Expression of the Yeast PIS1 Gene Requires Multiple Regulatory Elements Including a Rox1p Binding Site*

The PIS1 gene is required for de novo synthesis of phosphatidylinositol (PI), an essential phospholipid in Saccharomyces cerevisiae. PIS1 gene expression is unusual because it is uncoupled from the other phospholipid biosynthetic genes, which are regulated in response to inositol and choline. Relatively little is known about regulation of transcription of the PIS1 gene. We reported previously that PIS1 transcription is sensitive to carbon source. To further our understanding of the regulation of PIS1 transcription, we carried out a promoter deletion analysis that identified three regions required for PIS1 gene expression (upstream activating sequence (UAS) elements 1-3). Deletion of either UAS1 or UAS2 resulted in an ∼45% reduction in expression, whereas removal of UAS3 yielded an 84% decrease in expression. A comparison of promoters among several Saccharomyces species shows that these sequences are highly conserved. Curiously, the UAS3 element region (-149 to -138) includes a Rox1p binding site. Rox1p is a repressor of hypoxic genes under aerobic growth conditions. Consistent with this, we have found that expression of a PIS1-cat reporter was repressed under aerobic conditions, and this repression was dependent on both Rox1p and its binding site. Furthermore, PI levels were elevated under anaerobic conditions. This is the first evidence that PI levels are affected by regulation of PIS1 transcription.

In Saccharomyces cerevisiae, the PIS1 gene encodes the membrane-associated PI synthase, which combines CDP-diacylglycerol (DAG) and inositol to synthesize PI (7,17) (Fig. 1). The yeast PIS1 locus was originally defined by a mutant, pis, that required high levels of inositol (100 M) for growth because of a decreased affinity of PI synthase for inositol (18,19). This mutant phenotype was utilized to clone the yeast PIS1 gene (19), as well as the rat (20) and Arabidopsis (21) homologues. The PIS1 gene has been identified in many organisms, including Toxoplasma gondii (22), rat (20), Arabidopsis (21), Mycobacterium (6), Schizosaccharomyces pombe, and humans (23). PI synthase activity has also been characterized in Candida albicans (24) and several additional mammalian species (5). Disruption of the yeast PIS1 gene is lethal (7), demonstrating that PI synthesis is an essential process.
Yeast cells do not extensively regulate PIS1 expression or its product (25,26). Furthermore, PIS1 gene expression is not co-regulated with the expression of other phospholipid biosynthetic genes in response to inositol and choline (1)(2)(3)(4)25). However, PIS1 gene expression is regulated in response to different carbon sources where PIS1 expression is elevated in fermentable carbon sources (galactose and glucose) relative to nonfermentable carbon sources (glycerol) (25). Although PIS1 gene expression is uncoupled from the inositol choline response, PI synthase activity is affected by inositol levels. Increased levels of inositol have been shown to increase the rate of PI synthesis (27). The K m of PI synthase for inositol (0.21 mM) is 9-fold greater than the intracellular concentration of inositol (24 M). Additionally, inositol is a noncompetitive inhibitor of phosphatidylserine (PS) synthase (CHO1 gene product), the first enzyme in the phosphatidylcholine (PC) biosynthetic branch (Fig. 1). Therefore, when cells are supplemented with inositol, the amount of PI in cell membranes doubles at the expense of PC synthesis (27).
Hypoxic regulation of gene expression has been studied extensively in many eukaryotes including yeast, Drosophila, and mammals (28 -30). In yeast, the mechanism for regulation has been shown to be predominantly controlled by Rox1p (28,(31)(32)(33)(34)(35). In aerobic environments, ROX1 transcription is induced by a Hap1p heme complex. Rox1p then recruits the Ssn6-Tup1 repression complex and represses transcription of anaerobic (hypoxic) genes such as ANB1 (36). In anaerobic environments, intracellular heme synthesis decreases and, as a result, Hap1p complexes with additional proteins to repress transcription of ROX1 (37,38). Rox1p levels decrease, and anaerobic (hypoxic) genes are derepressed. More recently, the MOT3 and MOX genes have been shown to be regulators of the hypoxic response (31,39,40). PIS1 expression is induced 2-fold under anaerobic conditions and in a rox1⌬ mutant strain grown in chemostat or batch cultures (41)(42)(43). In this study, we identified three upstream activating sequence (UAS) elements in the PIS1 promoter and found that within the UAS3 element is a binding site of Rox1p. We also show that PIS1 gene expression is induced in anaerobic conditions through the Rox1p binding site. Anaerobic growth conditions were also found to influence phospholipid composition.
Plasmid Construction-A nested set of PIS1 promoter deletions fused to the cat reporter gene was created by PCR using appropriate oligonucleotides (Table I). The 5Ј-terminal deletion PCRs utilized the 3Ј primer: PIS1-3Ј Ϫ1 along with the 5Ј primer series PIS1 Ϫ918 to Ϫ127 (Table I). The individual PCR products were cloned into pGEM-T (Promega, Madison, WI). PIS1 promoter fragments were excised from pGEM-T by digestion with BamHI and BglII and inserted into pBM2015 (48). Fragments from appropriate pGEM-T derivatives were combined into pBM2015 to create PIS1 promoter internal deletions. For each plasmid, the name indicates the deletion endpoints. A PIS1-cat derivative containing a mutant Rox1p binding site was created by site-directed mutagenesis of pPIS1 Ϫ325 (to create pPIS1 Ϫ325rox⌬) using a QuikChange XL site-directed mutagenesis kit (Stratagene, Cedar Creek, TX) and primers PIS1sd Ϫ162 and Ϫ120. Specifically, the Rox1p binding site (PIS1 promoter sequences Ϫ149 to Ϫ132) was changed from 5Ј-GCCCCTCCTATTGTTTTT-3Ј to 5Ј-GCCCCTCCTC-GCGTTTTT-3Ј. The presence of the mutation was confirmed by DNA sequencing. An ANB1 promoter cat reporter gene fusion was created by PCR using primers ANB1 Ϫ1000 and Ϫ1. Yeast strains containing the promoter-cat reporters stably integrated at the GAL4 locus were created by transformation, as described previously (49).
Plasmid pANB1 was used as a control for Rox1p binding in electrophoretic mobility shift assay (EMSA) experiments. This pGEM-T derivative was constructed by cloning 150 bp of the ANB1 promoter Ϫ321 to Ϫ171 amplified by using primers ANB1 Ϫ321 and Ϫ171. Plasmid pPIS1-bend was generated for use in a circular permutation assay. This plasmid was constructed by cloning a tandem direct repeat of PIS1 promoter sequences from Ϫ247 to Ϫ39 using primers PIS1 Ϫ247 and Ϫ39.
Plasmid pGEX-KG-ROX1 was created for the purpose of over-expressing and purifying Rox1p. The ROX1 ORF was amplified by PCR using the ROX1 ORF 5Ј and 3Ј primers. Plasmid pGEX-KG (Amersham Biosciences) was linearized with BamHI and EcoRI, and a 1101-bp ROX1 fragment was cloned in-frame with glutathione S-transferase (GST).
CAT Assays-Five-milliliter cultures were grown to 60 -80 Klett units in appropriate media. Assays were conducted as described previously (49). Units of CAT activity were defined as counts/min measured in the organic phase and expressed as a percentage of the total counts/ min (percent conversion) divided by the amount of protein assayed (in micrograms) and the time of incubation (in hours).
RNA Ligase-mediated Rapid Amplification of cDNA Ends (RLM-RACE)-The 5Ј end of the PIS1 transcript was mapped using the FirstChoice RLM-RACE kit (Ambion). RNA was isolated from yeast by a glass bead disruption and hot phenol extraction procedure (50). Total RNA was treated with calf intestinal phosphatase to remove the 5Јphosphate from non-full-length uncapped RNA. The 5Ј cap was removed from full-length mRNA by treatment with tobacco acid pyrophosphatase (TAP), leaving a 5Ј-monophosphate to which a 5Ј RACE adapter oligonucleotide (5Ј-GCUGAUGGCGAUGAAUGAACACUGCG-UUUGCUGGCUUUGAUGAA-3Ј) was ligated using T4 RNA ligase. Random-primed reverse transcription and nested PCR were then used to amplify the 5Ј transcript. The outer PCR reaction used the 5Ј RACE outer primers, whereas the inner PCR reaction used the 5Ј RACE inner primers ( Table I). The 5Ј RACE inner PCR products were cloned into pGEM-T (Promega) and sequenced.
EMSA-Full-length GST-Rox1p was induced from plasmid pGEX-KG-ROX1 with 0.3 M isopropyl-1-thio-␤-D-galactopyranoside at 22°C in BL21 (DE 3) pLysS cells. Cells were lysed by a freeze-thaw procedure, and GST-Rox1p was purified by using 50% glutathione-agarose beads (w/v) (Sigma). GST-Rox1 fusion protein was solubilized with 20% Triton X-100 to a final concentration of 1%. A 1-ml bed volume of 50% glutathione-agarose per 100 ml of sonicate was incubated with cell lysate at 4°C. Beads were pelleted by centrifugation, the supernatant was removed, and the pellet was washed with 1ϫ phosphate-buffered saline. GST-Rox1p was eluted from the beads with glutathione elution buffer (10 mM reduced glutathione in 50 mM Tris-HCl, pH 8.0) at 22°C, and protein concentrations were determined by the Bradford assay (Bio-Rad).
EMSA experiments were performed with BamHI-BglII restriction fragments excised from pANB1, pPIS1 Ϫ325, and pPIS1 Ϫ325rox⌬, as described previously (32). Fragments for the circular permutation assay were excised from pPIS1-bend using the restriction enzymes listed in the legend to Fig. 7. Restriction fragments were end-labeled by means of an exchange reaction using [␥-32 P]ATP and T4 polynucleotide kinase (Promega), as described previously (51). Protein-DNA complexes were visualized by using a PhosphorImager (Amersham Biosciences).
Lipid Analysis by One-dimensional Thin Layer Chromatography-Steady-state levels of phospholipids were quantified after growing BRS1001 and FY23 at 30°C in either aerobic or anaerobic conditions with agitation (47) in the presence of [ 32 P]orthophosphate (ICN Pharmaceuticals Inc., Costa Mesa, CA) for five generations (52). 1-ml cell cultures were harvested in late logarithmic growth phase after 20 h of growth, and spheroplasts were prepared by treatment with 4 mg Zymolase 100T (U.S. Biological, Swampscott, MA) in 50 mM Tris, pH 7.5, 1.2 M glycerol, and 100 mM sodium thioglycolate. Spheroplasts were suspended in chloroform:methanol (2:1) and incubated for 1 h at room temperature with agitation. Lipids were extracted after adding 0.2 volumes of H 2 O. Lipids were separated on a Whatman LK5D silica gel 150 Å ϫ 20-ϫ 20-cm plate pre-treated with 2.3% boric acid (in ethanol). Phospholipids were separated in a single dimension using chloroform: ethanol:H 2 O:triethylamine (30:35:7:35), as described previously (53). Phospholipids were visualized with a PhosphorImager (Amersham Biosciences) and quantified using Imagequant software (Amersham Biosciences).

Deletion Analysis of the PIS1 Promoter Identifies Three UAS
Elements-UAS and upstream regulatory sequences are regulatory elements found within the promoters of yeast genes. A nested set of PIS1 promoter deletions fused to the cat reporter gene was used to delineate region(s) of the PIS1 promoter that are required for gene expression. This deletion analysis identified three UAS elements that are required for PIS1 promoter activity. Removal of sequences upstream of Ϫ225 did not affect expression of the cat reporter. However, removal of promoter sequences from Ϫ224 to Ϫ206 reduced CAT activity by 45% (compare pPIS1 Ϫ224 with Ϫ205 in Fig. 2). This suggests that a UAS element (designated UAS1) is present between se- quences Ϫ224 to Ϫ205. An internal deletion removing sequences from Ϫ184 to Ϫ149 also reduced CAT activity by 44% (compare pPIS1 Ϫ325 to pPIS1 Ϫ325 to Ϫ185/Ϫ149 in Fig. 2). We have designated this as UAS2. This region has been shown previously to bind Mcm1p (54). Removal of both UAS1 and UAS2 sequences resulted in a 94% reduction in CAT activity, suggesting that these elements function independently (compare pPIS1 Ϫ325 to Ϫ149 in Fig. 2). Another internal deletion that removed sequences Ϫ149 to Ϫ127 also reduced CAT activity by 84% (compare pPIS1 Ϫ325 to pPIS1 Ϫ325 to Ϫ149/ Ϫ127 in Fig. 2), suggesting the presence of a third UAS element (designated UAS3). When all three UAS elements were deleted (pPIS1 Ϫ127 in Fig. 2), CAT activity was completely eliminated, suggesting that all three UAS elements are required for PIS1 gene expression.
To aid in the interpretation of the PIS1 promoter deletion studies, the PIS1 transcription initiation site was mapped using 5Ј RLM-RACE (Ambion). This technique was chosen over more traditional 5Ј RACE protocols because it offers the advantage of specificity for full-length capped transcripts. Total RNA was treated with calf intestinal phosphatase to remove the 5Ј-phosphate from non-full-length uncapped RNA (degraded mRNA, tRNA, and rRNA). The 5Ј cap was then removed from full-length mRNA by treatment with tobacco acid pyrophos-phatase (TAP), leaving a 5Ј-monophosphate to which a 5Ј RACE adapter oligonucleotide was ligated using T4 RNA ligase. As a control, a sample of RNA was not treated with TAP. The presence of the 5Ј cap would prevent the ligation of the 5Ј RACE adapter, thereby preventing amplification in subsequent steps. Random-primed reverse transcription was used to generate cDNA. A nested set of primers (outer and inner) specific to the 5Ј adapter and an internal site on the PIS1 transcript were used to amplify sequentially the 5Ј end of the transcript. Amplification with the outer primers typically does not yield a specific PCR product (Fig. 3, 1°PCR). However, secondary amplification with the inner primers generated a product in reactions treated with TAP. (Fig. 3, 2°PCR, lane E). As expected, the control reaction carried out on RNA not treated with TAP did not generate a PCR product (Fig. 3, 2°PCR, lane C). The PCR products from the 2°PCR reaction were ligated into pGEM-T. Sixteen subclones were obtained, and restriction digestion revealed three different sizes of inserts. All 16 subclones were sequenced and identified three start sites in the PIS1 promoter sequence at Ϫ38 (9 subclones), Ϫ34 (2 subclones), and Ϫ25 (5 subclones) (Fig. 3).
To further elucidate the regulatory sequences defined by the PIS1-cat deletion analyses, the Saccharomyces cerevisiae Promoter Database was used to search for potential protein bind-  ing sites present in the PIS1 regulatory elements. The UAS1 element (Ϫ224 to Ϫ205) contains potential Ste12p and Pho2p sites, the UAS2 element (Ϫ184 to Ϫ149) contains a potential Pho2p binding site in addition to two Mcm1p binding sites, and UAS3 (Ϫ149 to Ϫ127) contains potential Rox1p and Gcr1p sites. The PIS1 promoter sequence from S. cerevisiae was also compared with that of other closely related Saccharomyces species. 2 The degree of sequence similarity between these species is in the following order: S. cerevisiae Ͼ Saccharomyces mikatae Ͼ Saccharomyces kudriavzevii Ͼ Saccharomyces bayanus Ͼ Saccharomyces castellii Ͼ Saccharomyces kluveryi (55). These comparisons revealed highly conserved regions in the UAS2 and UAS3 elements, particularly in the sequences known to bind Mcm1p and Rox1p (Fig. 4). The UAS1 element was also highly conserved in a region containing a potential Ste12p binding site ((T/A)TTTCAT). An additional repetitive sequence (GAGAT, Ϫ262 to Ϫ239) (Fig. 4) located upstream of the UAS elements was highly conserved among even the most divergent species. However, this sequence was not required for PIS1 expression under the conditions tested here.
PIS1 Is Regulated by Oxygen Availability through Rox1p-Rox1p is required for repression of hypoxically regulated genes under aerobic conditions. It has been shown that PIS1 expression is induced 2-fold under anaerobic conditions during growth in chemostat or batch cultures (41,43) and in a rox1⌬ mutant strain (42,43). Consistent with this, the region we defined as UAS3 contains a potential Rox1p binding site. This finding raises the possibility that this region includes binding sites for both an activator and the Rox1p repressor.
We used the PIS1-cat reporter to directly test the effect of oxygen on PIS1 promoter activity. PIS1-cat expression increased 1.7-fold under anaerobic conditions (Fig. 5, pPIS1 Ϫ325 in FY23). Because Rox1p is known to repress expression of the ANB1 gene under aerobic conditions (28), an ANB1-cat fusion was utilized as a control. CAT activity driven by the ANB1 promoter was regulated 26.6-fold in the FY23 (wild-type) strain and was constitutively elevated in the isogenic FY23-rox1⌬ strain (Fig. 5). To determine whether Rox1p is required for the regulation of PIS1, we assayed a PIS1-cat construct (pPIS1 Ϫ325rox⌬) containing a mutant Rox1p binding site element (ATTGTT to cgcGTT). The mutant Rox1p binding site was designed based on information from the reported consensus Rox1p binding site (33). The individual mutations have been shown to decrease the affinity of Rox1p from 8-to 29-fold. The construct containing the mutant Rox1p binding site over-expressed PIS1-cat under aerobic conditions. Furthermore, the level of PIS1-cat expression from the mutant Rox1p binding site construct was identical in the FY23 (wild-type) and FY23-rox1⌬ strains, suggesting that Rox1p represses transcription through a site in the UAS3 element (Fig. 5). However, it is interesting that the mutant Rox1p binding site construct yielded reduced expression under anaerobic conditions. This observation suggests that there may be another protein that binds in the vicinity of the Rox1p binding site of the PIS1 promoter under anaerobic conditions.
To determine whether Rox1p binds the cognate in the PIS1 promoter, an EMSA was done using a purified GST-Rox1p fusion. GST-Rox1p was found to bind a 325-bp restriction fragment from the PIS1 promoter (Fig. 6, pPIS1). However, GST-Rox1p did not bind an identical fragment containing the mutant Rox1p binding site (Fig. 6, pPIS1rox⌬). As a control, GST-Rox1p was shown to bind to the promoter of the ANB1 gene (Fig. 6, pANB1).
Rox1p binding is known to cause a bend in the ANB1 promoter (34). We used a circular permutation assay to determine whether Rox1p also bends the PIS1 promoter. This assay is based on the observation that fragments of equal length with a 2 P. Cliften, M. Johnston, and the Washington University Genome Sequencing Center, personal communication. FIG. 3. 5 RLM-RACE assay of PIS1 mRNA. A 5Ј RACE adapter was ligated to full-length decapped total cellular mRNA from a wildtype strain (BRS 1001) grown in complete synthetic medium containing 75 M inositol and 1 mM choline. Random-primed reverse transcription using Moloney murine leukemia virus reverse transcriptase followed by nested PCR amplified the 5Ј end of the PIS1 transcript. The 1°and 2°l anes contain products from PCR reactions that used the outer primers and inner primers (Table I), respectively. The PIS1-specific inner primer annealed to positions ϩ157 to ϩ180 (relative to the PIS1 start codon, A ϭ ϩ1). The E lanes contain reactions performed with RNA that were uncapped enzymatically with TAP, and the C lanes contain reactions performed on capped RNA that will not ligate with the 5Ј RACE adapter. The sequence at the right shows the nucleotide assignment of PIS1 mRNA start sites. Three start sites were identified and are indicated by arrows; the number of subclones for each start site is indicated above each arrow. protein binding site at different positions will migrate differentially if the bound protein bends the DNA (56). To test this proposition, we created a plasmid containing tandem inserts of the PIS1 UAS3 region (Fig. 7). Cutting this plasmid with different restriction enzymes yielded fragments of equal length that contained the Rox1p binding site at different positions. As observed with the ANB1 promoter (34), GST-Rox1p also bent the PIS1 promoter (Fig. 7).
Phospholipid Composition in Cells Is Altered by Oxygen Availability-PI biosynthesis has been shown previously to be regulated in response to inositol supplementation at the enzymatic level (27). The regulation of PIS1 transcription suggests that PI biosynthesis might also be regulated by oxygen. To examine the effect of oxygen availability on phospholipid levels, we quantified steady-state levels of phospholipids in a wild-type strain grown aerobically and anaerobically, both in the presence and absence of inositol (Fig. 8). PA, PS, and cardiolipin and phosphatidyldimethylethanolamine levels remained relatively unchanged regardless of growth conditions. Consistent with previously reported results, PI levels were elevated in cells grown aerobically in the presence of inositol (1-4). The increased PI synthesis in the presence of inositol occurs at the expense of PC synthesis because inositol also inhibits the first enzyme in the PC biosynthetic branch (PS synthase) (27). Consistent with the large increase in PI levels, CDP-DAG levels decrease in aerobically grown cells in the presence of inositol.
Anaerobic growth caused an increase in PI levels and a decrease in PC levels both in the presence and absence of inositol. The increase in PI under anaerobic conditions correlates with the increase in PIS1 transcription. Surprisingly, CDP-DAG levels remained relatively unchanged in cells grown anaerobically. This seems to be because of a greater overall decrease in the synthesis of PC. This finding suggests that under anaerobic growth conditions, the presence of inositol may inhibit PS synthase activity more than under aerobic conditions. DISCUSSION Genomic analysis of anaerobically induced genes in S. cerevisiae has begun to provide insight into the complex array of changes that occur when cells are grown anaerobically rather than aerobically. These studies reveal that Rox1p controls approximately one-third of anaerobically induced genes in S. cerevisiae (43). Many of these genes are involved in lipid, isoprenoid, and sphingolipid metabolism. Here, we show that anaerobic regulation of PIS1 expression resulted in changes in membrane composition. In general, PI levels were increased in cells grown anaerobically. However, the consequence of anaerobic growth was not limited to PI; PC levels and CDP-DAG levels were also altered. The consequences of this altered phospholipid composition cannot be predicted completely at this time. However, it has been suggested that the up-regulation of PI synthesis coupled with increased sphingolipid biosynthesis under anaerobic conditions probably alters the balance of phospholipids to sphingolipids (43).
In anaerobic conditions, Hap1p (in a high molecular weight complex) represses ROX1 transcription allowing hypoxic gene expression. Under aerobic conditions, Rox1p recruits the Tup1p-Ssn6p complex and represses transcription of hypoxic genes (31,32). In addition, transcription activators such as Upc2p/Mox4p (39,(57)(58)(59) and repressors Mox1p, Mox2p, and Mot3p/Rox7p (35,58) have recently been identified as regulators of hypoxic genes. Although Mot3p does not regulate PIS1 transcription, 3  FIG. 6. Recombinant GST-Rox1p forms a complex with a restriction fragment from the PIS1 promoter. Restriction fragments containing either a wild-type (pPIS1) or mutated (pPIS1rox⌬) Rox1p binding site from the PIS1 promoter were combined with recombinant GST-Rox1p. As a control, a restriction fragment containing the Rox1p binding site from the ANB1 gene was used. Bands labeled C refer to complexes formed with GST-Rox1p; F PIS1 and F ANB1 refer to free template DNA from the PIS1 and ANB1 promoters, respectively. a rox1⌬ strain. This result suggests that another regulator functions through the Rox1p binding site.
By using the circular permutation assay, Rox1p was found to induce a 90°bend in the ANB1 promoter (34). The DNA binding affinity of a protein that induces DNA bending depends on specific sequence recognition as well as the ability of a DNA sequence to be deformed (56). The difference in Rox1p-mediated regulation of ANB1 (ϳ27-fold) and PIS1 (1.7-fold) could be attributed to differences in the bend angle Rox1p creates when bound. Alternatively, the difference in regulation could be because of the presence of a single Rox1p binding site in the PIS1 promoter and four binding sites in the ANB1 promoter (Fig. 5, two each in Operators A and B). Additional experiments will be required to distinguish between these possibilities.
Although PIS1 is uncoupled from the inositol/choline response that regulates other phospholipid biosynthetic genes, inspection of the PIS1 UAS1 and UAS2 elements revealed potential protein binding sites that may provide a clue as to how PIS1 is regulated. It is particularly interesting that there is a potential binding site for Ste12p (UAS1) and two sites known to bind Mcm1p (UAS2) (54) in close proximity (33 bp apart). Ste12p is required for the induction of gene expression in response to pheromone treatment (60,61). Because Ste12p binds weakly to its target site, it typically requires two binding sites or interaction with Mcm1p (60, 61). Mcm1p has been shown to bind the PIS1 promoter in vitro and in vivo (54,62). Interestingly, a genome-wide analysis of protein binding in vivo did not identify Ste12p binding to the PIS1 promoter (62). However, this same study did not identify Rox1p binding. It is possible that both Rox1p and Ste12p are below the level of detection of the ChIP on CHiP. Our results showing that Rox1p regulated PIS1 transcription and PI synthesis suggest that the effect of Ste12p on expression of PIS1 should be examined. However, it should be noted that microarray experiments do not identify PIS1 as a pheromone-inducible gene (63).
A comparison of the PIS1 promoter sequence of S. cerevisiae to that of other closely related Saccharomyces species 2 (64) revealed highly conserved regions in the UAS1, UAS2, and UAS3 elements, particularly in the sequences predicted to bind Ste12p and known to bind Mcm1p and Rox1p. It will be interesting to examine PIS1 gene expression in these other species to determine whether regulation via these UAS elements is also conserved. An additional repetitive sequence (Ϫ262 to Ϫ239) located upstream of the three UAS elements was highly conserved among even the most divergent Saccharomyces species; however, this sequence was not required for PIS1 expression and does not include any known trans-factor binding sites, suggesting that it may not be involved in transcriptional regulation. Alternatively, it may be required for transcription under conditions not tested here.
Eukaryotic genes are subject to multiple levels of regulation. An example of this is regulation of translation initiation by upstream open reading frames (uORF) found in some transcript leaders (65)(66)(67)(68). Genes that are regulated by uORFs in S. cerevisiae include GCN4, CPA1, YAP1, YAP2, and INO2 (69 -72). Mapping of the PIS1 transcript 5Ј terminus revealed an AUG codon upstream of the PI synthase AUG codon. Whether initiation of translation at the PIS1 5Ј-uORF regulates PI synthase levels is yet to be determined. However, it is curious that the PIS1 transcript has a very long half-life (57 Ϯ 3 min) (73), because it has been proposed that transcripts containing uORFs could be targets of the non-sense mediated decay pathway that degrades RNAs containing non-sense codons (74). This has been shown to be the case for the CPA1 transcript (75). However, the GCN4 and YAP1 transcripts are protected from this pathway by a stabilizing protein (Pub1p) which binds to a stabilizer element found in the leaders of these transcripts (75). FIG. 7. Rox1p bends the PIS1 promoter. A circular permutation assay was used to determine whether DNA bending occurred when Rox1p binds the PIS1 promoter. The DNA fragments used were generated by digestion of plasmid pPIS1bend that contains direct tandem repeats of a 325-bp fragment from the PIS1 promoter. The enzymes used to generate the restriction fragments are depicted. The empty boxes depict the position of the Rox1p binding site within each restriction fragment. Further studies are needed to determine whether mutating the PIS1 uAUG affects PI synthase levels and if PIS1 RNA stability requires Pub1p and a stabilizer element.