three elements identical to the pentanucleotide STRE, having the central C of the core sequence CCCCT positioned at nucleotides Ϫ328, Ϫ284, and Ϫ32. STREs are found in many promoters and have been shown to be required for the general stress regulation of a number of genes (1). It was recently demonstrated by the use of GPD1 promoter constructs in which all three STREs had been The reversed Rap1p-binding consensus sequence is boxed, and the location is given by comparing the footprints with the Maxam-Gilbert G ϩ A promoter sequence. Protein addition to the binding mixture is indicated as ϪP (no protein extract added) and ϩP (20 g of protein extract added), and the concentrations of NaCl (0, 0.7, and 1.4 M) used to supplement the growth medium are indicated above lanes 2-4.
FIG. 4. Identification of Rap1p as the protein responsible for the formation of complexes II* and III* in the GPD1 promoter, but not of complex I* (see Fig. 2). EMSA was carried out using an extract from mutant cells with truncated Rap1p that is deleted of its most N-terminal 230 amino acids, YLS91 (Rt), which was compared with the corresponding wild-type full-length Rap1p, YDS2 (wt). A, probe VI, nucleotides Ϫ477 to Ϫ323; B, probe III, nucleotides Ϫ377 to Ϫ16; C, probe I, nucleotides Ϫ687 to Ϫ478 (see Fig. 2). Rap1p and ⌬Rap1p indicate the Rap1p-DNA complexes with the wild type and the truncated mutant, respectively. GUP stands for GPD1 promoter upstream binding protein.
For further explanations, see legend to Fig. 2. comp., competitor; FP, free probe. inactivated by mutations 3 that the response of the promoter to either salt stress or heat treatment was STRE-independent. Apparently, the presence of STREs is not a strict requirement for stress-induced expression, as is further supported by the lack of STREs in the salt-responsive DAK1 gene (9).
Rap1p Acts as a Major Activator in the Transcription of GPD1-We have identified the multifunctional DNA-binding protein Rap1p as an important activator of GPD1 expression under basal and saline growth conditions and shown that at least two nearby Rap1p-binding sites are occupied in vitro and are functional in vivo. The most distal site (core of consensus sequence at position Ϫ386) exhibited the highest affinity for Rap1p. The exact position of the weak Rap1p site between nucleotides Ϫ377 and Ϫ324 remains to be determined since there are two theoretical locations (indicated as RAP1b and RAP1c in Fig. 5A). However, the relative proximity of the RAP1b site to the strong RAP1a site makes RAP1b a less likely candidate. The distance between the central parts of the two core consensus sequences is only 14 nucleotides, which is in the range of the published minimum distance (39). Based on recent studies of the crystal structure of the DNA-binding domain of Rap1p in complex with telomeric DNA (40), it was concluded that the protein contains two similar DNA-binding domains recognizing a tandemly repeated DNA sequence. The repeated sequence in telomeric DNA responsible for the high affinity binding of Rap1p is ACACC with the intermediate sequence CAC. When comparing these recognition sequences with the nucleotide sequence of the three theoretical Rap1p-binding sites of the GPD1 promoter, it is clear that the RAP1a site has the best homology, followed by RAP1c. The RAP1b site almost completely lacks the second ACACC motif. The interaction(s) identified as Rap1p binding to site b or c were efficiently outcompeted by fragments containing the RAP1a site (Fig. 2C,  compare lane 7 with lane 3). This is in agreement with the finding that the RAP1a site is the most conserved in relation to the consensus sequence and thus promotes, at least in vitro, a stronger binding than the RAP1b or RAP1c site.
Rap1p is an essential and ubiquitous DNA-binding protein involved in a wide variety of cellular activities such as (i) silencing of mating-type genes, (ii) ensuring telomere function and structure, (iii) stimulating meiotic recombination, (iv) binding to the nuclear scaffold protein complex, and (v) regulating the transcription of an array of genes (25). Rap1p has been experimentally linked to the transcriptional regulation of genes involved in different cellular contexts (25), and homologies to a binding site consensus sequence have been found in Ͼ100 yeast promoters (35). Among the experimentally well verified examples are genes coding for glycolytic enzymes, where a heteromeric complex consisting of the glycolytic regulator Gcr1p and Rap1p is required to mediate an efficient transcriptional activation. Gcr1p was shown to bind a so-called CT box found in several promoters of glycolytic genes (41), and it was later concluded that the Gcr1p-mediated activity of the glycolytic ADH1 promoter was dependent on a functional Rap1p-binding upstream activating sequence element (42). The conclusion drawn from these experiments was that Rap1p increases the possibility of Gcr1p-specific transcriptional activation by physical interaction, thereby attracting the regulator to the promoter.
Since functional Rap1p-binding sites in promoters are often found close to additional elements that control transcription in a specific manner, one might hypothesize that Rap1p recruits osmoregulatory factors in a GPD1 promoter context. The ability of Rap1p to bind and perturb the chromatin structure of promoter regions has been proposed to be a prerequisite for binding of other transcription factors with low element affinity (43). Using a two-hybrid screen, Rep et al. (23) identified a protein named Hot1p as interacting with the mitogen-activated protein kinase, Hog1p. Hot1p displays sequence similarity to the DNA-binding C-terminal domain of Gcr1p, as well as to Msn1p, and deletion of either HOT1 or MSN1 reduced the magnitude of induction of GPD1 by salt stress. Since Gcr1p is dependent on Rap1p for proper control of certain glycolytic genes (44), the Hot1p and Msn1p members of the Gcr1p family might depend on binding of this general transcription factor in their partial control of the GPD1 gene.
Distinct Low and High Salt Mechanisms-This report provides the first documentation of a role for Rap1p in the regulation of a gene responsive to osmotic stress. In addition, we also demonstrated that the binding sites for Rap1p appear functionally distinct since a mutation in the distal high affinity FIG. 5. Impact from Rap1p-binding site mutations on the salinity response of the GPD1 promoter. A, base shift mutations (C to A) introduced in the central triple C core of the three putative binding sites for Rap1p. The nucleotides in the Rap1p-binding site core consensus are shaded. B, gel shift analysis with probe Ϫ687 to Ϫ377 with a wild-type Rap1p-binding site (WT) and the same fragment with a mutated RAP1a site (C386A). C, in vivo consequence of mutations at the putative Rap1p-binding sites. Strain YPH499 was transformed with GPD1 promoter-CAT constructs starting at position Ϫ687 with the indicated internal promoter point mutations in the putative Rap1pbinding sites: wild-type promoter pPE110 (nucleotides Ϫ687 to Ϫ16) (white bars); single point mutation at the strongest Rap1p-binding site, pPE110R1 (C386A) (gray bars); and point mutations at all three putative Rap1p-binding sites, pPE110R3 (C386A, C371A, C358A) (black bars). CAT protein was measured on an extract from exponentially growing cells in minimal medium (yeast nitrogen base) with the indicated amounts of NaCl. The amount of CAT protein was recorded by an immunoassay and normalized to the total amount of protein in the extract (expressed as parts/million). Data represent means Ϯ S.D. (error bars) of three independent experiments. site leads to a nonresponsive promoter at low salinity while leaving the high salinity response more or less intact. Indications of distinct low and high salt mechanisms for gene activation are also apparent from studies on the differential induction of ALD2 (low salt) and DDR48 (high salt) (45). The protein phosphatase calcineurin has been implicated in high salt activation of ENA1 (11). However, it is less likely that this signaling pathway is responsible for the observed high salt response of the GPD1 promoter since only marginal effects on the salt induction of normal chromosomal GPD1 were apparent in the calcineurin-deficient strain cna1⌬cna2⌬ (46). The high salt response mechanism appeared in the present study to require a threshold salinity in the range 0.6 -0.8 M NaCl to be implemented (Fig. 5C). Several published experiments on salinityinduced gene expression have been conducted at a salinity of 0.7 M NaCl (23,24,(47)(48)(49), thus potentially examining the combined effect of two different molecular mechanisms. It appears important to separate these effects for more conclusive future mechanistic investigations. Further experiments should address this issue and the way Rap1p allows for discrimination between a low and high salinity response.