Coordinate Control of Sphingolipid Biosynthesis and Multidrug Resistance in Saccharomyces cerevisiae *

Multiple or pleiotropic drug resistance often occurs in the yeast Saccharomyces cerevisiae through genetic activation of the Cys6-Zn(II) transcription factors Pdr1p and Pdr3p. Hyperactive alleles of these proteins cause overproduction of target genes that include drug efflux pumps, which in turn confer high level drug resistance. Here we provide evidence that both Pdr1p and Pdr3p act to regulate production of an enzyme involved in sphingolipid biosynthesis in S. cerevisiae. The last step in formation of the major sphingolipid in the yeast plasma membrane, mannosyldiinositol phosphorylceramide, is catalyzed by the product of the IPT1 gene, inositol phosphotransferase (Ipt1p). Transcription of the IPT1 gene is responsive to changes in activity of Pdr1p and Pdr3p. A single Pdr1p/Pdr3p response element is present in the IPT1 promoter and is required for regulation by these factors. Loss of IPT1 has complex effects on drug resistance of the resulting strain, consistent with an important role for mannosyldiinositol phosphorylceramide in normal plasma membrane function. Direct assay for lipid contents of cells demonstrates that changes in sphingolipid composition correlate with changes in the activity of Pdr3p. These data suggest that Pdr1p and Pdr3p may act to modulate the lipid composition of membranes in S. cerevisiae through activation of sphingolipid biosynthesis along with other target genes.

Saccharomyces cerevisiae cells contain three major classes of inositol-containing sphingolipids: inositol phosphorylceramide (IPC), 1 mannosylinositol phosphorylceramide (MIPC), and mannosyldiinositol phosphorylceramide (M(IP) 2 C). These lipids compose 30% of the total phospholipids present in the yeast plasma membrane, with M(IP) 2 C accounting for 75% of these yeast sphingolipids (1). Control of the biosynthesis of these inositol-containing sphingolipids ( Fig. 1) is critical to maintain normal function of the plasma membrane even though produc-tion of MIPC or M(IP) 2 C is not required for viability (2,3).
Although maintenance of normal sphingolipid levels is crucial for S. cerevisiae cell function, little information is available detailing the mechanisms that regulate their biosynthesis. Recently, microarray experiments have provided insight into a possible means of regulation of sphingolipid biosynthetic enzyme production. DeRisi et al. (4) profiled the genomic expression pattern of cells containing hyperactive forms of the Pdr1p and Pdr3p transcription factors. Pdr1p and Pdr3p are Cys 6 -Zn(II) transcription factors that act to modulate expression of genes involved in multiple or pleiotropic drug resistance in S. cerevisiae (see Refs. 5 and 6 for reviews). Previous data have demonstrated that single amino acid substitution mutations in Pdr1p (7) and Pdr3p (8) can lock these proteins into a hyperactive state, leading to high level expression of downstream target genes that include ATP-binding cassette transporters like Pdr5p (9,10) and Yor1p (11). The work of DeRisi et al. (4) identified genes that are transcriptionally up-regulated in the presence of hyperactive alleles of Pdr1p and/or Pdr3p. These loci included previously known Pdr1p/Pdr3p target genes like PDR5 and YOR1 as well as new members of this regulon such as the IPT1 locus.
IPT1 encodes the last step in biosynthesis of sphingolipids, inositol phosphotransferase, that produces M(IP) 2 C from MIPC (3). Microarray analysis indicated that levels of IPT1 mRNA increased by 2-fold in the presence of hyperactive forms of Pdr1p or Pdr3p (4). Examination of the 5Ј-noncoding region of IPT1 suggested the presence of a Pdr1p/Pdr3p response element (PDRE), the binding site for these transcriptional regulatory proteins (12). In the work described here, we provide evidence that this PDRE is required for the previously observed induction of IPT1 gene expression by hyperactive alleles of PDR1 and PDR3. Additionally, we have found that loss of the mitochondrial genome ( 0 cell) leads to activation of Pdr3p, but not Pdr1p (13). Transcription of IPT1 is induced in 0 cells in a fashion parallel to that of PDR5. Finally, the effects on biosynthesis of phospholipids are assayed in response to loss of the mitochondrial genome alone or simultaneous to loss of PDR3. These data strongly suggest that sphingolipid biosynthesis is a physiological target of Pdr1p/Pdr3p regulation and provide the first description of how production of these important lipids is transcriptionally regulated.

EXPERIMENTAL PROCEDURES
Yeast Strains, Media, and Methods S. cerevisiae strains (Table I) were grown in YPD medium (2% yeast extract, 1% peptone, and 2% dextrose) or minimal medium (14) supplemented with casamino acids (synthetic complete medium) at 30°C. Drug resistance assays were performed by addition of compounds to solid medium at the indicated concentrations or by use of gradient plates as described (15).

Plasmids
A fragment of the IPT1 promoter extending from Ϫ621 to ϩ4 was generated by PCR using the primer pair GCG GAT CCA TGC TTT CTT AGA CGT TGA AAG AC and CGG AAT TCC GCA TGG GGA AAT GCA AGG CAA GC. This DNA fragment was digested with EcoRI and BamHI and inserted into pBluescript KS II Ϫ to form pLL1. This same EcoRI/BamHI fragment was also transferred into pSEYC102 (16) to produce an IPT1-lacZ gene fusion (pLL2). Using pLL1 as a template, a mutant form of the IPT1 PDRE was generated by PCR employing the primer pair CTT CTC GAG GAA CAA AAA TGT GAA CGC and GTA AAA CGA CGG CCA GT (M13 forward primer) along with the primer pair GGC GTC AAG AAG CGG CG and AAC AGC TAT GAC CAT G (M13 reverse primer). These products were isolated, and aliquots were pooled and reamplified with M13 forward and reverse primers. The resulting PCR fragment was cloned as an EcoRI/BamHI fragment into pBluescript SK ϩ and pSEYC102 as described above. A subclone of the IPT1 promoter was prepared for use as a template in DNase I protection experiments using PCR and the primer pair CGG AAT TCG GCG AAG AAG GCG G and CGG GAT CCG CGG CGA TAA ACG G. This PCR product was digested with EcoRI and BamHI and cloned into pBluescript SK ϩ digested with the same restriction enzymes. This plasmid was designated pLL3. The fidelity of all cloned PCR fragments was verified by DNA sequence analysis. The isolated alleles of PDR1 (PDR1-2, PDR1-3, and PDR1-6) in pRS315 (17) have been characterized previously (7). The mutant PDR3-11 allele was obtained from W. Nichols (Emory University) and transferred into pRS315 as a SalI/BamHI fragment.
The ipt1-⌬1::URA3 disruption allele was constructed by cleaving a genomic clone of IPT with KpnI/AvrII to remove nucleotides 669 -1096 of the 1584-base pair open reading frame. This fragment was replaced with the URA3 gene, and the resulting plasmid was designated pipt1::URA3. The disruption allele was separated from vector sequences by EcoRI/BglI digestion and used to transform cells to Ura3 ϩ .

DNase I Protection Assay
Amino-terminal fragments of Pdr1p and Pdr3p were expressed in bacterial cells as described previously (12). Extracts were prepared from bacterial cells containing the empty expression vector pOTSV (18) as a negative control. Probes were generated from pLL3 cut with either EcoRI or BamHI, followed by treatment with calf intestinal alkaline phosphatase. DNA fragments were labeled by incubation with T4 polynucleotide kinase and [␥-32 P]ATP. Radiolabeled IPT1 fragments were released from the vector by secondary restriction digestion and gel-purified.

Northern Blot and ␤-Galactosidase Activity Assays
Total RNA was prepared using hot phenol extraction (19). Total RNA was electrophoresed through a formaldehyde-containing 1% agarose gel and transferred to nylon membranes using standard techniques (20). An internal PvuII fragment from IPT1 and a BamHI/HindIII fragment from the ACT1 gene were used as probes and were labeled by nick translation with [␣-32 P]dCTP. Hybridizations were carried out for 15 h at 42°C in buffer containing 40% formamide, 5ϫ Denhardt's solution, 0.5% SDS, 5ϫ saline/sodium phosphate/EDTA, and 20 g/ml denatured salmon sperm DNA. The membranes were washed at 55°C in 1ϫ SSC and 0.2% SDS and then exposed to x-ray film. Assay of ␤-galactosidase activity produced from fusion genes was accomplished either by preparing a whole cell protein extract (21) or through the use of permeabilized cells as described (22). All ␤-galactosidase activity values represent the average of at least two independent determinations on multiple transformants.

Lipid Methods
Short-term Labeling-Overnight cultures were grown in synthetic complete medium to late exponential phase. Aliquots corresponding to ϳ10 7 cells were harvested in sterile Eppendorf tubes, resuspended in 1 ml of fresh synthetic complete medium (lacking serine for serine labeling), and transferred to sterile 15-ml glass test tubes. The cultures were preincubated with shaking for 120 min at 30°C and 200 rpm and then labeled with either 3 Ci of D-erythro- [4,  Lipid Assay-Prior to harvesting, the cell density was determined by absorbance of a 1:20 dilution at 600 nm. The cultures were chilled on ice for 5 min and then harvested at 4°C by centrifugation at 3000 ϫ g. Preparation of whole cell lipid extracts was performed according to protocol IIIB as described (23). If necessary, lipids were deacylated by mild alkaline hydrolysis (24) and afterward partitioned between butanol and water as described (25). The dried lipid extracts were resuspended in 100 l (pulse-labeled cells) or 30 l (steady-state labeled cells) of chloroform/methanol/water (10:10:3, v/v/v) per 10 7 cells. Usually, 5 l/lane was spotted onto high performance TLC plates (Kieselgel 60, Merck, Darmstadt, Germany) and resolved in chloroform, methanol, and 4.2 N NH 4 OH (9:7:2, v/v/v). Tritium-labeled lipids were treated with EN 3 HANCE (PerkinElmer Life Sciences) and subjected to autoradiography/fluorography using X-Omat films (Eastman Kodak Co.). Quantitation of 32 P-labeled lipids was performed using a Fuji FLA2000 phosphoimager (Fuji Photo Film, Tokyo, Japan). All bands were identified using authentic standards, except for ceramide, which was identified due to its sensitivity to fumonisin B 1 in combination with its R F value. Authentic sphingolipid standards were provided by Robert L. Lester (University of Kentucky, Lexington, KY).

IPT1 mRNA Levels Respond to Changes in Pdr1p
Activity-A possible regulatory influence of the Pdr1p and/or Pdr3p transcription factor on sphingolipid biosynthesis was suggested by two complementary observations. First, inspection of the IPT1 promoter sequence indicated the presence of a sequence element identical to known PDREs located from Ϫ454 to Ϫ445 base pairs upstream of the IPT1 ATG codon. The IPT1 PDRE has the sequence TTCCGCGGAA, which is identical to known PDREs located in PDR3 (26) and PDR5 (12). Second, microarray experiments identified IPT1 as a potential target gene of hyperactive forms of both Pdr1p and Pdr3p (4). To confirm that IPT1 mRNA levels respond to alterations in the activity of Pdr1p, we examined the expression of IPT1 mRNA by Northern analysis. Total RNA was isolated from cells carrying five different alleles of PDR1. These alleles covered a range of activity varying from no Pdr1p (pdr1-⌬2) to the hyperactive alleles (PDR1-2, PDR1-3, and PDR1-6), which produce higher levels of drug resistance and target gene expression than the wild-type gene. IPT1 mRNA was analyzed along with ACT1 as a control for loading (Fig. 2).
This analysis is entirely consistent with the previous microarray data and demonstrates that IPT1 mRNA levels were highest in the cells carrying the hyperactive alleles and were reduced in wild-type or pdr1-⌬2-containing cells. We conclude from these data that the expression of the genomic copy of IPT1 is sensitive to the level of Pdr1p activity. To determine if the influence of Pdr1p was exerted at the level of the IPT1 promoter, we analyzed the regulation of an IPT1-lacZ fusion gene.
Pdr1p Control of IPT1 Expression Is Mediated through the IPT1 Promoter-To facilitate measurement of IPT1 gene expression, we fused a DNA fragment extending from the IPT1 ATG codon to 621 base pairs of 5Ј-noncoding sequences with the Escherichia coli lacZ gene carried in the shuttle vector pSEYC102 (16). This plasmid produces ␤-galactosidase activity under the control of the IPT1 transcriptional and translational regulatory signals. This IPT1-lacZ fusion gene was introduced into several different genetic backgrounds, producing different levels of Pdr1p activity, and expression of the fusion gene was assayed (Table II).
The IPT1-lacZ fusion gene directed production of 30 units/mg ␤-galactosidase activity in wild-type cells. This level of expression was not significantly influenced upon loss of either the PDR1 or PDR3 gene alone or in combination. However, introduction of the hyperactive alleles of PDR1 (PDR1-3 and PDR1-6) led to an increase in IPT1 expression, consistent with the behavior of the genomic copy of IPT1. Expression of the PDR5 gene, which is known to be highly dependent on the presence of either PDR1 or PDR3 (10), also showed no significant decrease when either transcription factor was removed alone, but was strongly inhibited upon loss of both. PDR5 was similarly up-regulated when the hyperactive alleles of PDR1 were introduced.
These data argue that, although IPT1 expression has a significant PDR1/PDR3-independent component, gain-offunction forms of Pdr1p are capable of elevating expression of this sphingolipid biosynthetic enzyme-encoding gene. To determine if this effect of Pdr1p requires the presence of the PDRE in the IPT1 promoter, a site-directed mutation was produced that removed this PDRE. This mutant IPT1 promoter (mPDRE-IPT1) was then placed in the lacZ fusion plasmid and introduced into the strains with differentially active PDR1 alleles.
Loss of the PDRE from IPT1 eliminated the response of this promoter to the hyperactive forms of Pdr1p. This analysis indicates that not only does Pdr1p regulate expression of IPT1 and that a single PDRE, located at Ϫ450 base pairs upstream of the ATG codon, is required for this regulatory effect, but that this single PDRE also mediates the response of IPT1 to Pdr3p as discussed below.
Pdr1p Directly Interacts with the PDRE at Ϫ450 -The presence of a PDRE in the promoter of IPT1 suggests that Pdr1p (and likely Pdr3p) will directly interact with this DNA element as seen in a variety of other Pdr1p/Pdr3p-regulated genes. To test this idea, DNase I protection experiments were carried out examining the ability of bacterially produced Pdr1p and Pdr3p to bind to the IPT1 promoter (Fig. 3).
DNase I reactions performed in the absence of protein or the presence of protein extract prepared from bacterial cells containing the empty expression vector were indistinguishable. Extracts prepared from cells expressing the DNA-binding domains of Pdr1p or Pdr3p were able to inhibit DNase I cleavage over sequences corresponding to the IPT1 PDRE. Both factors protect approximately equal regions of the promoter and induce strong DNase I-hypersensitive cleavage sites on both DNA strands between the PDRE and the IPT1 ATG codon. This analysis demonstrates that both Pdr1p and Pdr3p can directly bind to the IPT1 PDRE and supports the view that this binding is important in transcriptional control of IPT1. IPT1 Responds to Pdr3p and Mitochondrial Status-Recently, we have found that the activity of Pdr3p is strongly regulated by the mitochondrial status of cells (13). Loss of the mitochondrial genome or the F 1 F 0 /cytochrome oxidase assembly factor Oxa1p leads to strong activation of PDR5 transcription in a Pdr3p-dependent fashion. To determine if IPT1 expression might also be regulated in response to loss of mitochondrial function, we examined the expression of the IPT1-lacZ fusion gene in 0 cells (Table III).
Loss of the mitochondrial genome led to an increase in IPT1-lacZ expression from 15 units/OD in wild-type cells to 31 units/OD in 0 cells. Introduction of the hyperactive allele of PDR3, PDR3-11, into wild-type cells along with the IPT1-lacZ gene fusion led to production of 28 units/OD of ␤-galactosidase activity. These data argue that Pdr3p can influence expression of IPT1 and that this gene is induced in response to loss of the mitochondrial genome. This finding led us to examine if changes in sphingolipid composition/synthesis could be detected in 0 cells.
Sphingolipid Metabolism Is Altered in 0 Cells-The above  2. IPT1 mRNA responds to increases in PDR1 function. Total RNA was prepared from strains expressing the indicated alleles of PDR1. Equal amounts (30 g) of RNA were electrophoresed through a denaturing agarose gel, transferred to nylon membranes, and probed with radiolabeled fragments from the genes indicated on the left. The ACT1 signal served as a control to ensure equal loading of each lane. data indicate that IPT1 gene expression is both responsive to Pdr1p/Pdr3p control and induced in response to loss of the mitochondrial genome. These findings suggest the possibility that PDR3 activation following loss of the mitochondrial genome might alter the efficiency of the sphingolipid biosynthetic pathway. To examine this idea, we analyzed the rates of sphingolipid synthesis by short-term labeling of the cells with [ 3

H]DHS or [ 3 H]serine.
Wild-type cells incorporated exogenously added [ 3 H]DHS into ceramide, IPC, MIPC, and M(IP) 2 C (Fig. 4). Along with these complex lipids, [ 3 H]DHS was also recovered in its hydroxylated derivative phytosphingosine as well as in DHS 1-phosphate. The diacylglycerol-based phosphatidylinositol, phosphatidylethanolamine, and phosphatidylcholine were also labeled from DHS via conversion of DHS 1-phosphate into palmitate, due to the action of the sphingoid base phosphate lyase Dpl1p (27). The loss of the mitochondrial genome in the 0 strain led to a different labeling pattern compared with the wild type. The amount of MIPC was reduced, whereas IPC-IV and M(IP) 2 C were elevated in the 0 strain. Strikingly, elimination of PDR3 from the 0 strain returned MIPC and M(IP) 2 C levels back to those seen in the wild-type background, but did not restore normal IPC-IV levels. The deletion of PDR1 and DR3 in the wild-type background did not influence the lipid labeling pattern (data not shown), consistent with the lack of effect of the ⌬pdr1,⌬pdr3 background on IPT1 expression (Table II).
In contrast to endogenously synthesized DHS, exogenously added DHS is efficiently phosphorylated upon uptake (28) and must be dephosphorylated before it will be converted to phytosphingosine or ceramide (29,30). Moreover, exogenously acquired DHS and de novo synthesized DHS appear to form different pools in the cell that do not interchange. 2 To test whether the altered sphingolipid synthesis observed in the 0 strain and its dependence on PDR3 vary with respect to the different pools of DHS, de novo sphingolipid synthesis was monitored by labeling the cells with [ 3 H]serine. Fig. 4 shows the corresponding whole cell lipid extracts after treatment with mild alkaline hydrolysis. The [ 3 H]serine labeling emphasized The indicated lacZ fusion genes were introduced into several different strains with different complements of PDR1 and PDR3. Transformants were grown to mid-log phase, and ␤-galactosidase activities were determined in protein extracts as described (21). IPT1-lacZ is a gene fusion between the wild-type IPT1 promoter and lacZ, whereas mPDRE-IPT1-lacZ is the same construct, but with the PDRE altered by site-directed mutagenesis as described under "Experimental Procedures." PDR5-lacZ is a control for a known Pdr1p/Pdr3p-responsive gene (10

TABLE III IPT1 expression responds to PDR3 and mitochondrial status
Wild-type (SEY6210 ( ϩ )) or mitochondrial genome-deficient (SEY6210 ( 0 ) versions of SEY6210 were transformed with the indicated lacZ gene fusions and PDR3 expression plasmids. Transformants were grown to mid-log phase and assayed for ␤-galactosidase activity using permeabilized cells (22). 11 28 Ϯ 2 1 0Ϯ 1 the increase in IPC-IV and M(IP) 2 C and again showed the decreased MIPC amounts in the 0 strain compared with those in the wild type. Again, introduction of the ⌬pdr3 allele into the 0 background produced a strain that had near-normal MIPC and M(IP) 2 C levels, but still exhibited increased production of IPC-IV. These data confirm that sphingolipid synthesis is altered in 0 cells in a Pdr3p-dependent fashion. Finally, to confirm that the altered sphingolipid biosynthetic rates result in a change in steady-state sphingolipid composition, phosphate labeling was employed. Cells were labeled with [ 32 P]orthophosphate for an extended period of time, and total lipids were extracted. After mild alkaline treatment, the phosphate-labeled lipids were resolved by thin-layer chromatography and analyzed using of the phosphoimager. This analysis confirmed that levels of M(IP) 2 C increased in 0 cells compared with a wild-type strain. This increase was reversed by nearly 60% to near-normal levels upon introduction of a ⌬pdr3 allele into the 0 background. Note that although the increase in M(IP) 2 C was always seen, the magnitude of the change was modest in these phosphate labeling experiments. Possible reasons for this are discussed below.

IPT1-lacZ mPDRE-IPT1-lacZ
IPT1 Is Required for Normal Drug Tolerance Phenotypes-As the above data demonstrate that IPT1 expression responds to both Pdr1p and Pdr3p and that sphingolipid metabolism is altered in response to activation of Pdr3p, the role of IPT1 in drug resistance was analyzed. Isogenic strains carrying either the wild-type or deleted forms of the IPT1 gene were assayed by testing the ability to grow on medium containing various drugs (Fig. 5).
Loss of the IPT1 gene had differential effects on the various resistance phenotypes tested. Introduction of the ipt1-⌬1::URA3 allele into strains carrying either the PDR1-3 or PDR1-2 mutant form of PDR1 caused resistance to the translation inhibitor hygromycin to increase. Interestingly, deletion of the hyperactive allele of the PDR1 gene also caused a similar increase in hygromycin tolerance. Similarly, loss of IPT1 from the PDR1-3 strain led to an increase in resistance to cycloheximide.
A different pattern of phenotypic behavior was seen for the mitochondrial ATPase inhibitor oligomycin and the antifungal drug t-buconazole. Removal of IPT1 from the PDR1-3 strain led to a pronounced decrease in tolerance to these two compounds. The role of Ipt1p and levels of M(IP) 2 C in drug resistance phenotypes is complex and required for normal drug resistance.

DISCUSSION
Sphingolipids are important components of eukaryotic membranes and participate as signal transduction intermediates in regulatory pathways from humans to yeast. Rapid progress has been made in S. cerevisiae in the establishment of gene-enzyme relationships, but the understanding of the functional roles of these lipids is limited (1). Analysis of the IPT1 gene indicated that loss of this gene leads to a modest increase in calcium tolerance (3) and resistance to syringomycin E (31). Changes in calcium tolerance have been seen for many mutants lacking the normal complement of sphingolipid biosynthetic loci (32), and syringomycin E is believed to directly interact with mature sphingolipids and sterols (33,34). Here we extend the functional roles of M(IP) 2 C in the cell by showing a requirement for synthesis of this lipid to ensure normal drug resistance phenotypes. The effect of M(IP) 2 C varies in relation to the different drugs assayed, suggesting that this lipid may play different roles in determining the activity of various membrane proteins or the permeability properties of membranes.
The differential resistance phenotypes observed in cells lacking IPT1 suggest that lowered levels of M(IP) 2 C or altered levels of other sphingolipids may act to inhibit some transporters, like Pdr5p, yet stimulate others, like Yor1p. It is interesting to compare the resistance phenotypes of IPT1 or PDR1 disruption mutations in terms of the ability to grow on hygromycin-or oligomycin-containing medium. Loss of either IPT1 or PDR1 seems to have the same relative effect on these two drugs, suggesting that Pdr1p-dependent activation of IPT1 expression might be the cause of the sensitivity or resistance seen for hygromycin or oligomycin. Although the locus modulating hygromycin resistance remains unknown, it is not likely to be PDR5 (35). Activation of IPT1 by Pdr1p leads to a reduction in the ability to tolerate hygromycin, but an increase in oligomycin resistance. We have already demonstrated that Pdr1p-dependent activation of YOR1 expression elevates resistance to oligomycin (36) and have now shown that only in the presence of a wild-type IPT1 locus can this increased oligomycin resistance be seen (Fig. 5). In opposition to the effect on oligomycin tolerance, elimination of IPT1 leads to an increase Lipids were extracted and processed as described under "Experimental Procedures." Labeled lipids were then separated on TLC plates and visualized by autoradiography. The locations of labeled species were established by comparison with known standards. WT, wild type; CER, ceramide; PE, phosphatidylethanolamine; PHS, phytosphingosine; PC, phosphatidylcholine; PI, phosphatidylinositol; DHS-1-P, DHS 1-phosphate; FA, fatty acid. IPC-III and IPC-IV correspond to progressively higher states of IPC hydroxylation (52). B, shown are changes in steady-state levels of sphingolipids. The indicated strains (inset) were grown for Ͼ10 generations in the presence of [ 32 P]orthophosphate and harvested in late log phase. Equal cell numbers were lysed, and lipid extracts were prepared and run on TLC. The levels of 32 P-labeled sphingolipids were quantitated using a phosphoimager. wt, wild type.
in cycloheximide resistance, a phenotype significantly dependent on PDR5 function (37). Further experiments are required to determine if the changes in membrane sphingolipid content influence the activity, trafficking, or synthesis of these transporters.
Identification of IPT1 as a target gene for PDR gene regulation has important implications for the understanding of the physiology defined by the PDR regulon. Previous experiments from a variety of laboratories have determined that a number of different membrane transporter proteins are controlled by Pdr1p and/or Pdr3p at the transcriptional level (9,36,38,39). Although the drug resistance effects of these membrane transporters are clear, their normal functions are not, with the possible exception of Pdr5p and Yor1p. Experiments performed using a fluorescent phosphatidylethanolamine derivative have indicated that these two ABC transporters may be involved in controlling the phospholipid content of the plasma membrane (40,41). Further implication of the PDR system in control of membrane content has come from the finding that Pdr17p, a putative phospholipid transfer protein, is regulated by Pdr1p/ Pdr3p and is involved in phosphatidylserine metabolism (42,43). Coupled with the definition of IPT1 as a Pdr1p/Pdr3pregulated gene, these data strongly suggest that a physiological role of PDR genes may be to modulate the composition of membranes in S. cerevisiae.
The relatively small but highly reproducible change in M(IP) 2 C levels seen in 0 cells may be explained in several different ways. First, the rate-limiting step in biosynthesis of M(IP) 2 C is likely to occur upstream of Ipt1p at the first step in the pathway, serine palmitoyltransferase (1). Second, turnover of M(IP) 2 C may also be elevated in these cells to maintain a relatively constant level of this sphingolipid. It is notable that the elevated levels of IPC-IV seen in a 0 cell are not restored to normal upon the loss of PDR3 (Fig. 4). This indicates the presence of at least one additional regulatory pathway acting on sphingolipid biosynthetic loci that could also influence M(IP) 2 C synthesis. Third, the 300% increase in M(IP) 2 C synthesis detected by pulse labeling with serine compared with the smaller change in steady-state levels may reflect the contribution from salvage pathways that enter above the initial serine palmitoyltransferase reaction. Finally, measurement of M(IP) 2 C was accomplished using total lipids. Since M(IP) 2 C is highly enriched in lipid raft domains (44), the actual increase in the effective concentration might be much greater in terms of localized effects. Experiments are underway to distinguish between these possibilities.
Along with their role in maintaining the normal structure of membranes, sphingolipid derivatives have been found to be potent second messengers. Ceramide is a key modulator of the stress response, often serving an antiproliferative role in control of growth during an environmental challenge (see Ref. 45 for a review). In S. cerevisiae, heat shock of cells was observed to cause an elevation of ceramide levels and related metabolites like DHS, phytosphingosine, and their phosphorylated products (46,47). Ablation of the genes encoding proteins responsible for DHS 1-phosphate and phytosphingosine 1-phosphate breakdown caused large enhancements of heat resistance and supports the notion that these sphingolipid intermediates are involved in the heat stress response (27,48).
We have consistently observed a small but reproducible increase in ceramide production in 0 cells compared with the ϩ parental strain, which is unaffected by loss of PDR3. The availability of strains bearing different genetic blocks in ceramide metabolism will allow testing of the possibility that ceramide or some other sphingolipid metabolite might be the signal leading to Pdr3p activation in response to defects in the mitochondria. In larger eukaryotes, ceramide signaling is an important modulator of cell viability through control of entry into apoptosis (reviewed in Ref. 45). Our finding that Pdr3p both controls sphingolipid biosynthesis and in turn is responsive to mitochondrial status (13) suggests the possibility that this regulatory circuit also exists in S. cerevisiae. Further analysis of the relationship between PDR loci, mitochondrial function, and sphingolipid biosynthesis will allow this hypothesis to be tested.